🙋│image-help

[dim zodiac]

I usually gen 2 at a time as my starting point too

Then you can blast it bigger, and with image reference if you've got something workable

[broken elm]

I mean elephants already have tusks but what if i had picket a rabbit instead and wanted to add it tusks, a trex jaw and elephant trunk, that would of sucked i bet

[dim zodiac]

And final details in Canvas is always way easier than trying to fix it in prompt

Yeah some fusion animals are easier than others, that's 100% true

[broken elm]

Not sure about that, maybe i dont use it well

[thin tinsel]

Better to learn how to use all of Leonardo's features and functions with smaller projects than diving in with something so difficult, and wasting tokens.

[broken elm]

I started using 1 generation, just to figure out which style is more creative than others. Some presets really were restricted creativity; couldnt think out the box trying to fusion things, it would do one or the other. The i narrowed it down to a few style, moved to 4 small images, to tweak prompts

Then when i changed to a more portrait style image instead of 1:1 that really helped, tweaked the prompt over and over until i could get close to what i wanted

Then i went bigger scale 2 at a time, prompt tweaking, finaly 4 when i was close, until i got something i thought had most of the elements i wanted that i thought could be fixed with canvas

Then i had to figure out canvas XD

I hate it and i love it at the same time

If i typed “add carapace dragon like”, painted the back, it would keep popping dragon heads

[dim zodiac]

Ah yeah, this is a thing I see a lot. It's tempting to "talk" to canvas

But really you just want to prompt just like image generating

So "Dragon scale carapace" would be a better prompt

[broken elm]

I couldnt add big hooks to the trunk, i settled for small spikes, lastly decided to do more canvas to fix the anatomy and modify the tail, that was also a nightmare

[dim zodiac]

Along with masking the area you want to see it, and using Inpaint maxed to 1

For this type of fantastical generating, I would suggest using Sketch2Image mode

[thin tinsel]

Focus mode might be good for those types of changes.

[dim zodiac]

You could use the eyedropper to match the tail colour, draw in a really rough spike, and then prompt for "bone spike" or something

[thin tinsel]

@dim zodiac do you know where Zro Defects' guide about focus mode is?

[broken elm]

Didint work

I wanted a big spikey ball tail, it wouldnt do it, at some point i managed to get one but it was tiny, i could make it better, than i thought about moving the generated image thing, erase everything but the small ball tail, scaled it up, then rotate, etc, placed it back on, asked it to blend it , it worked

But it looked very blurry

[thin tinsel]

https://discordapp.com/channels/989166677390426132/1280845243561803838/1280851326628266057

@dim zodiac I found it

[broken elm]

I try over and over with different prompts to have it remake it, it wouldnt, then i did a big ball but no spikes and now that is where i am at XD

[thin tinsel]

and where is your mentoring thread?

[broken elm]

It’s a problem I have, I wish I could but I can’t motivate myself to do that, I get too bored from those. Which is why I thought about the mentoring thing 🤣

[broken elm]

I’m not sure how to link a thread and would also be nice if someone could tell me how i invite someone in there XD

[thin tinsel]

Ok, well if you know where your mentoring thread is, you can do @thin tinsel because I can't find it.

[dim zodiac]

Yeah I was just a bit slow on the copy paste

[thin tinsel]

no worries.
I can never remember where the guides are

I just thought of the bit where the bat had a frogs head

[dim zodiac]

Yeah, although Zro always prompts like he's talking to Canvas...

[dim zodiac]

You've posted this about 10x including in your own mentoring thread. I don't know what you want with it. It's fine, very basic, you need to explain what you want to achieve to get any feedback. Please do this in your thread

[nimble bay]

Question - Phoenix now supports certain guidance styles. Why is the platform charging tokens if you have the wrong input selected? Anyone?

Surely if a certain guidanc option in not avaliable on Phoenix we shouldn't be charged for it?

[thin tinsel]

You can report this using the Intercom.

@steel cosmos Happy to share your prompt?

[real lynx]

Phoenix only supports image x image and image Style reference

[steel cosmos]

Hi, so some help required. 😄 For example, this is the start of my prompt: Closed mouth, closed lips portrait of a mesmerizing, ethereal female emerges with enhanced beauty, preserving her original aesthetics.

Then.. I use a modified reference image (i generated it earlier but.. it was with open mouth once again) I modified it myself locally with LivePortrait because.. I want the mouth to be closed.. Then when I use it for content reference, the mouth is still open..

[thin tinsel]

Hi Cami 👋

[steel cosmos]

Closed mouth, closed lips portrait of a mesmerizing, ethereal female emerges with enhanced beauty. For exmaple.

