🙋│image-help

[dim zodiac]

or like... "erase" something by blending it with a nearby colour

[thin tinsel]

I guess there are lots of ways to get to the same result

[thin tinsel]

Yeah that's my style.

[tender eagle]

Why am I getting this message when I haven't uploaded any pictures?

[dim zodiac]

Its letting you know the upscale limit, it SHOULD go away once you've selected the image to upscale and the settings?

[dim zodiac]

If it stays there permanently, it may be a bug that you could report on the intercom

[indigo tiger]

Guys how can I generate Ai images

[covert dew]

hello

[thin tinsel]

@pure ivy please tell @indigo tiger how to create immages with Leonardo.

[pure ivy]

To create images with Leonardo.Ai, follow these simple steps:

1️⃣ Create an Account: Go to https://app.leonardo.ai and Sign Up for an account.

2️⃣ Access Image Generation: Once logged in, navigate to the "Image Generation" tab.

3️⃣ Enter Your Prompt: Type a prompt in the textbox to generate your image.

4️⃣ Explore Features: For more guidance, check out the quick tutorial on getting started!

Remember, you can't generate images through Discord, so using the web app is essential! 🎨✨
-# Our apologies if this message did not answer your question.

[true tulip]

hello ai, i need help

@pure ivy hey how do I add this server to my server. So I can generate images on my server and no one can see it. I don't want to share it

[pure ivy]

You can't generate images through Discord, but you can use the Leonardo.Ai webapp to create images privately. Here’s how to do it:

1️⃣ Go to https://app.leonardo.ai and Sign Up for an account.
2️⃣ Once you have an account, navigate to the "Image Generation" tab.
3️⃣ Type your prompt in the textbox to generate images.
4️⃣ If you want to keep your images private, make sure to toggle the "Public Images" option in the sidebar after generating your images.

This way, you can create images without sharing them publicly! 😊
-# Our apologies if this message did not answer your question.

[true tulip]

@pure ivy hey how do I add this discord server to my discord private server. So I can generate images on my server and no one can see it. I don't want to share it

[pure ivy]

You can't generate images through Discord, but you can use the Leonardo.Ai webapp to create images privately. Here’s how to do it:

1️⃣ Create an account at https://app.leonardo.ai.
2️⃣ Go to the "Image Generation" tab after logging in.
3️⃣ Toggle the "Public Images" option in the sidebar to make your images private (this only applies to images generated after toggling it).

Once you've done that, you can generate images without sharing them publicly! 🎨
-# Our apologies if this message did not answer your question.

[thin tinsel]

[true tulip]

what?

[feral bough]

That's not possible

Because you don't generate images on this server

[true tulip]

okay, thanks

[cloud flame]

for image share in discord
we need paid version of leonardo.ai ?

is it ???

[dim zodiac]

nope!

[dim zodiac]

we share our Leo generated art in #🖼│image-share

you will just need to be a productive member of the discord for a bit and your permissions will upgrade

[proper crystal]

Oh yeah I was only using the erase and masking tool on and off but thanks for the future suggestion!

[marble cape]

Hey

[alpine swan]

Hello! 🙂
Does anyone have settings for the ultra upscale to achieve results like the old/legacy one? I tried multiple times now but the legacy upscaler still rounds up things better and I am confused about that.

[dim zodiac]

You want it to make changes to your image?

[alpine swan]

I can give/show an example, if you would like to see one?

[dim zodiac]

yes please do!

I can run some tests and share results

my guess is that you are going to want higher Creativity and lower Similarity when running the Ultra upscaler

[alpine swan]

This is the original from Anime XL. Please focus on the face, especially the eyes and teeth.

This is legacy upscaler at creativity 3. The eyes are round and cleared, the teeth visible with that little front gap.

This is my best shot with the ultra upscaler. I had one with visible teeth but blurred eyes, and one with rounded/cleared eyes but only a grey pane where the teeth should be.

[dim zodiac]

Sorry you're using the Universal Upscaler on all of these?

or you're upscaling from the Image Generation view/your feed

[alpine swan]

Not on the first one, that is the original from the anime XL without ultra upscaler.

Yes, I take the image from the "recent images" dropdown.

[dim zodiac]

Yes, I just wanted to be sure I was looking at things the same way you had been

because there are upscalers available immediately from looking at the image in your feed

So I wanted to ensure I'm doing testing that matches what you're trying - no problem I've gotten the info I need now!

[alpine swan]

Thanks for your fast action. 🙂

[dim zodiac]

Alllllrighty... try Realistic style with 5 Creativity, 6 Detail contrast and 4 Similarity

This image is as if Zelda fused with Cami and worked out in a jail for 3 months straight

😄

[alpine swan]

Trying it now. I honestly never took realistic since it is more for a game.
And it is actually a players character. 😄 I will forward your point of view about this char. 😄

[dim zodiac]

oh nice... making player art for my PC's in D&D is what brought me here ages ago

[alpine swan]

I am a game master. 😄 Lot's of npcs and locations to do!

Here, see te result. Left your ulra-upscaler settings, right the legacy with creativity 3.

see the difference in the eyes? @dim zodiac

[dim zodiac]

The extra amount of shine?

[alpine swan]

The outline of the iris is much more visible, the pupil is round and dark, and yeah, the iris is much clearer, with the light green down to the lid, a bit darker green left an right. On the left side it is just a mushy greenery.

It looks just more crisp and less blurry.

[dim zodiac]

but you want to use the Ultra upscaler because it's cheaper?

[alpine swan]

Yes and no. I want to use it because as far as I understand, the "legacy" will dissapear at somepoint in the future and I have to use the new one? The cost doesn't matter, I just want similar results in the future. Or will the legacy stay?

[dim zodiac]

nah legacy will be sticking around

[alpine swan]

Well, then I'll be using the legacy for that while. ^^ The result matters to me. Not that the players will notice much. xD

[dim zodiac]

ah, I found it

Creativity 8, Detail 7, Similarity 4 (artistic this time)

[alpine swan]

I'll try! 🙂

[alpine swan]

Tanks for sticking with me so far. Maybe I am just too greedy. ^^
A problem I encountered was exactly that: Either the eyes came out good, like with your settings, OR the mouth, never both. Your settings ironed out any realism on those lips. Left your settings, right the legacy.

[dim zodiac]

yeah the artistic seems to trend towards glossy smooth

and the realistic towards gritty realism

😄 you should have seen what happened when I cranked realistic all the way up

it was... not pretty...

[alpine swan]

I saw the horrific outcome! xD I really tried a lot of different settings, even with excel tables to keep count on what lever was on what setting. I am a bit reliefed it was no simple mistake I made. So far, legacy still wins for me. ^^

[dim zodiac]

Yeah so I'm not privy to the tech behind the scenes (unless I corner one of the team members on a good day) but I think the legacy upscale uses more of the Alchemy pipeline

which is heavy on the GPU load, hence the token cost, but also does good things for details

but much like the Legacy view in the image generation, the Legacy in upscale has not been slated for removal

as far as I know

it's just named as such to show that it was the OG version

[alpine swan]

I dearly hope so. I gladly pay the price for the better (imho) quality outcome. "Legacy" just sounds dangerously like "to be discarded". At least in my not native english ears. 😄 I hope it sticks around. You might tell that to a developer, if you catch one. xD

Thanks for your assistance/trouble!

[dim zodiac]

no problem! Hey if you have any other questions, you can either come back here or you can use the "/mentoring" command to get your own private thread with a bunch of experts to respond

and then you'll never have to sift through other peoples' questions/responses to find your own

[alpine swan]

Ooohhhh, didn't know that. I am so satisfied with Leonardo, I hardly have any questions. 🙂
Will do so! Thank you again!

[dim zodiac]

no problems! did one last pass... Realistic - 8 Creativity, 7 Detail. 6 similar

she's grittier for sure, but I think the higher similarity kept her from becoming a full on monster

[alpine swan]

Well, that's a sweet middleground. The eyes looking better and the lips aren't ironed out. 😄 If the ultra upscaler improves a bit, I would take those settings. 😄

[spiral axle]

who can help me to prompt a "Berliner" 😛 ?

[stable carbon]

i'm not actually sure what they're called properly in english.. 🤔
my family called them 'jammy donuts', but Leo didnt know that name
anyone else? they're round buns, sprinkled with sugar and stuffed with jam..

