🙋│image-help

[real lynx]

You can create a private room with the /mentor command.

[thin tinsel]

Ok, let me re-read the conversation and I will try to offer some advice.

I see you asked about hands and Jawknee advised using Phoenix, as it does hands much better than the other models.
I see that you were trying to offset the issue by using the negative prompt option. Unfortunately negative prompting is not very effective and since AI images are known for hand being unusual, there is a bit of luck involved.
However if you are interested in being able to fix the hands on the images you particularly liked that is something we can help you.

Is that something you're interested in learning how to do?

Hi Louise, did you still have any questions about using Leonardo.ai ?

[agile raptor]

With regenerating parts of images

[thin tinsel]

Hi @rough needle, you indicated that you would like some assistance with using Leonardo. I'm just messaging to see if you got what you needed. If not, then please put your questions here, so we can assist.

[thin tinsel]

What steps did you try and which model were you using?

[agile raptor]

What do you mean what steps? And this last time I tried it was the Lightning XL but I have tried before with different models, not sure which ones but I just remember it almost always gives bad results.

[river scroll]

Yes if possible

[thin tinsel]

Ok great, please type
/mentoring
and fill in the form

[thin tinsel]

Ok, so when you're in focus mode you can use any of the XL models but otherwise you would need Absolute Reality or Dreamshaper.
There is a skill to it. I use it all the time and don't have huge issues with it. Yes it's AI and therefore you don't get what you're imagining 100% of the time, but you can still get good results with it. I can help you if you have a specific thing you're working on.

Hi, I saw that you requested help using the canvas editor, can you please reply to this message if you still require assistance. Thank you.

[agile raptor]

Well I am trying to get fingers of a woman grasping a coffee cup in a realistic styled image I created with Leo starting out. Just the fingers are giving me issues no matter what I try.. Really thought we were past this problem with AI lol

[thin tinsel]

Hands and fingers are still an issue with AI generation. Although imo I think not quite as bad as a couple of years ago.
Please type /mentoring to open a mentor thread and you can share the image you want to work on in there.

[agile raptor]

I started it.. now what?

[thin tinsel]

speak to me in there

[agile raptor]

Do I need to mention your name to get you in there?

[thin tinsel]

No, I am there now. But ideally you should reply directly or @ mention me so I get pinged.

[dapper tendon]

Hey im having issues with my api image to videos being 1 second long? Is anyone having that issue?

[acoustic siren]

this sounds like it would be more suitable in #💻│api might wanna ask there to see if someone can help 🙂

[ocean wind]

@acoustic siren

Yo le doy hasta el 1.00 más queda mal y hasta el 0.70 o 0.80

Eso hace que el prompt sea más fuerte que tu imagen y las modifique

[acoustic siren]

ah buenísimo! muchas gracias por compartir eso, voy a entrenar de nuevo el elemento mío y lo voy a estar probando

[ocean wind]

Si tenés que entrenarlo de nuevo así te lo saca de data base y te lo pone como elemento es la nueva evolución del viejo data base

Y está muy mejorado

[acoustic siren]

apenas termine en lo que estoy, voy a reentrenarlo 🙂

[echo mulch]

Why there are limits to how many words we can use, which is '1500', sometimes I need more words to describe my characters or anything else, but I can't because of the limits, which causes me to use other platform while I did not have to from the first place?

[viral oak]

Hey, is anyone here facing this problem? I can't write down prompt when upscaling in leonardo ai.

[sonic swift]

I am trying to make small flowers, like really really tiny flowers. I have tried this so far in the prompt, and the flowers are still big

pattern of very small overlapping watercolor flowers,
pattern of countless tiny watercolor flowers
pattern of thousands of ultra-small hand-drawn flowers
pattern of countless tiny watercolor flowers,
pattern of countless ultra-tiny flowers, each only a few millimeters wide.
pattern of countless ultra-tiny flat watercolor flowers, each no larger than a few millimeters
pattern of countless nano-tiny flat watercolor flowers, each no larger than a 2 millimeters,
pattern of countless surreal nano-tiny flat watercolor flowers

I have run out of ideas - all of them has to be as small as the red flower here:

Nevermind - I think I worked it out now 🙂 (The layout is structured in hidden micro-grid zones 20 by 20 mm for visual balance at large scale -- this helped)

[haughty plover]

Hi there, sorry probably stupid question but can you download PNG’s rather than JPEG of this app?

[jovial plover]

Hi, anyone knows if it is possible change the color of the background of this image but conserving the shadows? and that looks realistic? for example to blue?

what is legacy mode?

[sonic swift]

You can always do it manually

[versed ocean]

crea una escena de crimen

[thin tinsel]

You cannot create on Discord, please go to the website. http://app.leonardo.ai

@plush pond You indicated to say you were seeking assistance. What do you need help with? You can put your questions here.

@halcyon ore You indicated to say you were seeking assistance. What do you need help with? You can put your questions here.

@placid gust You indicated to say you were seeking assistance. What do you need help with? You can put your questions here.

[thin tinsel]

I believe Canvas Editor will let you download PNG

[versed ocean]

escena de crimen

[jovial plover]

oh wow! how did you do it?

[real lynx]

solo en la pagina web podes crear imagenes

[sonic swift]

I have ordinary photo software on my computer - and there is a lot online that you can use for free

[jovial plover]

thanks! Do you know, how can I create a voice ad with AI? for example giving the text and that it give me the audio of it?

[acoustic siren]

look into Suno for this, you can give it the text of what you want and it generates it based on what you prompt it

[echo trout]

guys the prompt for this?

[acoustic siren]

that one most likely is the image as first frame, and the prompt would possibly go along the lines of "camera slowly closes up on a flying castle, suspended above the clouds during dawn"

[sonic swift]

or Kapwing 🙂

[real lynx]

I managed to recreate the prompt?

[marble cape]

Hey is motion free?

?

[real lynx]

no

is premium

[signal terrace]

"Una representación realista de cuerpo completo del rey Tutankamón asumiendo el trono en el antiguo Egipto. Es un joven faraón vestido con atuendo real egipcio, incluyendo una corona nemes dorada, collar con joyas y bastón ceremonial. Está sentado con confianza en un trono dorado ornamentado, elevado sobre una plataforma. A su alrededor hay nobles, sacerdotes y guardias egipcios observándolo con reverencia. El fondo muestra columnas del palacio, muros con jeroglíficos y braseros encendidos, capturando la majestuosidad de la ceremonia de coronación

[real lynx]

hola

las imagenes las creamos en la web

no en discord

https://app.leonardo.ai/

proba por aqui

[steep bolt]

Hi, why can't I write prompt in Univeral Upscaler anymore ?

[next cedar]

Hola

[steep bolt]

Please help

[acoustic siren]

hi there, I believe the option is temporarily disabled as a few people have asked about it today

[steep bolt]

Damn it they keep mesh up the stable things !

[dreamy spindle]

"Seven angels stand before the divine throne, draped in radiant robes of light, their engraved golden trumpets reflecting the glow of a firmament pulsing in sapphire and celestial fire. The golden aurora dances between ebony clouds and shimmering incense, while ethereal echoes resonate across the vast cosmos. Captured in high-resolution 4K, 16:9 aspect ratio, with 35mm lenses, background blur emphasizing the golden luminosity. The scene is illuminated by soft beams of light contrasting with dramatic shadows in warm tones."

[lucid eagle]

#🙋│image-help

[haughty plover]

How do I download multiple PNG files? Thank you very much.

[valid pier]

#🙋│image-help How do I get his mouth not to move?

Hi how do I get his mouth to not move?

A 1990s cartoon action hero in a patriotic gym setting, drawn in bold cel-shaded style with thick outlines and flat colors. He stands in a heroic pose, clenching his teeth in a workout grimace with no lip movement. He wears aviator sunglasses, a brown chevron mustache, a flowing blonde mullet, blue parachute pants tucked into red boots. With his left hand, he tears off a white tank top with a blue star. The shirt rips cleanly away and is aggressively flung to the right in a dramatic arc and spin. The entire shirt leaves the screen — no fabric remains on his chest or shoulders. The animation is 2 seconds long, with quick, snappy motion, like a sudden burst of strength. Background includes wooden gym walls, patriotic decor, weights, and a punching bag. 16:9 format.

[cinder gorge]

Create a realistic image in the style of a traditional painting, depicting a symbolic opening ceremony inside Solomon's Temple, which is still under construction. The scene should include the following characters, all wearing simple white Masonic-style aprons:

Solomon, King of Israel, positioned in the East, wearing a royal robe and holding a scepter, with an aura of authority.

Hiram, King of Tyre, in the West, wearing a Phoenician dignitary robe, with a vigilant posture.

Hiram Abif, in the South, dressed as a master builder, holding a compass or square.

Joab, the Temple Guard, standing outside, dressed as an Israelite guard with a flaming sword.

Adoniram, to the right of Solomon, holding a staff in his left hand.

Benaiah, to the right of Hiram, King of Tyre, also holding a staff in his left hand.

Baruch, to the left of Solomon, holding a quill in his right hand and a small clay inkpot in his left.

Eliakim, to the right of Solomon, holding a key in his right hand and a small cloth pouch in his left.

Abiathar, to the left side of Solomon, dressed in simple priestly garments.

Asaph, in the North, holding a harp.

The Temple should be partially built, with columns rising, scaffolding, construction tools, and stone blocks visible. The atmosphere should feel solemn and sacred, with soft lighting resembling dawn or dusk to enhance the spiritual mood.

[pseudo rover]

I trained an element but I am not able to use it .. I am getting this error ....
Element input is not compatible with sdVeraion SDXL0_9
"

#🙋│image-help I trained an element but I am not able to use it .. I am getting this error ....
Element input is not compatible with sdVeraion SDXL0_9
"

[stiff bloom]

I need a photo of solar panels on the farm em HD

[night vault]

it is known issue. You need to train element with wither SDXL 0.9 or 1.0. It can then be used with all SDXL and lightning models

[pseudo rover]

Thanks. And what is this issue ?

Element input is not compatible with sd v1_5

What is the best way to train a model?
Basically I model to learn a particular style. Should I train a model or an element also how?

[night vault]

what is your trained element setting?

[pseudo rover]

I used leonardo vision xl which didn't work..
Then I retrain another with default settings aas flux dev.. which is working...

But the fine tuned model with stable diffusion is not working

[night vault]

when train use sdxl 0.9 or 1.0 base model. It can be used with any sdxl or lighting models.

[thin tinsel]

I'm pretty sure people can train with Flux, which would be better

[summer falcon]

Hello why is my fixed seed not working ?

[gaunt hull]

hi everyone, i m trying to improve a drawing could anyone help.
I want to make it more realistic and if possible with a nose ring like the ones in the movies.
I have been working with leonardo without success.
I would appreciate if someone could help.

[merry timber]

Hi there, is anyone able to tell me why the images are still being created after more than 1400 seconds? I thought it was only supposed to last a few seconds.

[ocean wind]

Hi

[ocean wind]

Your internet is not working properly because I just created it closed and reopened it or refreshed the page.

[merry timber]

Using the description I boxed at the top?

[merry timber]

Thanks for that... I wonder if it's my free membership that's prohibiting me

Still not working, do Leonardo.ai support read this forum, or is it an email sent to them that's needed?

[merry timber]

No such luck.

Finally worked... on the 2nd time of running it, 4 pictures appeared after a minute.

[night vault]

seems like the first one got stuck

[ivory tinsel]

transformar esse foto em uma cara de dragão

transform this photo into a dragon face

[ocean wind]

Quem é porque ele quer fazer um meme sobre ele 😅

[civic schooner]

is the Character Consistent Model going to work today? Said last week it was available Monday, and today is Monday.

[feral mica]

hi there, I guess you are talking about this, right?

we still have no fixed date for it's release, we want to make sure it up to your standards before doing so, sorry for the wait, it's worth it

[haughty plover]

Is there any way of just doing writing in this app? But like big bold colourful stuff.

[vocal portal]

You can use Phoenix for text 🙂

[civic schooner]

yes, last week it said it was going to be ready Monday, today. lol
definitely worth the wait, and i get it is so worth it. thanks

[haughty plover]

Thank you

[vocal portal]

My pleasure!

[chilly seal]

#❗│get-started

"A series of three connected scenes featuring a stylish baby and a large orange cat.
First scene: A confident baby wearing a gold chain, sunglasses, and a white-and-gold floral outfit, walking outside on a sunny day with palm trees in the background, looking like a cool boss.
Second scene: A baby in a white onesie standing indoors, facing and high-fiving a fluffy orange cat that's standing on its hind legs. The setting is a modern home with bright lighting.
Third scene: A baby and the same orange cat relaxing together on a blue pool float in a clear swimming pool. Both are wearing sunglasses; the baby has tropical swimwear and a drink in hand, while the cat also wears a tropical shirt and holds a drink with a straw and a slice of fruit. Sunny and playful vibe."

[zenith kelp]

I know that this is for image help. But I need help, because I have leveled up and my role is always displaying as novice

And this causes issue with a channel that I can't post in when it's novice

It only changes once I ask for the ranking. And after some time or when reloading (don't know) it goes back to novice again

[zenith kelp]

Is there an admin here?

[pseudo rover]

#🙋│image-help

What is this error?

[copper warren]

from the screenshot it looks like a bug of some sort. could you report it using /report_issue ?

[long ingot]

I keep getting errors after generating images and they don't show up, yet they are still taking tokens. I'm not quick enough to catch what the error says, but it's happened twice in a row now. I was able to generate 2 sets of images successfully before this. It stops and it looks like the images are going to pop up and then they go away with an error. I'm using an Android on the mobile app if that matters.

[night vault]

try click reset settings

[worldly socket]

make a black asian man

[stable carbon]

you're looking to generate in that kind of style.. let me have a think..

[lilac plaza]

yeah

everytime i try it comes out like this

[stable carbon]

do you have paid membership or only working with free tools?

[manic root]

To create an image of a black Asian man, you can use the Leonardo AI app. You'll need to provide a specific prompt that describes the appearance or characteristics you're looking for.

While I can't generate images here on Discord, I recommend going to the Leonardo AI Webapp and entering your prompt there. If you need help with crafting your request or have specific elements you'd like to include, feel free to ask!

[manic root]

It looks like you've shared some images of animated cats! If you're experiencing issues with generating images or have any specific questions related to these images or their creation process, let me know, and I'll do my best to assist you!

[lilac plaza]

paid

[marble cape]

Is it possible for me to generate the same image I have without subscription

[stable carbon]

it seems to be a fairly generic style.. i tried flowstate and used: "3d animation style fat orange cat sitting on the kitchen floor, fuzzy orange hair, warm sunlight"

got the image below, the AI stretched the prompt into: "Tight close-up, 3D animation style shot of a fat orange cat with fuzzy orange hair sitting on a kitchen floor in warm sunlight, iridescent colors overlaid on a soft cinematic haze, detailed focus, soft background focus, iridescent reflections."

I would remove some stuff from that enhanced prompt and sugest the following: "3D animation style shot of a fat orange cat with fuzzy orange hair sitting on a kitchen floor in warm sunlight, detailed focus, soft background focus"

[stable carbon]

you've posted this exact same image how many times across the server now?
what exactly are you trying to do? do you need it in a different style?

yes, you should be able to get that kind of concept in many of the available models

"trading card art illustration of a magical bronze jar, an ominous purple vapour leaks into the air. a spooky face forms in the vapour"

or if you want the exact recreation, just slap it in image2image

[marble cape]

i wanted to look the same but want to save more money

[stable carbon]

i dont understand what you want. you have the image. what do you want different?
would you prefer asking in a different language..?

[marble cape]

what?

i mean i want it to look the same im saving more money for buying subscription cayse image2image is not free and i would need subscrioption

i speak english sir..

some spanish

[stable carbon]

you need to explain better what you want from us. I dont understand what you want to do.
you want to make that exact same image again?
you want to make images in a very similar style..?
you're unhappy with that image you have and want it slightly different..?
we can only help if we know what you want

[marble cape]

i said im trying to save money

no i wanted the image the same but in ai

same style

like you said image2image

[stable carbon]

so you made that yourself, but you want it in your Leo Library now..?
basic Img2Img is available to free users, as far as I'm aware.. 🤔 last i heard free users have style reference and basic img2img..

[marble cape]

yah i design i from scratch on pad and trying to genreate it the same

[stable carbon]

ookay. let me hop on my free user account adn see what image guidance is available

[marble cape]

but what im saying all the other guidance are for subscription

[stable carbon]

if you just want that exact image in your Feed, then you should only need Image2Image.
as i thought. free users have Img2Img and Style Ref.
you should be able to get most of the image in your feed using Img2Img at maximum strength

[marble cape]

Can you show me where just in Case

[stable carbon]

sure, sec..

pick a model. lets say Concept Art, altho with maximum Img2Img strength it shouldnt really matter much..

then image guidance, select "View More" then select Img2Img

then select your image. then crank up the strength to maximum

after that, just add a prompt of your image and generate. note, i think with Img2Img you need to generate in the same dimensions as the reference image, for it to work

[marble cape]

I want it to make it more anime like

In like yugioh artwork

[stable carbon]

then choose an anime model and lower the strength a little. the lower the strength, the more the AI will improvise

[zenith kelp]

💁 Do you know if I can train an element with 1344x1344px images? Because I used the Phoenix model in quality and couldn't get 1024x1024px. The bot told me that I could use them anyway, but it failed to create the trained model.
Anyone knows about this one?

[vocal portal]

I think there is an issue with this today. I currently have 4 failed attempts and decided to wait

[zenith kelp]

ohhhh... makes sense!

[zenith kelp]

you're also trying to train an element?

[vocal portal]

yes 🙂

[zenith kelp]

cool! 😃 and by any chance, have you already tried to train an element with another format than 1024x1024? such as 1344x1344?

[vocal portal]

Today? Yes. and I also tried with 1024x1024 and it also failed.. 3 times lol. I'm stubborn 😆

In the past? A larger size worked fine.

[zenith kelp]

I totally understand that! it happens to me very often! LOLLL

[zenith kelp]

and great to know that ir worked in the past!

[vocal portal]

Hopefully it will be resolved soon 🙂

[zenith kelp]

I hope the same!

I would need help from someone please! I want to cover that part of the chest with hair, and add a white shirt or anything else! LOL And it's not working for me... 😢

[vocal portal]

Your inpaint strength is to low. Raise that to maybe .85. That should help

[zenith kelp]

it improved! but not getting a super good result.

[vocal portal]

Try masking more of the shirt and changing the prompt to white shirt instead of "add a" white shirt

[green perch]

Create a caramel-colored mongrel dog with completely human behavior, in the style of films like "Bud, the Friendly Dog" or "Cats and Dogs", but with a hint of Brazilian country humor. His name is Totó da Roça, he speaks with a Minas Gerais accent, and acts like a grumpy and wise old man, full of funny reflections and criticisms of everyday life. He walks on two legs, holds a coffee mug, sits with a human posture, and gestures with his paws like a human talking. He reacts with exaggerated facial expressions, crosses his arms, scratches his head, points his finger, and even sighs as if he were carrying the weight of the world. The setting is a simple country kitchen, with a wood-burning stove in the background, the sound of chicken and the radio playing a modão. The video is 60 seconds long, in vertical format, with a light, charismatic rhythm and a reflective message at the end. Humor style: funny rural chronicle with constructive criticism. Target audience: pet lovers, people who

[zenith kelp]

Let me try that!

[vocal portal]

Hi! Please create an account here https://leonardo.ai/

[green perch]

#conversando

[vocal portal]

He looks like a good conversationalist. Did you need some help?

[zenith kelp]

I always get weird images that don't go well...

[vocal portal]

He does look kinda mauled huh? Have you tried using sketch to image and drawing in a white patch and then prompting for one?

[zenith kelp]

yeah... how do I do that?

[vocal portal]

one sec

[zenith kelp]

sure!

[real lynx]

hi

[vocal portal]

You should be able to change the canvas mode to sketch in the top right corner. Then choose the size brush you want and the color

[vocal portal]

Hi Cami!

[zenith kelp]

So I paint white in the area I want to change? and add the prompt?

[vocal portal]

Yes

[real lynx]

yo tambien blurearia mas la parte rosa para que la ia entienda el completo de tu dibujo y donde queres que ponga la camiseta y cambiaria el modelo a dreamsheaper7

algo asi

[vocal portal]

Thank you Cami!

[vocal portal]

I'm going to have some dinner but will be back in a bit. Cami is great and Willowb is an expert with Canvas.

[zenith kelp]

Thank you so much for your help! I'm getting there now! 😺

[zenith kelp]

ahhh gracias! Y hola! No había visto tu mensaje, pensaba que era la imagen que yo había mandado! 😆

[zenith kelp]

voy a intentar eso. Porque me fui a Sketch2Img

[real lynx]

oki

[zenith kelp]

Y cómo sabes que modelo elegir?

[real lynx]

a la derecha

[zenith kelp]

Finalmente con el Sketch2Img logré algo no tan mal!

[zenith kelp]

siii! me refería un poco al tema que hay muchos modelos y es difícil saber cuál es mejor

[real lynx]

te habia dicho mas arriba :3

[zenith kelp]

osea, si se que me dijiste. Me refiero que sin ayuda, cómo saber? ahora es porque me lo dijiste

Y una pregunta. Ahora que está listo. Tengo que descargarlo? o cómo es?

[humble pendant]

I want to generate an image just like this I want it looking the same with the cats

[zenith kelp]

lo he descargado varias veces y me manda un archivo vacío... 😢

[vocal portal]

You're welcome! I'm glad to see you are making great progress! Cami is wonderful as well!

[real lynx]

a ver me fijo yo

[zenith kelp]

gracias! 😄

[manic root]

To generate images similar to the ones you've provided, you'll need to use the Leonardo.Ai web app, as image generation cannot be done through Discord.

You can visit the app at app.leonardo.ai, and you might want to describe the key elements you want in the images, such as the style and the cats' appearance.

If you need any more specific guidance on how to set up your generation prompts or any other questions, feel free to ask!

[humble pendant]

I know but the style of the image etc

i want to generate ones like it

like what do i put in to generate images like with the cats etc

[real lynx]

si me lo descargo

[humble pendant]

pls

[real lynx]

You can use the tool to describe images within the web and use this image in image style.

[humble pendant]

it doesnt turn out like the image though it looks all diffrent and weird

[zenith kelp]

ahh ya me funcionó! tenía que hacer "refresh" la página. Gracias!

[zenith kelp]

Yes! she helped me too! 😃

[humble pendant]

it made it look like this it looks like cartoony like still better though

compared to this more realstic version

[manic root]

To generate images similar to the ones you've shared with the prominent cats and their realistic features, here are some suggestions for your prompts:

1. Character Features: Include specific characteristics such as "a muscular white cat with well-defined muscles, wearing red gym shorts" or "a cartoonish cat with human-like features".

2. Style Descriptors: Mention the style you’re going for. For example, use phrases like "hyper-realistic" or "digital illustration with cartoon elements" in your prompt.

3. Context: Specify the environment, like "in a gym" or "walking in a cityscape", to set the scene that captures the essence of your reference images.

4. Pose and Actions: Describe specific poses, such as "standing confidently with arms crossed" or "holding hands with another character".

Here’s a sample prompt you could use:

• "A hyper-realistic muscular white cat with defined muscles wearing red shorts, confidently posing in a modern gym setting, emphasizing its cartoonish human-like facial features."

Feel free to tweak these options until you get the desired results! If the outputs are still looking off or too cartoony, you could adjust your descriptors towards a more realistic style. Let me know if you need further adjustments or guidance!

[humble pendant]

stil lcant get bro

damn

[manic root]

I understand that you're having trouble getting the images to turn out as you'd like. Here are a few tips to help improve your results:

1. Refine Your Prompt: Be specific about what you want. Instead of broad terms, try more detailed descriptions. For example, "a hyper-realistic muscular white cat posing confidently in a gym" may yield better results than just "muscular cat."

2. Experiment with Styles: If one style isn’t working, try another. You might want to specify "realistic" or "semi-realistic" in combination with "cartoonish" to find a middle ground that suits your taste.

3. Use Reference Images: If possible, use images similar to what you want to generate as reference images in the app. This can help the model understand better what you're aiming for.

4. Adjust Parameters: If the app allows for adjustment settings, try tweaking factors like "creativity" or "style strength" to see if that yields a better output.

If you're still not getting the results you want even with these changes, feel free to share specific changes you made or ask for further assistance!

[humble pendant]

i just want images just like the photo is there a type a style I need to use

[manic root]

To generate images just like the ones you've provided, you might want to focus on using specific styles and elements in your prompts. Here are some suggestions:

1. Art Style: Specify an art style that matches your desired outcome. You can try using terms like "digital painting," "3D render," "hyper-realistic," or "stylized cartoon."

2. Character Description: Include detailed descriptors. For instance, "a muscular white cat with exaggerated muscles and human-like features, wearing red shorts, in an athletic pose."

3. Environment and Action: Describe the scene and action. For example, "in a modern gym setting with fitness equipment visible" or "walking in a busy city street with taxis around."

4. Facial Features: Be specific about the features of the characters. You could say something like, "large expressive eyes, friendly smile," to capture the likeness of your reference images.

Here’s a sample prompt you could try:

• "A hyper-realistic 3D rendered muscular white cat with defined muscles, wearing red gym shorts, confidently posing in a modern gym with fitness equipment."

Feel free to adjust your prompt based on these suggestions until you achieve the style you're looking for! If you encounter issues or need more refinement, let me know!

[humble pendant]

it made this ugly

[manic root]

I’m sorry to hear that the generated image isn’t matching your expectations! Here are a few strategies you can try to improve the output:

1. Adjust the Prompt: Sometimes, being more specific or using different adjectives can help. If "hyper-realistic" isn't working as intended, you could try terms like "realistic cartoon" or "stylized realism."