[thin tinsel]

What model? Any other image reference ?

[steel cosmos]

Kino XL model

[real lynx]

So you are uploading a character image that has its mouth open and wants to close it.

[wanton belfry]

stop putting in mouth

[steel cosmos]

no, its closed on the image im using as content reference.

[wanton belfry]

the general outcome is closed mouth

[steel cosmos]

Even without.. still. most of them have open mouth\

[thin tinsel]

Give me your full prompt and I will re-write it for you

[wanton belfry]

which model are you using?

[steel cosmos]

Kino XL

[real lynx]

You have to give it less force so that the prompt can close its mouth.

[steel cosmos]

This was the full prompt i was using. But I tried many things. Believe me. 😛

Closed mouth, closed lips portrait of a mesmerizing, ethereal female emerges with enhanced beauty, preserving her original aesthetics. Her vibrant, rich purple hair flows down her back in luscious locks that cascade around her heart-shaped face, framing it with subtle, wispy strands. Her skin tone is a radiant, sun-kissed complexion, characteristic of a 22-year-old Korean woman. She wears a soft, pastel pink cropped cardigan with a matching mini skirt, complete with a tiny bell attached to the center of the cardigan. Her face is adorned with iridescent freckles, drawing the viewer's gaze to her hypnotic blue eyes. Her lips are subtly pursed, adding a touch of mystery to her captivating expression. Captured in a stunning, photorealistic 8K photograph, she exudes an otherworldly radiance that mesmerizes viewers.

the pursed lips part wasnt there before btw. I tried it without that as well

[steel cosmos]

Maybe I should use a different model? Idk. Really.. Ive been trying for hours already.

[wanton belfry]

hmmm probably might be better in portrait than landscape format

[steel cosmos]

ye but I want landscape because of what I want to do with it

[wanton belfry]

your looking for realistic or more fictional?

[steel cosmos]

realistic

[wanton belfry]

yea its not really returning completely closed mouth in most of outcomes

[steel cosmos]

ye because.. for what I want to do. It will make my life a lot easier when the mouth is closed heh.

[thin tinsel]

Here is your new prompt

A stunning realistic portrait of an ethereal 22-year-old Korean woman, Blue eyes, flowing purple hair, wispy strands. Wears a soft, pastel pink cropped cardigan, tiny bell in the centre, pastel-pink mini skirt,. Iridescent freckles. 8K photograph. Photo-realistic.

[steel cosmos]

lets give it a shot

[wanton belfry]

this is with vision than kino

[steel cosmos]

generated 8 images, all have teeth visible

But the Kino model gives me more of the result I want. Besides the mouth part heh

But I also tried lightning, diffusion and vision

[wanton belfry]

Kino is pretty cinematic

[thin tinsel]

Put Teeth in the negative

[steel cosmos]

yep but thats not in landscape

ill try

[steel cosmos]

somethinjg like this is perfect, but not the format I need unfortunately.

I guess I cant share images here right?

I can nm

on all of them the teeth are visible. Even when its slightly, it makes me hard for me to use.

[thin tinsel]

Still a little relaxed

the mouth I mean

[steel cosmos]

its better than when I used my own prompt

[wanton belfry]

sadly its probably trained on models with that sort of pictures with mouth slightly open

[wanton belfry]

it'll return it most of the time

[steel cosmos]

I see

[thin tinsel]

@steel cosmos You can also use the EDIT WITH AI option. Have you tried it

[steel cosmos]

ye

Ill try that now, but the prompt I shared was also modified with AI

[wanton belfry]

oh wth I keep forgetting we have phoenix now

agh

[thin tinsel]

When your prompt is so full of information, the AI does not know what to prioritise, so that it why I slimmed down your prompt, even though the prompt enhance tool itself gives you a lot of detail

[steel cosmos]

this part was added with the ai, but that made no difference. still all of them slightly open.

[thin tinsel]

Try this:

A stunning realistic portrait of Korean woman in her 20's, hypnotic blue eyes, flowing wispy purple hair. Her lips are pressed together, firmly closed, giving her an enigmatic smile. She's wearing a soft, pastel pink cropped cardigan with a tiny bell in the centre, with a pastel-pink mini skirt, and white sandals. 8K photograph. Photo-realistic.

[wanton belfry]

just put the prompt in Phoenix and run

[thin tinsel]

Phoenix

[wanton belfry]

oh and put in pursed lips

[wanton belfry]

he's looking for landscape ratio

[steel cosmos]

This might have done it. Let me give it another try

Ye Phoenix seems to listen to the prompt in that way. But I like the Kino model a bit more for what I need it for.

