🙋│image-help

[hybrid idol]

I didnt used it with text

[real lynx]

Perfect delete the text

It is also possible

But remember to use Phoenix to get the same result

[hybrid idol]

Can you paste an example of this prompt?

[real lynx]

Which

[real lynx]

I'm going to look for the prompt

In the majestic style of Studio Ghibli's .Miniature chibi cute Elf happy, face happy with tears falling from his eyes.1Lolita Blonde hair 1girl suit Pijama Totoro ,Bubbles ,.smiling and covering her mouth, multiple pink hearts, dark aesthetic,detailed facial features,smiley,anime, a fantastical illustration of the beloved character Leaving a cup gigant of coffee. adorned with mystical carvings, amidst a lush, vibrant forest with dappled light filtering through the leaves,Giants large colorful letters , "HOLA" .in shimmering gold ink, evoking a sense of whimsical fantasy, surrounded by lush greenery and soft, diffused light, with vibrant hues of emerald, amber, and turquoise, and ultra-high resolution, showcasing intricate textures and subtle shading.

This is another cartel prompt

Hola means hello in Spanish

[real lynx]

A 3-panel gag comic
Person=chibi cute 1Lolita 1girl Blonde Hair, studio giblhi style
Panel 1:
Person: *Smiling * "HOLA BIENVENIDO "

Panel 2:
Person: It would be thinking "MI NOMBRE ES CAMILA"

Panel 3:
Person: Crying tears fall from her eyes "Recuerda que solo puedes crear imágenes en el sitio web Leonardo.ai"

[low girder]

Hello, within Leonardo it is possible through the canvas editor or real time canvas if I want to delete a text, delete something, do it within the application because I don't know and in the end I download the image and I have to go to Photoshop and delete what I want or I do it on the canvas website and there with the help of the AI ​​it deleted what I'm not interested in but from within Leonardo either I don't know how to do it or I don't know if it is possible, for example, to delete a text and then replace it with the background that there is already in the image

[hybrid idol]

Thank you ill
try this

[real lynx]

Hola

Te envío un privado

[real lynx]

https://discord.com/channels/989166677390426132/989166678053113901

Te invito a nuestro canal en español

Bienvenido!

[hardy sentinel]

Hello, i don't understand your question

Would you explain it again step by step?

[open echo]

Perfectly possible 😄

There you go!

https://youtu.be/hKf2kxbgmhE?si=GI6VGj4uO5DJRm_s

[serene crow]

How can I make this picture realistic?

[real lynx]

I really liked it, you can use these models, they are more realistic

I added these words to your prompt

Cinematic masterpiece intricate meticulous

[serene crow]

ok.

[real lynx]

If you want to share your prompt

I can try it myself

So that you do not spend TOKENS

[serene crow]

ok I'll share ye.

[serene crow]

A solemn preacher delivers a sermon during Sunday service at a church, surrounded by rows of parishioners peacefully sleeping on wooden benches.

[sleek glacier]

😄

[real lynx]

Well the parishioners are sleeping....

But in their homes...

But it came out quite real

[obtuse sigil]

john cena be enjoying

[serene crow]

thanks for your effort.

[real lynx]

No one wants to fall asleep

[silent hawk]

Is it just me, or do almost all results created with Phoenix have the artist's signature and are therefore unfortunately not usable. Even adding a negative prompt does not change this.

This is just one example, unfortunately it happens with all prompts, regardless of their type

[real lynx]

Hello Phoenix is for creating posters or letters

That's why shapes appear, maybe take some letters from your prompt.

Although it is true it looks like a signature

I wrote in the negative prompt

Signatures so that they do not appear or letters

Why do I use it for posters

But you can also create images

[silent hawk]

Hi Cami, I added „Signature“ and „Text“ zu my negative prompts, didn’t help, which is sad because I really like Phoenix. Results are really good

[open echo]

If you get a really good picture but there is a detail you want to remove, you can use the Canvas Editor to highlight that part, and re-generate over it.

Saved me a bunch of times of a sixth finger or a logo 🤣

[real lynx]

🫰

It had two signatures

[silent hawk]

Thanks Tuxman. Removing the signature is not a problem, but I don’t want to get in trouble if I share a signed image online, even when the signature has been removed

[raw thicket]

Was the image actually generated with Leonardo.ai on it like that?
I've never seen that before

[fossil flame]

hi, im looking at the IMAGE GUIDENCE, but I dont see any option to select sketch lines similar to Controlnet. Previously I think I was told here there was lots more options once your paying for Leonadro, but seems just the same.....6 options

[raw thicket]

Image guidance currently isn't available with Phoenix enabled.
You'll have to use a different preset, then click the little image icon in the prompt box

[pulsar stream]

yes there are watermark in motion videos now

[raw thicket]

Or switch to legacy mode and the "image guidance" tab will be near the top

[raw thicket]

Have you tried more than one video?
Are all of them like this?

[pulsar stream]

all new motion videos have watermarks.

not sure if it is bug or it will be added permanently

[night vault]

something like watermark is intentional for sure. The issue is lack of communication from what I’ve seen.

[raw thicket]

Yes 100%

[fossil flame]

In Legacy, cant see a line art or sketch option. Whats the one to use?

[raw thicket]

Edge to image

If you put the weight higher, it will stick closer to your sketch.
If you put the weight lower, the AI will have more freedom to get creative.

[tender olive]

hi need help with cat ears no folding !! here is prompt : A vibrant mesmerizing masterpiece rough color pencil sketch in tattoo style of a Scottish Fold cat with radiant white-silver fur, lounging regally in a majestic medieval castle, surrounded by lush greenery and mysterious mist, showcasing whimsical, dreamlike pencil marks that evoke a sense of movement and life, on a textured paper that hints at a nostalgic, vintage feel, with a predominantly pastel color palette infused with hints of rich blues and emerald greens, and delicate, intricate patterns that add depth and mystery to this fantastical scene.

second image the same prompt: A mesmerizing masterpiece rough color pencil sketch in tattoo style of a Scottish Fold breed cat with ears to "fold", bending forward and down towards the front of the head. and striking color fur, a blend of white and silver hues, on high-quality textured paper that echoes the rugged, expressive strokes of the artist's pencils, with visible lead marks and subtle blending that infuses a sense of movement and energy, all set against a subtle, gradient background that hints at a dreamy, mystical realm, exuding a whimsical, mystical aura that draws the viewer in.

[thin tinsel]

The image Leonardo created for you is a composit of images from its training. It hasn't found someone's existing image and served that to you. It's a brand new image for you to use. The reason there are often signatures or branding logos, is because they must've been in the training dataset, and the AI thinks images are supposed to have that on there.
You are free to use your Leonardo.ai created images for your projects, you don't need to worry. 🙂

[thin tinsel]

If you choose Absolute Reality, or Dreamshaper v7 as your model, you will see the line-to-image and Sketch to image options in there.

[thin tinsel]

You have a lot of stuff going on in your prompt. The AI would not know how important the Ears are to you, and how important the other requirements are.

Prompt details
(((SCOTTISH FOLD))), An illustration of a white SCOTTISH FOLD cat with folded ears, laying on pink cushion in the castle grounds, pastel colours. SCOTTISH FOLD.

Preset: Illustrative Albedo
Preset Style: Sketch (Color)

[undone osprey]

why cant i upload my own image??..its telling to use a different preset...what should i do?

[night vault]

you can upload on any models but Phoenix model.

[silent hawk]

That‘s great, thank you 😊 🙏

[wide knot]

I need Something like this

But Leonardo can't do it

I use this prompt: A white tiger woman with a human body. Her skin is covered in white fur with black stripes distinctive of a white tiger. She has a slender, athletic figure, with tiger ears and piercing ice blue eyes. She is wearing a light blue shirt that leaves her shoulders exposed, highlighting the grace and elegance of her figure. The shirt has rose petal patterns on the chest and abdomen, adding a delicate and feminine touch to its appearance. He wears grayish, tight-fitting pants that complement his outfit and give him a modern and stylized look.

And I use this image for reference:

And Leonardo give to me this

What am I doing wrong?

[night vault]

what type of reference? style?

[night vault]

I’d try describe subject as “anthropomorphic white tiger with female figure” to the first sentence

[real lynx]

@nova sonnet

In the beginning there were the druids

And then the crows came with the owls

[night vault]

@nova sonnet I have been asking you few times to share setting so we can start understand what could be improved. If you only rant then I cannot have enough information to understand issue and help you here unfortunately. 🙂

At least, I know one thing that counting is real limitation in generation process. There is no way in their process to ensure result end up with exact 5 or 8 entities count.

[wide knot]

Yes

[wide knot]

Okay I will try

@night vault This created Leonardo...

The same thing :'v

[night vault]

phoenix model?

[wide knot]

Anime model

[night vault]

AnimeXL? or which one

Let’s try with really minimal prompt

anthropomorphic white tiger with a female body. blue tribal cloak

[wide knot]

9k

[night vault]

my crapy real time canvas

[wide knot]

Yes that

[wide knot]

I think that Leonardo ai don't do that for one reason

Well I don't know how to create a tiger woman

[tender olive]

thank u

[wide knot]

@night vault do you know what can I do ?

[night vault]

what is your prompt now? also did you use elements?

[feral mica]

try fewer subjects in the prompt, you could later add them using Canvas Editor

[wide knot]

@night vault The one you told me to use: anthropomorphic white tiger with a female body. blue tribal cloak, and I don't use elements

[feral mica]

did you see this?
#💬│general message

Is this kind of what you are looking for?

[wide knot]

No

Like this

[wide knot]

But I don't know why Leonardo can't do that

[night vault]

Let me have token recharge and then will try to get something like this.

[marble cape]

need asisstant

where is leonardo phoenix?

i dont see it\

?

[feral mica]

people are giving their time trying to help you, there is no need to be rude and we don't accept this lack of respect for the others in here

https://tenor.com/view/bane-no-banned-and-you-are-explode-gif-16047504

[feral mica]

hello!

[marble cape]

i dont see option leonardo phoenix

[feral mica]

you need to make sure you turn off the legacy mode

[marble cape]

oh ok thanks

[wide knot]

Okay thanks, I'll wait

[feral mica]

I keep trying, not there yet

[marble cape]

is quality free in leonardi ai not really huh

[wide knot]

Thanks for trying bro

[marble cape]

Steven p was helping me that other timne

qaulity is not free

[open echo]

And Leonardo Phoenix can count 🤣 (8 druids, all looking away)

But… what do we know, right?