2. Modify the Character Features: Instead of just asking for a "muscular white cat," you could specify additional details, such as "a white cat with a smooth, glossy fur texture" or "a friendly expression with a confident stance."

3. Experiment with Sizes and Proportions: Sometimes adjusting how you describe the body proportions can help, like "a muscular yet proportionate body" or "exaggerated muscles that still look natural."

4. Use More Context: Adding context can also shape the generation. You could specify things like "soft lighting in a gym environment" or "dynamic pose that conveys movement."

5. Iterate: If you have specific elements from your preferred images, describe each one separately in your prompt for clarity.

Here’s a revised prompt example you might consider:

• "A realistic cartoon-style white cat with smooth fur and an approachable smile, showcasing a muscular body in red gym shorts, standing confidently in a modern gym with soft lighting."

Feel free to tweak these suggestions to suit your vision better, and don't hesitate to share any further thoughts or desired changes!

[humble pendant]

I would like it to be the same art style as the image

and facial features build etc

[manic root]

To capture the same art style, facial features, and overall build as the images you've shared, here are some more tailored tips for your prompts:

1. Art Style Specificity: Include terms that reference the specific art style you want, such as:

   • "in the style of 3D animated characters"
   • "cartoonish yet muscular design"
   • "sleek and glossy rendering"

2. Facial Features: Describe the facial features in detail:

   • "large, expressive eyes with a friendly look"
   • "a smiling, playful demeanor"
   • "human-like facial structure with a cartoonish feel"

3. Body Build: Specify the body design you're aiming for:

   • "exaggerated muscular build with defined abs and arms"
   • "proportionate muscles that are realistically stylized"
   • "defined pectorals and large arm muscles"

4. Hold on Contrast: If you want a similar contrast in colors (such as fur and shorts), make comparisons like "vibrant red shorts" or "pure white fur that has a glossy finish."

Here’s an example prompt you might use to align closer with your desired output:

• "A 3D animated character style white cat with large expressive green eyes, smiling softly, featuring a muscular build with defined muscles and realistic proportions, wearing vibrant red gym shorts, set in a modern gym environment."

Feel free to modify these suggestions according to your vision, and don’t hesitate to ask if you’d like more help or different variations!

[orchid beacon]

Does any leonardo model know how to do two character reference in the same photo?
I’m doing a photo-story and in some images i need the couples in the same picture
I have been using overlay and manually putting them together with adobe

I just realized chatgpt can do two character references but not as close as Leonardo does it per character

Wondering if Leonardo can do it?

[manic root]

Currently, the Leonardo models do not support two character references in the same image simultaneously. You can use techniques like image overlay, as you've mentioned with Adobe, to combine images of two characters.

For character references, each model has limitations. While some may work well for single character references, combining characters or having them interact closely in the same image is not yet a supported feature in Leonardo.

If you have any further questions or need assistance with a specific aspect of your photo-story, feel free to ask!

[orchid beacon]

Thanks

[manic root]

You're welcome! If you have any more questions or need further assistance, just let me know. Happy creating! 😊

[deep cradle]

Has anyone had success using the canvas editor to replace a hand that has four fingers with something that has 5 instead? I'm trying to make some subtle changes to my image and maintain the exact same style but I'm not having much luck. When I attempt to replace a hand, I either get a distorted blob or it ends up looking nothing like a hand.

[manic root]

When using the canvas editor to replace a hand with four fingers with one that has five, there are a few steps and tips that might help you achieve better results:

1. Use the Masking Tool: First, make sure to use the 'Mask' tool to clearly define the area you want to change (the hand in your case). Properly masking the area ensures the AI has the right context to work from [【4:8†source】].

2. Adjust Inpaint Strength: Set the Inpaint Strength to a high value (around 0.7 to 0.9). This increases the AI's adherence to the changes, but keep in mind that a higher strength could lead to less fidelity to the original image [【4:8†source】]. Experimenting with this setting can help find a balance that works for your image.

3. Specific Prompts: When entering your prompt to generate the new hand, be specific about what you need. Instead of just asking for a "hand," you could describe it as "a realistic human hand with five fingers" to provide clearer guidance [【4:18†source】].

4. Image Selection: You may also want to experiment with different finetuned models if you’re not getting satisfactory results. Models like "Dreamshaper" are recommended for artistic images, while others might be better for realistic outputs [【4:15†source】] [【4:16†source】].

5. Preview Multiple Variations: When generating the images, select to generate multiple variations. This will give you options to cycle through, allowing you to find the best match to your needs [【4:14†source】].

[manic root]

6. Editing Original Images: If you're still facing distortion issues, consider editing the original version of the image and upscaling later instead of working with an already upscaled image, which can complicate details [【4:10†source】].

Following these steps should help you get closer to achieving a hand with five fingers that maintains the style of your original image. Don't hesitate to experiment with settings until you find the right combination that works for your specific image and artistic intent!

[plucky yarrow]

None of the images I’ve liked have shown up in my Liked Images in a week and it’s getting frustrating. No idea what to do

[manic root]

It sounds like you're experiencing issues with your Liked Images not showing up. Here are a few steps to troubleshoot the problem:

1. Check Your Account: Ensure that you are logged into the correct account where you liked the images.

2. Refresh the Page: Sometimes, simply refreshing the page can resolve display issues.

3. Clear Browser Cache: Clear your browser's cache and cookies. This can help resolve any loading issues with the website.

4. Check Storage Limits: Verify if there are any limitations regarding the number of images you can like or if you've reached any storage limits.

5. Reach Out for Support: If none of the above steps work, please reach out to customer support by emailing them at [e-mail] or using the support chat in the app.

If this issue continues, it might be a bug, and reporting it would be a good course of action. You can do this by using the command /report_issue in the app.

Let me know if you need further assistance!

[nimble falcon]

I'm having a hard time getting results to show my character being angry no matter what prompt I use. Please help!

[manic root]

To achieve improved results for generating an image depicting your character as angry, consider the following tips:

1. Use Character Reference: Start by uploading a reference image of the character that clearly shows their facial features. This image can help the model understand the specific traits of your character when generating the image. After selecting the image, adjust the strength of the reference in the prompt settings to ensure it influences the output more accurately [【4:5†source】] [【4:14†source】].

2. Detailed Prompt: Craft a very detailed prompt that describes not only the physical appearance of the character but also their emotions and actions. For example, you might say, "a male character with short black hair, expressive eyes showing anger, and furrowed brows, clenching fists." The more vivid and specific you are about the anger expression, the better the model can translate this into the image [【4:10†source】] [【4:19†source】].

3. Experiment with Strength Settings: In your Character Reference settings, use a higher strength level to ensure that the character’s likeness is maintained throughout the generation process. This helps in achieving consistency across various expressions [【4:14†source】] [【4:19†source】].

[manic root]

4. Adjust Weighting: If you're using any additional image guidance or references, experiment with the weighting settings. A higher weight can make the AI adhere more closely to the reference image but may require fine-tuning to avoid unwanted features in the output [【4:0†source】] [【4:12†source】].

5. Use of Mood Description: Emphasize emotions in your prompt, such as including phrases like “looking furious” or “with an intense glare,” to help convey the desired emotion more clearly to the AI, thus affecting how the character is depicted [【4:17†source】].

By following these tips and experimenting with different settings, you should be able to generate images that successfully portray your character expressing anger. If results continue to be suboptimal, consider reaching out for specific support or feedback on your current prompts and methods.

[raw thicket]

@shell drum Hi! How cani help you?

[raw thicket]

What model / settings are you using?
Are you using a guidance image?
What does your prompt look like?

[nova light]

I use my robot as a coffee maker

[manic root]

That's an interesting way to utilize your robot! Using technology creatively can really enhance our daily routines. How does your robot make coffee? Is it fully automated, or do you need to assist it in some way?

[real lynx]

hi

I also use my coffee machine robot

[north knoll]

image-help

[manic root]

Could you please provide more details about the kind of help you need with images? This could include questions about image generation, sharing images, or specific features of the Leonardo A.i app. Let me know how I can assist you!

[acoustic siren]

do you need help?

[real lynx]

hola necesitas ayuda?

[pale summit]

hi

[manic root]

Hello! How can I assist you today? If you have any questions about Leonardo.Ai or anything else, feel free to ask!

[pale summit]

why cant i see the whole server

[manic root]

If you're unable to see the whole Leonardo Discord server, here are a few common reasons and solutions:

1. Role Permissions: Sometimes, you might not have the appropriate roles to access certain channels. Make sure you've completed any necessary verification or role assignments.

2. Server Settings: The server might have certain channels set to private or restricted access. You can check with server admins or moderators for more information about specific channels.

3. Temporary Glitches: Occasionally, Discord can experience glitches. You can try refreshing your Discord app, logging out and back in, or clearing your cache.

4. Joined Recently: If you just joined the server, it might take a moment for all channels to load properly.

If the issue persists, consider reaching out to server support for assistance.

[copper warren]

#❗│get-started to get access to the rest of the server

[tall elbow]

muy creativa xD

[real lynx]

era cafe con leche

[pale summit]

tyyy

[tall elbow]

jajaja si me lo imaginé xD

[wheat bison]

Can anyone tell me how to see images I've created with flow state? Page was refreshed and everything went away.

[night vault]

flow state never save unless you click save buttons.

[manic root]

To view the images you've created with the Flow State feature, you'll need to go to your Library within the Leonardo.Ai platform. Here are the steps to follow:

1. Log into Your Account: Visit the Leonardo.Ai website and log in to your account.
2. Navigate to the Library: Look for a tab or section labeled "Library" in the menu.
3. Find Your Images: In the Library, you should see all the images you've previously generated, including ones from Flow State.

Do keep in mind that if you refresh the page or leave the Flow State feature, any current generations in Flow State will be cleared. So always be sure to save images you like before doing so [【4:17†source】].

If you have any more questions or run into issues, feel free to ask!

[night vault]

@copper warren not sure why bot step in and yet did not mentioned save button

[wise canyon]

Hi Zorgi 🤟
it should also autosave any images that you select more like this on 👍

[copper warren]

yeah, I think it meant to cover that with "if you refresh the page or leave the Flow State feature, any current generations in Flow State will be cleared. So always be sure to save images you like before doing so"

[night vault]

i feel if we can reorganize that first because that is more important that we need to save or more like this then viewing them they should already have know nkw to

[copper warren]

the message comes from gpt's analysis of the question, based on its knowledge base (mainly Leo help docs). so it cannot be re-organized manually, but I will do my best to improve the gpt instructions to form it

[alpine ore]

hi i need some advice how i promt my charater refrence has a emotion smile and angry ,, my character always flat face , i already prompt words smile and my character always have a flat face ,, any advice

[manic root]

To prompt your character with specific emotions like a smile and anger, you'll want to provide more context in your descriptions. Here are a few tips:

1. Detailed Emotion Descriptions: Instead of using only the word "smile," describe the type of smile you want. For example, "a wide, joyful smile" or "a sneering, sarcastic smile" can convey different emotions. Similarly, for anger, you might specify, "furrowed brows with a scowl" or "clenched teeth and wide eyes."

2. Facial Features: Include descriptions of the facial features associated with each emotion. For a smile, you could say, "eyes sparkling with joy" and "cheeks raised," while for anger, you might include "brow furrowed" and "lips pursed tightly."

3. Body Language: Incorporate body language that complements the emotions. For a happy character, you might say, "standing tall with hands on hips, radiating confidence," and for anger, "a tense stance with fists clenched."

4. Context or Scenario: Sometimes, providing a scenario can help set the mood for the emotion. For example, "After winning a battle, her face lights up with an exuberant smile" or "In a moment of betrayal, his face twists into a furious glare."

5. Iterate and Experiment: If your character still appears flat, continue to tweak your prompts and try different combinations of words until you achieve the desired effect.

Based on these suggestions, you can create a more comprehensive prompt that captures the emotions you want: "A character with a wide, joyful smile and sparkling eyes, standing tall, radiating happiness" or "A character with furrowed brows and clenching fists, lips pressed tightly, radiating anger."

Give it a try and adjust based on the results!

[timid ridge]

Design a professional, sleek, and user-friendly kiosk interface for a Smart Query Management System, for kiosks like metro stations or visa centers. The main interface should feature 6 distinct service options related to visa processing, such as "Apply for Visa," "Document Verification," and "Appointment Scheduling." When a user selects a service, the system should generate a token.

The design should incorporate a light mode theme using a clean, modern aesthetic with primary colors #ba962e (gold) and #828181 (gray) for branding accents. Use large, bold buttons, clear typography, and minimal text for quick readability. Include simple, intuitive icons for each service. Layout should be touch-optimized, fully accessible, and visually engaging, with balanced spacing and clear visual hierarchy. The overall look should convey professionalism and ease of use.

[manic root]

To create a professional, sleek, and user-friendly kiosk interface for a Smart Query Management System, consider the following design elements:

Interface Overview

1. Theme:

   • Light mode with a modern and clean aesthetic.
   • Primary colors for accents:
     • Gold (#ba962e)
     • Gray (#828181)

2. Overall Layout:

   • A touch-optimized interface.
   • Balanced spacing for a visually engaging experience.
   • A clear visual hierarchy for ease of navigation.

Main Interface

1. Header:

   • Logo of the Smart Query Management System.
   • A concise tagline (e.g., “Your Visa Processing Solution”).

2. Service Options:

   • Six distinct service options arranged in a grid layout for easy access:

     1. Apply for Visa
     2. Document Verification
     3. Appointment Scheduling
     4. Track Application Status
     5. Payment Processing
     6. Customer Support

   • Button Design:

     • Large, bold buttons with rounded corners for a modern look.
     • Clear typography (e.g., using a sans-serif font) for readability.
     • Icons next to each button for visual cues (e.g., a passport icon for "Apply for Visa").

3. Interactive Feedback:

   • When a service is selected, visually highlight the button (e.g., change background color to gold) and display a generating token message.

4. Footer:

   • Options for language selection.
   • Contact information or help icon for customer support.

Accessibility

• Ensure text has sufficient contrast against the background for readability.
• Buttons should be large enough to be tapped easily, with enough spacing to prevent accidental selections.
• Include alternative text for icons to assist users with visual impairments.

Visual Engagement

• Use high-quality icons that are intuitive and relevant to each service.
• Incorporate subtle animations (e.g., button shadows or hover effects) to enhance interactivity.

[manic root]

• Consider illustrations or background shapes in gray to create visual depth without being distracting.

Summary

The final design should evoke professionalism while remaining user-friendly and accessible. Test the interface with potential users to gather feedback and make necessary adjustments for optimal functionality and engagement.

[stoic lance]

help

[alpine ore]

for example like this how i should prompt that his expression face angry and mad , any tips how i prompt ? at least please give me a example

[manic root]

It looks like you've asked for help, but I'll need more specific details about what you need assistance with. Are you having trouble with generating images, logging in, or perhaps something else related to Leonardo.Ai? Please provide a bit more context so I can assist you better!

[alpine ore]

yes i having trouble and so sturggle that i need help for generating images that the face always flat although i already prompt a facial expressiom and keep like this ,,, hold a for second i would upload some images

example this ,,, i really want to generate images with expression mad or something like this picture how do i prompt ? i type prompt a black african women have a expreesion angry face is that correct ? or i should add something please help

[manic root]

To create a prompt that captures a strong angry expression similar to the characters in the images you've uploaded, you can use detailed descriptions. Here are examples of prompts that can convey anger effectively:

1. For the Female Character:

   • "A strong African woman in a police uniform, standing with arms crossed, her lips set in a firm line and eyes narrowed, displaying a fierce expression of anger as if confronting someone. The background shows a tense urban environment."

2. For the Male Character:

   • "A male figure in a military uniform with a furious expression, brow furrowed and lips tightly pressed in frustration, holding a rejection letter with his hand clenched in a fist. The setting should convey tension, with shadows emphasizing his anger."

General Tips for Prompting:

• Specify Facial Features: Include descriptions like "furrowed brows," "tight mouth," or "narrowed eyes."
• Body Language: Mention posture, such as "arms crossed" or "hands clenched."
• Contextual Details: Add elements that suggest a reason for their anger, like "standing amidst a chaotic scene" or "holding a letter of rejection."
• Emotional Tone: Use adjectives that convey intensity, like "fury," "rage," or "frustration."

By enhancing your prompt with these details, you'll help the generation process capture more expressive emotions.

[alpine ore]

A ferociously brooding African man is depicted in his home, his face contorted with unmistakable anger. The image is a vivid painting, capturing the intensity of his emotion. Every detail is impeccably rendered: his dark skin glistens with tension, his eyes smolder with a fiery rage, and his clenched fists reveal the depth of his frustration. The atmosphere is charged with raw energy, making this piece a powerful and evocative portrayal of human emotion.

that is my prompt ,,, i already type a word related to anger but still his face keep flat

[manic root]

It sounds like you're very close to achieving the emotional expression you want, but may need to refine your prompt a bit further. Here are some suggestions that can help evoke a more expressive angry face based on your prompt and the image you provided:

Refined Prompt Suggestions:

1. Focus on Specific Expressions:

   • Instead of just saying "anger," you can describe specific facial features associated with anger:
   • "A ferociously brooding African man in his home, his eyes wide and glaring, brows deeply furrowed, and lips curled in a snarl of anger. His dark skin glistens with tension, and his clenched fists show deep frustration, as he stands in a dimly lit room filled with shadows."

2. Add Contextual Emotion:

   • Providing context can help set the scene for the emotion:
   • "In a tense moment of frustration, a brooding African man is portrayed in his cluttered home. His face is twisted in anger, with intense eyes and gritted teeth, gripping his hair in frustration. The shadows cast by candlelight enhance his furious expression."

3. Use Action Words:

   • Action words can animate the expression more:
   • "An African man, visibly enraged, is depicted in his home, eyes blazing with anger. His brows are knit tightly together, and his mouth is set in an aggressive frown, fists balled, as he strikes a powerful stance that reflects his inner turmoil."

Additional Tips:

• Test Variations: Experiment with variations of adjectives and actions. Sometimes small changes can make a big difference.
• Avoid Flat Expressions: Ensure words like "flat," "calm," or neutral descriptors are not included, which might contradict the emotional intensity you seek.
• Iterate: If the first render doesn't capture the emotion, try adjusting the wording slightly and generate again.

Hopefully, by incorporating these adjustments, you'll see a more expressive angry face in your generated images!

[tacit fulcrum]

Is there a way to change the texture of an image whilst maintaining the original art?

I would like to take any digital image and change the texture to cotton or wool without changing the art itself.

[manic root]

To change the texture of an image while keeping the original art intact, you can utilize the Image2Image feature available on Leonardo.AI. This feature allows you to transform your image based on the input of another image, enabling you to apply a new texture like cotton or wool without altering the essential artwork itself.

Here's how you can do it:

1. Upload your original image to the Leonardo.AI platform.
2. Use the Image2Image feature to incorporate your desired texture. When you set up the Image2Image generation, you can provide a prompt that specifies the texture you'd like to be applied.
3. In your prompt, be clear about the aspects you want to maintain from the original artwork while describing the new texture you wish to introduce.

Keep in mind that adjusting the Strength setting can help you control how much of the original image remains dominant in the final result. A higher strength will make the original image more dominant, while a lower strength will allow more of the new texture to show through [【4:1†source】].

If you need more detailed control, you can also explore options like inpainting or outpainting, which allow for significant modifications while respecting the original context [【4:1†source】].

[unique grail]

Hello everyone. Can someone guide me on how to upload images generated on Google ImageFX to Leonardo ai?

[manic root]

To upload images generated on Google ImageFX to Leonardo AI, follow these steps:

1. Save the Image: First, save the image you generated in Google ImageFX to your local device. This is typically done by right-clicking the image and selecting the "Save image as..." option.

2. Upload to Leonardo AI:

   • Go to the LeonardoAI web app at app.leonardo.ai.
   • Log in to your account.
   • Once logged in, navigate to the image upload section of your library.
   • Click on the upload button and select the image file you saved from Google ImageFX.

3. Generate a Link (Optional): If you want to share the uploaded image on Discord, you can either right-click the image in your library to copy the image link or use the share button.

4. Paste in Discord: Finally, paste the link into the Discord chat. If you want to add a spoiler to the link, you can use the double lines || on either side.

If you have any issues or need further assistance, feel free to ask!

[unique grail]

Thank you for your response. In the library, I do not see an "Image Upload" tab anywhere.

[ocean wind]

Hi

Where you write your prompt has a symbol to upload images

[unique grail]

Hello Kami.

[unique grail]

I am using the free plan. Do I need to upgrade plan to use the image upload feature?

I got it. Thank you very much, Kami.

[manic root]

To upload images to Leonardo AI, you need to be aware of the following:

1. Image Upload Feature: The image upload feature is only available for users who have increased their rank on the server. This typically means participating in chat and completing daily missions.

2. Free Plan Limitations: If you're using the free plan, you will not have access to upload images directly. Upgrading to a paid plan will provide you with the option to upload images, among other benefits.

If you decide to upgrade, you'll be able to access the image upload feature along with other features. If you have any more questions or need clarification, feel free to ask!

[unique grail]

Thank you Botticelli.

[manic root]

You're welcome! If you have any more questions or need further assistance, feel free to ask. I'm here to help!

[ocean wind]

Yes

You can use some functions

But more better features are premium

[unique grail]

OK. I was able to upload one image. I have upgraded to a paid plan. Please walk me through the upload again. the method I used earlier does not seem to work anymore.

[manic root]

Great to hear that you've upgraded to a paid plan and are ready to upload images! Here's a step-by-step guide to help you with the upload process:

1. Log In: Go to the Leonardo AI web app at app.leonardo.ai and log in to your account.

2. Navigate to Your Library: Once you are logged in, look for the library section in the navigation menu. This is typically where you'll manage your images and uploads.

3. Upload Image:

   • Look for an "Upload" button or an icon that indicates adding or importing images (the exact wording may vary).
   • Click on it, which should prompt you to choose an image file from your device.
   • Select the image you wish to upload and confirm the upload.

4. Confirmation: After uploading, you should see the image appear in your library. If you don't see it immediately, try refreshing the page.

5. Sharing (Optional): If you wish to share the uploaded image, right-click on the image in your library to copy the link or use the share button for the same purpose.

If you encounter any issues during this process, please mention any error messages or details so I can assist you further!

[ocean wind]

For the simplest form use character image to imitate a face

Content reference to imitate a pose

And you can also create a parrot, a training element, but that requires 50 photos.

[wheat bison]

ok but why is it like this? I spent a bunch of credits on flow state and it all just goes away if I refresh or restart my computer?

[manic root]

I understand how frustrating that can be, especially after spending credits on generating images. The Flow State feature in Leonardo.Ai doesn't automatically save images once you refresh or leave the feature. It's designed this way because the images created in Flow State must be manually saved by clicking the save button to ensure they're stored in your Library.

This behavior is intended to give users the freedom to create freely without cluttering their Library with unintended images, but it can be inconvenient if you forget to save before refreshing or leaving the page.

Unfortunately, this is how the feature is currently set up, and I recommend always saving images you want to keep as soon as they are generated. If you have feedback about this process, it might be helpful to share it with the support team or use the /report_issue command to suggest improvements.

[wheat bison]

Another question - can you buy more tokens once your plan runs out or are you stuck with relax only?

[manic root]

Yes, you can purchase more Flow State tokens once your current plan runs out. You are not limited to just the "Relax" mode if you wish to continue generating images. You can buy additional tokens to continue using the features available in your plan.

If you find yourself running low on tokens frequently and you're looking for a more consistent experience, consider looking into upgrading to a plan that better suits your needs. If you have any other questions or need assistance, feel free to ask!

[alpine ore]

is so hard make edist face expression like smile or what is anything i need change style or something is so frustasted

always a flat face

[manic root]

I understand your frustration with trying to achieve expressive facial features in your generated images. Here are some additional suggestions that might help you get more dynamic and varied expressions like smiles or other emotions:

1. Be Specific with Emotions:

• Rather than just using general terms, be explicit about the type of smile or expression. For example:
  • Smile: "A bright, joyful smile with sparkling eyes, showcasing teeth and showing excitement."
  • Anger: "A fierce scowl with narrowed eyes and clenched jaws, expressing intense frustration."

2. Describe Facial Dynamics:

• Include detailed descriptions of how the face changes with emotions. For example:
  • "Her lips curl into a wide, cheerful grin, eyes crinkling at the edges with warmth."
  • "His face twists in rage, with furrowed brows and a tense jaw clenched tightly."

3. Use Contextual Cues:

• Setting the scene can help focus the expression:
  • "A woman laughing joyfully at a joke, her entire face lighting up with delight."
  • "A man shouting in anger, his face turning red with exasperation, fists raised."

4. Experiment with Different Styles:

• If you're not getting the desired results, consider experimenting with different artistic styles. Sometimes changing the art style can impact how expressions are rendered.
• For example, try asking for "a cartoonish style" or "a dramatic portrait style."

5. Iterate with Variations:

• Adjust your prompts slightly and try generating multiple times to see if you can hit the right expression. Sometimes, subtle changes in wording make a significant difference.

6. Feedback and Refine:

• If possible, assess the outputs you get and take notes on what works or doesn’t. This can help refine your approach over time.

Example Prompts:

[manic root]

• For a Smile: "A joyful African woman with a beaming smile, eyes sparkling with happiness and a wide grin revealing bright teeth."
• For Anger: "An angry African man furiously yelling, his face red with rage, teeth clenched and brows deeply furrowed."

Using these strategies can help you produce more varied and expressive faces in your images. Keep experimenting, and don't hesitate to tweak your prompts based on the results you're getting!