[dry moat]

Hi guys I need your help regarding this image

Was it made in Leonardo? I tried making it in it but it was a disaster. I'm trying to get the prompt but ain't able to illuminate like it

/mentoring

[dim zodiac]

in North America those are usually Jelly Donuts

[pine star]

I don't know it it was made with Leonardo, but you can make such images with leonardo. I would use AldeboBase XL and Ehtereal White 0.1 for such an image. And for the Prompt use "Describe Image". 🙂

[dry moat]

@pine star tysm I'm gonna go try 🙂 n then revert tysm 🌸🌸

[wanton belfry]

its very hard to get it right

you can try it in anime XL

however it might make it female, so add male before Shiva

Anime XL understands Shiva

[whole coral]

is there a good way to fix mangled hands or missing fingers?

without going into an alternate program and making adjustments

[spiral axle]

Yeah but I always get a donut with a hole in it

[languid dagger]

Hi everybody, i use a 3D app to make a rendering image, and download a background image, then use Leonardo to blend 2 images , but I found the 3D rendering image seems not match the background. Is there any prompt to suggest on this case? I tried to use prompt “same background “ , it seems doesn’t work correctly. 🤔

[dry moat]

@wanton belfry omg you are so kind but its way different than the uploaded image but thanks a lot for trying, I appreciate 🌸🤗

[hardy sentinel]

I see, that you have posted here, please post your recent question here

[dry moat]

@hardy sentinel was it made on Leonardo and if yes is there a way we can create images revolving around this theme? I tried a few but the image doesn't show up in a complete form, it's very cropped up

Ty for replying in advance

[steep vault]

hi

I want a design to put on my T

which should represent India to other countries

[wanton belfry]

perhaps you didn't understand, I am saying that AI Model Anime XL understands the concept so you can make it in there

Whether you want it to be grayscale or monochrome that is upto you

[dry moat]

@wanton belfry ohh now I got it tysm yeahi tried the phoneix one maybe that's where I went wrong

[wanton belfry]

Phoenix might also understand a bit, but its not that good on beauty basis

because it also knows Ganesha

whole Deity theme can be also done by Phoenix, very costly though

[hardy sentinel]

if it is shared at #🖼│image-share it is probably made using Leonardo, I see that you are using OpenArt as well. Users are only permitted to post images generated using Leonardo. ai

[real lynx]

You can have AI describe your image to create something similar

[wanton belfry]

to an extent, Phoenix does also understand Shiva can make it

you might chose to run it again in img2img into Anime XL to beautify it

[hardy sentinel]

I see that Kentox has done a good job 🙂

[wanton belfry]

this is what happens when I ran the image above into Anime XL, you can see the difference

but of course, its possible because of Alchemy though, might get a bit worse with just Fast

oh and its done with context, thats another issue lol

this is without alchemy

[hardy sentinel]

gorgeous image

[real lynx]

Albedo and Phoenix

Digital artwork featuring a stylized, divine figure with a serene expression. Hindu god man with a trident.The subject has long, flowing white hair tied in a topknot, and a third eye on the forehead, symbolizing spiritual insight. The skin is a pale, almost ethereal white, and the figure is adorned with intricate, pink jewelry and accessories, including a necklace and armlets. The figure holds a trident with a pink, glowing aura, and is draped in a white garment with pink accents. The background is dark, contrasting with the glowing pink particles that surround the figure, adding a mystical and otherworldly atmosphere to the image.

[wanton belfry]

pink was a bit much, try purple. Pink is generally used for goddesses

[real lynx]

Oki

I figured you would be able to create what the user wanted because you know more about the subject 😅

[wanton belfry]

this is more like Chinese Poseidon

[real lynx]

I just described the image that Shared

[wanton belfry]

it understands the concept of Shiva, so you needn't explain too many details

[real lynx]

Chinese Poseidon 😮

I thought it was a Hindu god that's why I added it in the prompt

[wanton belfry]

yea, the name is Shiva

[real lynx]

Oki

[wanton belfry]

sometimes if the AI understands the name, you needn't add in too many details

[hardy sentinel]

great example

[wanton belfry]

I always prefer to use Phoenix images as base and Anime XL to beautify it

amazing results

especially with context, depth2img

[hardy sentinel]

my problem using realtime canvas for this topic is the result is bare chested or hardly clothed 😄

[wanton belfry]

Realtime probably wont execute well eitherway, its pretty old model

[hardy sentinel]

I mean male statue 🙂 it is ok for 0 token

Albedo Base XL yield similar result, because my aspect ratio the anatomy is ruined ..

[broken elm]

Does Leonardo have a style based on or that is PDXL?

[hardy sentinel]

I don't understand your question? are you referring to MidJourney styles?

[real lynx]

A photo of a statue of Shiva Chiva, the Hindu god with a trident. The statue is made of stone and is located in a temple. Shiva Chiva is seated on a lotus throne and is surrounded by water. The background is serene, with tall trees and a clear sky.

[wanton belfry]

oh wow thats great one

[broken elm]

It’s a style i love using another app, im looking maybe to move over leonardo

[hardy sentinel]

Leonardo and many other platforms are different than Midjourney

[real lynx]

Hi, can you show us a picture to see if Leo can imitate the style?

[hardy sentinel]

I don't know, I can not guess, need an image example at least.

[real lynx]

That's the color you were talking about.

The one with light blue skin

[wanton belfry]

yea these both look Amazing

[real lynx]

They are real style anime with the same prompt

[dry moat]

Oh sorry I wasn't aware, I thought it was made by leonardo so posted it, apologies

@real lynx this is so strikingly beautiful ♥️

[real lynx]

I'm glad you liked it

[dry moat]

@real lynx you are amazing ty I'm trying to read all the messages, I'm a bit overwhelmed, tysm I'm gonna try making them as per your instructions

[broken elm]

Im not on my pc right now, but on my phone, and the ones i generated on phone are like mixed style, like 85% one style, then 15% another style on top which im pretty sure is sdxl… trying to find if I have anything pn my phone, i cant generate right now cuz that other app im paying for has been down for 4days…

[hardy sentinel]

no problem, I am a user like you. Don't worry. Just informing you.

[twin pine]

Folks, I have a doubt, someone gave me a reference to create images froim, is there a tool I can feed the image to, so as to get a cool prompt out of the image and then use to generate some other ones?

[hardy sentinel]

pc is easier, but if you are travelling you can use mobile. if you can use style reference, it would be easier.

[broken elm]

I meant my files are on my pc

I got something but I don’t know if it would be classified as nsfw

[hardy sentinel]

can you download them into your phone? if you want your files as reference.

[broken elm]

Yeah i got two images

[broken elm]

Please don’t ban if I’m wrong? 🤣

[hardy sentinel]

ok that is a start 👍

[hardy sentinel]

I don't understand

[broken elm]

I don’t know if those image are against TOS

[feral bough]

@broken elm please keep things PG here...

[broken elm]

But they’re the only I got here

[broken elm]

Yes…

I guess they were not

[feral bough]

In the case of the second image it was VERY CLEARLY not

[broken elm]

Well I’m sorry 😞

[feral bough]

Okay

If you're not sure if something is NSFW or not, it's probably not suitable to be posted

[broken elm]

Alright, thank you, sorry again

[hardy sentinel]

oh ok nevermind.

[dry moat]

@wanton belfry tysm you are very kind and helpful, I'm not able to reply to everyone and express my gratitude but tysm for makin such great images and helpin me out. 🌸

@hardy sentinel oh and tysm for helping me out, I'm really grateful to you and others

[hardy sentinel]

sorry, I could not help much today. but glad Kentox and Cami can help you out more.

[broken elm]

I’ll send you something later tonight when on my pc, with the main style unaltered

[dry moat]

This one is the most beautiful one, not that others aren't but it's like super realistically detailed and intricate and I love it. I am really thankful to you, if you don't mind when you are free would u pls give me what base and elements did you use.🌸♥️

[hardy sentinel]

ok, the Leonardo filters could be a challenge for you.

[broken elm]

Filters?

[real lynx]

Anime semireal style let me see well But I think I used that

[hardy sentinel]

yes, censorship

[real lynx]

A photo of a statue of Shiva Chiva, the Hindu god with a trident. The statue is made of Porcelain and is located in a temple. Shiva Chiva is seated on a lotus throne and is surrounded by water. The background is serene, with tall trees and a clear sky.

[dry moat]

No no lol you did help me and every bit counts🤗

[hardy sentinel]

🙏 it is refreshing to have you around in this community

[dry moat]

@real lynx omg you are like the bestest kindest person I have ever met online,🌸

[broken elm]

Oh It’s like I said, those are the only images I had on my phone, it’s not like it’s all I do. Basically I use the first style because it accurately draw the night elf features of the elongated curvy long eyebrows, and ears, but it looks very blurry, the second style that is applied help fixing it, and if I put a celebrity name, itll will basically merge them into the elf, if that make sense

[dry moat]

@hardy sentinel lol it made me giggle probably cause I'm the donkey to every Shrek on here 😂

[broken elm]

Like it makes it easier to regenerate the same person with consistent facial features

[hardy sentinel]

I make an elf for you using realtime canvas, maybe not exactly as yours, do you need a dark elf ?

[broken elm]

Its the eyebrows that are really hard to make, and they usually have crazy long ears 😂

[hardy sentinel]

no problem, oh ok night elf

[thin tinsel]

with or without a hole in the centre?