[feral mica]

no, that is a paid feature

[marble cape]

is steven p around he was showing me that other time

[feral mica]

have you tried Leonardo Phoenix?

[feral mica]

I believe its a bit late for him, but let me know what you need assistance with, we'll do our best to help you

[marble cape]

where is upload when legacy is turned off

[open echo]

Hey there @fiery vine 😄

We believe you, no need to post full frontal nudity (even with spoiler 😅)

Have you added “nudity” in the negative prompt?

[fiery vine]

no I didn't try that!

[feral mica]

Phoenix does not accept image guidance, yet, for the rest of the models

[open echo]

Oh right.. Legacy “off”… my bad 🤣

[marble cape]

oh

[fiery vine]

Where is that option located?

[open echo]

Look in the left bar, at the bottom for “Advanced options” and turn on “Negative prompt”

It will add a second line to the prompt bar

[marble cape]

sucks cause i wanted to try myself to make this design i made more anime like

[fiery vine]

Thank you! Now I am getting somewhere.

[marble cape]

also wanted to try myself on to make undernath the two lotuses red and blue strings more like an aura that makes one of the eye on the left red and right blue

[open echo]

Always a pleasure 😄

[marble cape]

also is this the type of prompt i should put in? Anime character that has two lotuses on each side of her hair and underneath the two lotuses has a red and blue strings that is attached underneath of the two lotuses has a mystical auras on each side of her hair and also effect the eyes. On the right eye has a red aura and on the left eye has an blue aura.

if you can assist me on this please

[night vault]

I suppose add more descriptive of their cloth also help

[marble cape]

can anyone fill me in on what to do?

cause i want to try it myself

[feral mica]

your prompt may be a bit too specific IMO

[open echo]

This type of prompt will work better with Phoenix, but for now it doesn’t support Image Guidance

Canvas Editor would be a better fit for now to change the color of each flowers

[marble cape]

but you understand what i want?

[feral mica]

mr @open echo has a good youtube video from a well known guy who gives you a hint on how to prompt better

[marble cape]

canvas editor ok ill give it a try

[feral mica]

check this for Canvas, https://intercom.help/leonardo-ai/en/articles/8092594-canvas-2-0

[marble cape]

then what should i put in based on the prompt i explained?

[open echo]

Something like this:

https://youtu.be/hKf2kxbgmhE?si=RinFNkU_9XSuBflC

Minimal, to the point

[feral mica]

I would go for the character with 2 lotuses on each side of the hair, then try to do the rest on canvas

[marble cape]

oh man ill try cause underneath those lotuses are thoses red and blue strings and i wanted it to be an aura like sensed and also that makes her eyes glow red and the other blue

right now watching video

[real lynx]

Look at what they are doing at the academy, maybe it can help you.

https://discord.com/channels/989166677390426132/1230345629776085023

[marble cape]

Right now just started

[real lynx]

You are with the canvas

[marble cape]

Yeah canvas editor someone recommended

Um

[open echo]

Just a quick exemple … it can do better but it’s just to show a quick edit 😅

[marble cape]

Hmmm

First what’s the size for my canvas?

I just started

[tame jacinth]

Hi, I also have trouble to use Canvas Editor, where can I find techniques? Thanks

[feral mica]

https://intercom.help/leonardo-ai/en/articles/8092594-canvas-2-0

[tame jacinth]

🙏 thanks

[feral mica]

sure!

[marble cape]

Ok now I see is this in the center is it good?

[feral mica]

looks good, you can also expand the size of the square with the image dimension options on your right

[marble cape]

what is the size reccomened for this design?

[feral mica]

its a matter of trying what is best for you, i would make it a bit bigger If I where to use the inpaint feature, because that takes into consideration the context that surrounds the generation

[marble cape]

and is this the right prompt to generate how i want it to look? "red and blue strings aura like and left eye blue and right eye red aura like as well".

like show me the size and all so i know circle it if it may be

oh and trying to make red and blue strings longer

[feral mica]

I would approach each eye in 1 generation, if I get something I like I'd use the same prompt for the next one

check this video, may help you: #1138504217087193118 message

[marble cape]

what video?

the one you posted?

[feral mica]

yes, its a small practical guide on how to use the tool

[marble cape]

and wht is the image dimension for my design?

512x?

[marble cape]

?

[marble cape]

hello?

[marble cape]

i need help with the setup

[subtle gulch]

The dimension there refers to the dimension of the workbox.

[marble cape]

Ok well now I’m making some improvement on my design now I’ll let you know when im
Done

[subtle gulch]

From what I've read above, it seems you want to change your draft into a better anime style.
How much do you want to alter the original design you have?
If you want to alter small portion of the image, then Canvas Editor is the most effective.

If you want to rework the whole image, it may be better to use img to imge/ sketch to img / depth to img from image reference.

[marble cape]

Can I add you in the meantime

[subtle gulch]

sure

[marble cape]

Thanks man all love

[subtle gulch]

Using Content Reference, Preset: Concept Art (Stylised Illustration), with Model: AlbedoXL, LightningXL and AnimeXL

Prompt：An anime style anthropomorphic tiger lady, adorned with tribal attire in a tropical forest, chiaroscuro, sgraffito oil painting style, studio light

I think AlbedoXL is your best bet. AnimeXL tends to generate anime human characters in most cases.

Also from AlbedoXL

[obtuse sigil]

hetrochromatic eyes really not possible with other models apart from phoenix

it almost always returns same color eyes

[marble cape]

Oh ok cause I was going to see if they could generate one eye red and the other blue you like glowing because of the lotuses strings that made out of aura like you know was hoping I could generate that idea as prompt

As you can see now what I’m doing is I’m making the strings little longer

Just experimenting

[obtuse sigil]

that kind of prompt only phoenix can handle and tbh it isn't even near best model to generate anime style.

sooo. only canvas editor can

[fiery vine]

what do you do when you apply motion but no motion was applied?

[obtuse sigil]

repent (jk)

[fiery vine]

it kind of feels like when you put your money in the vending machine, but it gets stuck.

[dire river]

So a question for you all, trying to do a image of a person entering the apartment with the camera in this type of shot as seen this screenshot, but the person head is titled away from the camera as he enters. So far everything has him facing it in some way. Suggestions?

[marble cape]

Yeah sucks phoenix isn’t on the free version

[subtle gulch]

It happens when your image does not have enough data for the AI to recognise to provide motion.

[fiery vine]

I tried it again but uped the motion strength. That worked in this instance.

[subtle gulch]

Even though Leonardo's Motion does not have any means to control, the basic principle is the same:

• the higher resolution the better
• the more realistic the better
• the clearer depth of field there is the better
• If there are clear elements that AI can recognise the better
  (like road, sky, sea, spiral, motion lines, etc.)

[subtle gulch]

One way to do it is not to put your character in the front of your prompt. AI will always treat the prompts in the front as the main subject and as the main subject the character will tend to look at the camera.

[open echo]

To the best of my knowledge, Phoenix is available with a free account

[subtle gulch]

Try describing the detailed setting first, and then place the description of the character in the later part and make it shorter.

Prompt: A living room of an apartment, cozy furniture, dim studio lighting, a man entering from the door looking to his left

[wide knot]

Thanks you and @night vault, @feral mica helped me a lot

[marble cape]

Oh it’s not a subscription thing for the phoenix option?

Oh thanks!

[open echo]

No, only the things with a small diamonds beside it

[marble cape]

Thank you

[pure ivy]

Prego, amico mio! If you have any more questions or need further assistance on your journey here, feel free to ask. 🎨🖌️
Our apologies if this message did not answer your question.

[dire river]

@subtle gulch so then if I wanted the first image to basically be the entrance way and camera angle, I would discribe it like you did, then put the character in after (second i ref image). Am I understanding that right?

[subtle gulch]

Yes.
If you want the scene to look exactly like the first image, you can simply use that as content reference as well.

[dire river]

Want close to the setting look and I want the guy to be entering, but head turned away from the camera. Sort of that short when someone's looking away as they open

[subtle gulch]

Just remember, if you use a character reference on an image with two characters, it's very likely that both characters will look the same.

Cinematic movie still, A living room of an apartment, cozy furniture, dim studio lighting, a close up half body shot of a man entering from the door looking to his left

[dire river]

And ya, I'm ok with that-the one closer to the door and the second image are the same person. I did find that if you turned on negative prompts, it did stop AI from taking into account the two characters in the setting image

[dire river]

Also what does high, low, and medium mean for character and setting options?

[subtle gulch]

It's how much it resemble the facial features of the reference

For example, if you want to change their expression or add facial hair, you may want a medium or low for the setting.

[dire river]

So high in the setting option means it would try to closely resemble the apartment walls and door?

[subtle gulch]

I thought you're talking about character reference. Character reference only focuses on the facial feature of the character, not the surroundings.

Content reference will retain the composition of the image, but the details can be changed.

[dire river]

Oh, was, sorry, my mind just combined the two. The apartment screenshot is what I want the AI to try to get close to for how the enteryway looks

[subtle gulch]

Then Content reference is what you want.

I did this with content reference just a while ago

the posture and composition stay the same, but everything else can be different.

[dire river]

So would still be setting first (content refence), then add my character (character reference) in order of discription/prompt layout.

[subtle gulch]

I think you can simply combine the two images you have with content refrence on the first image and character reference on the second.

You shouldn't need to prompt much differently than before unless you want to include other details or make modification.

[dire river]

Well as said, want the appartement entry way and the guy's look, but changing the pose he's end (entering the apartment with door open, hand on the doorknob, and head looking towards the door at a bit of an angle, but body itself is still facing forward)

[subtle gulch]

oh, then you can't use content reference

[dire river]

😦 Guess have to manually discibe that entryway

[subtle gulch]

another way of doing so is to use Canvas Editor

Then you can maintain the entryway, and just edit in the other characters you're happy with from other images.

[dire river]

Got a example?