[alpine ore]

is that i need change with element ? or i chnage elemetn like character prefreence ? maybe some advice

[manic root]

If you're considering whether to change elements or character preferences to achieve better expressive faces, here are some tips and things to consider:

1. Character Preferences:

• Refining Character Traits: Ensure that your character's traits are set to reflect the emotions you want to express. If your character is typically stoic, even a prompt for anger might result in a flat expression. Try adjusting the character's background or personality to be more expressive.
• Expressive Characteristics: Choose character features that are traditionally associated with strong emotions, such as features that might suggest a more exaggerated expression.

2. Element Types:

• Using Different Elements: If you're using specific elements (e.g., styles or themes that may influence the output), consider switching to an element that supports more dynamic facial expressions.
• Experimenting with Styles: Some styles may lend themselves more to expressive rendering than others, so trying different artistic styles for your character may help.

3. Image Guidance Options:

• If you're using models that allow for content or style references, consider employing those features to add more context to your prompts. This can guide the generation towards the type of expression you want.
• Use elements that are known to support facial expression rendering, such as models designed for emotional or dramatic artwork.

4. Iterate and Experiment:

• Test Various Combinations: Change one aspect at a time, whether it’s the character, the element, or the details in your prompt. This helps isolate what works and allows you to refine your approach.
• Review and Adjust: Regularly review the output and adjust your approach based on what you see. If certain prompts result in flat faces, tweak your descriptions to be more vivid.

Example Adjustments:

[manic root]

• If you find that a character always comes out flat, you might change to a more vibrant or expressive model or add new attributes to your character’s description.
• If using character preferences, try introducing traits like “charismatic” or “emotional” that encourage a varied expression.

In summary, consider adjusting both the details in your prompts and the character elements you're using. Experimentation is key, so don’t hesitate to try a variety of approaches until you find what resonates with your creative goals!

[alpine ore]

for example i need my character is like running and like act waving which choice that i should use ?

[alpine ore]

where word should i out emotional in which i should in front or back ?

[manic root]

Typically, whether you place the word "emotional" at the front or back of your character description can depend on what you're trying to emphasize in your prompt. Here are some examples to illustrate:

1. Placing "Emotional" at the Front:

• This emphasizes the emotional quality from the start:
  • Example: "An emotional African woman with a bright smile, her eyes sparkling with joy."
  • Example: "An emotional African man, visibly angry, with furrowed brows and clenched fists."

2. Placing "Emotional" at the Back:

• This places emphasis on the character first before introducing their emotional state:
  • Example: "An African woman with a bright smile, radiating emotional warmth and joy."
  • Example: "An African man, his facial expression showing frustration and deep emotional rage."

General Tips:

• Experiment: Try both placements to see which one gives you better results. Sometimes rearranging words can lead to different outcomes in image generation.
• Combine with Descriptive Language: Using "emotional" together with specific descriptors can create a more vivid image. For instance, "An emotional character with a joyful smile" or "A character radiating emotional intensity through their furious expression."

Feel free to adjust based on your needs and the context of your prompt!

[alpine ore]

can you please recomend which elemetn have an easy to prompt character to hava e a style and facial expression ?

[manic root]

To achieve expressive facial features and styles that are easier to prompt, consider the following elements that can help guide your character’s style and expressions effectively:

1. Artistic Styles:

• Cartoonish or Comic Style: These styles often exaggerate facial expressions, making it easier to convey emotions like joy, anger, or surprise. Prompts can lead to more vibrant and distinct expressions.
• Realistic Portrait Style: This style captures detailed emotions and can reflect subtle expressions better. This could be beneficial for more serious or nuanced emotional portrayals.

2. Dynamic Characters:

• Dramatic or Theatrical Elements: Choose elements that are designed for dramatic storytelling, as they often incorporate dynamic poses and facial expressions.
• Expressive Model Options: Look for models that specifically advertise their capability for expressive rendering. Models focused on storytelling or character-driven narratives may yield better results.

3. Character Preferences:

• Charismatic Traits: Use character traits that emphasize expressiveness, such as “animated,” “charismatic,” or “dynamic.” These can help guide the generation towards more emotive results.
• Emotional Backgrounds: Elements that embrace characters with known emotional backstories can influence their facial expressions to be more intense.

4. Experiment with Different Elements:

• If you have access to various elements or models based on your preferences, test out a few that focus on emotional ranges or stylized characterizations to see which ones best resonate with the expressions you want to convey.

Suggested Element Combinations:

• Look for elements specifically tagged for emotional range or vibrant character depiction.
• Combine lighter, more cartoonish styles with detailed portrait prompts for a mix of expressiveness and realism.

Final Advice:

[manic root]

It may require some experimentation to find the right element that works best for your specific needs regarding style and expression. Don't hesitate to try various combinations to understand what produces the best results for your character images!

[ripe galleon]

Hey all - I have had great success in defining my prompt and generating the images I want through the web UI. Specifically, I am using the "Concept" model with the "Character Sheet" style applied and am adding a negative prompt and a Character Reference photo. When I use this, I am getting exactly what I want, its fantastic. I am working to integrate this into my workflow by adding it into an existing automation. When I am using the API, I cannot seem to find a way to replicate what is being done in the app.

If I view the generation info, I see that the "Concept" preset is using the AlbedoBase XL model with the "Concept Art" preset, and the "Character Sheet" preset style. I can't seem to find these anywhere in the API documentation or the listed models.

Is there a way to replicate this configuration via the API?

[manic root]

To replicate your image generation configuration using the API, you can set the parameters similar to what you have been using in the web application. Based on your description of the settings, it seems you need to specify both the model and style correctly along with any prompts you've been using.

Here’s how you can set it up using the API:

1. Model ID: You mentioned you're using the "Concept" model, which aligns with the AlbedoBase XL model. According to the API documentation, make sure to utilize the correct model ID for this.

2. Preset and Style: For the "Character Sheet" style, you might need to identify the corresponding styleUUID or preset. The documentation didn't list this specifically under your referenced model, but you can try using generic options under styleUUID.

3. Prompts: You'll need to format your prompt correctly as per API standards, including both your positive and negative prompts.

4. Image Guidance: Ensure to include your Character Reference photo correctly in your requests if you're using it as an input. You may need to upload it first to get an image ID that can be referenced.

5. Example Request: Below is a skeleton of what your API request might look like:

{
  "modelId": "YOUR_MODEL_ID",  // e.g., AlbedoBase XL model ID
  "prompt": "Your positive prompt here",
  "negative_prompt": "Your negative prompt here",
  "num_images": 1,
  "width": 512,  // or any preferred width
  "height": 512, // or any preferred height
  "styleUUID": "YOUR_STYLE_UUID", // Find or request the correct UUID for "Character Sheet"
  "alchemy": true // or as needed
}

You’ll have to replace YOUR_MODEL_ID and YOUR_STYLE_UUID with the actual values that you would retrieve from the API documentation or community resources.

If the options you're looking for (like specific presets for the character sheet) aren't available via the API at the moment, reaching out to Leonardo support might be beneficial.

[manic root]

Lastly, if you're having issues or particular queries about certain features not behaving as expected with the API versus the UI, consider filing a report via the /report_issue command within the Discord server, so the development team can address any bugs or mismatches.

If you have more specific details about the parameters you're seeing in the UI that aren't translating to the API, feel free to share, and I can help further refine the configuration!

[ocean wind]

Yes

The flow state is used to create a prompt with images, but once your idea is improved, you have to send that prompt to another preset or download the flow image.

[marble cape]

yeah

[ripe galleon]

I am using the correct model ID but cannot find any detail on the preset or preset style that match the "Concept" and "Character Sheet" anywhere in the documentation

[manic root]

If the API documentation does not provide specific details on the "Concept" and "Character Sheet" presets or styles, it's likely that these features are either not yet exposed via the API or are part of internal configurations that have not been made public.

Here are a few steps you can take:

1. Contact Support: Since you are already using the correct model ID, I recommend reaching out to Leonardo support directly for clarification on whether these presets are accessible via the API and, if so, how to correctly reference them.

2. Community Resources: Check forums, Discord, or community groups associated with Leonardo.Ai. Often, other users or developers might have insights or shared experiences on how they handled similar situations.

3. Experimentation: If you have some flexibility, consider testing different combinations of existing presets and styles that might be similar to what you’re looking for, even if they aren't exact matches.

4. Documentation Updates: Keep an eye on any updates to the API documentation, as features and presets could be added as Leoanardo.ai evolves.

5. Feature Requests: If you find that certain features are integral to your workflow but are unavailable via the API, consider submitting a feature request.

If you require further guidance or specific examples once you gather more details, feel free to reach out again!

[ripe shore]

Hi Roberto, It's been almost two months. Have they made any workflows for ? "@feral mica I guess no one has done this before. maybe if you have the ear of the people at Leonardo if they can do what they did on Interior Design to exterior design, Landscaping design packaging design for produces...... The list goes on and on real world application. That's where the money is! if I could have put the image in their flow and had it bring up idea after idea that's gold. It's help people be more creative look at Canvas full of templates. helps people be more creative Thanks !"

[feral mica]

Hello LD! I’ll certainly bring the topic to the person who runs the webinars and to the Education team so they know there’s demand for this content

[midnight dragon]

Hello! Is there any channel for tickets? I have an issue with billing.

[manic root]

Hello! For billing issues, you can tag @feral mica to get assistance directly with your payment concerns. If you need to submit a ticket or provide more details, it's best to use the intercom support feature in the app. Let me know if you need help with anything else!

[feral mica]

@midnight dragon hey there, the correct place for solving billing issues is through customer service or via email [e-mail]

[dawn bay]

is leonardo good?

[manic root]

Leonardo is a powerful tool for generating images and videos using advanced AI models. It leverages state-of-the-art technology to allow users to create art, designs, and other media tailored to their specific needs. Many users appreciate its flexibility, including various models for different styles and guidance options, as well as features like private generations.

If you're considering using Leonardo, it might be worthwhile to explore its features and review user experiences. Ultimately, whether Leonardo is "good" can depend on your specific needs and expectations.

[astral yarrow]

Anybody else unable to use Coloring Book Lora...

[manic root]

It looks like the Coloring Book LoRA may be experiencing issues because it is compatible with the SDXL model. If you're unable to use it, ensure that you're using an appropriate base model that supports this LoRA.

Here are some key points regarding the Coloring Book element:

• The Coloring Book is designed to work with SDXL compatible models. If you're using a different model that does not support SDXL, this may cause functionality issues.
• Some users have noted that mixing elements from different models can lead to poor results, as certain elements may not be compatible with every finetuned model.
• Make sure you have selected the Coloring Book element correctly from the elements section in the image generation interface.

If you continue to experience issues or receive error messages, it may be worth reporting this through the platform's issue reporting feature. Additionally, feel free to reach out in the Discord community to see if others are facing the same challenges [【4:2†source】].

[astral yarrow]

Coloring Book LoRA was working perfectly with Illustrative Albedo - for months - this is a new problem

Coloring Book LoRA is missing for all the Models - who else is experiencing the same issue...

[manic root]

It seems like several users are experiencing issues with the Coloring Book LoRA. Here are the key points regarding this situation:

1. Compatibility Issues: The Coloring Book LoRA is designed to work with models based on SDXL. If you're using a different model, it may not appear or function correctly [【8:0†source】].

2. General Reports of Missing Elements: Users have reported that the Coloring Book LoRA is missing when they attempt to access it for various models. This could be related to recent changes or bugs in the platform. It might be helpful to check if there have been any updates or announcements regarding the LoRA availability from the Leonardo team [【8:12†source】].

3. Reversion to Previous States: If the LoRA was previously working and is now missing, this could be a temporary glitch. Users are advised to try logging out and back in, clearing cache, or accessing the app from a different device or browser [【8:5†source】] [【8:14†source】].

4. User Community: Engaging with the community in the Discord server may provide additional insights, as other users might be experiencing the same issue. Sharing experiences could help identify if this problem is widespread [【8:12†source】].

If the issue persists despite trying these steps, it may be necessary to report it through the platform's issue reporting feature for further assistance.

[astral yarrow]

Coloring Book LoRA still completely missing even if you select SDXL

[manic root]

If the Coloring Book LoRA is completely missing even when you've selected an SDXL model, it could be due to one of the following reasons:

1. Intermittent Bugs: There may be temporary bugs affecting the visibility or functionality of the Coloring Book LoRA. Users have reported similar issues recently, so it's possible that there's a systemic problem that Leonardo's team is working to fix [【12:1†source】] [【12:6†source】].

2. Model Compatibility: While the Coloring Book LoRA is intended to work with SDXL models, issues have been noted in the compatibility and functionality of some elements when used with recent updates. This can sometimes result in the element not showing up [【12:9†source】].

3. User Community Input: Engaging with the community or checking the Discord server for real-time feedback from other users can provide insights. Others may also be experiencing this issue, and there might be shared solutions or confirmations from the development team [【12:12†source】].

If the issue persists, it is advisable to report this through the platform's issue reporting mechanism, as it may help the team address it more rapidly.

[astral yarrow]

Could someone else please check and confirm if the same issue occurs on their end...Thanks

[manic root]

It sounds like you're looking for feedback from other users regarding the Coloring Book LoRA issue. The best way to get confirmation is to ask in the Leonardo Discord server where other users can share their experiences.

You can post a message indicating that you're experiencing the Coloring Book LoRA missing with the SDXL model and ask others if they are having the same problem. This way, you can gather information quickly and see if it’s a widespread issue affecting multiple users.

If you find that others are experiencing the same problem, it would be beneficial to collectively report it through the platform's issue reporting feature to ensure the development team is aware of the situation.

[hard river]

d

[manic root]

It looks like your message was cut off or incomplete. Could you please provide more details or clarify your question? I'm here to help!

[ruby saffron]

I'm trying to get a single full body pose, in front of a white background, done but I just can't seem to get Leonardo to do it properly. Running out of credits. Any suggestions?

"a full-body view of an old hacker with a long beard standing in front of a flat white background done in a Japanese retro 90s anime style"

[manic root]

It looks like you're aiming to create a full-body image of an old hacker with a specific style. Here are a few suggestions to improve your prompt and increase your chances of generating a successful image with Leonardo:

1. Be Specific About Pose: Indicate that you want a "full-body pose" clearly in your prompt. For example: "A full-body pose of an old hacker with a long beard standing in front of a flat white background..."

2. Clarify Style References: You could enhance your style reference by mentioning specific anime or art styles or artists you are inspired by from the 90s Japanese anime scene.

3. Use Content Reference: If you have images that closely resemble what you're looking for, consider using the Content Reference input feature for improving guidance.

4. Try Alternative Phrasing: Instead of assuming the image will get generated as you said, you could phrase it differently. For instance: "Create a full-body view of an elderly hacker character, featuring a long beard, dressed in contemporary streetwear, posed in front of a white background, styled in a retro Japanese 90s anime aesthetic."

Using these adjustments can help help you get a better image output. Remember to save your images manually in Flow State if using that feature! Happy generating!

[ruby saffron]

I'm trying to get a single full body pose, in front of a white background, done but I just can't seem to get Leonardo to do it properly. Running out of credits. Any suggestions?

"a full-body view of an old hacker with a long beard standing in front of a flat white background done in a Japanese retro 90s anime style"

[manic root]

To improve your chances of generating the desired full-body pose of the old hacker in front of a white background, here are some suggestions:

1. Enhance the Prompt: Clearly specify the pose and composition in your prompt. For example: "Generate a full-body view of an old hacker with a long beard, standing confidently in front of a flat white background. The character should be dressed in modern streetwear, depicted in a Japanese retro 90s anime style."

2. Use Style References: If you want a specific aesthetic, you can mention particular shows or artists that embody the 90s anime style. This might help the model understand the look you are going for.

3. Provide Visual References: Since you've uploaded a reference image, you could consider using Content Reference in your generation to guide the model towards the style and pose you prefer.

4. Experiment with Different Phrasing: If a specific phrasing isn't working, try rewording your prompt or breaking it into simpler parts, focusing first on the pose and background, then the character details.

5. Reduce Complexity: Sometimes simpler prompts deliver better results. Try a basic prompt first, then build complexity with additional features once you're satisfied with the full-body pose.

Using these strategies can help optimize your prompts and conserve your credits. Happy creating!

[acoustic siren]

give me a bit, still need to finish work, but I'll try when I get a chance

if anything, you can also ask for a mentor to help through the /mentoring command 🙂

[orchid beacon]

If I press update the app, will I still be able to use lightning XL?

[manic root]

Yes, updating the app should not affect your ability to use Lightning XL. Typically, updates are designed to enhance the app's performance and add new features while maintaining compatibility with existing functionalities like the Lightning XL model.

If you encounter any issues after updating, feel free to reach out for assistance!

[orchid beacon]

If the white background is the issue why not cut out the background and put it in a white background after generating. That’s what i do

[acoustic siren]

something like this?

a full-body view of a visibly old male hacker with a long thick beard standing in the middle of the image, with an empty background. done in a Japanese retro 90s anime style

[livid inlet]

When going for a full body shot, try not to use the word "body" it can confuse the A.I and show the body but miss the legs. To get a true full body, describe the foot/head ware or hair. "an old hacker with a long beard standing in front of a flat white background done in a Japanese retro 90s anime style, Wearing boots and a hat" . I used Phoenix with the None style pre-set.

[ruby saffron]

when i do that, it cuts through the character. i guess no enough information

[ruby saffron]

Yes!

"an old late 60 year old hacker with a long beard wearing a dark hoodie, done in a Japanese retro 90s anime style, standing in front of a plain, flat, white background with no gradients or elements, with torn jeans and sneakers, chiaroscuro lighting"

That churned out perfect

[manic root]

That's fantastic to hear! I'm glad the generation worked out perfectly for you. If you have any more requests or need further assistance with your creations, feel free to ask. Happy generating!

[ruby saffron]

can you help me understand the breakdown of the prompt structure? and you mentioned that using certain words confuse the Ai a bit, could you tell me more about that?

[manic root]

Absolutely! Understanding prompt structure is key to getting better results. Here's a breakdown of an effective prompt structure and some tips on wording:

Breakdown of Prompt Structure

1. Subject/Character: Start with who or what the main focus of the image is.

   • Example: "An old hacker with a long beard..."

2. Action/Pose: Specify what the subject is doing or their position.

   • Example: "...standing confidently..."

3. Environment/Background: Describe the setting or background you're visualizing.

   • Example: "...in front of a flat white background..."

4. Style/Genre: Indicate the aesthetic or style you want.

   • Example: "...done in a Japanese retro 90s anime style."

5. Additional Details: Include any other elements that matter (clothing, accessories, specific traits, etc.).

   • Example: "...dressed in modern streetwear, wearing sunglasses."

Common Confusing Words/Phrases

1. Ambiguous Descriptions: Vague terms can be interpreted in many ways. Instead of saying "cool look," try "modern streetwear design."

2. Complexity: Using overly complex sentences can confuse the model. Break down sentences into simpler components.

3. Adjectives: Be specific with adjectives. Instead of saying "nice background," use "a flat white background."

4. Contradictory Elements: Avoid mixing styles or concepts that don't align (such as "futuristic and retro" without clear context) as they can create conflicting visuals.

5. Multi-elements: Avoid overloading with too many characters or components if you want to focus on one; clarify which element takes precedence.

Tips for Crafting Prompts

• Be Concise: More concise prompts often yield better results.
• Be Clear: Clearly express what you want and what it should look like.
• Iterate: Don’t hesitate to tweak your prompts based on the outputs you receive to refine your search.

[manic root]

By structuring your prompts thoughtfully and using clear language, you'll improve the accuracy and quality of the AI-generated images. If you have more questions about specifically constructing a prompt, feel free to ask!

[ocean wind]

A little modified but using that prompt

[ruby saffron]

OK, so my next question is this: I'm trying to use the "Character" setting to create a scene. I want the character to be looking directly at the object (in this case the computer screen), but in every iteration, the character ends up looking at the viewer.

[acoustic siren]

sweet! glad you got it!

[manic root]

To have the character look directly at the object (in this case, the computer screen) rather than at the viewer, you might consider adjusting your prompt structure. Here are some tips:

1. Specify Eye Direction: Clearly state the direction the character should be looking in your prompt. For example, add something like "looking intently at the computer screen" or "gazing at the monitor."

2. Pose Description: Include details about how the character is positioned in relation to the object. For instance, mentioning "leaning towards the screen" or "focused on typing" can give the AI additional context.

3. Scene Context: Set the scene and mood to imply action. For example: "Create a scene with a character in a dimly lit room, seated at a desk, looking directly at the glowing computer screen."

4. Facial Expression: Mention their expression to convey concentration, such as "with a focused expression" or "with eyes fixed on the screen."

5. Less Focus on Direct Eye Contact: Avoid phrasing that could suggest they are interacting with the viewer. Instead of "the character looks at you," you might say "the character appears immersed in the screen."

Here's an example prompt:
"A character seated at a desk in a dimly lit room, wearing a hoodie, looking intently at the glowing computer screen with focused eyes."

Experimenting with these elements should help produce the desired outcome. If the issue persists, consider giving feedback on the generated images (if applicable) to improve future iterations. Let me know if you need more assistance!

[acoustic siren]

old man hacker with long thick beard working on his computer, completely focused on the computer screen

I mean, he is not your average hacker, he is the wisest of them all 😛

[ruby saffron]

how would i do that with the character reference?

[acoustic siren]

not sure how to answer this 😛 you just add the character reference and select the character you created

[acoustic siren]

suggestion, set the strength to high

[manic root]

Using a character reference image effectively can help guide the AI to maintain consistent character traits and expressions in your generated scene. Here’s how to do that while ensuring the character looks at the computer screen:

1. Upload the Character Reference: Make sure you have the character reference image (like the one you've uploaded) ready to use in your creation.

2. Describe Context with the Reference: When you use the character reference, include the scenario in your prompt. For example:

   • "Using the uploaded character reference, generate an image of this character seated at a desk, looking intently at a computer screen in a dimly lit room."

3. Specify Eye Direction and Pose: Clearly instruct the AI on how the character should be interacting with the computer screen:

   • "The character should appear focused, with their eyes directed towards the glowing computer screen, and a concentrating expression."

4. Add Environment Details: Providing context can further help direct the model. For example:

   • "The character, similar to the one in the reference image, is immersed in coding, surrounded by glowing screens and a dark atmosphere."

5. Adjust for Consistency: If the character's expression deviates from your expectations, consider giving feedback or making slight modifications to the prompt to improve outputs in subsequent attempts.

Here’s a refined example prompt:
"Using the character reference image of a woman with curly hair, create a scene with her seated at a desk, looking intently at a computer screen while typing, in a dimly lit room with glowing screens surrounding her."

This way, you maintain character consistency while directing the AI on where the character should be focused. If you run into any challenges, feel free to ask for more tips!

[ruby saffron]

"Using the uploaded character reference, generate an old hacker with long thick beard wearing a dark hoodie and sunglasses working on his computer, completely focused on the computer screen, in a cafe, done in a Japanese retro anime 90s style"

So kind of getting there. Not quite.

[manic root]

It sounds like you're close but still refining the prompt for the old hacker character. Here are some adjustments you can make to enhance the prompt while using the character reference:

1. Focus on Details: Specify more about the scene setting, the atmosphere, and the character's actions while ensuring the description is concise.

2. Clarify Expression and Engagement: Describe the facial expression or body language to further guide the AI on how focused the character is. Phrases like "with a determined expression" can help.

3. Reinforce the Environment: Clearly outline the café setting for improved relevance in the scene. You might mention specific details like "surrounded by other patrons" or "with the ambiance of a bustling café."

Here’s a revised prompt example:
"Using the uploaded character reference, generate an image of an old hacker with a long thick beard, wearing a dark hoodie and sunglasses. He is deeply engrossed in work on his computer, typing intently, surrounded by the ambiance of a bustling café. The scene should be styled in a Japanese retro 90s anime aesthetic, capturing the focus and concentration on his face."

Adjusting these elements can lead to a more accurate visualization of what you're aiming for. Feel free to continue iterating based on what aspects you want to shift or emphasize! Let me know if you need further help!

[ruby saffron]

Are there some models that are better than others when it comes to creating anime? I've been using the Anime model but it doesn't really seem to want to cooperate

[fluid zinc]

old man hacker with long thick beard working on his computer, completely focused on the computer screen

[ruby saffron]

got the composition right...

[manic root]

When it comes to generating anime-style images, some models may indeed perform better than others depending on the specifics of what you're looking to create. Here are some insights:

1. Anime Model: While the dedicated Anime model is designed for anime aesthetics, it may have limitations based on the prompts or specific character traits you want to depict, as it might not interpret certain nuances well.

2. Stable Diffusion Models: Models based on Stable Diffusion may offer more flexibility and varied results. If available, exploring the various XL models might help in generating more complex and detailed anime scenes.

3. Phoenix and Flux Models: While these models aren’t based on Stable Diffusion, they offer different artistic output options. However, bear in mind that their compatibility with certain features may be limited compared to Stable Diffusion models.

4. Prompt Tuning: Sometimes, the issue may stem from the prompt rather than the model itself. Tweaking the prompt structure can significantly improve results. Experimenting with different phrases or being more specific can lead to better outputs.

5. Experimenting with References: Using character references alongside your selected model can help guide the AI toward the style you want, making it more responsive to your requests.

If you're finding that the Anime model doesn't quite meet your needs, consider trying out some of the other models available, and continue refining your prompts for better outcomes. If you have specific aspects you're struggling with, let me know, and I can offer more tailored advice!