[thin tinsel]

I know. I'm just helping with the prompt, not making images for them

[wanton belfry]

hmmm phoenix does tend to look plastic

[steel cosmos]

I think I can work with that prompt you gave me. These seems better. Still not all of them have it closed, but I do see some better results.

[steel cosmos]

mye 😦

[wanton belfry]

oh another thing, forgot photoreal

[steel cosmos]

ill try that as well

[wanton belfry]

its costly, beware

[steel cosmos]

ye.. its okay. hehehehe

But this helped me a lot! 😄 Thanks!

[wanton belfry]

this is a rare problem I never realized was there

[steel cosmos]

ye.. i can tell you it was kinda frustrating.

hehehe

[thin tinsel]

We are still subject to luck when it comes to AI images.
But I think your're at least getting some closed mouth images in each batch of four

[wanton belfry]

of course, whenever it gets frustrating its best to ask for help than waste 100s of tokens to get nothing good

[half halo]

Its a lot of fun messing around with it, but yeah sometimes you go through tokens very fast

[wanton belfry]

sometimes you get notified that your model can't produce that...

then you feel dumb

oof

when I started out and waste 100s of tokens onto something only to realize it wasn't even possible

[steel cosmos]

yikes

[half halo]

often when i dont get what i want, and i can tell the AI is just not following my prompt, i switch presets

that can make a huge difference

[steel cosmos]

ye I tried switching around a lot as well, models, presets. Maybe its just really related to the landscape image size or something.

[thin tinsel]

I'm just trying out Lifelike Vision to see if that behaves any differently

[half halo]

I think each preset just has learned different things over time

and they focus on different aspects

at least thats what it feels like to me

i spent hundred of tokens trying to make a jacket sleeveless, and it just wouldnt do it

i switch presets, instantly got what i wanted

[thin tinsel]

The Infrared photography element works really nicely with your colour selection.

[steel cosmos]

Thats look pretty cool. Ill give it a shot as well

[half halo]

oh and a nice trick when you are generating, have a close look at the token cost vs how many images you generate, sometimes generating 1 is like 15, two is like 18, three is 19 and it goes up with steps from that. So your sweet spot while you are still refining your prompt would probably be three images per turn there

[steel cosmos]

I usually do 4 at a time.

[wanton belfry]

compare costs first

[half halo]

yeah for sure

[steel cosmos]

ahh I see

[half halo]

theres gains there

[steel cosmos]

Alright, noted.

[half halo]

took me a while to notice it too

[wanton belfry]

lightning models have lower costs

[half halo]

im down to 150/8500 now 😛

feels like im broke

[steel cosmos]

im at 0.. but sub refreshes soon. relaxed generations are still fairly quick tbh.

[half halo]

and tbh i spent like 6000 of those needlessly looking back

its just so easy to keep clicking lol

[steel cosmos]

eheheh yep

[wanton belfry]

takes time to gain experience and learn this as a tool

since there are many models, now presets too

[steel cosmos]

ye for sure

[wanton belfry]

me and many others can make alot of cool stuff, just because we know what to do and when to use what

[half halo]

yeah for sure

i took a break for like 8 months and its so different now

have to relearn everything

[wanton belfry]

I guess you were not here when XL lightning came out?

[half halo]

oh no

I was most active in beta

[wanton belfry]

Beta, oof only DS was good enough back then

[half halo]

Yeah its definitely come a long way since

[wanton belfry]

yeah I wont deny that

[steel cosmos]

Have you guys tried Kling AI as well?

[wanton belfry]

no, generally its best to stick to one

everyone has got different models

[steel cosmos]

ye but, for image to video

because the images you make in Leonardo are awesome to use for videos in Kling

[real lynx]

You can also combine elements

This combination is magical, I like it a lot.

[steel cosmos]

Oh nice!

[wanton belfry]

I am more about styles

[half halo]

thats a very cool image kentox

[thin tinsel]

white ethereal is lovely

[real lynx]

Copy your pose :3

[wanton belfry]

put in word "Edgy"

[plain moat]

hello

[wanton belfry]

it'll put you in a brooding pose

[plain moat]

i need help

[wanton belfry]

with?

[plain moat]

i need to change this character mouth so the mouth is closed and neutral

i am in canvas editor right now

but i don t know how to do it

[half halo]

I think i would do this using image guidance instead

[plain moat]

how

[wanton belfry]

@thin tinsel Someone needs help with canvas

[half halo]

i think one of the mentors can help you best 🙂 If they are not available ill help you out

[plain moat]

nah no willow please i talked to on room and he got upset

[wanton belfry]

no, it wouldn't really work image takes reference from existing image...