[subtle gulch]

okay

First I use content reference and character reference to generate this image

[dire river]

Neet. Right now seeing if I can erase the two, and then 'add' the missing aspects of the foor and walls

[subtle gulch]

hmm... I found that you original scene taken from YouTube has quite low res... you may want to regenerate one with higher res before you edit them.

[dire river]

Ya, sadlly not sure where else to get one as do not own it, show's not aviable in canada online far as I can tell in good quality

[subtle gulch]

I get this one with content reference and character reference again

[dire river]

Ok, so that good, so may still be not lost. Wil have to find a way to not screenshot it, lol.

[subtle gulch]

Good luck with your generation
I may take a few trial and error
but once you understand how it works
it'll work better for you 👍

[dire river]

Thanks for all your help

I guess my final question, does the high on characters affect their posture too or just face resemblnce?

[subtle gulch]

just face resemblance

I set both reference as high, and you can see even the expression is similar to the original

If you want it to deviate, set it to a lower value

[dire river]

Hn, well again, if I tell the ai to have him look away should not matter but maybe the high has it assume you want the face visable

[subtle gulch]

The first one is set on high as well and you can see the face's angle can be turned together with the content reference

The second one doesn't work mostly because of the low res.

[dire river]

So does canvas just erase stuff and can't be told to 'assume 'x' goes to distance. In this case, since the cloest figure in the appartment image is cover lower quartor of the door, to 'fill' it in where you know it goes based on what's visable?

@subtle gulch So based on my experimints, seems to be a hard no.

[subtle gulch]

You can use the inpaint funtion of the Canvas Editor to fill up the image with details you want. Just make sure you have adequate references within the workbox for the AI to generate.

[subtle gulch]

For instance, here I just paste the door from the other scene.
I erase everything except the door.
I don't need to worry about the uneven edges because I can simply mask it and inpaint with the prrompt "background" to smoothen the edges and merge the images.

[dire river]

I'll Do that and some other stuff.

[subtle gulch]

Now I paste the head of the second image on and do some more inpainting

And trim the extra part

[dire river]

@subtle gulch so masked the parts I don't want as seen here. So now if I remove them, can the AI know to take into account based on remaining image?

[subtle gulch]

Can you see the workbox in my image?

The AI can only work on the portion within the workbox.

[dire river]

Yes

[subtle gulch]

You need to make sure that there is enough information for the AI to deduce the part you don’t want.

A very common prompt I use to remove items in images is simply “background”

[dire river]

I say there is, most of the door is there and with the new image form another scene most of the other wall. Don't know why it would not

Close

I do think the quality of the images is why its doing that

[subtle gulch]

Right, resolution matters quite a bit

[dire river]

Ya, aslo seemingly the magic touch on prompt to get more than a forward facing pose

Like this is what the AI seems to really like to do with the face

And this is the prompt: A main entrance for an apartment from the resident's side, a short green wall to his left and another short wall to the right, dim studio lighting, a man entering from the door with head looking to his left with his hand on the doorknob. Full shot.

@subtle gulch sure you didn't put any negatives in your examples with the character reg image?

[subtle gulch]

Not at all

[dire river]

😦 Then there a tick in either my setting or prompt.

[subtle gulch]

Which model are you using?

Phoenix would give you the best prompt coherence.

[dire river]

Was cinimatic, now trying default/dynamic and seeing what contrast does

Will try that

[subtle gulch]

If you can’t get the style you want on Phoenix, you can still use that to get the base image for content reference.

[dire river]

Guess image ref can't be used in phoenix right now

[real lynx]

It can't be done

[worldly blade]

How do you get "cute emotes" to generate a cohesive character with different expressions. It keeps giving me multiple characters with too similar of facial expressions.

Prompt: (FOCUS: Cute chibi emotes of a gothic princess with dark brown hair and large green eyes)...(BLACK BACKGROUND)

[subtle gulch]

It’s not available yet, so for the time being you need to take the longer route and do an extra content guidance.

I think they’re working on it.

[wanton belfry]

this suddenly reminded me how simply this can intergrated to VFX work if AI tech is stable with it

removing wires etc would be so easy from scene

[wanton belfry]

put it on low strength

less than 2-3

do not put on landscape format

keep 1:1 or portrait ratio

[white delta]

i want to create a promotional poster for my twitter space

Create a vibrant and eye-catching promotional poster for an upcoming Twitter Space event. The theme of the event is "Memecoin Bonanza," and it will be hosted by Poseidon. The event will feature guest speakers who are experts in cryptocurrency and digital finance, covering topics such as the rise of memecoins, investment strategies, and the future of digital currencies.

here's the prompt i used

[unique anchor]

does the 'image to image' reference take a while to process? the generate button has been grayed out for a few minutes here

nevermind. i think i somewhat know what happened. ended up refreshing the page and redoing it

[rapid shore]

Hey gang - I need some help. I am trying to dial this prompt in and have consistently had poor luck. I am trying to apply a hand drawn look to a lawnmower (see reference image) and need high accuracy for the output. I've tried all manner of reference combinations. I get so close but the accuracy is off just enough to make it unusable. Also, by accuracy I mean physical representation of the forms and elements that make this mower unique, not things like the color, etc. Does anyone have any suggestions or recommendations for me? Thanks in advance for the help.

[livid inlet]

https://discord.gg/leonardo-ai?event=1255186107297955951

[open echo]

Hello there,
I see you want a banner to showcase an event.

The prompt is vague 😬

Even if I’d like to do it in Adobe Illustrator, I wouldn’t know what Colors or style to start with 😅

There’s also A LOT of text you want added. Leonardo Phoenix can do it, but expect a couple of writing mistakes that will need to be taken care of.

[open echo]

Hello there! 😁

Keep in mind image promoting is not ChatGPT. It instead takes in keywords and pair them to shapes, Colors and styles.

Remove all commands form your prompt and keep it strictly to the concrete result you are looking for “hand drawn orange lawnmowers”

[rapid shore]

That's helpful, thank you. I'll give it a shot

[worldly blade]

What do you mean by low strength?

[wanton belfry]

lower intensity

there is a bar

[worldly blade]

I understand the concept of low strength what I don't understand is in reference to which part of the process. Guidance Scale? Cute Emote Element? UU? I'm using Legacy mode. Nothing I do comes out well in the new mode.

[wanton belfry]

cute emote element

[worldly blade]

Mine is already set to 0.30

[wanton belfry]

then do .20

if that doesnt work, you might have to alter prompt

[worldly blade]

Is this with or without an image guidance? I can't get it to do emotes at all if I don't use an image guide. It just pops back a single character shot.

[wanton belfry]

without Img guidance

its very tricky to get img guidance right witht this subject unless you have perfect image

of course if you are premium then there is a way, get a good image and get a content reference of anime figure you want and then prompt for it

big hit or miss still

however, more control

its generally good idea to try without any img guidance or too much additional stuff first to see if the model can do it or not

this is without the Element

I admit it can be very tricky getting single character images out of cute elements, I have done it multiple times but I dont know if it will extend to your subject too

[night vault]

try simple icon instead of cute if you dont like character sheet. start with 0.3

[wanton belfry]

also possible...might be less cute

but if you add "Cute" in front of prompt, it might somewhat work

[worldly blade]

Ok thank you I'll keep trying

[tame jacinth]

@real lynx How?

[night vault]

it is what it is. It is new model.

[real lynx]

I told you that in Phoenix you cannot use the image upload function

For the moment I hope that at some point it will be great.

[real lynx]

Phoenix is great

[tame jacinth]

you are great ✨

[marble cape]

ok i managed to make the strings a little longer this time maybe the aura effects for the lotuses red and blue strings will work this time if someone can just give me the prompt for it

i put this prompt but it didnt give me the details i wanted.. "Anime character and has two lotuses on each side of her hair and underneath the two lotuses has a red and blue strings that is attached underneath of the two lotuses has a mystical auras on each side of her hair and also effect the eyes. On the right eye has a red aura and on the left eye has an blue aura."

look i have these drawings i did and im just trying to make it better

just want to try it myself

[dim zodiac]

heya! Which models are you trying this in?

[marble cape]

Sorry just watched a video more on canvas editor

I’m trying canvas editor I’m just trying to make this particular design I made more like anime

This time I made the strings a little longer. I saved the other image just in case this doesn’t generate the exact way but in more anime like I know it’s phoenix one but I would have to subscribe I have others but for now this one since this was my first design I did years ago..

I wasn’t going to do any of this as I got other important things but I only did all this out of my own boredom

[open echo]

It's coming along well.

Remember that if you want to use canvas editor, it's way better to focus on one element at a time

(Or use Image Guidance with either Depth to Image or Content Reference, depending on the model and UI version)

[dire river]

@subtle gulch So trying something new. I have wondershare fimora so and canvas, so think the AI should be able use these for setting reference now

[dire river]

Took a while to edit it though

[subtle gulch]

Very well done

[eternal urchin]

Hello, any tips on how to really limit the amount of characters generated ? I need a composition of 4 characters but I keep getting less or more !

I'm trying to generate 4 but not only does it create more than 4, but it's also mixing which character has what feature

[whole trellis]

I’m working on creating a 4K image and I’m really happy with how the upper part has turned out and I want to keep it like it is. However, I’m having some trouble matching the consistent style in the lower half. Does anyone have any advice or tips they could share? I would greatly appreciate any help! Thank you so much!

[lyric latch]

are you working with phoenix? Because I have no problem having only four charaters with phoenix

[eternal urchin]

Ahh right, no I'm using leonardo anime xl rn

[lyric latch]

Maybe use Phoenix, and add some "anime-inspired-style" in the prompt?

[eternal urchin]

I'll try !

Man the details are good and on the right character but the result isn't even close to leonardo anime xl 😦

[night vault]

one way is use some image reference like img2img or depth2img at very low weight 0.1-0.2. This can help ai put character in place of original image but then allow the face and clothing to follow what you prompt for

[zealous wave]

Hi AR. I am not sure I do understand what is your expected results, and what is not matching in what you get. Can you precise?

[night vault]

When you extend you canvas do you inpaint half of canvas (and leaving other half with image?) ?

[open echo]

Hey there 😄
Historically, image generation based on Stable Diffusion is based on keywords getting compared to patterns, Colors and shapes. So counting, or precise layouts, doesn’t work very well in older models.