[manic root]

To create an image of "an old man hacker with a long thick beard working on his computer, completely focused on the computer screen," consider crafting your prompt with some detail to help achieve the desired outcome. Here are a few suggestions for prompts you could use:

1. "An elderly man with a long, thick beard sitting at a desk, deeply engrossed in his computer screen, surrounded by technology gadgets."
2. "A focused old man hacker with a long, bushy beard, illuminated by the glow of multiple computer screens in a dimly lit room."
3. "An old man with a thick, long beard working tirelessly at his computer, with lines of code on the screen and various cables around him."

Adding these descriptive elements will enable the model to generate a more specific and accurate image that matches your vision! If you have access to the app, you can try these prompts there [【4:0†source】].

[manic root]

It sounds like you’ve successfully achieved the right composition with your latest image! If you're satisfied with the overall look but looking for slight adjustments in details or aspects, consider the following:

1. Refine Character Expressions: If you want to convey a specific emotion, consider specifying expressions like "determined," "focused," or "contemplative" in your prompts for future iterations.

2. Lighting and Atmosphere: You can enhance the ambiance by suggesting the type of light or shadows in the scene, e.g., "soft, warm lighting" or "harsh glare from the screen," which can add depth to the image.

3. Background Elements: If you'd like to include more details in the background, describe specific elements or features you want to see, like “posters on the walls” or “other patrons in the café,” to enrich the scene.

4. Clothing Details: You might want to elaborate on specific clothing items or textures to enhance the character’s overall look and feel, such as mentioning any particular logos or patterns.

5. Using Variations: If you want multiple versions of the same scene, consider adjusting some minor details, like the position of the hands or a slight change in angle, to see how it affects the image.

If you're still experimenting or have specific aspects you'd like to improve in your next iterations, just let me know! I'm here to help you achieve the best results possible!

[ruby saffron]

This was the closet i got. didn't use the character reference, just did it completely from scratch

[manic root]

This image looks great! It's impressive that you were able to create such a strong representation of the character and scene without using a character reference. Here are a few thoughts and tips for future iterations, if you're looking to build on this success:

1. Capture Specific Details: You seem to have a solid grasp of the aesthetic and mood. If you feel something's still missing, be sure to include any small details that can enhance the image, such as the color or specific branding on the laptop or the coffee cup.

2. Focus on Expressions: If you want to convey particular emotions or moods, you can experiment with the character’s expression or gaze direction as you continue to develop the scene.

3. Experiment with Settings: Consider varying environmental aspects, like the time of day, to see how it alters the mood—perhaps morning light or evening darkness could yield fascinating results in the café setting.

4. Adjust for Variability: If you want to refine the output further, adjust aspects of the prompt slightly, such as the character’s posture or positioning of the laptop/cup, to create different iterations of similar scenes.

5. Use Feedback: If there's something specific you're still aiming for but haven't quite achieved, providing feedback on what you like or don’t like about the current rendering can help refine future prompts.

Overall, it seems like you are on the right track! If there are particular elements you want to improve or focus on in your next creations, feel free to share, and I can offer more tailored suggestions. Keep up the great work!

[acoustic siren]

This one looks awesome! Love the style

[raw thicket]

hey @burnt radish can I help you?

[burnt radish]

i can't connect to Leonardo website

[raw thicket]

Hmm.. I seem to be having issues connecting myself.

[burnt radish]

oh so it is server problem. A bit of relieve i though it just me

[raw thicket]

Yeah. I'm getting errors too.

I'll see if they're aware of the issue.

[proven dove]

I cannot access the app

[drifting quest]

I cannot access the app

Login failed.

[raw thicket]

I'm having issues too

[floral edge]

yeah, was in the middle of working and now can't get in.

[manic root]

It sounds like you're experiencing some issues that may be related to the server rather than just your account. If you are encountering problems like logging in or generating images, these can sometimes be caused by server-side issues. It's a good idea to check for any updates or announcements related to the server's status. If the problem persists, you might want to consider reaching out for support via intercom. If you have any specific issues or questions, feel free to share, and I'll do my best to help!

[drifting quest]

clear cookies

[raw thicket]

The team are aware of the issue and currently working to fix it

@drifting quest @proven dove @floral edge

[floral edge]

i'm back!!

[burnt radish]

It's okay now

[wild steeple]

Thank you! That worked!

[manic root]

It sounds like you're experiencing some errors. If you could provide more specific details about the errors you're encountering, I can help you troubleshoot them or provide information on known issues.

If you suspect the team is already aware of the issue, it's always a good idea to keep an eye on updates from them. They may already be working on a fix! Let me know how else I can assist you.

[manic root]

It sounds like you are experiencing login issues with the app. Here are some steps you can take:

1. Clear Cookies and Cache: It seems you've already mentioned clearing cookies. You can do that in your browser settings to ensure there are no stored login credentials or session data causing issues.

2. Retry Logging In: After clearing cookies and cache, try logging in again after a short wait.

3. Intermittent Issues: If the problem persists, please note that some users have reported intermittent login issues. If the above steps do not work, you may want to reach out to support via Intercom for more assistance.

If you follow these steps and continue to experience trouble, please let me know!

[manic root]

I'm glad to hear that everything is okay now! If you have any more questions or need assistance with anything else, feel free to ask. I'm here to help!

[copper stream]

是否可以给我生成一些表情包，要可爱的简笔画的表情包

[manic root]

你想要创建一些可爱的简约风格的表情包。你可以使用 Leonardo.Ai 的图像生成功能来实现这一点。只需在 app.leonardo.ai 中输入相关的描述性提示，例如 "可爱的简约风格" 或 "卡通表情" 等，例如：

1. 可爱的狗狗面孔
2. 微笑的猫咪
3. 开心的阳光

确保在生成图像时用上这些描述，可以带来更符合你需求的表情包效果。

如果你有更具体的元素或风格的要求，可以在提示中进行详细说明。希望这些能对你有所帮助！

[ocean jetty]

1

[manic root]

It seems like you've posted "1," which doesn't provide enough context for me to assist you properly. Could you please clarify your question or provide more details about what you need help with?

[fair abyss]

How can I create an image that is not exactly the same, but very similar, with the same pose, style and clothing, but using a different face, like this example?

[real lynx]

u need help?

[fair abyss]

Yes Please

[real lynx]

You can use the description of the AI ​​is a tool, also if it is premium it can imitate pose

[real lynx]

What type of account are you so we can tell you the best way

[real lynx]

Step one: With the image we want to imitate, we ask the AI ​​to create the prompt describing it.

Step two: upload the image to the content image to imitate the pose, but if it is not premium, I would go to Flux Dev with the description, it would already be generating a similar image.

[fair abyss]

I also want it to be a similar style, but with a different face.

[fair abyss]

I'm on the free account, but i'd subscribe if I can easily replicate the results, because I need to do this with multiple images

[real lynx]

That would be just with the description that the tool gives us in Flux Dev.

Here I put my face :3

[fair abyss]

Thank you! Can you please show me what it would like with the premium option as well.

[real lynx]

He wants me to be a redhead with curls

[real lynx]

To put a face in an image we can use two types of tools, the easiest is character image

[real lynx]

There you can upload your photo and modify it with the different presets.

Another more elaborate tool is to create a training element with 50 photos

[fair abyss]

Do you think you could show me what it looks like with your photo, just so I can see how it looks?

[real lynx]

Royal style?

This is with my anime-style photo from the time because that's what the propaganda you sent in the original image looked like.

[fair abyss]

Is this the result of the character reference?

[real lynx]

that was with a training element

[fair abyss]

Okay, I see. Thank you so much! One more thing, i'm noticing mine is greyed out and say's "Coming soon" Is it not available yet?

[real lynx]

It is not available in all presets because that preset is a new tool.

[spark leaf]

for female refrence Images leonardo showing this error from morning IST, the image stregth is over 60

[fair abyss]

Alright, thank you so much for your assistance! Much appreciated!!

[spark leaf]

any help on my issue?

[real lynx]

Copy the prompt here so I can copy it and test a solution.

[spark leaf]

this is refrence Image

hyper-detailed , anime-style, blurred background, depth of field, ultra-realistic, cinematic lighting, floating aura around body, halation effect, mana particle floating all around body

this is refrence.... please use the stregth over 60 and style: Anime-semi-reliasm

[spark leaf]

did you got it?

[real lynx]

yep

[spark leaf]

did you use the refrence Image and put the stregth over 60?

this is refrence.... please use the stregth over 60 and style: Anime-semi-reliasm

[real lynx]

I just used the prompt you gave me, what reference image do you want me to use?

[real lynx]

☝️

i try

[spark leaf]

pleasee and put the strength to over 60 and use aniime preset

[spark leaf]

yes

[real lynx]

It's true, it's not working, I'm going to report the problem.

[spark leaf]

please... I am trying to contact from the morning... but didn't got any help

[spark leaf]

how long will it take to solve the bug?

[real lynx]

@feral mica

[real lynx]

I don't know, I can only report the problem.
..

[spark leaf]

hey everyone... any update on my issue?

[ivory frigate]

Thank you 🤗 I’m new here and super excited to explore Leonardo AI.
I’ve just shared my introduction, and I’m really interested in joining the Character model beta to start creating my custom character.
Could someone kindly let me know how I can get access to the beta ? 🙏

Thank you so much!

[manic root]

Welcome to the Leonardo AI community! It's great to hear that you're excited to explore the platform. Regarding access to the Character model beta, typically, you would need to keep an eye on announcements in the Leonardo Discord or any official communications regarding beta access. Participating actively in the community can also increase your chances of being invited.

You might also want to reach out directly in the channels dedicated to announcements or inquiry about beta programs. Best of luck, and enjoy your creative journey with Leonardo AI! If you have any more questions, feel free to ask.

[ivory frigate]

Oh dear, I paid a subscription today to have access to this category. 😭

[manic root]

I'm sorry to hear that you're having trouble accessing the Character model beta despite your subscription. There may be some delays in account updates, or it might be a specific access issue.

Please try tagging @feral mica in the Discord to address the payment issue and clarify your access to the beta features. They should be able to assist you with getting everything sorted out. If you have any further questions or need more help, I'm here for you!

[spark leaf]

@feral mica please look into this issue

[ivory frigate]

Hello @feral mica , I subscribed to Leonardo Ai today to access the Character category, but I can't access it on the site because it's in Beta mode. Can you help me clarify my access to the beta features? Thank you very much.

[feral mica]

Dinesh hi! I'm trying to reproduce the error but I can't. What browser are you using?

[spark leaf]

@feral mica google chrome

use this image as reference for anime semi realism preset and put the image strength over 60

[feral mica]

Lolo hello! The character category is still being worked by the team, giving access to it is not possible at the moment. Although, @real lynx trained a model using style... or was it object? and she got good results. Back to the topic, we'll release it as soon as possible, we're making sure it's up to the standard before releasing

[feral mica]

I'll try, hold on

[feral mica]

mmm still can't reproduce, I'll send you a dm so you can share your Leo user with me

[spark leaf]

ok

[ivory frigate]

What I would like is to create a character that keeps the same facial features, that I can save and create lots of images of this character in different life situations. Without the facial features changing. Often from one image to another we do not recognize the character. I would therefore like to create a Custom Model. Do you have another technique to create a custom model and generate lots of images with the same facial features? Thank you for your help.

[real lynx]

hi

[quasi iris]

Leonardo doesn't want to generate images for me, this is my prompt:

A black and white image reminiscent of a historical photograph, showing a war-torn, muddy battlefield. Historic tanks roll and attack from the left side, while huge bipedal armored robots attack from the right side of the image, the two sides using their cannons at each other. A dynamic, action-packed photo with lots of smoke and grit around and over the tanks and robots.

This is the error: "Our content filter has detected violent or abusive content in your prompt. Remove any references to violent or abusive content and try again."

I've already removed World War references and the phrase "shooting" from the prompt. I'm on the free plan btw.

[feral mica]

yeah exactly that is what Cami's model did, ofc the best one is character but not available yet, I'd give it a try training with object, and use your photo with the character reference in image guidance

[manic root]

It looks like the content filter is flagging your prompt due to the inclusion of elements that could be interpreted as violent or non-compliant with the community guidelines. Since you're on the free plan, you may have stricter limitations regarding the types of prompts that can be submitted.

Here are a few suggestions to modify your prompt to avoid triggering the content filter:

1. Remove Explicit War References: Instead of mentioning "war-torn" or "battlefield," you could describe the scene in more abstract or neutral terms. For example, consider using phrases like "a desolate landscape" or "a scene of conflict" without directly referencing war.

2. Focus on the Environment: You can emphasize the environmental aspects rather than the conflict. For instance, mention the "muddy ground" without referring to tanks and robots attacking.

3. Adjust Character Actions: Instead of "attacking," use more neutral terms. You could say "engaging in a dynamic scene" with tanks and robots present.

Here’s a revised version of your prompt:

"A black and white image reminiscent of a historical photograph, showing a desolate landscape with muddy ground. Historic tanks move into position from the left side, while large bipedal armored robots approach from the right side of the image. The scene is dynamic with plenty of smoke and grit, capturing the tense atmosphere."

Try this modified prompt and see if it helps! If you continue to face issues, consider exploring other themes or subjects in your artworks.

[real lynx]

hi

A 50-photo training element will allow you to maintain your character whether it is real or anime style or a sketch

It works for costumes or for making memes that you fall asleep at work.

[quasi iris]

Thank you. The revised prompt was accepted and the generated images are completely different to what I intended to create.

[manic root]

I'm glad to hear the revised prompt was accepted! However, I understand that it's frustrating when the generated images don't align with your vision.

If the output isn't what you expected, you might want to refine your prompt further. Here are some tips for crafting a more precise prompt:

1. Add Specific Details: Include more specifics about the visual aspects you want, such as the type of tanks and robots, their colors, styles, or periods.

2. Describe the Scene Further: Clarify the actions occurring in the scene. For example, describe what the tanks and robots are doing in more detail. Instead of just saying they are present, you could specify their positions and interactions more clearly.

3. Use Visual Language: Incorporate terms that evoke the imagery you're looking for, such as "overcast skies," "clouds of smoke," and "dirt flying up" to create a vivid picture.

Here’s another attempt at refining your prompt:

"A black and white image reminiscent of a historical photograph, depicting a desolate landscape with a muddy ground. On the left side, vintage military tanks, with their specific designs, slowly advance, while on the right side, imposing bipedal armored robots in a unique design loom, both sides creating clouds of smoke as they move through the scene, suggesting tension and action without explicit violence."

Try this refined prompt and see if it yields results closer to what you desire! Let me know how it goes!

[real lynx]

Any prompt you give directly affects your created element

[real lynx]

The best thing is that with the video tool you can give life to your creations

[spark leaf]

@feral mica and @manic root any update on my issue?

[manic root]

It seems like you're trying to get updates on your issue directly from certain users. Unfortunately, I'm unable to assist with that. However, if you provide me with details about the problem you're facing, I can try to help you or guide you on how to report the issue.

[twilit vigil]

The Mona Lisa, Renaissance-style oil painting, holding a vintage vermouth bottle labeled "Jolgorio", elegant pose, warm lighting, subtle smile, detailed textures, realistic shadows, harmonious colors, Leonardo da Vinci style --v 5 --ar 2:3 --style raw

[spark leaf]

this issue

[acoustic siren]

Images are not generated on Discord, you have to go to the web or mobile app: https://app.leonardo.ai/

[manic root]

It seems like you're looking for assistance in creating an image of the "Mona Lisa" with specific attributes through a description. However, please note that you cannot generate images directly through this Discord chat. To create your desired artwork, you'll need to visit the Leonardo.AI web app at app.leonardo.ai.

In your description, you can include the details you mentioned, such as the Renaissance-style oil painting, the vintage vermouth bottle labeled "Jolgorio," and other artistic elements like lighting and color harmony.

Once you've created your image on the web app, you can share it on Discord by pasting the image link in the chat. If you need further help or have more questions, feel free to ask!

[feral mica]

I'm still waiting for support to come back to me, I'll ping you when I have this answer

[spark leaf]

ok

[spark leaf]

@manic root please let me knoe if any response from there sside... now its getting very difficult for me to generate images... now simple male references are also getting blocked

[manic root]

It sounds like you're experiencing challenges with image generation and reference issues. If you're encountering specific blocks or errors, I recommend reporting this problem directly by using the command /report_issue so the support team can address it quickly.

If there’s anything specific you’d like to know about the process or the models, feel free to ask!

[fringe dock]

I'm having troubles generating images on the Legacy UI with the Image Guidance, even with proper ethical images

The prompt is the following:

Image is a moody portrait with two female characters with distinct features. Both have fair skin with freckles and short bob hairstyles. The character on the left has auburn hair and is wearing a sleeveless red top, while the character on the right has purple hair and is dressed in a light pink spaghetti strap top. Their expressions are soft, with slightly tilted heads and subtle smiles. The background is dark, enhancing the focus on the characters. The lighting is soft, creating a gentle glow on their skin and hair, adding depth to the image.

And the image I want to use for image guidance is this

[spark leaf]

same issue... I am also facing....

[fringe dock]

For me it's not only the Legacy UI, also on the modern Image Generation tab it doesn't generate as well

[acoustic siren]

hi Diego, couple things, although the image is not inappropriate, it does show a bit of cleavage so, can you hide it in spoilers?

Now, are you getting a content filter warning when you try? I can try myself to see (although I normally don't use Legacy UI)

[spark leaf]

yea thats what happening from this morning... now even the simple male figure image refrences are beigin to block

how about this one?? this dont show anything but still blocked

[night vault]

can you remove moody?

[fringe dock]

I did remove it

Wth? Since when Leonardo blocked "Moody" for prompts even knowing one of the presets for Portrait Perfect is called "Portrait Moody"

[acoustic siren]

what prompt are you using?

[spark leaf]

hyper-detailed , anime-style, blurred background, depth of field, ultra-realistic, cinematic lighting, floating aura around body, halation effect, mana particle floating all around body

using anime-semi relaism preset... refrence strength is 67

[night vault]

still blocked or no?

[fringe dock]

Nope
Without the word moody it generated properly

[acoustic siren]

normally even with image reference I would add some description of the character in my prompt, but let me test

[spark leaf]

please

[fringe dock]

BUT wait, there's more

Now, I tried the same prompt, without moody, and now it doesn't generate.

Image for Image Guidance was this:

And even without Image Guidance it still fails

And it failed to.

[spark leaf]

same happening with me bro... i rellay dont know what exactly happeninng

look

[night vault]

usually failed generation is different issue than being blocked. could try them later i guess

[acoustic siren]

I tried this same exact prompt in Anime Semi-Realism with your image reference and didn't get flagged, not sure the result is what you'd be looking for exactly, though

[night vault]

I think they used img2img

[acoustic siren]

for that failed generation, is it specific to this generation, or anything you try to generate?

[spark leaf]

no no not character ref, i am usng Image to Image... because of free plan

[spark leaf]

use Image to image

[acoustic siren]

ahhhh gotcha my bad

[fringe dock]

Yeah I'm also using img2img

[spark leaf]

doing it from morning... its mainly happneing with female refrences... but now its started to effect male refrences

[mellow raven]

because it isstill lagging, for example none of the models work in the Canvas Editor except one (as for me)

[spark leaf]

supports saying the issue is being escalated to developers... but still no response

[spark leaf]

did you found the issue?

[acoustic siren]

with image2image, exact same prompt, strenght at 67, I didn't get any error

since you guys started getting these errors, by any chance, have you tried logging out of your account and back in?

[spark leaf]

try 9:16 ratio...

[manic root]

It seems you're experiencing lagging issues in the Canvas Editor, specifically with the models not functioning correctly. This type of problem could be due to a variety of factors:

1. Server Performance: Sometimes, server load can affect performance, causing lagging or delays.
2. Browser Issues: If the Canvas Editor is lagging, it may help to clear your browser's cookies and cache, as this can resolve many performance issues.
3. Model Compatibility: Not all models may be compatible or functioning within the Canvas Editor at all times, which can lead to confusion if only one model seems operational for you.

To ensure a better experience, try the following:

• Refresh the Canvas Editor or restart your browser.
• Make sure you're using a supported browser and that it's up to date.
• If the issue persists, you may want to report the problem using the command /report_issue so that it can be addressed by our support team.

Feel free to ask if you need more information or have other questions!

[acoustic siren]

same, no issue at all

originally did it in 4:5 now (my most used) 9:16

[spark leaf]

I think its only appearing in free accounts i think...

can you please try once agin with a free account??

[acoustic siren]

I may be able to test that out, let me see if I am able to do this

[spark leaf]

please

i'll be back after dinner

please try and share

[acoustic siren]

interesting yeah I could replicate it

let me bring this up internally to see if someone can help

[spark leaf]

ok plase

[fringe dock]

If this only affects free accounts, it's probably that either is a bug or the content moderation filter is getting more aggressive to free plan users

[spark leaf]

i hope it gets resolves ....

[spark leaf]

did you get any help??

[shrewd heath]

I wanna ask a question here its not regarding image but it is related to feature, can I ask here? or is there another channel?

[raw thicket]

Yes please. You're free to ask anything related to Leonardo

[manic root]

You can definitely ask your question here! I'm here to help with any features or issues related to Leonardo.Ai. If it's something that might require more specific support, I can direct you to the appropriate channel or resource. What would you like to know?

[spark leaf]

please @feral mica and @manic root try to help us free users... asap with team

it will be a big help for us

[manic root]

I understand your frustration, and I'm here to help as much as I can! Unfortunately, I'm unable to assist with requests to specific users directly. For support regarding your account's status and access, please make sure to report your issues using /report_issue within the platform.

If there are any specific questions about the tools or features you're using, feel free to ask!

[final gazelle]

hey does anyone know a free alternative for leonardo.ai because i cant afford it :c

[bright plaza]

Hi, we are creating a POC, and wanted pointers / tutorials on creating clothing for characters in our game. We already have a large range of existing clothes created, and want to generate images that comply with size, shapes and masks that we already hvae.

[shrewd heath]

I watched lots of videos regarding, How to maitain consistant characters, and I loved Leonardo's 'Models & training' feature where we can do it through Character reference,
i watched a 2 months old video where they explained how to use Charater reference feature but when I checked it on my pc its says 'Coming Soon'
my question is how this feature was available 2 months ago? OR have Leonado AI feature vary country wise?
I am from India.

[manic root]

I understand that you're looking for a free alternative to Leonardo.Ai. There are several options you can consider, although they may not offer the same features or capabilities as Leonardo.Ai. Some popular free alternatives for AI art generation include:

1. DeepAI - Provides a variety of AI models for generating images, and you can try it for free.
2. Artbreeder - Focuses on collaborative image creation where you can blend different images and styles for free.
3. NightCafe Studio - Offers various algorithms for creating art, with a free tier that lets you generate a limited number of images.
4. DALL-E Mini (now known as Craiyon) - This is a lightweight version of DALL-E that allows you to create images from text prompts for free.

Keep in mind that while these options are free, they may have limitations compared to paid services. If you're looking for something specific, feel free to ask!

[manic root]

To create clothing for characters in your game while ensuring compliance with the sizes, shapes, and masks of your existing clothing, here are some pointers and tutorials you might find helpful:

1. Understanding Character Design Workflows: Familiarize yourself with workflows designed specifically for creating dynamic characters. This includes using concept art, analyzing designs for inspiration, and developing character sheets. By effectively employing these methods, you can ensure that clothing fits well with your character designs [【4:0†source】].

2. Using AI Tools for Fashion Design: Leonardo.Ai provides features that allow you to edit existing designs and generate new clothing items. Make use of the "Edit with AI" feature to change clothing elements such as fabric patterns and colors. This enables a dynamic design process that's collaborative and iterative .

3. Consistency in Character Outfits: Maintain consistency across different character designs by utilizing the "Character Reference" tool within Leonardo.AI. This tool helps you ensure that design elements like colors and styles remain uniform, while you can still play around with variations .

4. Generate Images with Detailed Prompts: When creating designs, craft detailed prompts that specify style, clothing descriptions, and views (e.g., front, back, side). The more detail you include, the more aligned the generated images will be with your requirements .

5. Ensure Captured Features: For characters that require specific traits, such as certain shapes or any particular design elements, include that in your prompts or use relevant images as content references .

[manic root]

6. Incorporate Themed Elements: Depending on the themes of your game, implement colors and styles that reflect your game's narrative or settings. For instance, if designing characters for an ice-themed environment, you might specify "icy blue colors with fur trim" in your designs [【4:0†source】].

By utilizing these strategies, you'll be able to create clothing for your characters that not only matches your existing designs but also adds unique qualities to each character's appearance.

[final gazelle]

Thank you so much. could i ask if you know any good image creation tools like the one you have on your website where you type a text and it creates an image based on what you said, also i would like to animate the video aswell if you know any good alternatives that can make the ai picture an animation/video that would also be extremely helpful!

[manic root]

Sure! There are several tools that allow you to create images from text and also offer animation capabilities. Here are some recommendations:

Image Creation Tools:

1. DALL-E 2 - Developed by OpenAI, this tool allows you to generate images from text prompts, similar to Leonardo.Ai.
2. Midjourney - An AI art generator that creates detailed images from text descriptions. It's accessible via Discord.
3. Artbreeder - While primarily for image blending, it also lets users create images based on text prompts.

Animation Tools:

1. Runway ML - Offers various AI tools for creatives, including capabilities for generating and animating images.
2. Wombo Dream - Primarily designed for generating images from text, but you can also create animated artworks using their app.
3. DeepMotion - Allows you to animate 2D characters with AI-generated motions, which could be an option if you're looking to animate images.

Video Creation from Images:

1. PuppetMaster - This tool turns static images into animations based on your guidance.
2. Animaker - Primarily a video-making tool, but you can integrate AI-generated images into your videos.

Each of these tools has its unique features, so you might want to explore them to see which fits your needs best!