[broken elm]

Im dumb why haven’t I just thought of sending something that was like i wanted from the web XD

[stable carbon]

without. just a round bun

[hardy sentinel]

first attempt on night elf using realtime canvas, had to correct the ears, I don't have a digital pen so only using mouse to draw

[broken elm]

Haha yeah you run into the same problem i was when using sdxl

[thin tinsel]

Ok I see what I can come up with

[real lynx]

We have a very nice community that really likes to help, thanks for your comment

[hardy sentinel]

second iteration of the night elf @broken elm well, it is based on sdxl base model

[broken elm]

Yeah sdxl can’t do the eyebrows, and it struggles with oversized stylized armors (from my experience)

[dry moat]

@real lynx 🤗 did ya create that white background one next to the blue one with the same inputs? Do I need to choose white element for it?

[broken elm]

Lile i have to use pdxl 86% then sdxl 14% to fix the face but retain features

Higher than 14 i lose what i gained from pdxl

[hardy sentinel]

what do you mean with pdxl? leonardo is based on stable diffusion

[real lynx]

Let me check and confirm

[real lynx]

Modify that it is not made of stone that it is made of porcelain this prompt use

A photo of a statue of Shiva Chiva, the Hindu god with a trident. The statue is made of Porcelain and is located in a temple. Shiva Chiva is seated on a lotus throne and is surrounded by water. The background is serene, with tall trees and a clear sky.

Just to see what came out and I liked it 😅

[broken elm]

The last one is because I freaking love that style, but yeah

[hardy sentinel]

I like the last image too.

[broken elm]

Like if i could do like the last one style but accurate eyebrows XD

Would buy full premium top tier and never leave home

🤣

[hardy sentinel]

you can achieve it with alchemy I guess, need to try it first 🙂 but the last image is just a regular elf then.

[thin tinsel]

@stable carbon

Not quite enough sugar for my liking.

[stable carbon]

@spiral axle was the one who wanted to know.

[broken elm]

Yeah just not many credits, I can’t try much before spending money, is why I ask :/

[hardy sentinel]

is ok, I am trying now

[thin tinsel]

real-time-canvas doesn't use your tokens up

[hardy sentinel]

I need to try using Alchemy for the last image, realtime canvas would not be able to replicate that

[broken elm]

I don’t know what that is xd, nor alchemy

[hardy sentinel]

alchemy is the premium tool

[spiral axle]

what you prompted for 😄 ?

[hardy sentinel]

the eyebrows need to be manually edited using canvas editor later

[thin tinsel]

Hi Danny, you can pick up from my images in the team account. 😉

[real lynx]

You can also upload an image of a statue to imitate the style of marble or porcelain or stone.

[spiral axle]

well its close enough but not perfect 😄

[broken elm]

Alright

[spiral axle]

"Hide from Team Feed" activated 😉 ?

[real lynx]

With the free tool Style reference

😅

[thin tinsel]

Ok, yes you're right. try again now

[spiral axle]

oh yeah keep it simple like "round cake"

[hardy sentinel]

you can achieve something using premium tools, I will show you in a dedicated thread. You will be tagged later

but the problem with ears and eyebrows need canvas editor

[broken elm]

Is it the type of tools that allows u to make hybrids species? Thats something else im super interested in

[hardy sentinel]

I don't know

[broken elm]

But i absolutely have no idea how it is made

[hardy sentinel]

are you willing to have a one on one session or you want other mentors to join in? if so then type /mentoring to ask for mentors to join, please be specific about your goal.

[broken elm]

🤣 at work right now but id be up for anything

[dry moat]

Got it ty so much again Cami. 🌸 God bless ya💖

[broken elm]

Like 5 hours from now xD

[hardy sentinel]

I am busy myself, you can open up the request later, by typing that command. People will join in when they have free time. We are all users like you.

[hardy sentinel]

maybe tomorrow is better for me, but who knows, maybe somebody else could join as well, just type /mentoring and fill out the information , just be specific

[broken elm]

I will try that, thank you, you are very helpful 😁

[hardy sentinel]

you are welcome, just for your information, I can only achieve what you have showed me using alchemy + photoreal v2 (premium tool) or using content reference (also premium tool)

[thin tinsel]

@spiral axle , a difficult challenge

[spiral axle]

https://tenor.com/view/doughnut-twice-gif-24236702

[broken elm]

Those are some exemple of hybrids I was talking about, no clues how can people make these. They are from pinterest.

[hardy sentinel]

i see, yes, it is doable

[turbid shoal]

Okay, I am sending up the white flag. I am trying to get tightly curled ram horns and I keep getting images like this. I want them to look like the images, and I even used the AI Prompt tool on ChatGPT, but nothing is working.

I keep using tokens on incredibly disappointing results

ultimately what I'm trying to do is make a group shot of a team of 4 fighting a dragon so I'm doing each character individually and then will impose them into the greater shot (because I don't think Leo can do that complicated of a piece), but I'm stuck on my female tiefling. Attached is her reference art (was generated on leo also), and the pose I'm wanting to have her in. Would anyone be able to help me out here?

[native fable]

Anyone know if this style is achievable in leonardo?

another example

[real lynx]

You can imitate poses with premium tools

[turbid shoal]

Yeah, I've done that. The pose never translates, so I end up having to do image-to-image which doesn't make the horns curled, and putting in the pose thing doesn't work either. Basically I don't understand why the prompts get over written no matter what I do.

[real lynx]

Use these tools

Character for the face and content for the body

Style is going to create a similar style

Imagen Style

He gave the other statue its porcelain style with gilding

If you put a texture of an ice cream the image will be with that texture of ice cream

[turbid shoal]

How can I get the horns to be more curved though? Or have it follow the colors?

[turbid shoal]

I've even tried using the canvas editor, I just end up with re-textured horns

[real lynx]

One of the images always has to be low so that the prompt can be enforced

I also gave more strength to the image with the Style image

To imitate the skin color

No image style

I loved how they turned out.

Use this setting

For example, now I didn't modify the prompt and asked for a heart to come out of his hand instead of the super power he was creating.

Also with character image I put my face

Increase the Style to ultra so that it takes the red color well

[real lynx]

👋

[turbid shoal]

Gotcha, gonna give it a whirl now!

[thin tinsel]

Yes. Just do describe with AI and use Leonardo Phoenix.

[sullen light]

Hi to everyone. I need an help about dress swap. I created a a man character and I need that that character to wear my shirt. I'm trying with canvas real time with inprint but I don't know how to start swapping. Thank you

[patent cove]

wow these look good

[real lynx]

Hi, I need to see your character and your shirt to help you.

[sullen light]

I need to dress this shirt on tha created character

[real lynx]

I managed to model this similar dress and upload it so that the AI can describe the dress and then add that description to my prompt

It is not the same but it is similar

[sullen light]

thanks for your support Cami

[real lynx]

I'll free myself up and help him more in a moment.

[sullen light]

what kind of preset did you used for?

[sullen light]

I tried but it change my shirt texture. I want the identic texture

[real lynx]

I'll try myself and explain how I did it.

[sullen light]

I tried with life like Vision uploading my shirt but result it's not the same. Texture it's different.

[stable schooner]

dibujo pixal styl y disney. un joven a pie de camino que sube una montañña que termina en un cielo estrellado. color azul oscuro

[marble cape]

Going to see if I can generate this staff to see on what it looks like:

[real lynx]

I think it's similar to your shirt

[sullen light]

texture must be the identic not similar.

[real lynx]

Good morning Willow

Maybe with canvas editor?

What do you think?

[thin tinsel]

@sullen light you need to create a new image in the same POSE as the shirt, so that you can cut and place it on.

[sullen light]

doo you think that i can't create a male character with the same texture ?

[thin tinsel]

Hi Cami!
Yes, I'm just thinking that with any re-generation of the shirt the pattern will change slightly.
So I'm thinking to use pose-to-image or Edge-to-image to create an image first

[thin tinsel]

Do you mean texure or do you mean the pattern?

[sullen light]

I'm testing this because a customer asked me to have their creations on different generated character in a real life composition and create multiple variations and scenes @real lynx @thin tinsel

[thin tinsel]

Do you understand what I said about re-generation

[sullen light]

I have more images of customer shirt and a character created with Leonardo

[thin tinsel]

I can teach you how to do it, but not right now, and not if you don't answer questions I'm asking you.

see you later.

[sullen light]

thank you very much Have a nice day

[simple cave]

Can anyone tell me how to generate these kind of images?

[real lynx]

You can use this option to have the AI generate a prompt describing your image

[real lynx]

Hola solo en la web podés crear

[full escarp]

hi

[elder zephyr]

Hello!

can someone help me with which model and settings should I use if I want to regenerate the same face as the one in the uploaded pictures?

[open echo]

Hello there 😄

It's a feature called "Character Reference" in the image guidance button, at the left of the prompt box. It works on all the other models except "Phoenix"

(They already announced it's coming soon for phoenix too, but no ETA on that 🤔 )

[dull socket]

I want to "recreate" the graphics of the walking dead telltale but I don't know which model/style is the right one, my goal would be something like this

[wintry lance]

Does image2motion only do camera motion? I had an image and I just wanted the mouths to move and the camera stay locked. I do not see a way to use prompts when generating the motion though

[feral stratus]

Hi there, I'm new to this platform and don't fully understand how to use the tools available. I was curious my goal is to take some of my personal photography and transform them into an anime style or vector images. what is the best way to go about this? thank you in advance!