[wanton belfry]

its not a he...

[plain moat]

then she

[wanton belfry]

thats weird, she's the nicest ones around

[half halo]

well i dont want to pry but as long as you are curteous and patient im sure they are willing to help 🙂

[half halo]

that reaction is not very surprising 😛

[wanton belfry]

yes, she's not a bro..

[plain moat]

i didn t know that she is a she and not a he

soo

[half halo]

so what did she tell you on how to use canvas?

[wanton belfry]

she's not angry, just expects you to spend your tokens and work to get what you want

[half halo]

she seems to think you already have the answer

[plain moat]

that s how the chat ended

[half halo]

i think she gave you information before that?

[plain moat]

she helped me before with some things but i tried and doesn t work and i asked again to help with this specific

[wanton belfry]

bah I dont use canvas much, but let me give it a try

[plain moat]

okay

[half halo]

yeah i would try and help as well but im down to 150/8500

[plain moat]

=///

you can tell me what to do and i do

[half halo]

So since you started the process you have made progress right?

[plain moat]

yeah

[half halo]

can you look back at what youve learned and see maybe if you already have the tools you need?

[plain moat]

i tried but doesn t work

[half halo]

what have you tried?

[plain moat]

with the things that i did before

[real lynx]

He also called me bro, I got to the point of thinking that it's normal and that there's no difference whether it's a woman or a man, everyone calls him bro, that's why I don't get upset.

[plain moat]

with erase then mask and use the promt or mask over and prompt or erase then draw then mask then prompt

[real lynx]

Willow She is always very patient, I don't think she is angry.

[plain moat]

things like these that willow showed me

[plain moat]

because i don t expect to be girls on this discord servers

and for me is normal to talk with bro

i mean is instinct

[half halo]

women are everywhere my friend 😄 its good

[plain moat]

i haven t see many womans on discord servers like this soo this is the reason

[half halo]

fair enough, live and learn

[plain moat]

yh

[half halo]

i think give kentox a little time to try a few things

[plain moat]

are u a she or a he? =))

[half halo]

i am a he

[plain moat]

ah okay

[wanton belfry]

I am pretty bad with canvas, I did get what I want however there are some pixelated mess in there

probably because image is too small

[plain moat]

how u did?

i have the full image so it may works at me

[wanton belfry]

just drag in the image, change to portrait ratio, fit your image in that rectangle, write "Anime, Closed lips" in prompt and its done...more or less

however there are some pixel messes there

not sure what to do with that

oh oops, I realize I forgot to mask it first

nevermind, I am dumb

not fully close but I have closed version too

you gotta mask the lips proper

I accidently erased the lips instead of removing the mask and thats why it was pixel mess, ugh

[wanton belfry]

its not that hard, but might take a few tries

[plain moat]

ok

i m gonna try now

[half halo]

i mean thats not a bad job at all

[plain moat]

doesn t work

it just change a little bit the mout smiling

but just a bit

maybe is because of the settings

[feral mica]

if you are working with the mask, the outpaint options should be off

like this

[plain moat]

is off

[feral mica]

if you are deleting completely the mouth, then turn outpaint on

[plain moat]

this is on

i just want the lips to be closed

like no smiling

[half halo]

you could delete the mouth first

and then remake it closed

[plain moat]

how? i tried that to but i m not doing it right

[feral mica]

select the eraser and just delete the mouth

[plain moat]

and then?

[feral mica]

turn on the outpaint, prompt the closed lips

and remember to pick an appropriate model, I'd go with dreamshaper or albedo, but that's up to you

[plain moat]

and before prompt and generate i have to put mask on right?

[feral mica]

no

if you already deleted, no mask is needed

[plain moat]

ok

[feral mica]

if you want to use the base image then you should mask and don't delete, here you'll need the outpaint option off

[plain moat]

worked the first one

thanks

[feral mica]

the result image doesn't need to be perfect, you can refine it with the same process, here it got some odd points around the mouth

so in this case i used the mask only,

voila!

[plain moat]

i did 2 gens of 4 images to get this

is ok

Thanks mate!

[feral mica]

and thanks @wanton belfry and @half halo too!

[plain moat]

yh

thanks guys

[half halo]

you are very welcome 🙂 Glad you got it sorted

[plain moat]

and also learned some things

[half halo]

thats the most valuable thing

[wanton belfry]

good, it just takes a few bit of tries

keep going

[half halo]

You can give a man a fish and he will eat for a day, teach him how to fish and he will eat forever 😄

[plain moat]

exactly

[dim zodiac]

well... glad that you got what you wanted out here

[azure rapids]

Hi. The upscaler isn't preserving the transparency. Is that by design?