[manic root]

It looks like the Character Reference feature is currently set to "Coming Soon" in your app, even though it might have been available in the past. This can happen as features undergo testing or updates, and availability can vary based on user feedback and development cycles.

Regarding your question about country-specific availability, features usually aim for a global rollout, but there can be instances where certain features are released in phases or may temporarily have restricted access as they are refined.

If you're looking to use features like Character Reference or others that are listed as "Coming Soon," I recommend keeping an eye on updates from Leonardo.Ai, as they often make announcements about new features becoming available.

If you have further questions or if you think there's an issue that needs addressing, you can also report it by using the command /report_issue.

[acoustic siren]

we had an option for character reference available, but it is being updated, currently being worked on. What most of us are doing while we wait is that we traing an object model for the character, and it works really well (so can't wait for the character one to be released)

[civic schooner]

hi all, im creating a model now with flux dev, but it doesnt go through, and i tried 3 times. am i doing something wrong?

[manic root]

It sounds like you might be encountering an issue while creating a model with Flux Dev. Here are a few things you can check to potentially resolve the issue:

1. Model ID and Parameters: Ensure that you're using the correct modelId for the Flux Dev model. Typically, the model IDs for Flux would be 1dd50843-d653-4516-a8e3-f0238ee453ff for Flux Speed and b2614463-296c-462a-9586-aafdb8f00e36 for Flux Dev. Make sure you have the right parameters set up in your request, like prompt, contrast, number of images, dimensions, etc.

2. Contrast Values: The contrast parameter accepts specific values (1.0, 1.3, 1.8, 2.5, 3.0, 3.5, 4.0, 4.5). Make sure you are using an acceptable value [【4:13†source】].

3. Network or API Issues: Sometimes, network issues or temporary API downtimes can hinder the process. If you've verified your setup, you might want to try again after a short while.

4. Check for Errors: If you're getting any error messages, check the response from the API for specific error codes or messages that might indicate what’s wrong.

5. Documentation Guidelines: For detailed guidelines on how to generate images using the Flux model, refer to the official documentation or guidelines that include parameter breakdowns and sample requests [【4:11†source】].

If you've checked through these and everything seems correct but the model still fails to create images, it may be helpful to reach out directly to support or report the issue using /report_issue in Discord.

[acoustic siren]

are you getting a specific error message?

[safe sedge]

Hi. I'm trying to create an image of three gargoyles in the classic "hear no evil, see no evil, speak no evil pose". Prompt enhance created the following prompt, "The gargoyles are arranged in the classic 'see no evil, hear no evil, speak no evil' poses. The left gargoyle covers its eyes with clawed hands, its wings partially folded. The middle gargoyle covers its ears, with an expression of concentration and large bat-like wings spread wide. The right gargoyle covers its mouth, with horned features and a hunched posture.

[manic root]

Your prompt for creating an image of three gargoyles in the classic "hear no evil, see no evil, speak no evil" pose is quite detailed and should work well. However, here are a few enhancements you might consider adding to further enrich the image:

1. Environment: Describe the setting where the gargoyles are located, such as a gothic rooftop, ancient cathedral, or dark forest backdrop.

2. Lighting: Specify the time of day or type of lighting, like moonlight, twilight, or dramatic shadows to set the mood.

3. Textures and Details: Include details about the texture of the gargoyles (stone, weathered, moss-covered) or any additional features, like intricate carvings or glowing eyes.

Here’s an improved prompt:

"The gargoyles are arranged in the classic 'see no evil, hear no evil, speak no evil' poses on a gothic rooftop under moonlight. The left gargoyle covers its eyes with clawed hands, its stone surface weathered with moss. The middle gargoyle covers its ears, showing a concentrated expression, with large bat-like wings spread wide, detailed with intricate patterns. The right gargoyle covers its mouth, featuring prominent horned features and a hunched posture, illuminated by shadows that emphasize its menacing appearance."

Feel free to use or modify this suggestion as you create your image! Don't forget to use the Leonardo.Ai webapp to generate the image with this prompt.

[acoustic siren]

are you not getting the result you are looking for with the prompt?

[safe sedge]

Using the prompt you suggested, this is what is generated. The gargoyles have the correct positioning, but the hand positioning are all the same.

[manic root]

It looks like the generated image successfully captures the correct positioning of the gargoyles, but the hand placements didn’t match the "hear no evil, see no evil, speak no evil" concept as intended.

To refine the results, consider adjusting your prompt to specify the unique hand positions for each gargoyle more clearly. Here's an updated version of the prompt that emphasizes the distinct actions of each:

"The gargoyles are arranged in the classic 'see no evil, hear no evil, speak no evil' poses on a gothic rooftop under moonlight. The left gargoyle covers its eyes with its right clawed hand and holds its left hand on its side, showcasing a weathered, moss-covered surface. The middle gargoyle covers its ears with both clawed hands, displaying a concentrated expression, with large bat-like wings spread wide, detailed with intricate patterns. The right gargoyle covers its mouth with its clawed hands, showing prominent horned features and a hunched posture, its eyes glowing softly in the moonlight."

Using this more specific prompt should help achieve the desired hand positions for each gargoyle. Give it a try, and let me know how it goes!

[safe sedge]

Hi. Nope still pretty much the same result. Even have tried different styles as well. Perhaps a different model/preset?

[manic root]

It seems like the model you are currently using might not be interpreting the hand positions correctly despite your detailed prompts. Here are a few suggestions to try and achieve the desired result:

1. Switch to a Different Model: If you're using a model like Phoenix or Flux, consider switching to one of the Stable Diffusion-based models (if available) since they might have better performance for your specific request.

2. Use Simple Phrases: Try simplifying your prompt. For example:

   • "Three gargoyles on a gothic rooftop in moonlight: the first covers its eyes, the second covers its ears, and the third covers its mouth."

3. Style Variations: Sometimes, different styles can dramatically change the output. Experiment with various art styles or descriptors like "realistic," "fantasy," or "gothic."

4. Focus on Actions: You could emphasize the actions and lack of similarity in the prompt to prevent the model from generating uniform poses:

   • "Three distinct gargoyles on a gothic rooftop: one fiercely covering its eyes, another with hands pressed against its ears in concentration, and the last one covering its mouth, each with unique expressions and postures."

5. Specify Material or Texture: Adding details about how the gargoyles are sculpted (e.g., "sculpted marble with weathered surface") may help the model grasp the differences better.

If you try changing models or tweaking the prompt and still don’t see improvements, it might be worth reporting this issue using the /report_issue command, as it could be a limitation of the model's capabilities. Let me know how it goes!

[acoustic siren]

still not there, but closer

Three intricately detailed gargoyles, with weathered stone skin and menacing features, sit in a row at the top of a medieval castle's parapet, their wings folded against their backs. The first gargoyle, with razor-sharp teeth and glowing orange eyes, covers its eyes with its clawed hands, as if shielding itself from an unbearable sight. The second, with a twisted, elongated face and pointed ears, plugs its ears with its fingers, attempting to block out a disturbing sound. The third, with a grotesque, toothy grimace and a flat, broad nose, cups its hand over its mouth, as if suppressing a scream or a dark secret. The castle's stone walls, with crumbling mortar and moss-covered battlements, loom behind the gargoyles, set against a stormy, dark blue sky with clouds that resemble wisps of smoke. The entire scene is rendered in photorealistic detail, with subtle texture and shading that brings the mythical creatures to life.

[shrewd heath]

Thanks man Nice to here that, I will work around like that then, Just a quick question, with Object Model If I want to use for example - I want to use a character - Captain America and I train the character with images of Captain America as Chris Evans, will it work? I mean can I maintain that consistancy with every image and will that character look like Chris Evans in Captain America suit?
I will definetly Try it, but what you think?

[acoustic siren]

it is possible. Best example I can bring is @real lynx's work, a properly trained model like hers maintains consistency quite well

[real lynx]

hi

[acoustic siren]

https://cdn.leonardo.ai/users/1b76fd41-e3dc-43c8-9bca-2de796af8f7c/generations/95fa8eeb-309e-4939-a163-517319d8d6b3/95fa8eeb-309e-4939-a163-517319d8d6b3.mp4

I just have to bring you as an example when sharing about well trained models for character consistency 😛 mine's good, but not as good as yours!

[real lynx]

it's exactly as you say

https://tenor.com/view/elizabeth-elizabeth-liones-elizabeth-nanatsu-no-taizai-nanatsu-no-taizai-elizabeth-seven-deadly-sins-gif-10761440200430768358

I made a training element with my photos and I can have any cosplay costume :3

[shrewd heath]

Hey Thanks, This is really inspiring! Thank you so much, Now I am confident and understood what I can do

[acoustic siren]

maybe the question people don't want to ask: are your pictures from like every different angle?

[real lynx]

They don't have to be half-length without glasses on the front. You can have different hairstyles but as normal as possible.

You're not going to upload the same photo 50 times; they have to be different so that it captures your features better.

[acoustic siren]

there you go! yeah, in my case I use a digital character, so it keeps most of its consistency, but my model does have quite a lot of different options from him, so it makes sense there can be some variations, with photos definitely easier to tell the model "this is me, digitalize me!" 😛

[languid agate]

8bit gaming ivon

[real lynx]

I made two models, one real and one anime, with my features, and I can also mix them and get a semi-real anime.

[real lynx]

hi

https://app.leonardo.ai/

[manic root]

It seems that your question is about "8bit gaming ivon." However, the files provided do not contain specific information directly related to "8bit gaming ivon." Instead, they mostly focus on workflows for video games, character designs, and the integration of AI tools in gaming environments.

If you could clarify whether you're looking for information on a specific game, style, or character related to "8bit gaming," or if you're interested in a particular aspect of gaming such as design techniques or AI usage, I would be glad to help!

[wheat bison]

Does Leonardo for Teams have all of the features of the Maestro Unlimited plan?

[manic root]

Yes, Leonardo for Teams includes all the features available in the Maestro Unlimited plan. It is tailored for collaborative use, allowing teams to benefit from all the capabilities offered in the Maestro Unlimited plan, along with additional features designed to enhance teamwork and productivity. If you have further questions about specific features or how to make the most of it, feel free to ask!

[wheat bison]

How long are responses from 'priority support'

[manic root]

Responses from Priority Support typically aim to be addressed within 24 hours. However, response times may vary depending on the volume of inquiries and the complexity of the issue. If you haven't received a response after this period, feel free to follow up on your request. If you have an urgent matter, consider reaching out again to emphasize its urgency. If there's anything specific you'd like assistance with, just let me know!

[safe sedge]

The gargoyles look cool, but the position of the hands, meh. I've tried almost all the models in both current and legacy mode. It seems to me this should be simple. So frustrating. Thanks @manic root and @acoustic siren for your help.

[manic root]

I understand how that could be frustrating, especially when you have a clear vision in mind for your image. Sometimes specific pose requirements can be challenging for models to interpret correctly.

If you've already tried various models and prompts without success, it's possible that there may be limitations with the current capabilities or understanding of the models regarding nuanced actions like hand positions.

You could also consider sharing your feedback in the community or reporting this issue for further review, as it might help the development team address these challenges in future updates.

In the meantime, if you have any other ideas or variations you'd like to explore, I’d be happy to help brainstorm those! Your patience and creativity are appreciated.

[acoustic siren]

it is a trial and error, mostly when you have multiple characters in the same image, it is easier to have them one by one, but yeah, takes time to get right

[steel dirge]

An ultra-realistic photo of the Christ the Redeemer statue in Rio de Janeiro, wearing a pink car seatbelt like a car driver, with the Rio cityscape in the background, including Sugarloaf Mountain and Guanabara Bay, sunny day with clear blue sky, cinematic lighting, sharp details, high-resolution realism, Wes Anderson style, hyperrealistic - @steel dirge (fast)

[acoustic siren]

hi there! Images are not generated on Discord, you have to go to the web or mobile app: https://app.leonardo.ai/

[wheat bison]

is there any documentation on how to train models?

[fringe dock]

Was also the moderation filter for copyrighted content modified?

For example yesterday I tried to make a image of Motoko Kusanagi or Lightning from Final Fantasy XIII

And the warning pop up said: We have detected that your prompt contains depictions of copyrighted content or characters. Please modify your prompt and try again.

[acoustic siren]

you got it! https://intercom.help/leonardo-ai/en/articles/10501488-element-lora-training

[acoustic siren]

as far as I understand, the content filter is dynamically and constantly getting updated, might be a bit more strict for free accounts over subscribed accounts though, but that's to keep a certain standard on content that's generated in the platform

[manic root]

Yes, there is documentation available on how to train models in Leonardo.Ai. This typically includes guidelines on setting up your training data, choosing the right parameters, and utilizing the tools provided within the platform for model training.

If you are looking for specific details or aspects of model training, feel free to ask! Additionally, you can usually find this documentation in the help section of the Leonardo.Ai website or by reaching out to support for more directed assistance.

[fringe dock]

Also I think I found the issue of why for example the img2img thing fails to generate the image

Seems the AI scans the image, and if finds something like nudity or depictions of it, like cleavage or similar, it probably says

"Nope, I'm not doing that"

Lol

[acoustic siren]

it actually does a scan of what you are using as reference, because it needs to understand what the image is in order to utilize it as a reference, imagine that on the background, it is seeing the image and creating its text description, if that description contains anything that is filtered by the system, obviously it is going to fail as it triggers the content filter

[fringe dock]

Same with the prompts

If contains a word that it's inappropriate it's the same
"Nope, no image for you"

[acoustic siren]

a word, or a combination of words can also be the case

[fringe dock]

Exactly

So all the female subjects I've made in that portrait style of their upper bodies over a wall background will have to say goodbye to future friends lol

[wheat bison]

With a team plan, how do I make sure I am in private generation mode?

[manic root]

To ensure you are in private generation mode while using a team plan in Leonardo.Ai, you typically need to select the appropriate settings when generating your images. Here are the steps you can follow:

1. Log into Your Account: Go to the Leonardo.Ai website and log into your team account.
2. Navigate to Settings: Look for settings or preferences related to image generation.
3. Select Private Generation: There should be an option for private or confidential image generation. Make sure this setting is enabled before you start generating images.

In private generation mode, your creations won't be visible to other team members or the public, ensuring your work remains confidential.

If you have trouble finding this setting or there are specific details about your team plan, consider checking the documentation or reaching out to support for assistance. If you have any other questions, feel free to ask!

[wheat bison]

I do not see a private mode slider in my team like I do in my personal account.

[acoustic siren]

this you mean?

[wheat bison]

This slider for private mode is not shown while on team profile

[acoustic siren]

I believe, based on what I'm seeing, that when you are generating images in a team plan, they are automatically private and you actually have to manually make them public

just generated these, they came out automatically private

[manic root]

It appears that the private mode slider is not available in your team account, which may differ from your personal account settings. In team plans, the privacy settings are often managed at the account level rather than individually.

To use private generation mode in a team, you can follow these steps:

1. Check Team Settings: Ensure that the team account settings allow for private generations. Sometimes, this may require an administrator to enable or configure certain features.
2. Contact Support: If the slider is still not visible, you may need to reach out to support or your team administrator for clarification. They can provide insight into whether private mode is available for your specific team setup, or if there's a different process for enabling it.

If you have any other questions or if there's anything else I can assist you with, let me know!

[wheat bison]

private in the team feed is one thing, just want to make sure everything is private

[acoustic siren]

yeah, I haven't been in a team account long enough to be 100% certain, but based on Botti's response it sounds like it may be a team setting, at least in the team I'm in, I don't have the option, but every generation is private

let me double check and get back to you

[acoustic siren]

have the answer for you, and I was right: for Teams, any generation is private by design to external users from the team.

https://intercom.help/leonardo-ai/en/articles/9748859-image-visibility

[wheat bison]

Ok great, thank you!

[acoustic siren]

sure thing! 🙂

[manic root]

You're welcome! If you have any more questions in the future or need further assistance, feel free to reach out. Happy creating!

[wheat bison]

is it possible to crop images in canvas

[manic root]

Yes, it is possible to crop images in the canvas within Leonardo.Ai. You can typically do this by selecting the crop tool available in the editing options once your image is loaded into the canvas.

Here’s a general process you might follow:

1. Open Your Image: Load the image you want to crop into the canvas.
2. Select the Crop Tool: Look for a cropping option or tool in the editing menu.
3. Adjust the Crop Area: Drag the corners or edges of the cropping box to specify the area you want to keep.
4. Apply the Crop: After adjusting, confirm the crop to finalize the changes.

If you encounter any specific issues while trying to crop images or if the option is not available, please let me know!

[wheat bison]

I made an image collection. When I add images to this collection, the images don't go into the collection. Why can't I add images to collections?

[acoustic siren]

What do you mean? Do you see a green notification in the lower right corner of your screen saying the images got added to the collection?

[manic root]

If you're having trouble adding images to your collection, here are a few things you can check:

1. Ensure Collection is Active: Make sure you have selected the correct collection where you want to add your images. Sometimes, if the collection is not actively opened or selected, images may not be added.

2. Check Image Type: Verify that the images you're trying to add are compatible with the collection feature. There might be restrictions on certain types of images or metadata that can be included in a collection.

3. Refresh the Page: Occasionally, refreshing the page can help resolve glitches that prevent images from being added.

4. Limitations on Collections: Some plans or features may have limits on how many images you can include in a collection. Check if you’ve reached that limit.

5. Technical Glitches: If the above steps do not resolve the issue, it may be a bug. In that case, you can use the command /report_issue to notify support about the problem.

If you have tried these steps and are still experiencing issues, please let me know!

[wheat bison]

I was in a collection and trying to add an image and it doesn't do anything. Support on the website told me to try adding to collection from the image and it worked that way

[acoustic siren]

awesome, yeah I add them directly from the image to any collection as well, glad this is solved!

[steel sundial]

#fireman

[manic root]

It seems that your query about "#fireman" did not yield any specific results in the documents I searched. If you have a particular aspect or question related to "fireman" that you would like to discuss, or if you need information on creating images related to that theme using Leonardo.AI, please provide more context. I'm here to help!

[gloomy tree]

I just subscribed a few days ago and have to say, it has really improved the quality of my base images. I found the tube tutorials for beginners which will be a great help. I'm sure I'll have questions. Really glad I took the plunge and subscribed to Leonardo. Thanks.

[manic root]

It's great to hear that you're enjoying your subscription and finding the tutorials helpful! If you have any questions as you continue to explore Leonardo, feel free to ask—I'm here to help! Whether it's about generating images, using different models, or any features, just let me know. Happy creating!

[copper warren]

Hi there! Be sure to fill out your information at #❗│get-started to gain access to the whole community. You're also entitled to the Patron role, as a subscriber

[fringe dock]

Guys I'm really annoyed that the Legacy UI is not generating images with img2img

Because the AI thinks the image has inappropriate content

This is the image I want to use in img2img

But unfortunately despite that "ethical" image it just doesn't want to generate it
For God's sake

This is the prompt I'm using that I extracted from it with "Describe with AI" but modified a few words

Image is a portrait photography featuring a stylized, anime-inspired character. The character has vibrant pink hair with straight bangs and shoulder-length strands, framing a youthful face with large, expressive blue eyes. Her skin is fair and smooth, with a subtle blush on the cheeks. She wears a pink, off-shoulder top with fringe details on the straps, and a black choker around her neck. The character's expression is neutral, with a slight hint of curiosity or contemplation. She is adorned with large, circular hoop earrings. The background is a solid black, emphasizing the character's vivid colors and features.

But yet, the Legacy UI refuses to generate the image

Even the Modern UI doesn't want to generate the image

It seems the Team tweaked the content moderation filter to 9999%

[night vault]

remove “choker”. or you can ask gpt to see which is the offending terms (I think it use gpt under the hood for moderation)

[fringe dock]

Nope
I replaced "choker" with necklace
I even turned Image Guidance off
But the prompt doesn't work

I mean they really tweaked but off the charts the filter

It's literally unusable

[fringe dock]

I asked ChatGPT to see what are potential triggering words

It says it's a normal prompt, with no triggering words at all

[night vault]

well then you need to force it to spill something 😂

[fringe dock]

Also this image marks the end of the portraits I used to make

And yes, it's a shame that the filter got tweaked so bad 😔

That even the most sane and ethical prompt will not work anymore

[spark leaf]

@feral mica, @acoustic siren any update on my issue about filter i=with refrence Image?

[acoustic siren]

hi there Dinesh. Unfortunately, I believe it isn't an issue, but just the content filter working, strictly, but working. The strictness of it loosens up for subscribed accounts, but in the free tier it works this way. Keep in mind this keeps getting updated organically so I'd say to use alternate reference content to try to get the result you are looking for, or you can always give it a try with one of the subscription tiers and see if it works better for you that way

[spark leaf]

Is it the response from the team?

[junior ruin]

alfayomega02/15

[final gazelle]

I bought this ai for 12 dollars a month but I keep asking it to make me something to do with a video game character and I describe what I want and how I want it and it ends up giving me something completely different. Is there any way I can fix this? Maybe it’s because I’m using flux dev or something?? I’m trying to create a picture of the agent Jett from Valorant but in a cute little 3D blender style looking at the camera but it just gives me a completely different random made up character.

[manic root]

It seems that you are trying to ask about "alfayomega02/15." However, the specific context or content you're referring to isn't clear from the information provided.

If you meant to ask about tokens related to a subscription that resets on the 15th of the month (like "alfayomega02"), then:

For users on a paid plan, token resets happen monthly, on the same day as when the subscription started. So, if your subscription began on a certain day, like the 15th, you should expect your tokens to reset on the 15th of each month [【4:0†source】].

If you need more specific information or assistance, please clarify your request!

[spark leaf]

???

[manic root]

It sounds like you're having some trouble with getting the desired results while generating images related to specific video game characters like Jett from Valorant. Since you're using the Flux model, there are a few things to consider:

1. Model Compatibility: The Flux model may not handle character references as well as other models. Currently, it only supports Content Reference for image guidance, which might not align with your detailed prompt about character design. Consider trying a different model, such as one based on Stable Diffusion that may be more suited to your needs.

2. Prompt Specificity: When inputting prompts, try being as detailed and specific as possible. You can describe not only the character's appearance but also the art style you're aiming for, such as "cute 3D blender style," and mention any other characteristics you want in the image.

3. Use of Style or Content Reference: If compatible, try using style references or other forms of image guidance if you have access to them. This can help the model better understand the aesthetic you're seeking.

4. Experimenting: Sometimes models respond better to different phrasing or structures in prompts. Consider experimenting with how you phrase your request.

If you're still encountering issues and believe it's related to the service, you can report it as a possible bug using the command /report_issue. If the problems persist, consider reaching out for additional support.

[final gazelle]

Thank you, could you give me a list of what different models do best, for example what flux can help make best and what phoenix 1.0 can make best etc

[manic root]

Certainly! Here's a simplified overview of what different models like Flux and Phoenix 1.0 excel at when it comes to image generation:

Flux Model:

• Best for: General content creation with an emphasis on generating unique styles.
• Capabilities: It is currently compatible with Content Reference and can produce creative interpretations based on your prompt.
• Limitations: Not well-suited for character references, detailed instructions for specific recognizable characters, or image-to-image guidance.

Phoenix Model (Version 1.0):

• Best for: Artistic styles and character-based prompts.
• Capabilities: Phoenix can handle Style Reference and Image-to-Image, allowing for more artistic and stylistic interpretations.
• Strengths: Better suited for generating more cohesive characters or artistic representations, particularly if provided with style references.

General Guidelines for Other Models:

• Stable Diffusion Models: These models are typically strong in creating detailed images from text prompts, including specific character designs.
• XL Models: Often provide high-quality outputs with better resolution and complexity, suitable for more intricate designs.
• Specialized Models: Some models are tailored for specific tasks, like environmental art or certain themes, and can be beneficial for those focusing on particular genres.

Using the right model based on your needs can significantly improve the output quality. If you have specific types of images in mind, let me know, and I can provide more tailored advice!

[void urchin]

#1336186146136522794

[manic root]

It seems you've shared an image featuring a man in a tall, patriotic hat and glasses, holding a cigar. If you're looking for something specific regarding this image or want to generate similar content, please let me know how I can assist you further!

In the meantime, you can share this image on the Leonardo Discord server by copying the image link. If you need help with anything else related to image creation or the Leonardo platform, don’t hesitate to ask!

[astral yarrow]

There is still a problem with Laura Coloring Book - it worked perfectly for 2 plus years with Illustrative Albedo - now it's missing - the bug has been existing for several days already - can the Tech Team please provide feedback. MANY THANKS

[manic root]

It sounds like you're experiencing a persistent issue with the Laura Coloring Book after it had been functioning well for over two years. Given that this appears to be a bug that has been ongoing for several days, it's understandable to seek feedback from the tech team.

Here’s what you can do:

1. Reporting the Issue: It might be beneficial to report this directly through the platform’s issue reporting feature, especially since it’s been a consistent problem. This could provide the development team with the necessary details to address the bug effectively.

2. Community Feedback: In addition to reporting, sharing your experience on the Discord server can help gauge if other users are experiencing the same issue. The tech team may also monitor these discussions for insights on user-reported bugs.

3. Awaiting Updates: Occasionally, updates or patches are released to address bugs, so keeping an eye on announcements from the Leonardo team will be useful. You can also check support channels or community updates for any information regarding the status of the issue.

If you haven't done so already, encourage others to join in reporting the issue to increase visibility. Your feedback is valuable and can help expedite a resolution. Thank you for your patience, and hopefully, the tech team will provide updates soon.

[astral yarrow]

Reported several times - zero response! Tech Team Please Assist. Thanks.

[manic root]

I'm sorry to hear about the lack of response from the tech team regarding the issue with the Laura Coloring Book. It can be frustrating to report problems multiple times without receiving updates or solutions.