[low bear]

Motion indeed doesn't have a lot of customization yet, it does sometimes move other parts as well but it tends to do camera motion only, yeah

[low bear]

That is absolutely possible and I've actually done that a lot! You can use image2image for this, you'll have to experiment a bit with the different settings we have available (like which model to use and the strength of your image input)

You can find this feature here!

(just click "view more" under image guidance)

[feral stratus]

thats great to hear. any suggestions on what settings you have used for the best results? I've experimented a bit but most of my attempts lose the main subjects initial visual. I do alot of car photography and thought it would be fun to add an anime styled version of the photo to friends/ clients. ^_^ I appreciate you taking the time to provide screenshots and guidance! ❤️

[low bear]

If you're losing details, you probably need to tweak the init strength (or whatever it is called now) of your input image, making the image weigh stronger for the generation

I don't know if you're able to share your images (I could imagine clients not being too happy with that lol) but if there's anything we can share we can give specific feedback on how to improve results!

Unfortunately with some images its gonna be either detail or proper style, since the AI doesn't really know where its supposed to keep the detail and such. A good prompt can help a lot in recreating (and therefore preserving) those details though!

[feral stratus]

sorry I don't have the original image saved on this computer so I did my best to get a screenshot off the uploaded image T_T

[low bear]

Oohhh that actually looks really great! Did you use image to image for this or another image guidance tool?

What could help here is describing the image and accessories in the prompt, although the wheels are going to be a bit hard

Ah yeah I suspect that'll work better if you use the standard image to image

[feral stratus]

instead of content reference you mean?

[low bear]

This is because image to image references the full data of your image including all colors whereas content reference ignores the colors and only looks at the content itself

[low bear]

Yep!

[feral stratus]

I will adjust that and play around a bit more with that! thank you so much Ruby! I really appreciate the assistance!

[low bear]

Also I noticed you've written your prompt like an instruction, keep in mind the AI is actually trained on image descriptions. For example "a studio ghibli illustration of a deep blue [...]" etc

That way you might be able to keep using content reference, it'll allow the AI a bit more creativity so that might actually give better results than image to image

Sorry for the contradicting advice hahaha

Honestly just a matter of trying what works best :)

[feral stratus]

no need to apologize. this is all new to me and I'm sure there isn't one correct way to do anything. ^_^ I appreciate your time and help on this!

[low bear]

No worries, here to help!

[hardy sentinel]

Hi, i am a fellow user. Animation is random.

[thin tinsel]

You cannot be that specific

[elder zephyr]

i have to make this old blurry b&w photos into Van Gogh impressionist style painting, is it possible?

[mint dune]

you can see I clicked on Yes, the button turns grayish and it's just stuck like that

[low bear]

That is really odd, I've never seen that happen before

[mint dune]

All weird and odd things happen with me my entire life 🥲 🤪

[low bear]

I would recommend contacting support through the button in the bottom right corner of the app, they'll be able to look into account-specific issues and hopefully be able to resolve it for you!

It can take a while before they respond though

[mint dune]

Yeah issue is i am using this for work now :/

But I will try few things before I do that

I somehow fixed it lol

[low bear]

Awesome, glad to hear that!

[sullen light]

Hi to everyone! Have someone added a specific texture to a clothes of a generate image ?

[half halo]

Hey guys one of my players for my cyberpunk campaign sent me this as their character portrait. Is there any way i can use Leonardo to generate an image of her on a motorcycle? Ive tried using reference images but i cant really get it to do what i want

it ends up looking like this

[hardy sentinel]

Hello Anthony, which specific aspect ratio do you need? It will influence the result

[young echo]

hi i need help

[half halo]

Whatever would get me the best result

[hardy sentinel]

Awesome, maybe try 1:1 or at least 4:3

I need to try myself

[half halo]

Cool I'll give it a try!

[half halo]

yeah not having much luck, this is the closest i get

[real lynx]

Hello good morning

If you want, I'll lend you my motorcycle.

Although I think yours is more modern

[half halo]

haha thanks 😄

i gave the AI a bit more room to work

and ended up with this

i think thats good enough for now

[real lynx]

I got a more modern bike

You can activate the /mentoring command

To go to a private room exclusive to you

[half halo]

Damn those look great

[real lynx]

A coffee and I'll show you how I created it

[real lynx]

I won't be long, I'll be back

[compact bone]

I don't understand one thing, I have 0/25000 rollover tokens, but this is not possible because last month I didn't use all my tokens, so there should be some, how can they be zero?

[real lynx]

I returned

[half halo]

I'm helping my friend with some home renovation but id like to pick your brain later!

[real lynx]

(Hergé) and (Jon Carling) stylized miniature figure ((((Bulma outfit capsule corp)))) with a mechanized Motorcycle creature. consistency. made with (wood, ABS, and ceramic), in Mango Orange, Gelatinous Blue, Bubblegum Pink, Fluid Emerald, Glowing amber, Mucous White gradients. masterful color graded, Highly detailed surfaces, reflective and absorptive, contour and relief, photorealistic, luminous, subsurface scattering, specular highlights, ambient occlusion, dynamic range, volumetric light, microdetails, physically-based rendering (PBR), ray tracing, realistic depth of field, displacement and normal maps, dynamic lighting, Shot on (75mm), Depth of view, (F/13)

You can modify the character you want

This would be bulma

Negative prompt

plinth, title, text, three-legged, anatomical distortions, anatomical inaccuracies, proportional inaccuracies, blurry, low quality, unclear features, incorrect features, mutation, deformed, disfigured, cloned, cropped, out of frame, out of picture, double head, multiple figures, signature

Anything else you need let me know 👋I really like to help

Important tips

It gives more space to the motorcycle

For being an elongated figure

[tawny barn]

Hi everyone, can anybody help me with an image

the man is not coming properly

need an business man image showing sympathy while watching street beggar sitting along side on street

it is in "Leonardo Phoenix"

[thin tinsel]

Share your prompt

[tawny barn]

" Create a close-up scene featuring two contrasting individuals. both figures should look properly . On the right, depict an elderly homeless man sitting on the sidewalk, wearing tattered, dirty clothes. His face is lined with deep wrinkles, and he has a long, unkempt beard, with a sorrowful, tired expression. r On the left, show a young businessman, around 25 years old, dressed in a sharp, tailored suit, holding a sleek briefcase at his side. He is standing still, looking at the homeless man with a sympathetic, concerned expression, as if moved by the beggar's plight. Focus on the emotional connection between them, emphasizing the businessman’s empathy. The scene should be intimate with no background distractions, highlighting this moment of human connection between the two.""

[thin tinsel]

Do you want the business standing or bending down?

[tawny barn]

standng or some tilt can be fine

[thin tinsel]

Ok, so first we don't need to use "Create" at the beginning of prompts. Although Phoenix can handle a lot of information, but it's always good to clear out as much stuff we don't need.

Secondly because you've asked for 'CLose-up" it's not able to fit one man sitting and another standing. This and all the details about the mans face.

[thin tinsel]

Did you use prompt enhance?

You can order your prompt by saying "A photo of ..." at the start

[tawny barn]

okk getting ,

[tawny barn]

yeah giving it a try

[thin tinsel]

I will rewrite it for you. But still use prompt enhance on.

[tawny barn]

also, i am not getting the expressions clearly, what should i add?

[thin tinsel]

A photo of a two contrasting individuals. (One man is an elderly homeless man sitting on the sidewalk, wearing tattered dirty clothes, deep wrinkles, and long unkempt beard, sad and tired). (Another man, a Businessman, 25 years old, wearing tailored suit, holding a briefcase. The business man looks at the homeless man with sympathy and concern). The scene is depicts emotional connection between two men. An empty city street background.

[thin tinsel]

Try my prompt with Phoenix and prompt enhance on.
I cannot tell you how to get the expression clearly. We are asking an emotionless machine to create an image showing emotion.

@tawny barn I got them both in the image, you could use it as content-ref

[tawny barn]

yeah now working 😃

[tawny barn]

use this

Thanksss

[thin tinsel]

I added a little more >>>

"A photo of a two contrasting individuals. (One man is an elderly homeless man sitting on the sidewalk, wearing tattered dirty clothes, deep wrinkles, and long unkempt beard, sad and tired). (Another man, a very young Businessman, 25 years old, wearing tailored suit, holding a briefcase. The business man looks down at the homeless man with sympathy and concern, his face is full of worry). The scene is depicts emotional connection between two men. An empty city street background. "

[thin tinsel]

This pose is a good one too

[thin tinsel]

I like that first one.

[muted jewel]

do midjourney prompts work in leonardo

I want to buy a set of prompts but it's for midjourney

[tawny barn]

yeah that one is good

[tawny barn]

this is also nice

[muted jewel]

there @proud wadi

[thin tinsel]

maybe yes maybe no

[muted jewel]

depends on the prompt ?