On the other hand, these more mature models are also more finetuned. Which mean they have a better defined style and will provide better results within their area of expertise.

That said… solutions… 😄

You can use the image generated in Leonardo Phoenix and use Image Guidance with Leonardo Anime XL, AlbedoBaseXL or Anime Pastel Dream (in Legacy UI) to guide the generation with the positioning you had in Phoenix but apply the style from the other models.

[eternal urchin]

Thanks for all the explanations and solutions, super useful !

[open echo]

Quick exemple from Phoenix to Leo Anime XL (mind the legible text is also a pretty new detail that older models cannot understand)
((And it blows my mind Phoenix does it so well!!))

[eternal urchin]

Yes, that should greatly help

[solemn shale]

I have two images. One made by ChatGPT and one made by Leonardo.ai
But leonardo ai somewhat doesn't listen to my prompt as ChatGPT does. Any idea?
Prompt: A medieval realm of memory, where dark tentacles and eerie eyes emerge from everywhere, creating a sinister and unsettling atmosphere. The landscape is a twisted grassland, with patches of dark, twisted grass interspersed with ominous shadows. Dark tendrils weave through the grassy terrain, and countless eyes of various sizes peer from every surface, watching and waiting. The overall ambience is dark and unsettling, evoking a sense of dread and unease, while the medieval elements add a touch of ancient mystery and foreboding.

I'm expecting something like the chatgpt's, the first one

[subtle gulch]

If you use the Phoenix model, you should be able to get the tentacles and eyes correctly.
However, as you've mentioned about grassy terrain, you in fact get it from Leo.

Dall-E/ChatGPT is the one which did not follow your prompt completely.

[solemn shale]

lemme try phoenix really quick, thanks for the recommendation

it worked actually, thanks!

[potent hound]

Hey, guys. Quick question. Is there a tool in Leonardo AI for reverse engineering prompts? Basically, an image-to-prompt generator? I know there's an image-to-image workflow but tutorials that I watched on Youtube didn't have examples where external images were brought in. I want to use it to learn genre terms and how to be more descriptive.

[open echo]

Image to prompt is REALLY fickle because looking at a starting image it could have come from pretty much anything… it depends on the model, the randomness of the seed… god forbid they used Image Guidance (like Style Reference) because then no prompts will give the same image back 😅

My best suggestion is to read on artistic and photography terminology. Exemple: can’t prompt for “Bokeh” if you never heard the term before

[potent hound]

Thanks for the reply. I'm an artist, so I do know a lot of artistic and photography terms, but when it comes to the subgenres, there are so many to learn, but the AI generators seem to be able to work with them. I'm very new to AI, and I have a lot to learn.

[open echo]

We’ve all been there 😄

Prompt-based systems is indeed all about learning HOW to ask something 😅

Then again, some models will know different genres, or even recognize them under different names sometimes.. so as much as I would love to say “this works 100% of the time”… there is still no standardized formats for prompts and training.

It’s all so new !

3 years ago, image gen looked like this and we were all blown away by it 😅

Can’t wait to see where it will be in 3 more

[subtle gulch]

There are plenty of resources for that. On Leonardo, you can simply visit the community feed and see the prompts others use.
I think any platform providing ChatGPT 4.0 service can carry out the image-to-prompt function.

However, just like what Tuxman said, the same prompt can lead to different results from different model, not to mention different image generation platforms.

If it's for learning terminology, I think that will suffice.

[rough glen]

Maybe to add a question to the current conversation:

Should prompts be Keywords or Conversation. Which makes for better prompts with Leonardo?

[livid inlet]

Keywords for non Phoenix, more of a chatgpt style for when using Phoenix imo 🫂

But short prompts with auto off still work great with Phoenix so I'd do a little trial and error to find your preference 👍

[rough glen]

Thanks! I've been looking at a lot of tutorials on YouTube and they are using the (now) old interface. Yup, experimentation is on the agenda.

[livid inlet]

I've only been using Phoenix since it came out, I do need to use the uni upscale sometimes to get the best quality but I've been loving it.

@brisk wing this is the best place to get help with images 🫂

[brisk wing]

how to create this type of thumnail can you tell me

@livid inlet

[livid inlet]

I'm not sure you will get a copy of that but you might be able to get the same feel using the new Phoenix model. Let me try a few things and I'll see what I can get.

[brisk wing]

these type of thumnail i want to make can any one help me

[brisk wing]

ok i am waiting here

[livid inlet]

Got this so far, I'll try and add a character now

[livid inlet]

??? this any good ???

[brisk wing]

wow so good

[livid inlet]

A stunning illustration of Scary Thumbnail in an old colour graphic novel style, depicting a Man In A Baseball Cap, standing eerily in front of a darkened, mysterious background, With Large, bold, worn-out words saying "Scary Stories" in a retro, distressed font, rendered in a cinematic, nostalgic aesthetic reminiscent of classic 80s pulp horror, with a dominant palette of deep, foreboding reds and blacks, accented by flickers of eerie, pulsing orange, set against a backdrop of weathered, vintage paper texture, with ultra high resolution and intricate details.

[brisk wing]

you are the best

[feral mica]

dude, awesome!

[livid inlet]

Use that prompt with Phoenix and illustration style 🫂

You just need to change the words and the character to suit

[livid inlet]

Phoenix makes it easy ❤️‍🔥

[brisk wing]

i have a question how you just separate text and man in Your Thumnail

[livid inlet]

Put text inside "" 👍

It will be some luck as well

[brisk wing]

You Just Exectly Copy Man Face Samilar to Thumnail

how i can do it

[livid inlet]

I got some like this but I still think they would make good thumbnails

[brisk wing]

no way

amazing

[livid inlet]

That was luck lol

and Phoenix doing its magic

[brisk wing]

Haha Could You Please Make a short Video For me 😀

[livid inlet]

Jump into https://discord.com/channels/989166677390426132/1082641517245628526 and I'll show you what I did on share screen if you want?

[paper iris]

Hi, Can someone assist me or possibly share link of some tutorial on how to generate image of custom size. Specifically 3000*3000

[livid inlet]

We don't go that high, But if you set up a 1:1 at the highest res then do an upscale you should get more than that

[paper iris]

I am trying that but upscaler is not letting me add my own values

Is that how it's supposed to be

[livid inlet]

Yes, this bit shows the output size, there is no way yet to set your own output but the devs are always making upgrades so I'd put a #💡│suggestions in maybe?

[paper iris]

Oh no that's no good but thanks for helping. I will put up suggestion for devs,,,

[livid inlet]

Thinking about it the universal upscale does do a x2 so you can make you images 1500x1500 then run them through the upscale to get 3000x3000 👍

[paper iris]

That would work but even 1500 is not there or maybe I am still new and trying to find that

I am seeing 1536

[livid inlet]

Damn yeah leo works in sets of 8 so no 1500x1500, sorry for that

[paper iris]

No problem 🙇‍♂️

[marble cape]

Hello

[proud summit]

Hey there!

[marble cape]

what is draw mask in canvas editor?

what does it do?

[open echo]

"Draw Mask" will tell the canvas editor "I want to change THIS part"

You draw the pink zone where you want the modification to occur, place the purple square around it, enter the prompt... and voila! 😄

[marble cape]

Cause I’m trying to copy my original artwork and implanting on the left side

You know what I mean can you run me through it step by step?

[night vault]

I’ll be out for a sleep but hope this stream should get you some idea how to use canvas.

https://www.youtube.com/live/T_8R4PrqnvM?si=grzzcNOEyQXXe1O4

[open echo]

Ah!! Going through old guides I found this one with lists and lists of keywords ideas, styles and subgenres to try

I hope it'll tickle your creative flow 😄

#1087753675822342205 message

[open echo]

While I can absolutely drop in and answer some questions, at the moment I can't do one-to-one mentorings 😅

But from what I'm seeing, if I undertand what you are attempting to do, Canvas will not be used to transfer a style from one picture to the other.
You would have to use Style Reference or Image to Image for that. 👇

[marble cape]

Ok so it’s image generation then ok

I can tell you like my original design

[marble cape]

I don’t need help I think I got this

Anymore

Wait if I can do that I’m I can sketch but if I do image to image what would be the prompt for it when cooling the same outfit of my design

Did I do this right? It’s still generating

Taking long to generate but should of added more in prompt

[subtle gulch]

When you do img to img, the AI will take the original image (that is inside the workbox) can reference to regenerate a new image based on your prompt.

You can imagine this like a person is squinting their eyes really hard on your image
and then try to draw a new one based on that according to your description in your prompt

[marble cape]

So do I write this as my prompt grey outfit, light brown hair and two lotuses on each side of her hair with red and blue strings attached like that?

[subtle gulch]

What you'll get is two individual characters that are entirely different from the original (becuase of the low strength of 0.4)
but will have a similar color palette and contour

[marble cape]

cause right now im sketching color it the outfit and doing the hair

oh i see yeah it wont transfer the colr style to the one on the left side

[subtle gulch]

Of course not. It's not how the AI works

[marble cape]

well right now im sketching the outfit grey now

[subtle gulch]

If you want to transfer the style of one imge to another, use style reference in image generation

not in Canvas Editor

[marble cape]

As you can see

[subtle gulch]

If you're subscribed, then you have access to content reference and you can use it together with style reference to generate what you want much easier.

[subtle gulch]

no... It won't work like that

[marble cape]

Oh man

[subtle gulch]

Do you know how to set the transparency of the grey colour?

[marble cape]

So that in order to transfer my art style to this image I would need to subscribe then?

[subtle gulch]

Make that a bit more transparent before you brush it over so that the texture and pattern underneath is still there.

Then you can change the red and black outfit to grey

[marble cape]

How do I make it transparent

I’m lost here

[subtle gulch]

In one of the images I've DMed you, I showed you how to extract colors from the image.

You can see at the bottom of that pop up window there is a slider for transparency.

[tender olive]

hi need to be number 4 on tshirt (prompt:A vibrant, masterful rough color pencil sketch of running Boss Bit Baby in tshirts with big number 4 in red sneakers , situated amidst the rustic charm of tesla factory , wearing a gleaming steel cap adorned with the bold BTC logo, in a chaotic burst of energy, surrounded by scribbled pencil marks and visible paper texture, exuding a playful, edgy, and whimsical aesthetic, with a palette of radiant gold , electric black, and sunshine blue , evoking a sense of frenetic excitement and joy, with subtle coffee shop background nuances and scattered hamburger crumbs..)