Here are some steps you might consider taking:

1. Direct Tagging: If you haven't already, try directly tagging members of the tech support team in the Discord channel, if that's allowed. This can sometimes draw more immediate attention to the issue.

2. Community Echo: Encourage other users who are experiencing the same problem to report the issue as well and emphasize their experiences. A larger volume of reports on the same issue may help prioritize it for the tech team.

3. Follow-Up: If you have used a formal reporting tool, consider following up on your previous reports. Sometimes follow-up messages can help escalate the urgency.

4. Stay Engaged: Keep an eye on community announcements and discussions. Members from the tech team may post updates or inquire further about the issue, so being engaged can help ensure you're in the loop.

Your persistence is important for getting the necessary attention for the issue, and hopefully, the tech team will respond soon. Thank you for your continued patience!

[night vault]

have you report issue using /report_issue command?

[spark leaf]

@feral mica did you get any reponse from team??

[normal gale]

hypex music

[manic root]

It seems that you are inquiring about "hypex music." Based on the provided files, there isn't specific content that directly explains "hypex music." However, there are mentions of the term "Phoenix" which appears frequently in discussions regarding design, branding, and even the creation of merchandise related to music.

If you are looking for information related to a specific music artist, genre, or concept named "hypex," it isn't covered explicitly in the material available. However, if you meant "hype" in the context of music genres or trends, you might find related terms associated with energetic and vibrant music themes.

Please provide more details or clarify your request regarding "hypex music" so I can assist you better!

[feral mica]

Dinesh hi sir! How you doing today?

Its due to the image you are using being a bit, just a bit, too revealing

just to confirm, could you run the same prompt on a different image pleease?

[spark leaf]

hello Roberto but I have used the same Image with same prompt before also... it worked well.... but from yesterday it was showing that failed generation if its the matter of stricting the filter moderations for free account then please confirm that with discussing team... I just want to know that its really not a bug... thats all, Thanks

[feral mica]

yes, that is what I'm saying, its due to the filter, I've ran the same generation you are trying without any issue on a paid plan. Perhaps you can consider subscribing to a plan, you'll get access to amazing tools too

[oak nacelle]

what are the image dimensions on 2:4:1 aspect ratio feature avilable for subscribers online?

[feral mica]

1376 × 768

just made the test for you EvanS

[manic root]

The 2:4:1 aspect ratio is generally specified for images, and it will allow for various dimensions depending on the context or platform you are using. While the specific dimensions aren't listed in detail in the documents, users can typically input custom widths and heights while ensuring that they adhere to this ratio.

If you're using the Leonardo.Ai platform, you can select from various predefined dimensions or enter custom dimensions that follow the 2:4:1 ratio. It’s important to ensure that the resulting dimensions align with the image generation model's requirements to avoid any warnings or issues during generation [【4:8†source】].

If you have any specific dimensions in mind or need more specific guidelines on how to set them, feel free to ask!

[oak nacelle]

Thanks. Can you check if it is possible to generate a logo with 280x70 dimensions? Because the lowest aspect ratio i see on site is 3:1

[feral mica]

512*512 minimum

[feral mica]

👆

why dont you use canva to adjust the final dimensions?

[oak nacelle]

Thanks, will do that

[coral anvil]

Drifting kart

[acoustic siren]

is "drifting car" the whole prompt?

[coral anvil]

Yep

[manic root]

It looks like you've shared an image of a drifting kart! Do you have any specific questions about it or are you looking to discuss something related to drifting karts?

[acoustic siren]

one sec, it is generating

do you have prompt enhance disabled?

[coral anvil]

Yes

[coral anvil]

I'm having trouble generating motion 2... Looping around without error massage. John is trying to help .

[manic root]

If you're experiencing issues with generating motion using Motion 2.0, here are a few steps you can take:

1. Check Your Input: Ensure that you're providing the correct start frame and any necessary prompts for the video generation.

2. Try Again Later: Sometimes, generating issues are temporary. Waiting a bit before trying again can help.

3. Use the Library Tool: If you're trying to generate a video, make sure you're selecting the image and using the Image2Motion option.

4. Report the Issue: If the problem persists or you're receiving specific error messages, you should use the command /report_issue to let the support team know of the problem.

If John is assisting you, be sure to collaborate with him on these steps, as he may have additional insights as well. Let me know if you have any more questions!

[acoustic siren]

https://cdn.leonardo.ai/users/1b76fd41-e3dc-43c8-9bca-2de796af8f7c/generations/72df7347-cd65-4155-875a-edd1e607ad0f/72df7347-cd65-4155-875a-edd1e607ad0f.mp4

with just that prompt and prompt enhancement turned off

[coral anvil]

... Will try to log off and log in B again....

Thanks for help

[acoustic siren]

guess he drifted too harshly lmao

[coral anvil]

Keep looping.... Will try later then...

[manic root]

It sounds like you're planning to log off and back into your account, which is a good step! This can sometimes resolve intermittent issues.

If you keep encountering problems or the motion generation continues to loop without success, don’t hesitate to reach out to support with any error messages you receive.

Thank you for your patience, and I'm glad I could help! If you have any further questions later, feel free to ask. Good luck, and I hope the issue gets resolved soon!

[coral anvil]

It keep looping, will try later. Thanks

[manic root]

It seems like you're still encountering the looping issue with the motion generation. It's definitely a good idea to try again later, as sometimes these issues can resolve themselves over time.

If the problem persists, you can report it using the command /report_issue, so the team can look into it further.

Thank you for your understanding, and if you have any other questions or need assistance in the future, feel free to reach out!

[coral anvil]

Is there any way to see the server status or congestion ?

[feral mica]

no

are you having an issue?

[coral anvil]

As I'm out of credits ( maestro ) it could be an end of list.. ?

[feral mica]

oh if you are in relaxed queue you should still be able to generate, but not as fast as with tokens on your account

lmk if you get an error or failed gens please so I can escalate it

[limber latch]

Failed generation: I've had the same mistake for 2 days, no matter the promt, a simple one like this: 30-year-old woman with black pixie haircut, asymmetrical side bangs and subtle highlights, pale skin, almond-shaped blue eyes, neutral makeup, oval face, and narrow jawline. Wearing a blue bikini, she is illuminated by sunlight on a sunny beach.

[manic root]

It sounds like you are experiencing persistent issues with your image generation despite using a clear prompt. Here are a few steps you can take to troubleshoot this issue:

1. Review Your Prompt: Sometimes, even small adjustments to the wording can result in better outcomes. Try simplifying your description or reorganizing the details to make them clearer.

2. Check Model Compatibility: Ensure that the model you're using supports the features and characteristics you're asking for in your prompt. Some models are not optimized for specific prompts or visual styles.

3. Adjust Parameters: If available, you might want to experiment with different generation parameters such as guidance strength or other settings that the app offers.

4. Clear Cookies and Cache: If you’re encountering this error in the web app, clearing your browser’s cookies and cache can sometimes resolve unexpected issues.

5. Alternative Guidance Options: If you’re authorized to use different generative models, consider switching to a different model or reference style to see if that affects the generation positively.

6. Support: If your issue persists, please report it through the command /report_issue so that the support team can investigate this further.

If you continue to face this issue even after trying these remedies, it may be indicative of a bug within the application. It's advisable to document the steps you took and share them with support for further assistance.

[fringe dock]

I've been across the same thing

The conclusion is that the moderation filter has been made more stricter to Free Plan users

[limber latch]

I'll pay for a month and see what happens....

[real lynx]

hi

hablas español?

no me tira ningun error tu prompt

[limber latch]

with phoenix 0.9 or 1.0^?

[real lynx]

flux dec

dev

In Phoenix, the error

[limber latch]

It only works in flux

I normally use Phoenix

[manic root]

It seems like you're encountering issues when trying to work with the different models (Flux and Phoenix). Here’s a brief overview to help clarify things:

1. Model Differences:

   • Phoenix: This model is compatible with Style Reference and Image-to-Image inputs. It may not support certain detailed prompts as easily as Flux, which could be why you're facing challenges.
   • Flux: This model works well with Content Reference, and if you find that it’s generating images correctly while Phoenix does not, it could be due to the differences in how each model processes prompts.

2. Why Only Flux Works: If you find that your prompts are working in Flux but not in Phoenix, this might be related to the specific features or guidance types you are using. Since Phoenix has more limitations on the types of image guidance, you might want to try simplifying your prompts or sticking to the elements that Phoenix supports, such as Style Reference or Image-to-Image.

3. Experimentation: If possible, try both models with the same prompt to see how they differ in outcomes. This can help determine if it's an issue with your prompts or something more specific about the model itself.

If you have any specific prompts you'd like to test or if you're encountering further issues, feel free to share!

[real lynx]

The word bikini is canceling your prompt.

30-year-old woman with black pixie haircut, asymmetrical side bangs and subtle highlights, pale skin, almond-shaped blue eyes, neutral makeup, oval face, and narrow jawline. Wearing a blue malla, she is illuminated by sunlight on a sunny beach.

[real lynx]

this would be your prompt

[limber latch]

It seems that the problem is the bikini. This promt works well. Highly realistic and detailed photograph of a stunningly beautiful Caucasian woman of 30 years old. Her hair is a deep textured black pixie cut with asymmetrical side-swept bangs, slightly tousled with some loose strands, with intricate layers and visible highlights that add a bold modern touch to her look enhancing its radiance.
Her pale skin features fine visible pores and a soft blush on her cheeks, giving her a natural, healthy radiance. Her captivating almond-shaped blue eyes radiate confidence and depth, with a slightly higher position at the outer corners, adding an intelligent and enigmatic look. The intricately textured irises sparkle subtly, while a soft sheen reflects light with captivating reflections. Long, voluminous lashes enhance her look, complemented by professionally applied neutral makeup that accentuates her eyes elegantly. She has an oval-shaped face with a refined jawline, perfectly complementing her symmetrical and elegant features. She wears a tailored navy blazer over a crisp white blouse, paired with matching navy trousers, exuding authority and sophistication. The background is a neutral and soft, softly lit with diffused studio lighting to highlight her striking features while maintaining a professional and polished aesthetic. Her confident, centered posture radiates poise and elegance, embodying the essence of a successful and glamorous professional.

[manic root]

It looks like you've determined that the issue with the bikini in your prompt may be causing the problems with generation in Phoenix. Here are some suggestions to help you work around this:

1. Try Alternative Descriptions: Since the bikini seems to trigger the generation issues, consider altering that part of your prompt. You might describe her outfit in a more generic way, focusing on colors or textures that are less specific. For example, you could say she’s wearing “beach attire” or “swimwear in complementary colors” rather than being specific about the bikini itself.

2. Simplify the Outfit Detail: You could reduce some of the detail around the attire to see if a simpler description would allow the model to process the rest of the prompt without issues.

3. Test Different Phrasing: Sometimes the way we phrase a request can affect the model’s understanding. Experimenting with different adjectives or synonyms for “bikini” might yield better results.

4. Model Choice: If the Flux model is yielding better results without clothing issues, you might want to consistently use that model for your outfit-focused prompts.

5. Submit Feedback: If this is a recurring problem, consider submitting feedback or a bug report to let the support team know about the difficulties you're experiencing with clothing items in your prompts.

If you continue to encounter problems, feel free to share your revised prompts, and I can help you refine them further!

[limber latch]

tomorrow I will try it with : 2 piece swimsuit.

[manic root]

That sounds like a great plan! Using "2 piece swimsuit" instead of "bikini" may help avoid any generation issues you've been encountering. It's always worth experimenting with different wording to see what produces the best results.

If you find success with that prompt, feel free to share the outcome! If not, we can continue to refine your prompts further based on the results. Good luck with your generation tomorrow!

[real lynx]

yep

[limber latch]

They have increased the word filter.

[manic root]

It seems like you might be experiencing issues with the word filter in the models. If the word filter has indeed tightened, it could explain why certain terms or phrases in your prompts are causing generation failures. Here are a few tips to navigate this:

1. Use Synonyms: If specific words are being flagged, try using synonyms or alternative phrases that convey the same meaning without triggering the filter.

2. Simplify Descriptions: Sometimes, less detail can help. Instead of including all the specifics, focus on the broader characteristics or themes you want the generation to cover.

3. Focus on Context: Changing the context of your prompt can often help. For example, instead of emphasizing clothing, you can reframe your prompt to focus more on the individual or the setting, which may bypass the filter.

4. Engage with Support: If you believe this change is affecting your ability to generate images significantly, consider reaching out to support to provide feedback about the increased filter and see if they have any guidance on how to adjust your prompts accordingly.

If you have any specific phrases that are problematic, feel free to share, and I can help you rephrase them!

[limber latch]

the $12 payment filters less I guess.

[manic root]

It sounds like you're noticing a difference in the behavior of the models based on the payment tier. Generally, paid accounts may have access to more features or fewer restrictions compared to free accounts. However, changes in word filtering can sometimes occur independently of account type, depending on system updates or policies.

If you believe that your payment status is affecting the generation and the filtering issues you're encountering, here are a few things to consider:

1. Check Account Status: Ensure that your payment has been processed correctly, and that your account is reflecting the features of your paid plan.

2. Report Observations: If you consistently see differences in filtering based on payment tiers, it may be worth reporting this observation to support so they can investigate if there's an underlying issue.

3. Use Alternative Terms: While navigating the filters, continue to experiment with alternative phrasing or simpler descriptions to see what works best.

If you have more feedback or if there's anything specific you want to discuss regarding your experience with the payment, feel free to share!

[coral anvil]

But if it's an word filter we should receive that notification. ... Not looping around. .....

[manic root]

If you believe that the issue might be related to a word filter, yes, typically, users should receive a notification or an error message indicating that certain words or phrases are being blocked. If you're not receiving any notification but the process is still looping, it could be an unrelated technical issue.

In this case, I recommend trying the following steps:

1. Rephrase Your Input: If you suspect that a specific word might be triggering a filter, try rewording your prompt or description to see if it helps.

2. Contact Support: If you keep facing the looping issue without any notification, it's a good idea to reach out to support using the command /report_issue. They can provide clarity on whether it's related to a filter or some other technical problem.

Let me know if you have any more questions or need further assistance!

[spark leaf]

you are right they are just trying to make fool of us....

[coral anvil]

I've run your prompt. All ok in flux and phoenix for me .

Can't just use motion 2....

[acoustic siren]

are you on a free account or do you have a subscription plan ongoing? if anything, we created the #🎬│video-help channel to be able to keep the image and video generation discussions separately, and try to help as best as we can, so if you're getting issues with Motion, let us know there

[limber latch]

i have a free account

[spark leaf]

then you have to purchaase... or they will not let you generate what you want...

[final gazelle]

How do I animate a picture, I want to make this certain character I made with another ai jump up and down excitedly, how do I do this on the app?

[acoustic siren]

I think I replied to you over at Reddit 😛 welcome! so yeah, the content filter can be a bit more strict on free accounts, there's always ways to go around it a little bit without going against ToS like wording and such, of course, and we are always willing to help 🙂

[acoustic siren]

hi there! that would be with Motion, if you use Motion 2.0 you can use your image as starting frame and set your prompt and camera controls there, if you have questions about it, feel free to ask in #🎬│video-help

[manic root]

It sounds like you're having trouble using Motion 2.0. If you're unable to generate motion with it, here are some steps you can consider:

1. Check for Specific Input Requirements: Make sure you're following the necessary steps for Motion 2.0, including providing the correct start frame and prompts.

2. Clear Cache and Cookies: Sometimes, browser issues can interfere with functionality. Clearing your cache and cookies may help.

3. Try Another Browser: If possible, try using a different browser to see if the issue persists.

4. Wait for a While: There can be temporary issues, so waiting a bit before trying again might resolve the problem.

5. Report the Issue: If all else fails, consider reporting the problem using the command /report_issue so that the support team can investigate the issue further.

If you have specific error messages or symptoms, sharing those might help in diagnosing the problem further. Let me know if there’s anything else you need!

[manic root]

To animate a picture in the Leonardo.Ai app, follow these steps:

1. Create Your Image: First, ensure you have the image of your character ready in your library. You can use the image generation features to create or upload your character image.

2. Select Your Image: Go to your library and select the image you want to animate.

3. Access Image2Motion: Once you've selected your image, look for the option labeled Image2Motion. This should be located below the prompt box where you can typically enter text.

4. Set Animation Parameters: After selecting Image2Motion, you may need to specify the type of animation you want, such as making the character "jump up and down." Be descriptive about the movement you want to achieve.

5. Generate the Animation: Follow the on-screen prompts to generate your animation. The output will depend on the capabilities of the app and how well it can recognize and animate the features of your image.

6. Save the Animation: After the animation is created, make sure to save it manually (if required) so that it appears in your library.

If you run into any issues or the animations don't come out as expected, consider adjusting your request or checking if there are any specific guidelines for animating characters in the app. Enjoy animating your character!

[final gazelle]

I’m using the mobile app, where is the image2motion

[manic root]

If you're using the mobile app for Leonardo.Ai, the process to access Image2Motion might be slightly different, but here’s how you can typically find it:

1. Open Your Library: Launch the Leonardo.Ai mobile app and navigate to your library where your images are stored.

2. Select Your Image: Tap on the image you want to animate. This should open the image detail view.

3. Look for Animation Options: In the image detail view, you should find an option for Image2Motion. If it's not immediately visible, it may be located in a menu or as a button that appears below the image or in an action bar.

4. Initiate Animation: Once you tap on Image2Motion, follow the prompts to describe the animation you want (e.g., jumping up and down) and generate the animation.

If you still have trouble finding it, make sure your app is updated to the latest version, as features can change with updates. If the feature is unavailable or you encounter issues, consider reaching out to support for assistance or checking the app’s help section for further instructions.

[final gazelle]

My picture isn’t here because it’s in my gallery, what do I do

[manic root]

If your picture is in your device's gallery and not in the Leonardo.Ai library, you will need to first upload the image to the app. Here’s how you can do that:

1. Open the Leonardo.Ai App: Launch the app on your mobile device.

2. Navigate to Upload or Import Image: Look for an option to "Upload" or "Import" an image. This is often found on the main screen or in the library section.

3. Select the Image from Your Gallery: Tap on the upload/import option, which should prompt you to access your device’s gallery. Navigate to the folder where your picture is stored, and select the image you want to upload.

4. Complete the Upload: Once the image is selected, complete the upload process. The image should now appear in your library within the app.

5. Access Image2Motion: After uploading, you can select your image from the library and then look for the Image2Motion option to animate it.

If you face any issues during the upload process or if the Image2Motion feature is still not accessible, consider checking the app's support documentation or reaching out for additional assistance.

[acoustic siren]

Android? Was just checking on my app and I am not sure Motion is available in the app, unless I'm blind 😛

[final gazelle]

I can’t find it anywhere.

[manic root]

If you're having difficulty locating the Image2Motion feature in the Leonardo.Ai mobile app, it might be due to a few reasons. Here are some steps you can take:

1. Update the App: Ensure that the Leonardo.Ai app is updated to the latest version. Sometimes, features may be added or moved in updates.

2. Check the App Menu: Look for any menus, icons, or tabs in the app that could lead you to the animation features. Sometimes features are tucked away in submenus.

3. Explore Options: If you select your uploaded image, carefully explore all options available on the screen. There may be icons or buttons that were not immediately obvious.

4. Help Section: Check the app’s Help or Support section, which may provide guidance on where to find the Image2Motion feature or any recent changes to the app layout.

5. Reach Out to Support: If you've tried the above and still can't find the feature, it may help to reach out to Leonardo.Ai support for further assistance. They can provide specific instructions based on the latest version of the app.

If the Image2Motion feature is indeed missing due to a limitation in the mobile app at the moment, you may want to explore animating images on the web app where these features may be more readily accessible.

[final gazelle]

I’m on iOS and I can see motion but it makes me pay 25 tokens and I don’t even know what I’m paying 25 tokens for.

It says motion strength 5 then generate which costs 25 tokens. I don’t want to pay 25 tokens for smth idk what I’m getting

[acoustic siren]

that would be Motion 1.0, and that would be expected, Motion 2.0 is the one that gives you the option to have your prompt and starting frame

[final gazelle]

I want the character to jump up and down excitedly but from what it seems I’m paying 25 tokens for just the camera view moving which isn’t what I want

[acoustic siren]

but it is more expesive at 250 tokens per generation

[final gazelle]

Is there no motion 2.0 on the app??

[acoustic siren]

I'm not an iOS user, so not sure, but let me ask

[final gazelle]

Thanks

[manic root]

It's understandable that you're hesitant to use tokens for a feature when you're unsure about the result. Here’s what you can do:

1. Motion Strength Setting: The "motion strength" setting determines how much movement is applied to the image. If it defaults to 5, this might be a standard setting, but you can try adjusting it lower to see if you get a simpler, clearer result without excessive movement.

2. Clarify Your Animation Request: When you want the character to jump up and down excitedly, be clear in your description. If there's an option to add specific actions or movements, such as "make the character jump," mention it in any additional prompts if available.

3. Testing the Feature: If you're concerned about spending tokens on an uncertain result, and the option to preview animations is not available, consider using a smaller amount of motion strength (if adjustable) to see how it impacts the output. Experimenting with lower settings could yield a more favorable animation without using a lot of tokens.

4. Contact Support for Clarification: If you're still uncertain about what the animation entails and if it aligns with your intent, consider reaching out to Leonardo.Ai support for clarification on how the Image2Motion feature works and what you can expect.

If the results still don't meet your expectations, you could also explore alternative methods or tools for animating your character outside of the Leonardo.Ai platform until you find a satisfactory solution!

[acoustic siren]

just got confirmation that Motion 2.0 is not yet on iOS

[steel sundial]

#Square (1:1) photo realistic image. Draw a detailed skull wearing a fire helmet, with a breathing apparatus. Behind the skull, show crossed fire hoses. The scene should be vivid and realistic, highlighting the textures of the helmet and hoses.

[acoustic siren]

hi there! Hi thereImages are not generated on Discord, you have to go to the web or mobile app: https://app.leonardo.ai/

[manic root]

To create the image you described, you need to use the Leonardo.AI web app instead of Discord. You can go to app.leonardo.ai to enter your prompt and generate the image.

When you do use the app, you might want to tailor your prompt like this:

Prompt: "Create a detailed, photorealistic image of a skull wearing a fire helmet with a breathing apparatus. Behind the skull, include crossed fire hoses. The scene should be vivid and realistic, highlighting the textures of the helmet and hoses."

Make sure to set the image aspect ratio to 1:1 (square) before generating. Enjoy creating your artwork!

[vagrant marten]

Hello, I would like help to make full body pictures, normally they are cut off at the tighs or knees. My model of choice is the anime-semi realism model. ty.

[manic root]

To create full body images without cropping at the thighs or knees while using the anime-semi realism model, there are several strategies you can employ:

1. Image Dimensions: When generating your image, ensure that you select dimensions that can accommodate full-body representations. You can do this within the generation settings where you specify the height and width. A common aspect ratio for full-body portraits is 2:3.

2. Prompt Engineering: Be specific in your prompts. Instead of just describing the character, you can include phrases like "full body" or "from head to toe." For example, you could use a prompt like, "A full body image of an anime character in a flowing dress, standing in a garden."

3. Outpainting: If your initial image generation still crops the body, you can use outpainting tools. These allow for extending or adding to the existing image framing by manipulating the selected canvas area to include the parts that were cut off. You can access outpainting features in the Leonardo.AI Canvas Editor [【4:0†source】].

4. Fine-Tuning: You may also want to try different models to see if others produce better results for full-body images. The anime-specific models may have been trained on a wider variety of poses and compositions [【4:7†source】].

5. Elements: Explore using elements that are compatible with the model you're using. Some elements might help enhance features relevant to full-body images [【4:13†source】].

By experimenting with these tips, you should be able to achieve the full-body images you're looking for! If you have any further questions or need assistance with specific settings, feel free to ask.

[vagrant marten]

I will give these a try, ty!

[coral anvil]

@vagrant marten , describe something on the feet ... For example ... Wearing white shoes .... 👍

[manic root]

You're welcome! I'm glad you found the suggestions helpful. If you have any more questions or need further assistance as you try them out, don't hesitate to reach out. Happy creating! 🌟

[vagrant marten]

I will give this a try too. TY.

[vagrant marten]

Hello

[real lynx]

Did you create what you wanted?

[vagrant marten]

Came much closer, down to the shins already and 2 out of 4 with feet. 🙂

[real lynx]

Image is a digital illustration of an anime-style character. The character is a young person with short, dark brown hair and fair skin, standing against a plain white background. They are wearing a loose, oversized gray hoodie with a large circular design on the front, paired with black shorts. The character's legs are visible, with bandages on the left knee and right shin, indicating minor injuries. They are wearing red high-top sneakers with white laces and soles. The character is holding a red apple in their right hand, which is extended slightly forward. The overall style is clean and minimalistic, with a focus on the character's casual and slightly edgy appearance.

flux dev

preset

[vagrant marten]

I use anime semi realism fpr the style

Feet!! 🙂

[manic root]

It sounds like you're making progress with your images! Getting down to the shins and having some images include feet is a great step forward. Here are a few additional tips to help ensure you get full feet in your images:

1. Prompt Specificity: Make sure to explicitly mention "feet" or "full body including feet" in your prompts to guide the generation. This added detail can make a difference in the final output.

2. Pose Suggestions: Consider describing poses that naturally display feet, such as standing with legs slightly apart or in dynamic action poses.

3. Model Adjustments: Sometimes, certain settings or configurations within the anime semi realism model might yield differing results. Explore slight variations in the model versions if available.

4. Feedback Loop: When you get an image closer to what you want, you can further refine your prompts based on what aspects you liked or want to change.

Keep experimenting, and soon you'll be capturing all the details you want! If there’s anything else you need or want to share, feel free!