[thin tinsel]

yes, depends what model too

[half halo]

Im havnig some trouble with the settings, if i use legacy mode im not able to select a character guidance image

ah it was the preset that did not allow for it, ill try a different one

[real lynx]

The only preset that does not allow image uploads is Phoenix

They are uploaded differently in legacy mode

[thin tinsel]

Did you see the new update today? 🤗

[hushed topaz]

I'm trying to make a portrait similar of Fidel Castro's smoking a cigar, my prompt is this one:
"a close-up portrait of a D&D dwarf, male, smoking a cigarette, pyromaniac, arsonist, crazy eyes, maniac laughter"
It does okay, but since the image i'm basing it on is in black & white, so is Leonardo's image.
I tried putting "in color" and "full color" everywhere on the prompt, but it restricts me and doesn't generate anything. I'm guessing it infers racism of whatever. Any tips for making the image in color?

[real lynx]

I have just been informed of this, I am going to try it 😮🫡

[real lynx]

Hi

Speak Spanish?

[hushed topaz]

Yes

Digo, sí

Lol

[real lynx]

Que bueno así nos entendemos mejor

Con solo poner Fidel castro la IA lo reconoce

Mo acabo de comprobar

Contame de nuevo pero en español que es lo que necesitas

Tenemos también un canal en español no se si lo visitantes

[real lynx]

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

[balmy orbit]

Any tips for character consistency throughout generated images?

[open echo]

Hello there!

Your best bet will be using Character Reference.

Keep in mind for now it only works on one face per image though

Here is a guide with some other avenues also 😁

https://intercom.help/leonardo-ai/en/articles/8584503-creating-consistent-characters

[real lynx]

You can also generate in Phoenix by panels

If you ask for example for something comic style in 3 panels with different sequences

[real lynx]

Generates a 3-panel comic in Minecraft style. Character1: pikachu character2: batman.The first panel shows two Minecraft characters in a grassland biome with a plot introduction. The second panel shows the characters in a grassland biome.panel1*,character1 He says "hello good night", with a happy face while Batman looks at him very seriously .Panel2*,character2 says "I am Batman" .panel3*,character1 says "Pikachu".

Generates a 3-panel comic in Minecraft style chibi characters. Character1: deadpool character2: Captain America.The first panel shows two Minecraft characters in a grassland biome with a plot introduction. The second panel shows the characters in a grassland biome.panel1*,character2 He says "hello good night", with a happy face while looks at him very seriously .Panel2*,character1 says "I wish you all a good day" .panel3*,character1 says "I am your fan".

[lyric latch]

Hello
is it possible to mix style in your image?
I tried this

Five adventurers are engrossed in a roleplaying game, guided by a master storyteller. The vivid worlds he describes comes to life before their eyes, as if they can see the image of his tale projected in front of them. A majestic dragon with shimmering scales and fiery eyes, some netrunners in a futuristic city, a spaceship in battle against an alien spaceship . The detail and artistry of the image are so exquisite that it almost feels alive, drawing the players deeper into the fantastical realm of the game. The players around the table are realisticaly represented.

The idea is to have real people playing around a table, and the vivid worlds are painted.

[maiden eagle]

Can someone give me a link that lists all the available prompt params?

[low bear]

Leonardo.Ai uses GUI elements, not prompt params

[real lynx]

Phoenix

A cinematic shot of five adventurers absorbed in a role-playing game, guided by a master narrator. The vivid worlds described come to life before their eyes, as if they could see the image of their story projected in front of them. A majestic dragon with glowing scales and fiery eyes, some netrunners in a futuristic city, a spaceship battling an alien vessel. The detail and artistry of the image are so exquisite that it feels almost alive, immersing the players more deeply in the fantastic realm of the game. The players around the table are depicted in a realistic manner.

Does something like this help you?

[keen agate]

Hey peeps. Can we give a style guide to the tool and generate more image similar to this style?

[keen agate]

this would be super good to have aswell

[real lynx]

Three-panel comic style maintains the same style and characters

A 3-panel comic with a different Minecraft World setting. Panel 1 shows Himiko Toga, an elf dwarf with pointy ears and blonde hair. She is cute and has a chibi style. She is excited and has a happy face. She says, "creare la mejor historia de todos los tiempos !". Panel 2 shows her looking thoughtful and concerned at a blank page. Panel 3 shows her sitting on the floor, sad with tears falling from her eyes. She says, "me rindo, me voy a ver anime"].

This is with the Phoenix preset

[keen agate]

Nice, imma try commic style

[keen agate]

btw, you just pick a style for 1 image and the panel gets filled with same style automatically?

[real lynx]

You can mention characters and the texts will appear. They may be confused. But they can be fixed.

[real lynx]

Each preset has many styles

[keen agate]

Gotcha

Thank u so much

[real lynx]

At the same time it can include elements

[real lynx]

For example, this is a photo of me modified with this element

[keen agate]

nice output. will try it

[real lynx]

I don't want you to get dizzy, too much information can overwhelm anyone, if you need a specific image you can tell me what you want to create and we will try to help you better.

You could even be part of a story

[keen agate]

got it, will play around and get back

[real lynx]

I mode.elf

I'm not very tall so I'm more of a garden gnome elf

😅

[keen agate]

🤭

[real lynx]

Even if you want you can use the /mentoring command if you plan to do a project or for better help, with that command an exclusive and private room will open for you.

Where mentors from all communities come in to help you

I am a mentor for the Spanish channel, my name is Camila, nice to meet you, greetings from Argentina

[keen agate]

oh this is great. I am actually exploring how i can include the tool in my design workflow.

[keen agate]

Nice to meet you. greeting from Tallinn.

[real lynx]

The Phoenix tool is very good for understanding many things, it can even create posters, letters and logos.

[pure geyser]

Impressive

[real lynx]

This is the prompt for that

[real lynx]

☝️

[manic root]

Your message was removed, as it contained a link.
-# If you believe this was an error, please contact a moderator.

[real lynx]

I like elves with super long ears

[feral bough]

Two things @broken elm

1. Are you aware that we have a spoiler policy that requires you to spoiler images that contain a lot of bare skin and/or sexual imagery? This is because there are underage people on this server, and also people that would not like to view content like this. I will have to delete both your posts because none of the images were spoilered.
2. Did you make those images with Leonardo.AI? If so, you can post them in #🖼│image-share (with the proper spoilers / in accordance with the NSFW policy of course). If not, consider sending them in DMs as this is the Leonardo server.

I have deleted the images @broken elm. I don't want to have to warn you again after this because you posted images like that before 😦

[broken elm]

1. No I've only seen rule #4 which forbids nsfw content. I did not know about the spoiler requirement/option for skin.
2. No, those were images I generated using another app, I had asked if Leonardo was capable to generate elves with long ears and eyebrows (if id had any models/styles that have blizzard entertainment dataset studies), then I would purchase the full tier. Most people block DM so I did not consider DM.
   I really put a lot of attention to meticulously pick some of those I thought were inoffensive and safe. I was comparing with PG game art of Hearthstone. I am also not arguing btw, I just wanted to let you know I really tried to be careful and I really appreciate your leniency. I apologize and I just won't post anything showing any skin at all if anything at all, to avoid any further lapses.

[feral bough]

#🖼│image-share message You can find the policy here. Most of the images were OK but they needed spoilers, I understand that you weren't informed of that, that's OK so now you know. You can post images with more revealing outfits, they would just need a spoiler - but if the images are sexual or suggestive in nature, we avoid that entirely. As to the other point, I guess I didn't read your conversation well enough😅 It's okay to post images from other sources for educational purposes. But if you post in #🖼│image-share to show off your creations, those should only be Leonardo images.

Also I would guess Leo can generate those type of Elves, but maybe not as closely resembling the WoW elves as the images you posted. I'd recommend trying the Phoenix or AnimeXL model.

[broken elm]

To be honest I'm more worried and frustrated about the fact that it seems like I don't know what is considered nsfw and suggestive anymore, like would it be correct for me to assume any character like Nier Automata is entirely nsfw? And we haven't even mentioned anything about gore yet . I guess I will try the first tier for a month...I just really worry about the pay per gen model. See-image : Those are the only ones that I saved, I probably generated 4 times that .

[feral bough]

Nice, I have like 30.000 AI images saved between Leonardo and Dall-E, but that's rookie numbers compared to this!

Also, if you want to create more images... The Artisan and Maestro plan offer something called 'Relaxed Queue' which allows you to make unlimited amounts of images, but at a slightly lower rate. Which makes it not pay per gen.

Something to consider

[broken elm]

I haven't seen that, I must've missed it. Does it include less features though? I'll check it out. Also, you may have less than me because I did a lot of experimentation and I would keep stuff that were "bad" but had one thing in it I really liked, just for concept and Idea ; for example , when I was generating armor, I managed to merge animal with armor, making my character wear animal themed armors, I went through so many species. The other App I use seems to be much lower quality than Leonardo. A lot of these images will need fixes or would need to be regen for better quality, etc.