[night vault]

Could you try with Phoenix model?

[tender olive]

will try

[tender olive]

done i phoenix mode thx bro

[night vault]

are you happy with that?

[thin tinsel]

You can potentially do this canvas editor. If you let me know more about the types of image you're creating then I could do some instructions.
You can definitely do images of 1504 x 1504 and then upscale as Dee suggested.
Let me know if I can help.

[marble cape]

hey guys can you show me an example of making a image size bigger?

[open echo]

Of course! It’s called the Universal Upscaler

There you go for the explanation and examples:D

https://intercom.help/leonardo-ai/en/articles/8756274-universal-upscaler Universal Upscaler | Leonardo.Ai Help Center (intercom.help)

[marble cape]

Thanks

[nimble bay]

hey anyone has any suggestions on how to use canvas to draw strings for bows?

[night vault]

@dim zodiac bow master, please 😁

[nimble bay]

Thanks guys. I just used a editing tool for that. Looks to be easier...

Some minor edits now only needed

[fleet timber]

Top

[dim zodiac]

I was going to say - there are some tricks you can use but when it comes down to it, the AI is still bad at getting bowstrings 100% correct

if you have the editing skillz, getting a generation that you love and then editing in the strings is THE way to go

[nimble bay]

I appreciate the reply and feedback. Thank you Bowmaster. 😉

[stone fern]

Make picture bot

[spiral axle]

you cant make pictures here

please go to the Webapp 🙂

[marble cape]

help

[nimble bay]

Me again XD Is there a way to isolate certain elements in an image so that they are not changed in Universal Upscaler?

So I want to maintain the overall features on this image

While keeping the level of detail on the rest like so

Any tips or is that not something that is possible yet?

[clear tinsel]

Can someone tell me what's inappropriate about this prompt? (Photography:1.3), (((masterpiece))), (Photorealism:1.3), (1woman), (beautiful face:1.5), ((delicate facial features)), ((gorgeous)), (eyeliner:1.3), ((jawline)), 21, (lean muscle:1.5), ((vascular)), (sitting on a pillow), (looking straight at camera:1.3), (seductive), (glamorous), (shy), (large blue eyes:1.3), (eyelashes:1.3), pink lips, cheekbones, (long blond hair:1.5), (thick flowing ponytail:1.3), (eight pack abs:1.6), (button nose:1.4), (slim waist:1.5), (slim hips:1.5), (toned thighs:1.3), (muscular calves:1.3)

[soft harbor]

Hi! I am seeking help on how to add a shiba inu dog posing with my consistent chacter that i created. Her name is Danyelle, and I love the way she looks, but when I add the dog to the prompt it changes her appearance even with the fixed seed. I am currently training a model for her. How can I get the dog and her consistent in multiple settings and poses?

[night vault]

seed cannot helped with character consistency if two images are too different like full body shot and too half.

options are

1. could you try prompt that she carry shiba up in medium shot?

2. you can use canvas to outpaint from character you like

3. premium only - character reference

[soft harbor]

I have premium

The seed has actually been what has helped

Character reference not so much.

I'm trying to get a dog and girl consistency.

My prompt for reference: photorealistic 21 year old woman named Danyelle, fit, green eyes, long straigh blonde hair, American, white skin, standing on a miami beach wearing an american flag thong bikini, posing like a model

[subtle gulch]

That's because Character reference only retain facial features and not the clothing.
If you want the clothing to be consistent as well, it is going to be a challenge.

For one, you can try Character+Style reference at the same time, but even that the clothes may not be entirely consistent and may need post-editing.

[soft harbor]

I was able to get a some similar ones one with her gym location

[subtle gulch]

So do you mean you want the character to stiay the same and only upscale the background and the flame?

[subtle gulch]

It's also hard to maintain the consistency of two characters in the image at the same time.
It may work if the dog is a nondescript inu shiba, but anything more than that would need post editing.

[clear tinsel]

The censorship on the platform is mind-boggling.

[subtle gulch]

It depends on what you want to create.

[clear tinsel]

Mostly simple scantily clad women. I was led to believe the censorship was lighter with a paid subscription. I think that was false information.

[subtle gulch]

Paid or not, such image is still possible to create.

or else the streamers here won't be worried about using "NSFW" in the negative prompt everytime they are live lol.

[clear tinsel]

I have no doubt that it's possible, some way, somehow, but I posted a prompt above that is blocked in multiple different models without a single reference to nudity, or parts of the female anatomy, or anything adult.

[subtle gulch]

Tip: Try to avoid the word "girl" at all cost.

Leonardo is very strict in avoiding the generation of inappropriate underage images.

or "young"

[clear tinsel]

zero interest in generating underage

[subtle gulch]

Ikr, but the filter is there to lower the chance (or at least increase the difficulty) for such generation.

[subtle gulch]

If you still encounter difficulty, just ask.

[clear tinsel]

Using a character reference (generated on the leonardo platform itself) makes literally 100% of my prompts blocked

[subtle gulch]

You mean the prompt's okay if the character reference is off?

[clear tinsel]

soametimes, yeah

but the character reference isn't nude or anything, and the prompt has no references to nudity

[subtle gulch]

Care to share your prompt?

[clear tinsel]

sure

(Photography:1.3), (((masterpiece))), (Photorealism:1.3), (full body photo:1.5), (1woman), (beautiful face:1.5), ((delicate facial features)), ((gorgeous)), (eyeliner:1.3), ((jawline)), 21, (lean muscle:1.5), ((vascular)), (sitting on a pillow), (looking straight at camera:1.3), (seductive), (glamorous), (shy), (large blue eyes:1.3), (eyelashes:1.3), pink lips, cheekbones, (long blond hair:1.5), (eight pack abs:1.6), (button nose:1.4), (slim waist:1.5), (slim hips:1.5), (toned thighs:1.3), (muscular calves:1.3)

[subtle gulch]

The prompt has no problem and shouldn't trigger the filter

[subtle gulch]

Do you know what through brackets and numbers are for?

[clear tinsel]

yes

[subtle gulch]

I'd say those are unnecessary, especially if you're using the XL or lightning models

emphasising everything is like emphasising nothing, after all

No, didn't trigger the filter at all

[clear tinsel]

did you try adding a reference image? literally any reference image

"Stricter prompt filtering is in place when using Character Reference. Your prompt appears to contain suggestive or inappropriate content and has been blocked. You may modify your prompt and try again."

but it was literally the prompt I pasted

[subtle gulch]

Just remove the word "seductive" and you'll be fine

It's there to prevent abuse of that function for deepfakes

[clear tinsel]

looks like that might have been it

[subtle gulch]

You can imagine how people would want to abuse that function.

[clear tinsel]

"seductive" 🙄 thanks for helping with that

I mean, I guess. Though I'd probably be one to argue that the medicine is worse than the disease

especially if it comes down to "seductive"

anyway, I do appreciate the troubleshooting

[subtle gulch]

You can still be creative with words and prompts, if you get what I mean.
I'll leave that for you to explore.

[clear tinsel]

thank you

[night vault]

I’m sure you will run into “accident” for sure

[subtle gulch]

plenty

[rough glen]

Is it possible to have two reference photos and use image generation to merge them? Or another means? I have a great ranger and a great lynx as an animal companion. I just can't get them into the same image. Any thoughts?

Personally, I am not a fan of the ranger in the last photo.

[wise canyon]

Hi Greg,it is possible to use multiple images in "Image Guidence". When there is more than one image, sliders will appear for setting the % weight of each image, they share the overall image guidence weight (low, mid, etc). This would require you to play with those % weights a bit to get the outcome you want.
As your Ranger and Lynx images are already great, I would recommend using the Canvas Editor to layer them together (remove the background from the Lynx and scale it into position on the path), then after downloading the result, run it through Image Guidence as a single image using Content Reference, for a clean blended final generation.

[night vault]

you can use character reference on the first image, using prompt on the third

[open echo]

@slate grail Since your question was answered in #💬│general I’ll just go ahead and delete the duplicate here 😄

[buoyant oasis]

I have a problematic prompt for fantasy game I run...

I want to **make humanoid mice WITHOUT EYES. **( Not just blind, wihout even eyeballs. ) It's for this description on sepcific fantasy race in game I run

Got to those - but they clearly HAVE eyes. 😦

Preset: Lifelike Vision
Prompt: ( Blind:1.4 ) Humanoid mouse farmer from medieval fantasy. (Mouse have no eyes: 1.5). The space where their ocular cavities might be is covered with smooth skin. Zoom on face.
Negative Prompt: multiiple mice

[dim zodiac]

heya Chomik! There mayyyy be a prompting solution for this, but I bet you'll have way better luck just prompting for fur over their eyes in canvas

like, get the entire generation correct with all the stuff you want to see the mouse doing, and then take to canvas, mask the eyes and prompt for skin or fur

sketch2image with the fur colour is probably the strongest way of achieving it in Canvas

😅 I've had days like this where I spend hundred to thousands of tokens trying to perfect a prompt to give something, and then only to realize it takes 3 seconds in a different tool

[real lynx]

I merged my photo with this photo and also modified it to be a kind of Totoro

[nimble bay]

Kind of. I want to maintain the facial features of the character and upscale everything else. Doable?

[real lynx]

With character he maintains his features and puts them wherever he wants

[nimble bay]

What? I didnt get that...

[real lynx]

They are premium tools

Character reference + content reference

Original image

I upload it to reference content to maintain the figure

And in character reference I put what I want to keep the face and facial features. In my case I uploaded a photo of myself and this was the final result.

Your features work with the image and it looks pretty good

[nimble bay]

Hmmm. So use the uscaled image for content and the non upscalded image for character reference?

[real lynx]

My Tinkerbell prompt + characte reference

I modify my photo

But I don't modify my pose or my body

Referent content modifies my body and merges my features into the image

Character reference =face
Reference content = body

Reference content + prompt Tinkerbell

Maintained body reference

[real lynx]

I took the second photo and uploaded it again to content reference + character image with my photo + Tinkerbell prompt

He took my features and kept the shape of the body

[nimble bay]

okay will give it a go. Thank you. Seems like Universal Upscaler could have a feature to isolate elements. Sounds easier

[real lynx]

Anything let me know 👋

[nimble bay]

Yeah that definetely did not work exacty how I had in mind. I guess an easier way is simply to edit the features from my first image into the upscaled one

[real lynx]

Well do it with canvas

[nimble bay]

Yeah gonna give it a go and let you know how I get on 😉 thanks again

Looks like easier in Canvas after all.