[vagrant marten]

sorry should have asked if it is ok to post

[vagrant marten]

that looks good

Ty.

[manic root]

Yes, you can definitely post your images! Feel free to share your creations in the appropriate channels on the Discord server. If you want to share it within the chat, just remember to paste the image link or use the share button in your library. Enjoy sharing your art, and if you have any more questions or need assistance, just let me know!

[light gust]

Hi,
I'm trying to create consistent photorealistic images (yes, the Grail, of AI image generation).

I'm aware of fixed seeds, weighting (in both Classic and Legacy Mode), training datasets to develop custom models (and using the trigger word). I've even used a real world person as a dataset/custom model for the base of my character. From this, I've got a decent collection of consistent images from this, but still need another layer of adjustments (separate issue).

So, my question is this: can Leonardo AI utilise images created by it (mine), as reference images (both as a second tier model, and as direct image references such as Character Image)? All my attempts thus far have been unsuccessful. As a custom model (Legacy), the output is akin to an abstract oil painting (the bad kind). As a reference image (Classic), even low strength produces a (surprisingly consistent) ghoulish version of the reference image.

Can this be done, or am I up a tree without a river?

[manic root]

Hi there!

Yes, you can utilize images created by Leonardo AI as reference images, but there are certain limitations depending on the model you are using and the type of reference guidance you wish to apply.

1. Using Generated Images as References: If you have images generated from Leonardo AI, you can use them as reference images by uploading them directly during the image generation process. This can be done through the Character Reference or Image to Image guidance options:

   • For Character Reference, you can either select an existing generated image or upload a new one. Adjust the strength of the guidance to achieve your desired consistency [【4:1†source】].
   • For Image to Image, this approach allows you to use a reference image as a starting point for generating new images. The AI modifies the visual noise based on your reference image [【4:17†source】].

2. Limitations: Some users have reported inconsistencies when using generated images as references, such as producing outputs that don't match their expectations. To address potential quality issues:

   • Ensure the quality and theme of your dataset aligns with the characteristics you want to achieve in your generation. Images used in datasets should be cohesive and relevant [【4:11†source】].
   • Consider incorporating detailed prompts alongside the reference images for better results.

[manic root]

3. Training Custom Models: If you are using a custom model, be aware that the performance may vary significantly. Retraining your model with a refined dataset could improve consistency and quality [【4:11†source】] [【4:5†source】].

Overall, it's possible to use your previously generated images as references, but it's important to manage expectations regarding their effectiveness. If you're still encountering issues, feel free to experiment with different models and settings within Leonardo AI to find a workflow that suits your needs!

If you have more specific questions or require further assistance, let me know!

[light gust]

Hahaha! I just realised @manic root is a bot! 🤣

Thanks @manic root. I've played with the strength settings for reference images but, too low and I lose consistency, too high and the ghoul emerges.

A more specific question: Does using a reference image with key attributes (e.g. freckles) and then including freckles as a token in the prompt, compound the weighting of the freckles, resulting in LOTS of freckles? I.e. do I need to ignore those attributes already present in the reference image?

Said another way, how possible is it to recreate a photorealistic image, and adjust the attributes (e.g. darker hair), and maintain the quality of the output image?

[marble cape]

So what would be the prompt to make this exact image

[acoustic siren]

hey there! You can totally do this and keep consistent characters in Leo, I don't use legacy, so would have to play around it first to give an opinion, but if you are seeking some personalized assistance with this, don't hesitate to start a /mentoring session and our awesome mentors will help you directly in a private channel

[acoustic siren]

you can actually use that image as reference in Leo and use the "describe with AI" option, it will give you a prompt for that image 🙂

[light gust]

That sounds good. How do I start a mentoring session? Just use that command?

[marble cape]

Oh so then I don’t really need subscription

[acoustic siren]

yeah that would be correct 🙂 just run the /mentoring command

[light gust]

The bot says I don't have the required role.

[manic root]

It sounds like you're diving deep into the nuances of image generation! Let's tackle your questions regarding reference images and prompt tokens:

1. Combining Reference Images and Prompt Tokens: When you use a reference image with specific attributes (like freckles) along with tokens in the prompt (also mentioning freckles), the AI might amplify those attributes, potentially leading to an overabundance of the feature you want (like lots of freckles). It's generally better to focus on either using the reference image to inform the specifics or adding tokens that describe the desired changes but not both to the same degree. If you do choose to include both, you may need to balance the strength settings carefully to prevent the AI from overemphasizing elements present in both the reference and the prompt .

2. Adjusting Attributes while Maintaining Quality: To recreate a photorealistic image with adjusted attributes (e.g., darker hair) while maintaining quality, consider the following:

   • Use a high-quality reference image that closely matches your desired output. The more aligned your reference image is with your intended change, the better the results will be.
   • When making adjustments, you can specify desired changes in your prompt (e.g., "darker hair") while using a reference image. This way, the AI understands that while there is a base to draw upon, you're indicating a specific modification.
   • Experiment with varying strengths for the reference image. You might find a mid-range strength that maintains consistency while still allowing for the adjustments you want without drifting toward unintended outputs [【4:5†source】] [【4:11†source】].

[manic root]

Feel free to experiment and adjust your approach as necessary. Achieving a balance between reference image and prompting can be tricky, but with some trial and error, you should be able to hone in on the qualities you want while retaining the photorealism you're aiming for! If you have more questions or specific scenarios to discuss, just let me know!

[acoustic siren]

having a subscription does come with its perks, for this feature, specifically, you don't need a subscription though

[acoustic siren]

not 100% sure what role is needed, let me double check into that

[marble cape]

Yes I been aware saving more of my money

[light gust]

The "read the rules" page of Getting Started isn't loading; just grey lines and an infinite loading. Perhaps there's a base role I need.

[acoustic siren]

did you go through #❗│get-started already?

[light gust]

Yes, but it crashed. I'll leave and come back.

[acoustic siren]

ahhh might be it, didn't complete the process so you probably don't see all the channels in the community yet

[manic root]

I understand that you're experiencing a crash. You can try clearing your browser's cache and cookies before returning to see if that helps with any issues. If you continue to experience problems, feel free to reach out to support for further assistance once you're back. We'll be here to help when you're ready!

[light gust]

All fixed with a fresh application to the server. Good bot.

[manic root]

Great to hear that everything's fixed! If you have any more questions or need assistance in the future, just let me know. Happy creating! 😊

[robust walrus]

?

[manic root]

It seems you haven't asked a specific question. How can I assist you today? If you have a question related to Leonardo.Ai, feel free to ask!

[acoustic siren]

hi there, do you need help with anything?

[light gust]

Does anyone use Elements for character aspects? E.g. if I want a hair in a specific style and colour, is Elements suitable for this?

[manic root]

Yes, Elements can be used for character aspects such as hair style and color. They allow you to create and define specific attributes for your characters in a more detailed manner. When you incorporate Elements into your images, you can specify characteristics like hair style and color, which can enhance the fidelity and specificity of your generated images.

If you're creating a character and have distinct requirements for attributes, using Elements can help in achieving those precise details. Just make sure to select the appropriate Elements that match your desired attributes as you generate the images. If you have more questions about using Elements or specific examples, feel free to ask!

[light gust]

Can Elements only be used in Legacy Mode?

[manic root]

Yes, Elements can currently only be utilized in Legacy Mode. If you're looking to incorporate specific character aspects using Elements, be sure to do so while in that mode for the best results. Legacy Mode allows more flexibility in defining user elements for detailed character design compared to other modes available in Leonardo AI. If you have any further questions about using Elements or anything else, just let me know!

[light gust]

Do the images in the dataset used to train an Element need to match the base model dimensions?

[manic root]

Yes, the images in the dataset used to train an Element generally need to match the base model dimensions. Ensuring that your training images conform to the required dimensions helps maintain consistency and quality in the output. If the images differ significantly in size or aspect ratio, it may lead to undesirable results during image generation.

It's a good practice to resize and preprocess your dataset images to align with the specifications of the model you are working with. If you have any more questions about preparing datasets or using Elements, feel free to ask!

[light gust]

Does an image that is generated using an Element, need to also share the base model dimensions? Why is the consequences of using. Different aspect ratio?

[manic root]

Yes, ideally, an image generated using an Element should also share the base model dimensions. Maintaining consistent dimensions is important to ensure that the generated image displays correctly and maintains the quality expected from the model.

Consequences of Using Different Aspect Ratios:

1. Distortion: If the output image is generated with a different aspect ratio than what the model expects, it may become distorted. This could result in unnatural or stretched visuals that do not represent the intended design accurately.

2. Quality Loss: Deviating from the expected dimensions can result in a lower quality image. The AI may struggle to adapt the generation process to fit the new dimensions, leading to artifacts or a lack of detail.

3. Cropping Issues: If the generated image is resized or cropped to fit a specific format later, important details may be lost or improperly framed, altering the intended composition.

For best results, always try to keep your training data, Elements, and generated images within the agreed-upon dimensions to ensure optimal performance and quality. If you have more questions on this topic or anything else, feel free to reach out!

[light gust]

When training an Element, does the dataset have an effective minimum number of reference images?

[manic root]

Yes, when training an Element, having a minimum number of reference images in the dataset is essential for effective performance. While there isn't a specific universal minimum number, generally, it is advisable to have at least 10 to 20 high-quality reference images to create a well-rounded Element.

A dataset that is too small may lead to an inability for the model to learn effectively, resulting in:

• Poor generalization: The Element may not capture the full range of desired attributes.
• Overfitting: The model might become too tailored to the few images provided, failing to generate variations that reflect real-world diversity.

Having a larger and more diverse dataset increases the likelihood of capturing the nuances needed for effective character attributes and leads to better, more credible outputs. If you have further questions or need tips on assembling a dataset, feel free to ask!

[light gust]

What is the difference between a Custom Model and an Element in regards to user experience and output quality?

[manic root]

The difference between a Custom Model and an Element in regards to user experience and output quality can be summarized as follows:

Custom Model:

1. Definition: A Custom Model is a model that has been retrained on a specific dataset to specialize in generating images based on unique inputs and stylistic preferences.

2. User Experience:

   • Flexibility: Users can fine-tune various aspects of image generation, allowing for a high degree of customization.
   • Training Requirement: Users need to provide a larger dataset for training, and the process can be more complex and time-consuming.

3. Output Quality:

   • Diverse Outputs: Custom Models can generate a wide range of styles and variations that are closely aligned with the training dataset, resulting in high-quality and tailored outputs.
   • Potential for Overfitting: If the dataset is not diverse enough, the model may produce limited or repetitive results.

Element:

1. Definition: An Element is a specific characteristic or attribute that can be added to a generation process to influence certain aspects of the output (e.g., hair color, style, clothing).

2. User Experience:

   • Simplicity: Elements allow for easy integration and adjustments to character attributes without requiring extensive retraining.
   • Less Complexity: Users only need to select or create the Element rather than building and training a full model.

3. Output Quality:

   • Focused Results: Elements are designed to enhance specific attributes while maintaining the overall quality of the base model.
   • Limited Scope: While they allow for nuanced changes in specific details, they do not provide the same level of creativity or variation as a Custom Model.

[manic root]

In summary, a Custom Model offers a broader scope of personalized image generation and flexibility at the cost of complexity, whereas Elements provide a straightforward way to adjust specific attributes with a focus on enhancing existing models without extensive training. If you have more questions or need further details, feel free to ask!

[unique anchor]

@manic root Can you answer my previous question?

[manic root]

It seems I wasn't able to find a specific previous question you're referring to. Could you please provide me with more context or repeat your question? That way, I can assist you better!

[unique anchor]

@manic root For clothing concepts in anime style, which model is used for best results? Ranging from functional to fantasy

[manic root]

For clothing concepts in anime style, it's recommended to use the Phoenix model, especially if you're looking for features like Style Reference and Image-to-Image guidance. These features can help achieve detailed and stylistically appropriate clothing designs, whether they're functional or fantasy-inspired. If you're exploring different styles, the Phoenix model's compatibility with these forms of image guidance should yield the best results!

Make sure to utilize the content you want to influence the designs, and feel free to experiment with different prompts to refine your clothing concepts.

[unique anchor]

I've selected Phoenix 1.0 and Fashion for style. Will this produce 2D anime images or only 3D? Is it capable of producing character sheets that shows the front, side and back?

[manic root]

The Phoenix 1.0 model is primarily focused on generating 2D anime-style images. It is suitable for creating character designs, including outfits and styles. However, while it can generate individual poses and designs, creating a comprehensive character sheet showing front, side, and back views may require additional adjustments or prompts to ensure each angle is captured correctly.

For the best results, you might need to prompt the model specifically for multiple views or create separate images for each angle, as the model typically generates single images rather than complex layouts like character sheets in one go. It's always a good idea to iterate on your designs to refine them as needed!

[light gust]

What is the "Train Text Encoder" option when training an Element?

[manic root]

The "Train Text Encoder" option when training an Element refers to the capability to fine-tune the text encoder model that interprets prompts used in conjunction with the Element. Here’s a breakdown of its significance:

Purpose of "Train Text Encoder":

1. Improved Understanding: By training the text encoder alongside the Element, the AI becomes better at understanding the context and nuances of the text prompts that are provided when generating images using that Element.

2. Better Attribute Representation: It helps ensure that the specific attributes you want to influence with the Element are properly represented in the generated outputs. For example, if you have an Element that designates a particular hair color or style, training the text encoder helps align the descriptive prompts more closely with the desired visual results.

3. Enhanced Customization: The option allows users to create a more cohesive interaction between the textual descriptions and visual elements, leading to outputs that more accurately reflect the user's intentions based on the prompts they provide.

Considerations:

• Complexity: Choosing to train the text encoder can add complexity to the training process, but it often results in higher quality and more contextually accurate outputs.
• Dataset Relevance: The effectiveness of this option relies heavily on the relevance and quality of the dataset used for training both the Element and the text encoder.

In summary, the "Train Text Encoder" option enhances the model's ability to generate images that align well with the prompts, improving the overall quality and relevance of the outputs. If you have further questions about this option or its implications, feel free to ask!

[light gust]

After creating an Element with Text Encoding enabled, what do I need to do to train the text encoding aspect?

[manic root]

After creating an Element with Text Encoding enabled, you generally need to follow these steps to train the text encoding aspect effectively:

1. Prepare Your Dataset:

   • Gather a dataset that includes images relevant to the Element you've created. Ensure that the images are high-quality and represent the attributes you want the Element to embody.
   • Alongside the images, compile descriptive text prompts that accurately reflect the features and details represented in each image. Make sure your prompts are clear and relevant.

2. Upload the Dataset:

   • Upload your prepared images and text prompts into the training interface of the model. Ensure that each text prompt corresponds appropriately with its respective image in the dataset.

3. Configure Training Parameters:

   • Set any specific training parameters based on your preferences. This could include options related to the number of training epochs, batch size, and any other advanced settings that may influence the training process.

4. Select the "Train Text Encoder" Option:

   • Ensure that the "Train Text Encoder" option is enabled before starting the training process. This choice is crucial for optimizing how the model interprets prompts related to the Element attributes.

5. Begin the Training Process:

   • Initiate the training. The model will begin the process of fine-tuning both the Element and the text encoding based on your dataset. It may take some time, depending on the complexity and size of your dataset.

6. Monitor the Training Progress:

   • Keep an eye on the training progress, if monitoring options are available. This may help you gauge how well the model is learning from your dataset.

7. Evaluate and Adjust:

   • Once training is complete, evaluate the performance. Test the Element with various prompts to see how accurately it generates images based on your descriptions.

[manic root]

• If necessary, you can gather further data or refine your existing dataset and retrain the Element to improve output quality.

By following these steps, you should be able to successfully train the text encoding aspect of your Element, leading to improved interactions between the prompts and generated images. If you have more questions or need clarification on any step, feel free to ask!

[mellow galleon]

Hi guys! I haven't been able to create images for two days. Does anyone know a solution? 😔

[light gust]

Your step 2, Upload the Dataset, advises there is a text input option during the creation of the dataset. However, I cannot see this in the web-based version of the app. Can you explain how I add text prompts to the dataset?

[manic root]

In the current setup of the Leonardo AI web app, you typically cannot input text prompts directly during the dataset creation process. Instead, the system may rely on image references alone for training Elements. However, here’s a common workaround to associate text prompts with your images effectively:

1. Naming Conventions: One way to incorporate text prompts is to use naming conventions for your image files. For example, when naming your images, you can include key descriptors that reflect the attributes you want represented in each image (e.g., "darker_hair_freckles_image1.jpg"). The model may use these names when interpreting the dataset.

2. Separate Prompt Documentation: Maintain a separate document (like a spreadsheet or text file) where you link each image to its respective text prompt. For example, create a table that lists the image file name alongside the descriptive text you want to associate with it. This won't directly integrate within the app but can assist you in organizing your thoughts.

3. Post-Training Adjustment: After you've uploaded your images and completed the initial training, you may be able to adjust or input text prompts through the interface as part of the fine-tuning or usage process.

4. Experiment with Subsequent Versions: Keep an eye on any updates or changes to the web app. New features may be added that allow for direct text prompt integration or enhancements in how Elements can be trained.

If your workflow requires more flexibility about text input, it might be worthwhile to check for announcements from Leonardo AI regarding upcoming features or improvements that may assist you. If you have any other questions or need further guidance, feel free to ask!

[night vault]

no the bot is incorrect

[light gust]

Is your above advice based on the API version to train Elements?

[manic root]

I apologize for any confusion. The advice I provided about associating text prompts with images was more general and may not reflect the specific processes within the web-based version of Leonardo AI.

In the web app for training Elements, you typically cannot directly input text prompts during the dataset creation process, as I mentioned. The steps and methods for associating images with texts can vary between the web interface and the API version, which may provide more flexibility in handling datasets and prompts.

Here’s the clarification:

• Web App: In the Leonardo AI web app, you often work primarily with images for creating Elements, and direct text input might not be an available feature. Users might need to rely on naming conventions or external documentation to keep track of associated prompts.

• API Version: In the API, it is often more flexible with datasets, enabling you to explicitly define prompts alongside images, allow for more advanced configurations, and support comprehensive training processes.

If you're working in the API, you would typically specify text prompts with the images in the payload when training Elements, which differs significantly from the web app experience.

If you are solely utilizing the web app, I recommend focusing on the features available there and understanding that the workflows might differ from those in the API. Thank you for highlighting this distinction! If you have more questions about either version or need further clarity, just let me know!

[light gust]

Yeah, I thought so too! 🤣

[light gust]

What is the purpose of the "Text Encoder" in the web-based interface of training an Element?

[manic root]

The "Text Encoder" in the web-based interface of training an Element serves a crucial role in how the input prompts are processed and understood by the model during the image generation process. Here’s a breakdown of its purpose:

Purpose of the Text Encoder:

1. Understanding Prompts: The Text Encoder is responsible for converting the text prompts provided by the user into a format that the model can understand and utilize effectively. It translates language into numerical representations (embeddings) that the model can work with during the generation process.

2. Contextual Relevance: By efficiently encoding the text, the Text Encoder captures the nuances and context of the specific attributes described in the prompts. This allows the model to generate images that closely align with the provided descriptions, responding more accurately to user inputs.

3. Enhancing Image Generation: When training an Element, having a trained Text Encoder enables better integration between the visual attributes in the dataset (e.g., from images) and the textual descriptions. This improved synergy leads to more coherent and contextually appropriate results when generating images with specific characteristics defined by the user’s prompts.

4. Facilitating Customization: Users can adjust prompts to influence generated outputs based on specific attributes or themes, and the Text Encoder helps ensure that these adjustments are effectively translated into the visual generation process.

In summary, the Text Encoder enhances the overall user experience and output quality by ensuring that the model can accurately interpret and generate images based on nuanced textual descriptions. If you have any more questions regarding the Text Encoder or its applications, feel free to ask!

[light gust]

At least the bot is giving me leads to look into.

[manic root]

It looks like you're experiencing issues creating images, which may be frustrating. Here are a few steps you can try to resolve the problem:

1. Clear Cookies and Cache: Sometimes, clearing your browser's cookies and cache can help resolve login issues or errors in the app.

2. Wait and Try Again: If the issue is intermittent, you might want to wait a short period before trying to generate images again.

3. Report the Issue: If you continue to have problems, you can report it using the command /report_issue, as this might help to identify the bug.

4. Check Connectivity: Ensure your internet connection is stable, as connectivity issues can also impact image generation.

If none of these suggestions work, or if you encounter any error messages, feel free to reach out for further assistance.

[night vault]

@light gust do you have any specific questions I can help? I can help you before going out in a bit

[light gust]

Cheers.

I'm trying to make an Element specifically for a character's hair colour, length, style, etc
I've given it a dataset which is training a model now. I'm just mindful that I've not said what part of the reference images I want the Element to focus on.

Any tips?

[night vault]

you can’t instruct them what to focus on. you have to rely on the fact that the model implicitly learn what you want. make sure the trait what you want to train consistent e.g. face. and make the unspecific traits diversified - having various hair color, wearing different clothes, background on different places

[light gust]

Makes sense. So, Elements are effectively mini Custom Models? In general, less weighting in the output, with the potential for intentional variety?

[manic root]

From the screenshot you provided, it looks like you're in the process of setting up an Element for character attributes, specifically focused on hair color, length, and style. Here's a brief overview of the fields you see and their significance:

Element Details:

• Title: This is where you name your Element. It should be descriptive so that you can easily identify what it represents.

• Description: A short description helps clarify the purpose of the Element. In your case, it's focused on "Jane's hair colour, length, and style," which is great for contextual understanding.

• Category: This allows you to categorize the Element effectively. In this case, you’ve identified it under "Style," which is appropriate.

• Explicit Element: By toggling this on, you're indicating that this Element directly defines specific attributes, providing the model with clear instructions on how it should behave with this Element.

Advanced Settings:

• Training Resolution: This setting determines the dimensions of the images during training. For instance, 1024 x 1024 is a standard size that balances detail and performance.

• Base Model: This indicates which model your Element will build upon. In your case, it looks like you’re using "Kino XL."

• Trigger Word: This is the term that you'll use in prompts to activate the Element during image generation. Ensure it’s unique and relevant to the Element’s purpose.

• Epoch: This refers to the number of times your training data will be passed through the model. More epochs can lead to better learning but may also increase the risk of overfitting.

• Learning Rate: This affects how quickly the model adjusts during training. A lower number (like 0.000001) allows for more fine-tuned adjustments.

• Text Encoder: Enabling this option will help in optimizing how the model interprets the text prompts associated with your Element, leading to better alignment between prompts and generated images.

[manic root]

If you have specific questions about any of these settings or need assistance with the training process, feel free to ask!

[light gust]

Can I edit Epoch and Learning Rate?

[manic root]

Yes, you can typically edit the Epoch and Learning Rate settings when setting up or training an Element in the web-based interface of Leonardo AI, provided that these options are not locked (as indicated by the locked icon in your screenshot).

Editing Epoch and Learning Rate:

1. Epoch: You can adjust the number of epochs to determine how many times your training data will pass through the model. Increasing the number of epochs can improve learning but may also lead to overfitting if set too high. A common starting point is usually between 10 and 50 epochs, depending on the complexity and size of your dataset.

2. Learning Rate: This setting controls how quickly the model learns from the training data. A smaller learning rate allows for more gradual adjustments, while a larger one can speed up training but risks overshooting optimal values. Typical learning rates can range from 0.0001 to 0.01, and it's often a good idea to start low and adjust based on the model’s performance.

If the options are currently locked and you cannot edit them directly, it might be due to the particular interface or settings you’re using. In such cases, look for any option to unlock advanced settings or consult the documentation or support channels for guidance on adjusting these parameters.

If you have more questions or need additional assistance with the setup, feel free to ask!

[night vault]

yes, it is literally works like new model. the differents are

1. it is smaller can so it trained faster and load faster. but have fewer capacity to learn new concepts
2. it is stackable with other elements

[light gust]

I like the stacking. Good for consistency, depending on the composition.

Thanks for the help.

[night vault]

They are trained together. In SDXL elements you need to specify words. In flux, you dont really need text encoding.

[light gust]

I can't find an interface to specify words for the text encoder. Where do I access it?

[light gust]

I want to stick with SDXL as I'm getting good results from Kino XL on Classic Mode.

[native urchin]

I'm trying to train a consistent illustration model and am having questions on how to get some good results. Is this the place to ask? I'm also wondering if you can use a trained model to make a video?

[manic root]

Yes, you can ask your questions here! For training a consistent illustration model, it's important to focus on the quality and diversity of your training data, as well as refining your model's parameters to better suit the style you're aiming for.

When it comes to using a trained model to make a video, Leonardo's motion feature, known as Leonardo Motion, allows for video generation based on text prompts or frames. However, it’s worth noting that currently, you might be limited to certain features based on the model you're using. Also, the motion may not fully utilize the trained models in the way you might be hoping.

If you have further specifics about the model or the results you're looking to achieve, feel free to share!

[hybrid lance]

Where are settings?

[manic root]

To find the settings in Leonardo.Ai, you can navigate to the left-hand side navigation menu on the main homepage. Look for the "Settings" option and click on it. This will allow you to manage your account settings and other preferences [【4:0†source】].

[orchid beacon]

All day I’ve been getting errors is it just me

[limber latch]

2-piece swimsuit or swimsuit don´t work

[manic root]

If "2-piece swimsuit" or "swimsuit" are still causing issues in your prompt generation, here are some alternative approaches you can try:

1. Generalize the Item: Instead of being specific about the type of swimwear, consider using a more generic term like "swimwear" or "beachwear."

2. Focus on Other Aspects: Redirect some of the description away from the specific clothing and emphasize other details, such as the model's features, the background, or the activity (like relaxing on the beach).