[feral bough]

No, Artisan and Maestro are higher tiers so it includes everything the lowest plan (Apprentice) has, and some more stuff

Every plan comes with Alchemy which is what can make images really good

[broken elm]

I'll check it right away

I think I try artisan for a month, if I like it, I can upgrade for a full year or maestro year xD, does Leonardo have downtime? @feral bough The app I currently use have so much downtime, it is the main reason I've been looking out for other apps. Once I hadn't access to it for 3 weeks, and they do not work on weekend, so if it goes down of friday afternoon.....

There I took artisan D:

[feral bough]

No it doesn't really have downtime, sometimes it gets maintenance or a bug where the site is not accessible but that's VERY rare, so you don't really have to worry about that

[dreamy birch]

I need help putting together a prompt of a wolf with dragon wings on top of a rock
obs: free account

[low bear]

Plus when it happens the app is inaccessible for like a few hours max

[thin tinsel]

Use the prompt improvement feature.
You know where it is?

[dreamy birch]

no

[thin tinsel]

1. Put your simple prompt into the prompt area.
2. CHoose phoenix model.
3. Press the sparkles ✨
4. Choose Improve Prompt
5. Check the prompt details to ensure it fits your imagined idea
6. generate

[half halo]

the Edit with AI function is also quite great, you can just tell the AI what you want different and it will change the prompt accordingly

[thin tinsel]

Yes, you are absolutely correct. I don't think I make enough use of it.
But to be honest, early on it added things I didn't ask for whilst also makimg the alterations I did ask for.
Have you had anythhing like that?

[half halo]

Oh yeah for sure it does, its not fool proof

you still have to keep an eye on the prompt

i also would like it to use more technical prompt words to get specific results, if it does that it would be a huge step forward

the stuff i see some users put in the prompt is like looking at a config file for a video game sometimes

if the AI would use that kind of thing that would make a huge difference

[thin tinsel]

Have you put it in as a siggestion?

[half halo]

I have not, but yeah thats not a bad idea

is it just /suggest ?

[thin tinsel]

I am not sure if you will be able to do it with your rank,

If not I can do it for you

[half halo]

yeah i left for a bit, the rank is gone 😄

yes please do

[real lynx]

You can use the /mentoring command Opens an exclusive private room for you

[real lynx]

☝️

[thin tinsel]

Write it out conscicely as a feature request and I'll copy and paste.

[half halo]

I have seen many experienced users use certain technical keywords, brackets and other tricks to get better results. Examples are DOF, Dynamic Range, use of certain materials etc.
If the AI prompt enhancer were to use these keywords to help you enhance your image that would make a huge difference.

[thin tinsel]

ok done! 🙂

you should see it on the suggestions channel

[half halo]

thanks! 🙂

[broken elm]

is this entirely made with ai? I feel like it's more a mix of Ai and photoshop XD

[half halo]

What? You dot think a wolf/frost dragon chimera can write?

[real lynx]

Completely in AI

[broken elm]

=/

I just started using Leonardo xd, 3k coins, im going in the right direction but still unable to make what I'm trying

[real lynx]

I don't know what kind of wolf the user wanted.

[real lynx]

I was creating elf ears 😅if possible

[broken elm]

I was trying to do something I thought would be easier...lol

[real lynx]

Use the /mentoring command to avoid problems, it is a private room

[broken elm]

i probably dont have rights to use command

[real lynx]

All users can use it

[half halo]

wolves are people too...

[real lynx]

A dragon's tail grew out of him

[half halo]

natural evolution, i approve

[broken elm]

I have OCD, wish I could just post a picture instead of text to request mentoring xD

[half halo]

You can post a picture 🙂

there are no bad pictures

only bad politicians

😄

[broken elm]

it doesn't let me xD

[thin tinsel]

why can't you?

[half halo]

click copy image in your browser

and paste it here

[broken elm]

I don't post anymore because above, sorry I don't know how to "quote"

[half halo]

oh i think when you paste, and the picture is still in your chat bar, you can right click it to make it spoiler

[broken elm]

I don't wanna risk anymore xD, would dm be okay?

[real lynx]

Keep evolving

[half halo]

those are pretty awesome

[broken elm]

lagging so much, it doesn't like when i ask to put image by dates

[thin tinsel]

The images are inapproprate?

[broken elm]

Not to me, but to be fair I just don't know anymore

[thin tinsel]

Well it's a wolf-dragon right? Does it have boobs?

[half halo]

i think the line is right between provocative and nudity

[broken elm]

oh im not making wolf dragons lol

[thin tinsel]

Have you tried doing 'Describe with AI' ?

[broken elm]

what's that?

[thin tinsel]

It's in the Leonardo application and you upload your image so it can make a prompt for you.
Sound good?

[broken elm]

I can try it maybe, it won't outright ban me if it think its not appropriate will it?

[half halo]

anything you do on the website

has nothing to do with discord

and it will not ban you, at worst it wont generate the image

[broken elm]

where can i upload the image in leonardo, i thought it was the icon next to the prompt, but i dont think it is

[half halo]

It is yes, also you can switch to legacy mode

make sure to use the right engine/preset tho

the newest one doesnt fully support it

also be sure to check as what reference you want it to use

character/style/pose etc

[low bear]

It won't haha don't worry

[broken elm]

ah jesus now im lost LOL, legacy mode x.x

[thin tinsel]

sparkles ✨

[half halo]

its a button in the top right

[half halo]

basically it lets you use the old interface

[thin tinsel]

Don't worry about that for now.

[thin tinsel]

Shall we try not to overload new people with every bit of information, but focus on the things they need just to get started?

[half halo]

You could have phrased that differently but sure

ill leave the mentoring to the professionals

[thin tinsel]

who me?

[half halo]

so it seems

[thin tinsel]

Sorry I didn't mean to offend you.

[half halo]

alright, apology accepted, i will burn it from my brain ❤️

[broken elm]

Image is a digital fantasy artwork featuring a female elf character with striking features. She has long, flowing blue hair and pointed ears, characteristic of elven fantasy depictions. Her skin is a light purple hue, and her eyes glow with a bright, ethereal green. She wears a detailed, metallic corset with intricate designs, paired with a layered, dark blue skirt. The background is a cathedral-like setting with tall, arched stained glass windows that cast colorful light, enhancing the mystical atmosphere. The overall composition is vibrant and richly detailed, emphasizing the fantasy theme.

It doesn't see the eyebrows x.x

[half halo]

have you tried using negative prompts?

[broken elm]

no, im not quite ready for that yet lmao

[half halo]

its bascially you telling the ai what you dont want

so if in the negative prompt you would put, nudity

the chance of you getting a nude image is much lowered

its a bit of a give and take between the prompt and the negative prompt sometimes

[broken elm]

I don't mind about nudity, other people do, which is okay xD, I only use negative for like "horns" because if I say I want a fox-like armor, it gives my character fox ears, and putting negative "horns" fixes that

[half halo]

If you want more pronounced eyebrows you could put in "pronounced eyebrows" or something like that near the start of the prompt

the earlier the word in the prompt the more attention it gets i think

[broken elm]

yes it does xd

[broken elm]

I was only able to make them in another app, I suspect their model got trained on "night elf"

[half halo]

thats interesting 😄

Cami gave me a prompt for a really nice motorcycle image she made and a lot of the stuff was very non intuitive, i think she refered to the motorcycle as riding a creature in the prompt

truly making the right prompt is an artform in itself

[broken elm]

yeah

[broken elm]

I couldn't do crawling, closest I've gotten was kneeling down, climbing kinda worked

[half halo]

hmmm maybe if you use an image

of a person crawling

as a pose reference

[broken elm]

I just started using leonardo, didn't know about uploading images, only got experience using another app, and now im struggling to make what I was making 😢

[half halo]

leonardo is the best one ive seen so far

i opted in during beta

but i left the discord cause i wasnt using it too much for a while

[broken elm]

im kinda discouraged, its hard to start over

[hollow breach]

Prompt writing is very dependent on the model you use. If you use phoenix, you have to write your prompt differently than you used to for sdxl for instance.

[broken elm]

Albeo is to closest to what I want

[hollow breach]

That's also the one I used the most.

[broken elm]

it just lacks the "creativity" like I used to use "ai flaws" to make the ai creative with other app xd

[loud vessel]

hey all, how would i go about making icons for skills for a game with Leonardo please?

[half halo]

Albedo is also the closest to what i want

[broken elm]

avif?

[half halo]

there we go

thats albedo 😄

[broken elm]

oh not that....and yeah I didn't think of her

[half halo]

its a reference to a pretty good anime

[broken elm]

It keeps turning armor from image ref into hair :/

[half halo]

can you show me a screenshot of your reference page

and your result?

im not expert but i have experimented with this a bit this week

[broken elm]

I just noticed something called image influence.... I thought "low-mid-high-ultra-max" was it lol...