[real lynx]

What did you want to modify?

[nimble bay]

Keep overall upscaling of image but original features. Examples are above not to far off from when you suggested I used content and character 😉

[buoyant oasis]

In the end - I used Photoroom free AI removing elements feature to make those eyeless mouse- took about 5 seconds 🙂

[subtle gulch]

In fact, you can also combine the two images simply using Canvas Editor.

[nimble bay]

Yeah I think that was the simplest solution in the end. Just edited the facial features of one image into the upscaled one

Thank you guys.

[clear tinsel]

I have what I think is are some noob questions (although I did read all the relevant wiki information). I'm trying to take a face/portrait shot I generated in one image and generate a body around it. I've been messing around in Canvas/Out-paint to try to do that (it may not be the best way, I don't know...) but using the canvas/out-paint has given rise to a bunch of questions. (1) Can I create multiple outpaint boxes? If so, I can't seem to figure out how to do that. (2) Does the end result image get saved somewhere on the site so I can come back to it, or is it simply a download or lose it forever situation? (3) Is there a way to sort of... resize the canvas?

[night vault]

Not at all, in fact, you ask deep questions that made me think you have done good research before asking.

1. No any draw is a single rectangle box but you have multiple masks to identify where to paint and that does not need to be rectangle

2. Depends on generation types but for sake of simplicity treat them as never saved online and you need to download them to save progress.

3. did you see this one?

[clear tinsel]

Thanks, Zorgi. That's helpful. It's too bad there's only one box. On a portrait shot, it seems like I'd need multiple out-paint boxes to really get closer to a full body shot, at least that's how it seems. Gotcha on #2, that's how it seemed to me, but wasn't sure if I was missing something as that seems like useful functionality to save my shots somewhere to re-enter canvas, etc. without uploading again. I actually have had more luck with going into the canvas feature with an online-stored photo than re-uploading again. The upload feature consistently blocks my somewhat NSFW image uploads, even the ones that were generated by the app.

(One of my primary use cases is generating avatar pictures for character cards for use in other AI chat/RP apps.)

On #3, I did notice that... I'll play around with it some more, although I'm not entirely sure in what way that affects the out-paint box. On my last attempt, the out-paint box looked like it was around 75% the size of the original image.... and the output (which I most like) has the canvas looking like a square on top of a rectangle. Any way to sort of make the whole thing into a square? I'm sorry, I'm probably not clearly stating that question.

[night vault]

I believe that even the image is not saved as a file online. They are saved internally on your machine so you can continue from where you left.

@clear tinsel maybe share some context of what you want to do? and perhaps share your current canvas and how your goal is looking for?

[dim zodiac]

I'm just passing through, but wanted to note that it might help to think of Outpaint in canvas as an iterative process

[clear tinsel]

don't want to share anything that's NSFW or in violation of posting rules. But I have generated several portrait-like photos (head and shoulders, sometimes chest) and was looking to fill out the rest of the body. Current canvas (at least with just one box) hasn't been able to do that, but maybe there's a way to make the original shot smaller than it is to give more room for the rest of the body? Something like that. Ideally the end result photo would be a square (for insertion into AI/chat character card profiles), and the closest I've gotten/downloaded now is an odd-shaped pictures that with black around where the outpaint box didn't reach

[dim zodiac]

the other thing you could do, is use that face that you generated as image input Character Reference

and then regenerate a square image using that face

for example, I get this image:

and then go here

[night vault]

It is premium feature, btw. Can you do that?

[dim zodiac]

^^^ good point

[night vault]

Ideally the end result photo would be a square (for insertion into AI/chat character card profiles), and the closest I've gotten/downloaded now is an odd-shaped pictures that with black around where the outpaint box didn't reach

Yeah, you need to keep track yourself which part is generated which part is transparent.

[clear tinsel]

Yes, I'm a paid member. I'll try it.

[clear tinsel]

(Sorry - my internet went out as we were talking.) Thanks, ooo. When you say iterative process in this case, you mean outpaint image, download it, reupload, outpaint it again, etc.? Is that what you mean?

[dim zodiac]

no, I mean doing chunks of it at a time

and build it out slowly until you get to your full size image

this was ultimately what I painted that original image out to:

[clear tinsel]

So, just trying to understand (noob that I am). Take image that was created. Outpaint it to get a new portion of the image. Save that image. Upload it and outpaint it again to add more to the image? Download, upload, re-outpaint, etc.? Is that what we're talking about

[night vault]

no need to download, reupload, you can continue on that piece

[clear tinsel]

That's what I'm not understanding

[dim zodiac]

here's a thread from long ago so the UI has changed, but it explains it in detail https://discord.com/channels/989166677390426132/1101117155401015398

and I'll try and post a few screenshots here

say I wanted to have her full head in frame

image uploaded into canvas

outpaint her head only at this point:

choosing what I feel is the best fit for the image from the 4 options I generated

for now I'm only using the same prompt that I generated the image with - can debate whether it was all necessary, but I was lazy so just copy and pasted from my original generation

then I'll move the generation window over:

and fill in the gaps

if I ultimately want it to be a square aspect ratio, I can choose a square ratio that it all fits into at the end and do one final generation

then once it's all cleaned up, click download

[nimble bay]

Or instead of doing portions at a time just resize image and reupload and select a large dimension for the square. Lets say image is 1000×1000 and u make the square 1200x1200 place it over the picture write in prompt "same background" and will fill all empty spaces automatically without affecting main image

[dim zodiac]

that works occasionally, but you may get errors like text in the edges of images etc doing it all in one shot

also the prompt box in canvas works the same as image generation, you should be describing what you want to see rather than trying to "speak" with the AI

this is more important to know in Inpaint mode though, outpaint is mostly working off the image data - so many things still work. Blending a background can even be done with only a space in the prompt!

[brazen raptor]

Hello, quick question: when using Fixed Seed, is there a way to ask Leonardo.AI to show the character with the same clothes in all the photos it generates?

Goal is to have my character wear the same clothes but with different poses

[subtle gulch]

Clothes are hard to keep consistent. You may get close results but they will not be identical.

Fixed seeds won’t help as it can only allow minor changes. Drastic changes in prompt like pose will change the whole generation.

The best bet you have is style reference, and use either nondescript clothings or very specific clothings. Either way, you still have to expect some post-editing.

[dim zodiac]

extremely well demonstrated Siloa!!

[brazen raptor]

Thanks for the explanations! I'm still very new to this and haven't played with style reference yet. Will definitely give it a try!

[west pivot]

can someone help me with this, i really like this image, but i wanted it to be colorful, i wanted it this exact image

i tried depth, Image2image, style, content

[night vault]

if you want exact color but more saturation, it is better do that with other tools

depth2img or img2img can use if you want to recolor and happy to tolerate some changes

[subtle gulch]

You're welcome. Just let us know if you encounter any other difficulties

[subtle gulch]

Content reference would be better. If you keep the original prompt and just add color descriptions, it should do the trick.
(Content reference is premium though)

[west pivot]

Why is it that when I use depth, Image2Image, and content together, only the content reference appears? I think in the past I could use them together at the same time

[white delta]

best model to make realistic human portraits?

[west pivot]

i tried it with style ref. but it change like this

[west pivot]

i wanted the hair made of butterflies like the original image

[subtle gulch]

Another way to do it is with Canvas Editor

Choose the color that you want

turn that color more transparent

paint over the items you want the color changed

then sketch to image with a relatively high strength

set the prompt to exactly what you want its color changed

[wanton belfry]

same seed?

[real lynx]

Add a colorful image with a style image

[subtle gulch]

No

[real lynx]

We have a texture channel

[wanton belfry]

oh so you used canvas to change into similar clothes...

[real lynx]

I gave these images platinum color them with this texture

To these inagenes

[subtle gulch]

The clothes are mostly similar already

[wanton belfry]

I haven't been able to find ways to reliably get ways to keep clothing consistent

[subtle gulch]

I just need to fine tune it on Canvas

[wanton belfry]

I see, only way I know is by pigeon-holding.

[subtle gulch]

With character + Style

[wanton belfry]

using generic clothing

keeping it simple

very hard to replicate intricate designs

[subtle gulch]

I can show you the original and my prompt later

Now I’m doing interpretation on the stream now

[west pivot]

😭😭 I'm not really a fan when it comes to Canvas, really. Btw, thank you for your help. Maybe I'll just wait for Phoenix to have Elements and image guidance

[west pivot]

tried it already but some details lost, like the hair

[subtle gulch]

Can you let me know how you want to change the colors?

I may try after I’m done with what I’m working on now.

[west pivot]

any color, just make it colorful 😆

but with the same details

[west pivot]

heres some but the hair, all the tiny butterflies been lost, not like the original image

[subtle gulch]

I see

[west pivot]

i want this, but colorful

[subtle gulch]

There is a stream right now

And a flash contest

You can come if you want to join

The theme is: dreamscape

[real lynx]

Maybe with your prompt it will be more similar

At 40 strength I lost the butterflies

This item is new

Does magical things

And maintain your image

[west pivot]

i didn't try it yet, i will. tnx

[real lynx]

I hope your image works for you, I love it

[nimble bay]

Works 95% of the time for me. If you start getting text exit Leo and come back to it... i find it works better than doing bits at a time

[oblique latch]

I was wondering if you can refine images on the free plan?

[real lynx]

You can refine the images you created

But if you want to refine another image outside of AI you can upload it with image x image first so you can refine it after uploading

[oblique latch]

How do I refine images for the ai ones, because the only refiner I can find for them is the alchemy refiner that I need premium for

[low girder]

Hello, in the image I just uploaded the text of The Vampire Diaries is not well written but I do like the image. From within Leonardo there is a way to delete this text and introduce another text so that the image is not deformed and I can take advantage of it or do I have to do it from the third-party tool, for example I have 3d text created with eonardo that I could superimpose

[thin tinsel]

Yes, you can do that with Canvas Editor

[low girder]

I don't yet master the canvas enough, I have to practice it

[thin tinsel]

Do you have some guides or video tutorial that you are learning from?