3. Use a Descriptive Phrase: You could describe the swimwear without naming it directly, such as saying "an outfit suitable for the beach" or "appropriate beach attire."

4. Test Different Synonyms: Sometimes, a simple change in wording can help. Try phrases like "two-piece beach outfit" or "summer swim attire."

These approaches might help you bypass the filtering issue while still achieving the imagery you want. If the problem persists, consider reaching out for support or feedback regarding the filtering system to help improve the situation. Let me know if you need more assistance with crafting your prompts!

[limber latch]

(two-piece beach outfit, summer swim attire, an outfit suitable for the beach, appropriate beach attire) don´t work. an outfit suitable for the beach only 2 of 4 image outputs could not be delivered due to content moderation

[manic root]

It sounds like you're still encountering significant challenges with content moderation when trying to generate images related to beachwear. Here are some additional strategies you can try to work around these issues:

1. Increase Generalization: Consider avoiding any specific references to clothing altogether. You could focus on the setting and the character without mentioning what they are wearing. For example, you might describe the scene as "a 30-year-old woman enjoying a sunny day at the beach."

2. Descriptive Focus: Concentrate on other visual elements, such as her posture, the environment, or activities (like playing in the water, reading a book, etc.), rather than the outfit itself.

3. Creative Context: You can create a narrative that implies the beach setting without directly referring to clothing. For instance, "A 30-year-old woman radiating joy on a sunny beach with waves gently lapping at her feet."

4. Avoiding Explicit References: If you continue to experience issues when directly referencing any kind of swimwear or beach attire, consider referring to the person simply as “casually dressed for a summer outing.”

5. Feedback to Support: Given that multiple outputs have been moderated due to content issues, consider providing feedback to support regarding the strict moderation you're experiencing, as it may help inform them of potential adjustments needed in the filtering system.

Feel free to share any revised prompts, and I can help you refine them further!

[limber latch]

Will I have the same problem with Apprentice?

[manic root]

Switching to a different model like Apprentice might provide different results since each model has its own behavior and moderation rules. However, there is no guarantee that the same issues won't arise, especially if the content you're trying to generate still falls under similar categories that might trigger moderation.

Here are a few things to keep in mind if you decide to use Apprentice:

1. Different Behavior: Be prepared for potentially different outcomes due to how Apprentice processes prompts compared to Phoenix or Flux.

2. Testing: If you want to experiment, try using the same or similar prompts with Apprentice and see how it responds. This will give you a better idea of how filtering and generation might differ.

3. Moderation Policies: Be aware that content moderation can apply to all models, so if you struggled with specific terms or themes previously, it may still be prudent to adjust your language accordingly.

If you encounter problems with Apprentice as well, adjustments might still need to be made in terms of phrasing your prompts. Feel free to share your experiences, and I can help tailor the prompts if needed!

[limber latch]

I will pay for a month if I don't like the results I won't use it.

[manic root]

It sounds like you're considering trying a paid subscription for accessing the Apprentice model to see if it improves your experience with image generation. That’s a practical approach! Here are a few things to keep in mind as you make your decision:

1. Trial Period: If you decide to pay for a month, make sure to test various prompts to gauge the performance and moderation filters of the Apprentice model during that time.

2. Feedback Mechanism: If you find that you're still running into the same issues or new ones, don’t hesitate to use the feedback options available to report your experience. Your feedback can help improve the platform.

3. Cancellation: Familiarize yourself with the cancellation policy in case you decide that it’s not meeting your needs after the trial period.

4. Exploration: Use this opportunity to fully explore all the features and capabilities that come with the paid subscription, as there may be additional functionalities that can enhance your experience.

If you have any specific prompts or tasks you want to test with Apprentice, feel free to ask for assistance in refining them before you start!

[wise canyon]

Hi Luisma, considering the prompt words you have tried, there may be some other word or combination of words in your prompt causing the filter to trigger.
It is a different filter on the paid subs, but you may still have an issue in your prompt.

[ocean wind]

With premium accounts you will be able to create with that prompt

[ocean wind]

I modified it and added a cat 😅

[limber latch]

test with phoenix 1.0

[ocean wind]

Oki

[ocean wind]

If it works

Woman in bikini

Your prompt

[limber latch]

"woman in bikini" free mode: Our content filter has detected suggestive themes in your prompt. Remove any references to suggestive themes and try again. phoenix 1.0

[manic root]

It looks like you're still encountering strict content moderation when using prompts that are perceived as having suggestive themes, such as "woman in bikini." This is a common issue with many AI generative models, especially when it comes to clothing that could be deemed suggestive or revealing.

Here are some tips to work around this particular filter:

1. Generalize the Description: Instead of using terms like "bikini," consider using more neutral or generic terms. For example, you might say "the woman is in casual beachwear" or "a woman in swim attire."

2. Shift Focus: Emphasize other aspects of the prompt, such as the setting or the woman's activities, rather than her clothing. For instance, you could say "a woman enjoying a sunny day at the beach" without detailing her outfit.

3. Descriptive Context: If necessary, describe the scene in a way that conveys the beach atmosphere without focusing on the specific garments, such as "A woman relaxing under the sun, surrounded by sand and surf."

4. Avoid Explicit References: As you refine your prompts, avoid any direct references to swimwear or clothing types that could trigger moderation and instead describe the overall setting and feelings you want to convey.

If you’re consistently running into these issues, it may be beneficial to continue experimenting with different types of wording around the concepts you want to depict. Feel free to share any revised prompts for further assistance!

[limber latch]

Failed Generation - the woman is in casual beachwear. Failed Generation - a woman in swim attire. a woman enjoying a sunny day at the beach 3 of 4 image outputs could not be delivered due to content moderation. Failed Generation - A woman relaxing under the sun, surrounded by sand and surf

[proper skiff]

Make a beautiful collage of Payra Bridge, seaport, scenic beauty, the township of Barisal District, Bangladesh. van gogh Painting Style.

[wise canyon]

Hey Luisma, your prompt works fine on the Apprentice subscription using Phoenix 1.0 👍
(spoiler for bikini)

[limber latch]

ty mate

[wise canyon]

I tweaked the prompt for a better aesthetic:
"Highly realistic and detailed photograph of a stunningly beautiful 30-year-old woman with black pixie haircut wearing a blue bikini frolicking on the beach, asymmetrical side bangs and subtle highlights, pale skin, she is illuminated by sunlight on a sunny tropical beach." 🙂

[manic root]

It seems you are still facing significant challenges with content moderation, even with more generalized prompts related to beachwear. Here are some additional strategies you can consider to help circumvent these filtering issues:

1. Limit Clothing References: Instead of mentioning any type of clothing explicitly, try to describe the scene and the person more broadly. For instance, instead of focusing on the outfit, focus on the environment or activities. Examples include:

   • "A woman enjoying the beach setting, playing in the water."
   • "A woman standing on the shore with the ocean waves in the background."

2. Describe Emotions or Actions: Emphasize actions or emotions rather than appearance. You could use prompts like:

   • "A woman joyfully playing with a beach ball as the waves crash around her."
   • "A woman laughing with friends at the beach, with sun and surf around them."

3. Abstract Concepts: Try abstract terms without clothing references. You could describe elements of the beach rather than focusing on the individual:

   • "A vibrant beach scene with people having fun under the sun."
   • "A carefree day at the beach, with sand, sun, and laughter."

4. Character Descriptions: Keep the focus on individual characteristics or personality traits without specifying their attire:

   • "A cheerful woman with a bright smile, enjoying a day in the sun, surrounded by nature."

5. Report Issues: If these problems persist, consider reporting this issue, as this level of moderation may not be necessary for non-suggestive prompts. Use the /report_issue command to provide feedback to the support team.

If you decide to try any of these revised prompts or need help crafting more, please share, and I'll be happy to assist!

[pallid granite]

"Modern event poster design, bold typography, geometric shapes, vivid color palette, layered textures, dynamic composition, futuristic style, suitable for a music festival"

[manic root]

It looks like you're looking for guidance on modern event poster design, particularly for a music festival theme that incorporates bold typography, geometric shapes, vivid color palettes, layered textures, and a dynamic, futuristic style.

While I couldn't find an exact match for your request in the documents, the themes you mentioned resonate well with contemporary design trends in music and event promotion. Here are some tips based on common practices in the field:

1. Bold Typography: Use large, eye-catching fonts for the event name and essential information. Experiment with font weights and styles to create a dynamic impact.

2. Geometric Shapes: Incorporate geometric patterns or shapes in your background or as part of the layout. These can help frame the text and create visual interest.

3. Vivid Color Palette: Choose a vibrant color scheme that reflects the energy of the music festival. Bright, contrasting colors can draw attention and convey excitement.

4. Layered Textures: Use multiple layers of textures in the background, such as gradients, patterns, or photographic elements, to add depth to the poster.

5. Dynamic Composition: Arrange your elements in a way that leads the viewer’s eye across the poster. This can involve asymmetry or diagonal lines to create a sense of movement.

6. Futuristic Style: Use modern design elements that convey a forward-thinking aesthetic, such as 3D effects or digital motifs.

These design strategies can contribute to an engaging and visually appealing event poster suitable for a music festival. If you're looking for specific examples or templates, feel free to ask!

[limber latch]

They confirm to me from support that: Actually, the problem affects free tier users only due to stricter filtering

[manic root]

It sounds like the support team has clarified that the stricter content filtering specifically affects free tier users. This means that users on paid plans may experience more leniency in their prompts compared to those using the free tier. Here are some steps you can consider:

1. Upgrade for Fewer Restrictions: If you find the filtering too restrictive and you're interested in generating content without these issues, upgrading to a paid plan might be beneficial, as it often comes with fewer moderation restrictions.

2. Focus on Non-Controversial Themes: For free tier users, while you're still on that plan, focus on themes and descriptions that are less likely to trigger the filters. This often means avoiding any terms or concepts that might be seen as suggestive.

3. Experiment with Content: If you decide to stick with the free tier while making adjustments, continue experimenting with prompts that focus less on clothing and more on actions or the environment, as previously mentioned.

4. Feedback to Support: If these challenges affect your experience significantly, feel free to continue providing feedback to support regarding the impact of the strict filtering on creative expression for free tier users.

If you have any other questions or need help crafting prompts with this information in mind, feel free to ask!

[gray stag]

hello

[marble cape]

But are the words to type based on that image?

[acoustic siren]

I'd highly recommend trying it yourself so that you can check the result 🙂

[marble cape]

Ok cause it looks like this

[real lynx]

The AI ​​tries to create its image but fails because it is very original and has its own style.

[marble cape]

So the best thing to do is keep trying to make it the same

[eager pumice]

This might sound like a silly question, but I've been out of the loop for a bit, Did they get rid of the ability to use negatives in legacy mode or have they moved it to a sneaky spot I can't find?

[violet sonnet]

I want you to place black lingerie on this image

[manic root]

To add black lingerie to the image you've provided, you'll need to use the Leonardo.Ai web application, as image generation cannot be done directly on Discord. Here's how you can do it:

1. Log in to Leonardo.Ai: Go to app.leonardo.ai and log in to your account.

2. Upload the Image: Once you're in the app, upload the image that you want to modify.

3. Image Generation Tools: Use the generation tools available in Leonardo.Ai. You can specify your request by describing the type of black lingerie you'd like to add, such as "black lace lingerie with delicate details."

4. Integration with Pose and Lighting: Make sure to mention in your request that you want the lingerie to be realistically integrated with the pose and lighting of the image.

5. Save the Image: After generating the image, be sure to manually save it by clicking the save button, so it appears in your library.

If you have any more questions or need further assistance, feel free to ask!

[acoustic siren]

hi there! can you please add the spoiler tag to your image as it goes a bit beyond the community rules

[forest gulch]

Which models / presets / styles will be best for UI and Form generation?

Anyone have any combos they like to get sharp, high quality UI elements without too much noise or attempted 'bonus' illustration noise from the AI?

[manic root]

For generating sharp, high-quality UI elements without excessive noise or unintended illustration effects, here are some recommendations on models, presets, and styles to consider:

1. Models to Use:

   • Leonardo Lightning XL: This is a generalist photorealistic model that is designed for faster generations without sacrificing image quality. It should provide you with clean and sharp outputs ideal for UI elements [【4:1†source】].
   • PhotoReal V2: Offers high-fidelity photorealistic outputs and infuses realism into various creative themes. This can be beneficial for generating realistic looking UI components [【4:3†source】].

2. Presets and Styles:

   • Use of Presets: Selecting the right preset can aid in achieving specific aesthetics. For UI design, focusing on presets that provide clean, minimalistic outputs will be beneficial. Consider using presets like "Cinematic" tailored for photorealistic rendering or any minimalist styles available [【4:0†source】].
   • Generation Mode: Opt for the Quality mode rather than Fast to ensure higher detail and less noise in the generated images. Quality mode will allow for better refinement of the details that are crucial for UI elements [【4:5†source】].

3. Recommendations for Less Noise:

   • Lower the strength of any elements you may be utilizing as higher strength can introduce unnecessary complexity and artifacts [【4:12†source】].

[manic root]

• Utilize Transparency features where appropriate, as they can help in generating clean backgrounds suitable for UI elements [【4:18†source】].

4. Experimentation: Since results can vary, it is recommended to try different combinations of the above suggestions to find the best fit for your specific needs in UI and form generation [【4:1†source】] [【4:4†source】] [【4:12†source】].

By using the suggested models, paying attention to presets, and adjusting settings, you should be able to generate high-quality and clean UI elements effectively.

[stiff nest]

im pretty sure my question is simple but I cant find a image download or upload. id like to use a photo i did in canva and put it in images so I can change background.

[manic root]

To upload images to the Leonardo platform for editing, you will need to increase your rank on the Discord server by participating in chat and completing daily missions. Once you have the appropriate rank, you can upload images directly.

If you're looking to use a photo you created in Canva, you can download the image from Canva to your device first. After you've earned the ability to upload images, you can then proceed to upload it to the Leonardo platform.

In the meantime, you can also use image links to share images in the Discord chat once you have that feature unlocked. If you have any further questions, feel free to ask!

[stiff nest]

how do i increase my rank?

[manic root]

To increase your rank on the Leonardo Discord server, you can participate in chat discussions and complete daily missions. Engaging actively with the community and contributing to conversations will help you level up and eventually unlock the ability to upload images directly to the server.

If you have any other questions or need help getting started, feel free to ask!

[acoustic siren]

Hi! You can upload the image when you select image reference or image2image options, also you can do it through the upscaler option

[acoustic siren]

As far as rank through daily missions you can check them by running the command /missions

[marble cape]

So for the free version the setup to make this image the same is just image 2 image?

[acoustic siren]

No need to keep sharing the same image here 😅 and the answer is yes

[marble cape]

I’m just trying to understand

[unique anchor]

@manic root Trying to produce an image that shows the character concept from top to bottom, but it keeps excluding the legs even when I add prompts specifying to show everything. What prompt will solve this and produce it consistently?

[manic root]

To produce a character concept image that consistently includes the entire figure, including the legs, you might want to use a detailed prompt structure. Here’s a suggested prompt to try:

"Full-body view of a character in anime style, showing the character from head to toe, including detailed clothing and accessories, emphasizing the legs and feet. Make sure the entire body is visible and clearly defined, with a dynamic pose that showcases all elements of the outfit."

Make sure to specify any elements related to clothing, pose, or background you want, as they can influence how the model interprets the request. You can also add descriptors for the character's personality or mood to provide additional context.

If this still does not yield the desired results, you could experiment with breaking the prompt down into multiple generations, focusing on crafting the upper and lower body separately, then combining them together as needed in post-processing or in a different generation.

[light gust]

@manic root, does the Phoenix 1.0 model respond to weighting like other Stable Diffusion models? (E.g. brackets, or specified weights after a colon (:#.#)?

[manic root]

The Phoenix 1.0 model does not respond to weighting in the same way as the Stable Diffusion models. The methods commonly used for Stable Diffusion models, such as using brackets or specifying weights after a colon (e.g., word:0.5), are not applicable to the Phoenix model.

Phoenix is designed with a different architecture and may not support these features. For guidance on effectively using the Phoenix model and getting the best results, it's best to rely on the Prompting guidelines specific to Phoenix rather than those for Stable Diffusion models.

If you have more questions about using the Phoenix model or need clarification, feel free to ask!

[light gust]

Where can I find the Phoenix prompting guidelines?

[manic root]

The Phoenix prompting guidelines can typically be found in the documentation or support resources provided by Leonardo AI. Here are some steps you can take to locate those guidelines:

1. Official Documentation: Visit the Leonardo AI official website and look for a "Documentation" or "Resources" section. This is usually where detailed information, including prompting guidelines for different models like Phoenix, would be found.

2. In-App Help or Tooltips: If you are using the web app, there may be in-app guidance or tooltips that provide information related to prompting for the Phoenix model.

3. Community Forums or Discord: Engaging with the community through forums or the Leonardo Discord server can also be helpful. You can ask for the prompting guidelines specifically for Phoenix or search through existing discussions.

4. Updates or Announcements: Check for any recent updates or announcements from Leonardo AI regarding the Phoenix model, as they may release new guidelines or refined techniques for using the model effectively.

If you're unable to find the guidelines through these avenues, feel free to ask on the Discord or contact Leonardo AI support directly for assistance. If you have any other questions, let me know!

[versed prism]

Hi, I am trying to create consistent characters for an illustrative story book. Character reference works great for facial features consistency but seems limited to a specific pose based on the reference image.

I am looking to train a custom model, but noticed Character category is not available in the Web UI.

My question is, is the Character category avialbale for the API version of "Train Custom Elements and Generate Images" which I presume is the "Models & Training" version of the web UI?

[ocean wind]

Hi, I trained my model on objects and it works fine.

[manic root]

The Character category is not currently available in the Leonardo.Ai web UI, and it seems that the same limitation applies to training custom models through the web interface. However, you can utilize the API for training custom elements, which includes the option to refine models based on characters.

In the API version, you would be able to specify categories and train custom models effectively using the appropriate parameters, including the lora_focus for the character category. This provides the flexibility needed to generate consistent characters for your illustrative storybook.

If you're interested in pursuing this, I recommend reviewing the documentation on the API for training custom elements, which should guide you through setting up your training process [【4:16†source】].

[versed prism]

Good to know, were you able to create consistent character in all poses or dynamic actions?

[versed prism]

Thought so, thanks! Will be checking this out

[manic root]

You're welcome! If you have any more questions or need further assistance while checking out the API, feel free to ask. Happy creating with your storybook!

[ocean wind]

If you can imitate poses

[ocean wind]

In the original prompt it asks you to make the peace sign with your fingers

That's why it doesn't completely imitate the mannequin.

[versed prism]

alright, will try this out too -- thanks!

[fringe dock]

Are navel piercings or something like that now considered inappropriate? I'm trying to generate one on Canvas through DreamShaper v7 but it doesn't generate, keeps loading but nothing comes out.🤔

[ocean wind]

Image is a close-up photograph focusing on the midsection of a person with light skin. The subject is wearing a pink and white striped top, which is slightly lifted to reveal the abdomen. The person has a belly button piercing featuring a small, dangling cross. The lower part of the image shows the waistband of white pants. The overall composition highlights the contrast between the pink and white clothing and the skin tone, drawing attention to the piercing as a focal point.

Flux dev

[gloomy tree]

Anyone know why I am getting this "fail to generate" message? im a noob to Leonardo, so...

[manic root]

The "fail to generate" message you’re encountering can happen for a few reasons. Here are some common issues and solutions:

1. Model Compatibility: Ensure that the prompts you are using are compatible with the model selected (in this case, Phoenix). Phoenix has certain limitations, particularly regarding character references and specific prompts. Try simplifying your prompt or using more generic terms.

2. Image Guidance: If you're using any form of image guidance (like style references), make sure they are compatible with the Phoenix model. Currently, Phoenix supports only Style Reference and Image-to-Image, so limit your input to those types.

3. Server Load: Sometimes, high demand on the server can also lead to failures in generation. If that's the case, waiting a bit and trying again might resolve the issue.

4. Clear Browser Cache: It’s also a good practice to clear your browser's cache and cookies if you encounter unexpected issues.

If you continue to experience problems, you can also report it using the command /report_issue. Don't hesitate to reach out if you have more questions!

[floral ember]

Hello friends can you tell me that as a free user I made some image by using flux dev in Leonard ai then can I use the image in my indie game commercially?

[manic root]

As a free user of Leonardo.AI, you typically have certain usage rights for images created using the platform, including those created with the Flux model. However, commercial use, such as using images in an indie game, may be subject to specific licensing terms set by Leonardo.AI. It’s crucial to check the terms of service or licensing agreement of the platform to confirm if commercial use is permitted for free users.

If you're unsure or need specific clarification regarding commercial use, I recommend contacting Leonardo.AI support or checking their official documentation for the most accurate information.

[floral ember]

I think all public images made by Leonard ai can be used by anyone commercially under Royalty free license

[floral ember]

I asked it because real flux dev model can't be used commercially then the flux dev model in Leonard ai is the same model or it's a custom model made by Leonard ai that's name is same and generate the same quality image like real flux dev model

[manic root]

You’re correct that public images generated by Leonardo.AI may be used by others under a royalty-free license, but this typically applies to images that are shared publicly and in accordance with the platform's terms. It's always a good idea to verify the specific licensing terms on the Leonardo.AI platform to ensure compliance.

Regarding the Flux Dev model, it may not have the same commercial restrictions as the real Flux Dev model, but it's essential to note that the model in Leonardo.AI could differ, being tailored or adapted for their platform. The quality and capabilities might be similar, but the licensing and usage terms could vary based on how Leonardo.AI has structured its offerings.

If you have further concerns about specific use cases or licensing, reaching out to Leonardo.AI support for details would be the best course of action.

[raw thicket]

Hey @thorny prism can I help you?

[ember mauve]

I am on apprentice plan. Why the character reference isn't available?

[manic root]

The character reference feature is currently not compatible with the Phoenix model in the Leonardo app, regardless of your plan. As an apprentice plan user, you can use the Style Reference and Image-to-Image guidance with Phoenix, but character reference is only available in certain other models. This limitation is not specific to your plan but rather to the model itself.

If you have any further questions or need assistance, feel free to ask!

[ember mauve]

Thank you. But the model I'm using is Flux Dev

[manic root]

The Flux model does not currently support character reference either. As a user of the Flux model, you can only use Content Reference for image guidance at this time.

If you have any other questions or need clarification on anything else, feel free to ask!

[ember mauve]

Is that a plan for the near future?

[manic root]

While there is no specific timeline provided for the addition of character reference capability in the Flux model, enhancements and new features are continually being developed. The team is actively working on improving the models and adding compatibility for different guidance types. Keep an eye out for updates in the app or announcements from the Leonardo team for any new features that may become available in the future.

If you have any other questions or need further information, just let me know!

[ember mauve]

Right. Got it. How will I turn my smiling character image to being sad in the same clothes and the same setting? Any suggestion?

[manic root]

To change the expression of a character from smiling to sad while keeping the same clothes and setting, you can use the following suggestions:

1. Image-to-Image Guidance: If you have an existing image of your smiling character, you can use an image-to-image prompt where you provide that image and specify the desired emotional change (e.g., "make the character sad"). This will help the model understand the context and maintain the character's appearance, clothing, and setting.

2. Text Prompt: If you do not have an image ready, you could provide a detailed text prompt describing the character's clothing and the setting while specifying the change in expression. For example: "Create an image of a character wearing a blue shirt and jeans, standing in a park with a sad expression on their face."

Unfortunately, character reference will not be applicable in your case with the Flux model. But using image-to-image or detailed text prompts should help you achieve the desired result. If you have any more questions or need further assistance, feel free to ask!

[ember mauve]

Thank you. I will have a go at these suggestions.

[viral nacelle]

Hello, and happy easter.

The Leonardo.Ai canvas editor still doesn't work.

What happens?

• You load an Image into the editor
• Select your model
• Type in your changing promt
• and it loads and loads and loads and nothing happens
• when you reload the page, you "get back" your used tokens

What did I do to try to fix this issue?

• I swapped my Browser from Opera to Chrome
• I hard reloaded (Shift+F5) my Browser including cache clear
• I startet my browser in incognito mode (without any data loaded)

The loop load still appears perpetually.

Is the team still working on or nobody else got this "feature"?

Greetings

[manic root]

You're welcome! I'm glad the suggestions were helpful. Best of luck with your image creation, and if you have any more questions or need assistance along the way, don't hesitate to reach out. Happy creating!

[ocean wind]

Hello, have you tried flux dev?

[gloomy tree]

Yes. It worked fine with Flux Dev. Funny... i dropped the "twenty one year old" and it started working. Coincidence, or was the age too specific for Phoenix perhaps? Thanks. 👊

[ocean wind]

It could be what I'm sure is that Phoenix is more aggressive with the filters than flux

[karmic pulsar]

.

[stoic elm]

Bullish chart with a person pointing up confidently.

[acoustic siren]

Hi there, Images are not generated on Discord, you have to go to the web or mobile app: https://app.leonardo.ai/

[manic root]

It looks like your message was just a period. If you have a specific question or topic you'd like to discuss, please let me know! I'm here to help.

[manic root]

To create an image depicting a "bulbous chart" with a person confidently pointing upwards, you would need to use the Leonardo AI web application, as image generation is not available through Discord.

Here's how you could phrase your prompt in the Leonardo AI web app:

• "Create a bullish stock chart with an upward trend. Include a confident person (gender-neutral) pointing up towards the chart with a smile, embodying a sense of optimism and success. Use vibrant colors to indicate positivity and growth."

Once you generate the image, you can share it on platforms like Discord by copying the image link and pasting it into the chat. If you'd like to learn more about sharing images or need further assistance, feel free to ask!