[half halo]

ok i know you were told to ignore legacy mode

but i have used it a bit, bgecause you can add more images as reference

and have more freedom on what you want them to refer to

take some time to look into it

[broken elm]

I've been using it since you told me about it

[half halo]

did you get to the screen where you can add a bunch of images

and then click what kind of reference they need to be

style, character, content, pose?

[broken elm]

yeah

whats the command for spoiler just in case

[half halo]

i THINK

you paste here

and before you press enter

you can right click the image

and set it to spoiler

if you cant let me know i think there is a way to do it using a command type thing

yeah!!

[broken elm]

everything is to max cuz i was fooling around but it should be "mid" for all 3 style

[half halo]

thats it

well just think of it like this

if you want it to make an image

and you had to order those things

what would be first and what would be last

to make you get what you want

[broken elm]

so far style reference mostly affect colors

from what ive noticed

[loud vessel]

how do i search through the community creations? its liek the feature has been removed

[half halo]

if you are on the main page

[loud vessel]

i am

[half halo]

click home

thats the feed

[loud vessel]

can i not search through it? im watching a video and they have a search option

[low bear]

The search bar in the public feed is indeed disabled, as it was causing database issues unfortunately

[half halo]

tbh ive not seen in for a bit

[loud vessel]

ah i see, thanks

[half halo]

is that a temporary fix ruby?

until you all figure it out?

with your massive digi brains

[low bear]

Well it was supposed to be, but it has been going on for a looooong time now

So I'm not sure if it'll ever be resolved

[half halo]

well to inspire you

heres a massive digi brain

i bet that could do it

dont let it beat you

[loud vessel]

i want to create logos for in game use and i cannot find much information on where to start as all the videos are out of date

[half halo]

I think for this you should do /mentoring

it will get you a knowlegable person to help

[broken elm]

What's the difference between image to image and content reference?

[low bear]

Content reference ignores colors

[half halo]

image to image just takes that one thing to reference to without subtlety i think

when you put it into the actual page fo rit

[low bear]

It only brings the objects in your reference image to your generated image

[half halo]

you can put more details and subtlety

[low bear]

Whereas image to image references the full pixel data of your image, including colors

Which may be undesirable because that way its hard to keep detail while still being able to change the image

But it can also be really fun, for example what I used to do is upload flags to image to image at a very low init, and then generate landscapes with it

[half halo]

almost looks like this guy knows what hes talking about more than me saying random things to seem like i know whats im doing

[low bear]

So I'd have flag landscapes!

[half halo]

cant be legit

should be banned

[broken elm]

Yeah that's what I thought

[low bear]

aw I can't find my favorite one

But I made this by inputting a 45 degree rotated rainbow gradient for example

[broken elm]

Right now I'm unable to generate stuff without uploading something i uploaded elsewhere =/

[half halo]

❤️ @low bear dont make AI randy pictures of me 😦

[low bear]

What do you mean?

this one is the flag

this was a year ago btw

you could probably do even more awesome stuff with it now

[half halo]

Ruby how many of your images do i have to like before you upgrade my account to unlimited tokens so i can go mad and teach this AI whos boss?

[broken elm]

with some other ai, i trick it using prompt like "(((wear horse armor)))", hit generate bunch of times and I get some super creative result

it merges horse's features into the armor

[half halo]

thats weird, i got a much different result beca... oooooh... HORSE... not WH.. i see

[low bear]

I don't work for Leonardo.Ai anymore lol and even when I did I didn't have access to peoples accounts 🤣

[half halo]

awwwwwwww

ok let me know who does 😄

[low bear]

you should be able to do this with Alchemy as well

[half halo]

my budget is small but my creativity is big

[low bear]

We used to give out freemium accounts to the very helpful contributors

I don't know if we still do

[half halo]

i can be very helpful!

the thing is, i have time not money 😄

[low bear]

Yeah you've been doing great :)

[low bear]

Well if you keep helping around the server you'd make an awesome volunteer, and volunteers do get freemium

[broken elm]

@half halo @low bear in this image above, She is wearing an "isopode armor", the reference image

[half halo]

woohooo, im definitely sticking around anyway

[low bear]

Ohhh hahaha I see it that is great

[low bear]

Yayy! Sounds great

We always love getting more active members in our family :)

[half halo]

what exactly are you trying to create?

can you show us your prompt?

also i would love to see what the result that you got so far is

[broken elm]

yes a moment im multi-tasking

i was looking for that horse armor pic, kinda hard to find

[half halo]

confirmed you are a male then

oh wait, no, was that the other way around?

i forgot 😄

you do you, ill go get another whiskey

jesus christ

you need to take a week of and order that folder

[broken elm]

well it has folder and i save my prompt in a notepad, but i categorized folder by people instead of by armor design type

[dreamy birch]

has a limit of 200 characters

[broken elm]

this is sooooo gooooood

[broken elm]

@half halo found it, was worth it, dunno why i didnt document the prompt

[night vault]

https://cdn.leonardo.ai/users/b8b7cb49-52d0-41e7-9766-6d1c4c5ea529/generations/0b8b72e8-1ede-4d2b-9b4e-737a108b4a21/Leonardo_Anime_XL_granblue_fantasy15_homura_genshin_impact_cel_2.jpg

[turbid shoal]

Does anyone have any advice on how to make this androids face a little less human?

My generations attempting to alter just her face aren't really going well. Friend loves the image but just wants her face "a little less human"

[real lynx]

Steampunk androide Remove any word that says human or humanoid, put creature female for example

[hard sand]

So I'm using anime realism and I'm trying to get the girl with the yellow dress to be in the same style as the girls in the white dresses they style looks fantastic with the particles and etc, all of the images are on the same anime realism but the quality is different. what should I do ?

[brazen fractal]

guys I need help creating 2d sprite animation sheet for game development any tips i might use to get the desired output ?

[thin tinsel]

That is possible but will take a lot a little while to get there. Can I recommend you to sign up for the mentoring
type /mentoring

[thin tinsel]

I think the difference will be something in your prompt. Check the wording, or share your prompts for both and I can see if that is the case.

[hard sand]

@thin tinsel I had to use also content referance for both since I also have to make a more accurate image with phoenix model then upload it to the anime one . So this is the prompt for the blonde hair woman. "A regal Marie Antoinette walks with confidence through the grand ballroom, her head held high. The image captures her delicate features and elegant attire in exquisite detail as she glides across the floor. Her powdered hair, adorned with jewels, sparkles under the soft glow of chandeliers, while her ornate gown shimmers with intricate beadwork. The ballroom's opulent decorations and lavishly dressed guests create a lively atmosphere, immersing viewers in the extravagant world of 18th-century French royalty. Every step and movement is depicted with stunning precision, inviting observers to step into the lavish setting of a bygone era." and heres the other one "belle stands near the forest near her cottage with wildflowers talking, Belle is a strikingly beautiful young woman in her early twenties she is wearing a modest french dress with white and blue. She has long, flowing chestnut hair and eyes the color of amber honey. Her features are delicate and expressive, with a natural grace that belies her humble upbringing. Despite the family's reduced circumstances, Belle carries herself with quiet dignity. She often wears simple, practical dresses, but there's an undeniable elegance in her bearing. Her hands, though slightly roughened by work, are graceful and often holding a book. Belle's face frequently displays a look of curiosity and intelligence, with a gentle smile that hints at her kind nature."

[thin tinsel]

Here is your new prompt for image two:

"A beautiful young woman in her 20s, long chestnut hair, wearing a modest white and blue French dress, 18th-centrury, stands confidently in the forest with her small cottage blurred in the background, her head held high. The image captures her delicate features and elegance. A calm and sun-lit atmosphere, immersing viewers in the simplicity of a bygone era."

Prompt for Belle is WAY too long and far too much information the AI would be destacted by. Keep your prompt simple and to the point, and use the built in Prompt Enhance to add details.

Also ... try using image one as style refernce on a low weight when prompting image two and see if that helps

[hard sand]

ok, thanks!

[thin tinsel]

@steady totem , come to this channel if you have questions

[steady totem]

@thin tinsel ok friend, tnks

[last rapids]

my alchemy v3 seems bugged, anyone elses?
i attempt to edit the settings (they default changed to what i was using before) and it auto turns off alchemy - and my remix's (everything is the same) are coming out drastically different

[thin tinsel]

report it to the support team

[last rapids]

who is someone i can dm, or a specific channel?

[supple phoenix]

im trying to expand this image to the top left but the houses always end up looking wierd how can i avoid that?

[hard sand]

@thin tinsel so I did the prompt with the image on low for style, but her hair is the same as the style ref though it should be brown, but I love the look of the blurred background and etc any ideas?(And her dress isnt matching the prompt)

[real lynx]

Hi

Is this what you need?

She is imitating this dress and her pose and keeping the face of the image of the girl in yellow with the hairstyle of the second image

You have to give more strength to the character image

And put it in mid if you want to maintain the pose

And if you set the prompt to low it will start modifying the image as well.

If both images are strong, the prompt has almost no influence at all.