[low girder]

no

[thin tinsel]

Do you prefer written instructions step by step or do you prefer to watch a video?

[real lynx]

@vast patio

Hello how are you

[vast patio]

Fine cami

@WillowB were is video?

[open echo]

It's because Willow asked if they preferred written or video instructions on the Canvas but they never answered 😄

[real lynx]

Oki

[azure sigil]

where can I find written or video instructions on image creation?

[real lynx]

There are many on YouTube

If you need something more specific we can help you here

I saw that you have problems with a shield

[azure sigil]

yes i state it is on the left arm and it consistently gets placed on the right arm or floating in mid air

[azure sigil]

working on a character for a rpg : "Ultra 8k image of an incredibly handsome 25 year old white man, His gun metal blue eyes hold a great secret that he is keeping from everyone, he has platinum white hair, he is muscled, ripped and fit. he is without weapons a War Priest of Irori, strapped on his left arm [an intricately etched large steel shield adorned with mystical runes that shimmer with a soft ethereal blue light], he is wearing a full set of chainmail and a utility belt holding a healer's toolkit, dagger [an intricately etched large steel dagger adorned with mystical runes that shimmer with a soft ethereal blue light]. he evokes a sense of ancient wisdom, mysticism, and heroism, his right hand is raised in benediction[glowing with a soft ethereal blue light]."

[real lynx]

There are several ways to solve that.

The easiest way is in premium

But with free tools you have the canvas tool

If you go in parts you can repair those things

Premium tools with another external program I put a shield and a weapon and a creature

That would be the easy part But it's premium

To use canvas I am going to find you a video in English, I speak Spanish

[azure sigil]

all three show the shield on the left hand what did i do wrong with my prompt that it puts the shield on the right side of the characters body

[real lynx]

The power of Leonardo AI Canvas 2.0 (Tutorial)
https://www.youtube.com/watch?v=aw10ykp5n5A
Best YT Alternative Apps # Vanced Official

[real lynx]

Look at this video

With this tool you will be able to put the shield and the sword wherever you want

[azure sigil]

thank you just loaded it so time towatch and learn. thanks again.

[real lynx]

Any doubt comes back

One more thing

This tool is better to use on PC

Not in cell

[wise canyon]

Hi Sam, some extra info for you,
Prompts are weighted in strength from the front to the end, meaning that things at the beginning are weighted more heavily.
So when the AI sees "left arm" early in the prompt it thinks you want to see the left arm.
You will need to reword your prompt and make the instructions more direct, e.g- "holding a sheild, right arm raised".
Also note that square brackets cause a negative weighting, use the rounded brackets for + weighting.

[azure sigil]

I am on an all solid state PC with an 8 GB GeForce RTX 3060Ti Graphics card

[azure sigil]

Thank you, I am totally new to the AI generated art program so I am trying to learn.

[wise canyon]

Anytime, we are all here to help 👍

[white delta]

i want to put this logo on a individuals head and i want him on my podcast on the promotional poster

[paper iris]

Yes please so I am looking forward to create music album artwork

[night vault]

sorry, could you reframe the question again?

[warm forge]

Any logo makers. Please help.

[wanton belfry]

this sounds more like Photoshop work honestly

[wanton belfry]

you have to post what you need help with

without details you might not get much

sometimes just being told the right model to choose or setting is enough

[wanton belfry]

okay so what you want to do with this?

you want to add something? change something? create new?

[warm forge]

I'm typing it out and it says content is blocked

Have to retype it

I’m looking who can make it a high quality, all sizes and I need a few changes.

1. White bad croppings removed.
2. The trees need to be restyled and slightly simplified.
3. The orange dots of ore need to be removed. And replaced with the same orange coloured “Sharp” symbol

[wanton belfry]

you means moderation filter

[warm forge]

“Sharp” symbol is from CSharp and FSharp symbol looking like a hashtag or cross.

Like this. But orange. I’ll send my colour palette too

[wanton belfry]

by Csharp you mean this

?

[wanton belfry]

so you want to superimpose it?

or make something like this?

[warm forge]

Yes

[warm forge]

No.

[warm forge]

Read this please.

[warm forge]

Like to do what I described to my logo!

[wanton belfry]

you might have to use Canvas for that

[warm forge]

Idk how. Please do it

https://tenor.com/view/relatable-stressed-gif-27439196

[wanton belfry]

Its honestly not my speciality. I dont use canvas that often @thin tinsel @open echo might be able to help

both know canvas better than me

they can show you the way, but you have attempt it yourself

[open echo]

Hmmmm.... as always, for profesional use, combining image generation with photo editing is a SUPER power 😄

For this, I would use a multi-step process.

1- In a photo editing app, we erase the background (A logo needs a transparent background anyway, and it will help us for the next step)
2- Now that the U shape is separated from the background, we can use a selection to refill the blue color over the oranges dots
3- In Leonardo.Ai, use "Image to Image" combined with the Graphics Design:Vector preset to regenerate a cleaner version of the logo. We can use a simple prompt like "blue U shape, mountain, minimalist, orange highlights" (We'll change the forest later anyway, so no need to prompt for it yet)
4- We bring the update base logo into the Canvas Editor. Make sure to be in "Focus Mode" in the top right. Highlight the forest with the Mask tool and use a simple prompt like "vector, simple pine trees". Bring down the "Inpaint Strenght" just a little bit (like 0.9) so the new generation keeps the general shape.
5- Use the "Sketch' tool to draw the # sign at the bottom, copying the same orange than on the mountain. Change the mode to "Sketch2Img" and make sure "Automatic Masking" is on.
(You might have to regenerate 2-3 times as symbols can look funny)
6- Download Artwork and remove the background again (Which is even simpler since it's white)

We have a lot of guides on Canvas either written here or in video form on the Youtube channel. If you'd like a VERY quick primer for the Canvas, I recommend this article from the community: https://leonardo.ai/wiki/how-to-refine-details-using-leonardo-canvas/

For photo editing, I like to recommend "Photopea" as it's free and online (so no installs necessary).
I have a basic guide here where I show quick ways to cut out subjects from the background:#1101232396248023061 message

Don't expect to do it all in 10 minutes the first time, but practice makes perfect 😄

[warm forge]

Thank you a lot.

[subtle gulch]

Those tiny butterflies are indeed very hard to keep unchanged using image guidance or Canvas.

It would be best to start with the prompt generation with the prompt you used.

[open echo]

I think the easiest way to keep the same details is to use a Hue/Saturation tool from a photo editing software ... That's gonna ensure the details will be 100% exact

I've saved a TON of cool gen that simply had bad colors with these tools:
https://discord.com/channels/989166677390426132/1113604362254700614

[hardy sentinel]

@wary ferry nice to meet you, your question would get more answer here.

[wary ferry]

Ok Thank You "Hello, I created a logo with Leonardo Phoenix, but it came out in perspective. I would like some advice on how to change it to the foreground. I tried using Canvas Editor, but Leonardo Phoenix is not available among the options. Can you explain how to do it? Thank You"

[night vault]

try add minimalist logo banner design. it could be just of luck to get correct perspective.

[west pivot]

Thank you ❤️😊. Yes, I really need to specify each color in the prompt to achieve this result, rather than just using the word 'colorful'

[west pivot]

ok. thanks for the info, but its ok now😊

[subtle gulch]

Wonderful results

[subtle gulch]

I really like all the butterflies in your original image and it’ll be a shame to lose any of those.

[west pivot]

Thank you! I’m trying to get the exact same pose as in the first image by using the same seed. However, even adding one word to the original prompt changes the result. Still, I love it because it’s really colorful now. 😊😊😆

[subtle gulch]

Seed doesn't really fix posture. Any changes in the prompt will lead to variations of different extent.

[white delta]

is it possible to make this using leonardo?

[wanton belfry]

with Phoenix, you can make something Similar

[real lynx]

I just did this in Leo

Hello good morning says :3

In Spanish

[pastel surge]

Please help me ..
How can i get right username?

[karmic pulsar]

No matter what I try. I can not seem to get Leonardo to make a ghost or spirit. It always gives me a human

[night vault]

like this?

[karmic pulsar]

no. like a transparent ghost floating down the stairs

Photo realistic

[wanton belfry]

try the word hologram...transparent is sort of blacklisted

"Floating ghostly hologram floating down the stairs"

[subtle gulch]

Like these?

[wanton belfry]

or this?

I have a better result but its sort of too exposed sadly

[subtle gulch]

You also try “translucent”

[wanton belfry]

semi-transparent also works

I also advice to use cel-shaded lora with it, it works best

man this is the first time I got ghost this good

I am almost proud of myself

cel-shaded too good

[night vault]

He want a realistic one too

[wanton belfry]

hmmm not semi realistic?

[night vault]

#🙋│image-help message

[wanton belfry]

well no worries it still works with Kino

keep cel-shaded still

cel-shaed helps in keeping it transparent

[viral wadi]

hi guys, is there anyone who can help me improve my prompt?

i would like to generate image with retinue of heroes walking by in fantasy valley.. but iam struggling with multiple figures..

i want them to be kind of different from each other

[hardy sentinel]

Hi do you have access to Leonardo Phoenix?

[viral wadi]

yap

[hardy sentinel]

then use leonado Phoenix for this purpose

[limpid valve]

What are some negative prompts for someone or something that I made with Leonardo Ai Phoenix (anime)?

[night vault]

sorry, I don’t get the question

[subtle gulch]

It depends on what you don't want to see.
At this point, don't expect Phoenix to generate any anime images that could be on par with AnimeXL.
My suggestion is to generate the character and composition that you like on Phoenix, as it gives you more control. Then, you can use that as reference by img2img/ content reference or other means to generate your image with AnimeXL.

[jovial plover]

Hi community! I'm new, but I need your help, I have a photo that I need to put inside other image. Like a mockup, that can do Leonardo for me?

[livid inlet]

You could use the Canvas Editor Tool to do that

[jovial plover]

I'm in Canvas Editor, What I have to do next in order to upload my image as an imput?

[livid inlet]

if you click this it will give you the options of where you want to upload from

[jovial plover]

I'm not sure if my image is bigger, but I see a square that says something about "snap" , what does it means?

[livid inlet]

Snap is there for the generation box, it allows you to move the box in set amounts.

To make your image bigger or smaller click here and then click on you image to bring up the sizing

[jovial plover]

well, for now I don't need to rezise, I just uploaded the image that I want as the model, what is the next step? Can I provide a second image maybe in my prompt in order to merge both image in one?

[livid inlet]

You can upload your 2nd image, Erase the place you want the 2nd image to be. Use the mask tool around the edge and put the generation box over that place. Then generate and it should merger the image into one.

I'm really sorry but I have a meeting in 2 minutes so won't be able to respond for at least an hour

[jovial plover]

Thanks , I thought you were a bot, but actually you're an human, right?

[jovial plover]

I'm not sure if I could follow this part, is possible to join us in a live channel? voice channel?

[night vault]

There are couple bots @pure ivy and @manic root. All others are lively human folks 😄

[pure ivy]

My apologies, something went wrong when responding to your request.
Our apologies if this message did not answer your question.

[jovial plover]

lol! Thanks! (And my apologies for the confussion @livid inlet )

[jovial plover]

Hey @livid inlet when you come back, could you write me? 🙂

[livid inlet]

lolololol, Its my fault for talking like a bot I guess 😂

[jovial plover]

You did so well! that I thought you were a bot hahaha

[livid inlet]

I'm free to jump in the voice channel for a quick demo 👍

[jovial plover]

yes 🙂 great! I go there

[livid inlet]

https://discord.com/channels/989166677390426132/1082641517245628526

[sleek sail]

Hi guys, I'm trying to generate images of this type but I can't find the correct prompt. Do you have any advice?

[night vault]

Try this I got from bing

a digital illustration of Jesus Christ standing in front of a vibrant, colorful sky filled with clouds. He is wearing a crown of thorns and holding a large blue social media "Like" button sign with the text "Subscibe & Like". The overall style is a blend of traditional religious art with modern digital and social media elements, creating a striking contrast between the sacred and contemporary themes. The color palette is bright and inviting, enhancing the surreal and whimsical nature of the artwork.

use phoenix model

[wanton belfry]

it works well

3d preset is the best though...

add smile to the prompt, otherwise it looks like jesus is forced to advertise lol

there is a slight mistake in prompt, I've changed the first "like" to thumbs up button "a digital illustration of smiling Jesus Christ standing in front of a vibrant, colorful sky filled with clouds. He is wearing a crown of thorns and holding a large blue social media "Thumbs up" button sign with the text "Subscibe & Like". The overall style is a blend of traditional religious art with modern digital and social media elements, creating a striking contrast between the sacred and contemporary themes. The color palette is bright and inviting, enhancing the surreal and whimsical nature of the artwork."

result becomes better

and jesus looks happier

[sleek sail]

@night vault @wanton belfry Wow! Thanks for your help, I didn't expect all this collaboration 🙂 I tried these prompts, they're not ugly but it's as if Jesus always had a "porcelain" face while in the photo I sent he always has "porcelain" skin but in a different way.. as if it were more opaque (I don't know how to describe it) I'll send you another photo, maybe you can better understand what I mean. (Don't pay attention to the dash in the middle of the screen) But it feels more "real" and I would really like to have a prompt like this

[night vault]

if you use phoenix, you can use edit with ai and put command such as “make skin more shiny” https://x.com/leonardoai_/status/1805503754504171794?s=46&t=HOEK0ov9GDj0ZD9-4sXqJQ

[sleek sail]

In your opinion, the correct prompt is "shinier skin"^ I don't know how to describe that clear change in image that can be seen, is it only me who notices it? lol (Thanks for the advice, I will use it)

[wanton belfry]

its just a matter of taking your favorite one and putting it in different model like Kino and generating it with content reference

[sleek sail]

But can I insert this image (the one I sent) on Leonardo.ai and ask them to make me a similar/same one? based on my image?

[wanton belfry]

sure but you have to careful to remove uneeded elements because they have probability of creating artifacts in your scenes that you wont like

I do understand no matter how much you try, to a extent it is going to look 3d than realistic

its inevitable

[karmic pulsar]

Yes! Thankyou. I dont know what I was doing wrong

Thankyou to all who helped me with spirits and ghost

[tender olive]

hi what does it means

[night vault]

set weight of Wood cutting element to 0.80

[tender olive]

where i can set ?

[night vault]

#🖼│image-share message

[abstract pier]

@wanton belfry

Bỏ nền

[hardy sentinel]

hi, nice to meet you, you need to register at the website of leonardo ai

if you are using pc the website should look like this

if you have google, microsoft or apple accounts you can choose to register with those accounts (A)

[abstract pier]

Yes

[hardy sentinel]

alternatively you can register with your e-mail adress (B)

[hardy sentinel]

have you registered ?

please kindly show me your screenshot

are you using PC or mobile phone ?

[abstract pier]

Đã đăng ký thì phải vô đó như lạc vào đám rừng không biết đường đi

[abstract pier]

Cả 2

[hardy sentinel]

ok, let's use PC then, it is easier

[hardy sentinel]

I understand, please share your screenshot so that I can guide you

[abstract pier]

Link vào web?

[hardy sentinel]

once you have registered, this is home website of leonardo ai

[hardy sentinel]

Cami already gave you the link, do you trust me ?

This is the link https://app.leonardo.ai/

[hardy sentinel]

yes, you are correct 👍

ok please state your goal, what kind of image do you want to create ?

[abstract pier]

Ok để tui tập làm. Cảm ơn bạn rất nhiều

[hardy sentinel]

ok, just ask questions here

[abstract pier]

Tôi chỉ muốn tạo bức ảnh có sẵn thành bức ảnh or clip 3d

[hardy sentinel]

ok like the monk photo above? #🙋│image-help message

first of all, do you mean an animation ?

what do you mean with 3D photo?

if you want to use the photo above #🙋│image-help message as an image guidance then you need to change your preset

right now you are using Leonardo Phoenix preset group

[hardy sentinel]

ok that is an animation

this is video is made using various software and in multiple steps

[abstract pier]

Như clip này

[hardy sentinel]

Animation in Leonardo is limited to 4 seconds

[hardy sentinel]

this need a lot of other software. If you want to make a photo Leonardo. ai is suitable

And from that photo generated with Leonardo you can then make a 4 second animation

[abstract pier]

Mình có ghép nhiều cai đoạn clip 4s thành 1 clip dài 1 phút qua capcut
Vẫn được chứ

[hardy sentinel]

ok, I see that you have understood how to edit the video yourself.

[hardy sentinel]

awesome work by the way

now, do you want the face of that monk in different pose or what do you want to do with the monk photo? #🙋│image-help message

[abstract pier]

Vậy mình tạo ra bức ảnh trong app này và 4 bức ảnh sẽ thành clip 4s sao đó thể ghép với app capcut xử lý các hiệu ứng ảnh

[hardy sentinel]

correct

you create one photo, from that one photo you can animate it to get a 4 second clip based on that photo

one photo => one animation

[abstract pier]

Ok tui hiểu vấn đề chút chút, mấy bữa nay ko biết app discord này là gì ? Cảm ơn bạn hướng dẫn cho tôi

[hardy sentinel]

you are welcome

[abstract pier]

Tui sẽ thử trên pc sau đó coi thế nào, chúc bạn luôn may mắn và thành công

[hardy sentinel]

now, if you want to use that monk photo as an image reference, you need to switch to other preset group

[hardy sentinel]

Good luck and thank you for your wishes. If any problem arise, just ask here.

We are all user like you, who volunteered, if I am not here, maybe someone else can help you.

[robust iris]

Anyone can help me to find this art style??

https://youtube.com/shorts/b9gqV2BOXTw?si=PDDaC1X7FwsSHVda

[hardy sentinel]

the story is very touching.

this is my first attempt realtime generation

[hardy sentinel]

using anime xl

[robust iris]

Better

But not like the one

[hardy sentinel]

Albedo Base XL

[robust iris]

This is nice

[hardy sentinel]

no, it is different

[robust iris]

Send me the screenshot of details

Please

[hardy sentinel]

look I am also a user here. 😄

[robust iris]

😆

[hardy sentinel]

and it is not even similar to the style you are striving for

[robust iris]

But its nice

[hardy sentinel]

good, Albedo XL without anything, just use the Legacy Mode

[robust iris]

Ok

[real lynx]

Kino, alvedo, lighting

[pure geyser]

Check this https://cdn.discordapp.com/attachments/1052363298370236527/1258170252227903662/Default_A_closeup_portrait_of_a_woman_with_vibrant_green_and_y_0.jpg?ex=668711db&is=6685c05b&hm=52338cbc9a895bf3b258deb33ff23cbca5a198d6cbb5aedc2b7967fa9c3d2fe7&

[drowsy raptor]

Hi Everyone. I would love to speak with an expert to help me create a specific model for my copmany. Basically I would like to have a model where you upload a picture of yourself and then it applies a filter on your sking adding a layer of tan. If someone could help me to do this it would be gerat, even if we need to pay for the time. Thank you!

[strange iris]

Sup sup! Does anyone have any idea how to recreate a pose like this but in anime style?

(ignoring face expression, clothes etc)

[merry dome]

any idea on a negative prompt that might get rid of the black shading lines?

[open echo]

"flat colors" in the positive should help

"Shading" in the negative also?

It also depends if an Element is active or if you are using Phoenix's Auto enhance feature (As it will change the prompt)

[buoyant dirge]

@open echo please help

[raw thicket]

How can i help you?

[distant sierra]

im having issues with my prompt

A colossal worm monster erupts from the cracked earth, its segmented body twisting against a stormy, lightning-filled sky. The beast's maw, filled with razor-sharp teeth, reveals a glowing interior. Dark, oily scales reflect the lightning, while spiny appendages ripple along its sides. The desolate landscape features jagged rocks and skeletal remains. A sense of impending doom permeates the air. Inspired by H.R. Giger and Wayne Barlowe, with the dramatic intensity of a Guillermo del Toro film. Cinematic lighting, 4K resolution, photorealistic textures, digital art, dystopian horror concept.

im getting a filter which says the prompt is innapropriate