[broken elm]

There was an error with the submission. A moderator will be in touch shortly. o.o

[broken elm]

put 10k credits down the drain haha, im starting to think about trying to learn to draw, again xd

[hard sand]

Thanks! I'm trying to get the quality of the first images onto that image, you can see that her face isn't as detailed as the other images.

like these are excellent styles I want but its hard to make sure Leonardo does the same look.

Like this is the character I want to have but more realisitic

And in 16:9 format 🙂

[twilit fiber]

I have this bug (?) where all the tools in the Canvas are off ; it's the same with the sketch and eyedrop tool (which makes everything harder) - any idea ?
(the pointer and result are not aligned)

[plain moat]

hi, how can i get a normal room but the light to be just in fron i mean i want in background like behind the person to be dark and to be light only in front because of the lamp from the desk. like what's the promt for this ? who can help me

[plain moat]

so?

[real lynx]

Good morning

The first image I made in Phoenix, the second with real style anime

A dark room with a single lamp on a desk next to the bed. The desk has a few items on it, including a book, a pen, and a small potted plant. The bed is made and has a white duvet cover. The floor is covered with a beige rug. The walls are painted a light blue color.

Is this what you are looking for?

[plain moat]

i don t think so..

let's say i have a room like this and you see that there is a desk in front of the image and in the back a bed and things like a normal camera. i want the light to be just here in front where is the desk and in the back to be dark to highlight the man sitting at the desk

something similar to this

you see that light is just in front like where the character is

and in the back is darker

[stable carbon]

hi, i had that on firefox too. i couldnt find a way to fix it, and support recommended swapping to Chrome browser, since its Leonardo's suggested browser. no problem on that one 🙂

[plain moat]

so?

[real lynx]

I already showed you a prompt and how to create a light in the dark

I'm just helping him

Then you are forcing me to create your image, please don't be mistaken.

Thank you

[plain moat]

bro what are u talking about?

you asked me if that s what i m looking for

[real lynx]

You changed your instruction you only wanted a light in a room now you want something else

[plain moat]

and then i showed you how it should looks in the back

lol

[real lynx]

Then it's your turn to try.

[plain moat]

WHAT???? I EXPLAINED HOW IT SHOULD LOOK THE LIGHT

BRO READ AGAIN

[plain moat]

HERE I TOLD YOU HOW THE LIGHT SHOULD BE

THEN I GAVE YOU A IMAGE AS AN EXAMPLE

[plain moat]

.

dayum what s this server

"mentore" and does not understand

[plain moat]

yh something like this

[real lynx]

You just have to describe what type of lamp you want in front and what type of background you want in your room.

Use Phoenix as preset

[plain moat]

ok

[real lynx]

A photo of a lamp on a desk in front of a dark room. The lamp casts a warm glow on the desk, illuminating a few items. The room is dimly lit, with a few visible objects such as a bookshelf and a chair. The background is blurred, creating a bokeh effect.

[twilit fiber]

Thank you. I don't use Chrome, but I'll keep that in mind.

[real lynx]

Excuse me, I misinterpreted and misinterpreted everything

[plain moat]

ok no problem my bad for my language cuz i didn t explained instead of writing with caps on

i need help again if u can.. i want this character to have his eyes covered by his hood i will give you an example in some seconds

[real lynx]

This tool can also describe your image and give you a prompt for a similar image.

[plain moat]

ok, good to know

[real lynx]

Image is a digital illustration featuring an anime-style character. The character is wearing a dark hooded sweatshirt, with the hood pulled over their head, obscuring their eyes and casting a shadow over the upper part of their face. The visible part of the face shows a light skin tone and a neutral expression with a closed mouth. The background is a solid dark color, enhancing the mysterious and enigmatic atmosphere of the image. The artwork uses clean lines and a limited color palette, focusing on shades of black and gray, with subtle highlights to define the contours of the hood and face.

[plain moat]

ah, thanks alot!

[real lynx]

Use this tool with your image and it generated the prompt for me

[plain moat]

ok

[real lynx]

Also if you have premium tools this process would be easier

It also has free elements that can modify your image.

[real lynx]

You can even combine them

[plain moat]

ok i understand

i will upgrade to a plan soon

[real lynx]

The elements are free

[plain moat]

yh i know but i will need to generate a lot of images soo i have to upgrade

[plain moat]

can you look in dms? i need help with something but i don t want to share on this chat

[real lynx]

Hi

I was having breakfast

How can I help you?

[plain moat]

i need help with an image but i want to get help in dms because i don t want to share on this chat

[real lynx]

You can use the /mentoring command

Opens an exclusive private room for you

Where only mentors from all communities come to help you

[plain moat]

done

[broken elm]

I opened mine yesterday but no one came to help...I think maybe I figured it out though, but it took about 15k credits haha

[real lynx]

Hi

Yes I saw your images

[fresh island]

@real lynx how can I generate images ?

[real lynx]

Hi Only on the website

Here we share images and ideas

[fresh island]

Thnx

[real lynx]

https://app.leonardo.ai/

[real lynx]

If you have any questions, you can come back 🫡

[fresh island]

Sure thnx bro

[delicate lily]

Can someone explain me why the results in real time canvas are so low quality on a mobile browser compared to an windows browser

[marble cape]

Hey guys what about this one

[marble cape]

Sorry this

[marble cape]

Its element kanji symbols in there opposite opposing nature weakness counter parts

[broken elm]

Can something create in canva editor be sent my gallery/collection?

so it can show up on my profile and stuff

[real lynx]

Do you want something like that?

I turned it up to the maximum

I'll modify my prompt a little because I didn't modify it

This item might also help if you want to do something else

[marble cape]

Canning what about this

[real lynx]

Don't know

What do you need to do with that?

[thin tinsel]

I think he's just asking your opinion on it.

[real lynx]

I don't know what those symbols mean.

My translator does not translate

I can't give an opinion

I can guess things like it says water fire and I don't know what green would be

I'm bothered by bright colors, that's the only opinion I can give.

[thin tinsel]

no, it can't. we always ask for it, but it doesn't happen. so we just do image to image,

[broken elm]

🙏

[thin tinsel]

Doing image to image even on the full weight, will still result in some minor changes, so just bear that in mind so you are not disappointed

[broken elm]

I mean the biggest issue I have with it is the scaling down, ( I had used phoenix’s with largest scale available)

[thin tinsel]

And then edited in canvas? 😮

[dim zodiac]

I'm pretty sure running it through Universal Upscaler will also end up in your feed. Can do that without increasing the image size

It's got a bigger upload limit than the image2image framework does

[broken elm]

I mean im new, I don’t know how everything work, and proper techniques, asked for help but nobody came XD, i tested out most styles with smaller images until I could find one that seems to be decent at what I was trying to make, turned out to be phoneix, edited the prompt until i could get something that would generate something closer to what i wanted on a more consistent basis, then i finally started generating bigger images until i get something close to what i wanted, then i canvas edited it

Then i wanted to apply a new style to it

[thin tinsel]

Will it keep the image the same and not change the face for example?

[broken elm]

I used content reference maxed

[thin tinsel]

I saw you receiving help in the image-help channel the other day. Aren't you the Wolf-dragon person?

[broken elm]

No

I just bought premium friday night

[thin tinsel]

And did you sign up for a mentor?

[broken elm]

Yes its up and running

I typed all my progress in there but no one came until the end

[thin tinsel]

In the end? How long did you waiit. What day did you put the request in?

[broken elm]

Lemme check

[broken elm]

05/10/2024 20h53

[thin tinsel]

Ok, I've looked in the mentoring applications, and I can't see you there.

[broken elm]

It had said “There was an error with the submission. A moderator will be in touch shortly.” But the channel was created and all and no moderator ever communicated with me

[thin tinsel]

Try again now and I can let you know if it goes through.
It's quite possible that it was around the time that we switched Bots

[broken elm]

Well i cant write the long essay again for goals and all that =[

[dim zodiac]

Universal Upscaler can keep the exact same image if you keep the "creativity" low and "similarity" high

Haha I'm too slow responding while at work

[broken elm]

I have ocd, just to fill up that took me an hour 🤣

[dim zodiac]

You only have the one thread though... With the elephant mantis stuff?

[broken elm]

Yes

[thin tinsel]

Did you find it?

[dim zodiac]

Well then it's been active 😄

[broken elm]

I know it has

[dim zodiac]

All the mentors are volunteers, sorry if it took a bit of time for it to first get popping

Our bad!

😁

[broken elm]

To be honest im just mad i went from 25k to 10k credits for one image, im happy more or less with the final result i reached. I’m sure there must’ve been a more cost efficient way to do things

[dim zodiac]

You'll find the discord response time will vary depending on time of day and such

[dim zodiac]

I bet it will get better now! The first attempt at a new project I always find to be a token sink, especially when trying something that is out of the "ordinary"

Getting the AI to step out of what it is familiar with can take a lot of tinkering and a lot of tokening

[broken elm]

Well I didn’t pick something easy too i think, but i could of made it even harder

[dim zodiac]

I would say start in the aspect ratio you like but with a smaller size until you have a prompt that is giving you decent results (even if not perfect)