[half halo]

Have you tried another preset?

I dont know if its by design but sometimes features are not available on the new models

[azure rapids]

it's aaaalways something

[half halo]

well you know, constant development 🙂

were living on the cutting edge here ya know 😉

[thin tinsel]

Which upscaler are you using?

[steel cosmos]

Hi all, I'm back. 😛 Been getting a bit more consistent results, which is good. Got another question. Any way to fix 500 toes on a realistic generated image in the Canvas?

like.. make it 5. :p

[thin tinsel]

hello. There is a video on the YouTube channel about fixing hands

https://youtu.be/hKf2kxbgmhE?si=ZKToDPj-P7t37N5P

[steel cosmos]

I'll take a look, thanks

[azure rapids]

Hello. I would like to generate a photo of a person's back with a tattoo of this image in the middle of the back. I would like to make it a close-up photo, so basically just the white background replaced with the skin of the back. Is that possible? Thanks!

[cold geode]

#image-help create a girl with brown hairs take tea on park

[rustic tree]

hey all, i generated an art image of a character based on an image guidance input. i was able to edit some aspects of the character using canvas, but when it came time to replace the character's long white hair to a shorter black hair i'm having these weird yellowish lines that pop up along the areas marked by the eraser (as seen in this picture)

is there a way for me to fix this?

i removed the brackets and the 1.3 from the prompt and it's mostly gone but it's still visible... 🤔

[real lynx]

https://app.leonardo.ai/

Good morning Only on the web you can create images

[tender olive]

hi quest i need to train model eg girl or boy to wearing my clothing brand stuff how to do it ?? i need models and only changig clothes ?? something like this pic dont editing in other app

[thin tinsel]

Didn't Botticelli tell you how to use Leonardo before you were given access to the server?

[thin tinsel]

What was your process? Did you erase and mask over the whole hair?

[rustic tree]

to clarify when you say mask you mean like painting the hair over with the eraser tool right? if so then yes, then i used outpaint

[thin tinsel]

No the mask tool is above the one up from the eraser on the left-hand toolbar

[rustic tree]

ah in that case no, i only used the eraser tool

[thin tinsel]

Ok. Well to be honest it will work like that, but I like to mask as well.

I can see all your other settings are fine.
Sometimes you just get weird results, and if you come out of it completely, and delete everhing off the canvas and go back in it will work as you envisage

I did mine just using Focus mode. @rustic tree

[open echo]

@vague nymph
Hello there 😄

to answer your question about features and availability, indeed Alchemy (or « Quality » mode in the new interface) is only for paid users.

There are sometimes trials and free grace periods where free users may be able to use them for a time, but mostly these are premium features

I hope it answered your question 😁

[vague nymph]

Yes thank you.

[rustic tree]

i see, i'd like to know what is focus mode and what's it used for?

[thin tinsel]

The focus mode lets you draw around a specific area of your image, then prompt for what you want to see instead.
See the guide: https://discordapp.com/channels/989166677390426132/1280845243561803838/1280851326628266057

[rustic tree]

ahh okay, isn't that similar to say like the mask tool? or are they quite different?

[thin tinsel]

nope, it's just a bit quicker maybe

[rustic tree]

got it, well i'm out of tokens atm so i'll give it another shot tomorrow, hopefully there's none of those weird graphics again.

but if they do show up, i read that guide you linked it says something about just prompting to "remove object" while masking over the unwanted graphics, so i could try that as well

[thin tinsel]

Oh Ok, no, instructions don't always work. It's better to use describing language to say what you want to see.
But I use "BLEND" a lot it seems to understand that

[thin tinsel]

Let me know if you need any help with it tomorrow

[rustic tree]

i see, gotcha, will let ya know, thanks 👍

[slate dune]

Is that the case for anyone else here?

[dim zodiac]

I would guess you’re using the Phoenix model?

The other models will have Character Reference already, but it has not yet been implemented in Phoenix

We JUST got style and Image2Image though!!

What kind of image are you looking to make, I can suggest which model would work well for it

[slate dune]

Ah, I see. 😊

[slate dune]

That's very kind of you, but I'll try and figure it out myself for now. 🙂

If I get stuck, I'll get back to your offer.

[dim zodiac]

Cool!

[acoustic trench]

I want to make something like this can anyone hell please

[thin tinsel]

I said the same in general chat.

[real lynx]

Hi

[half halo]

thats the spirit!!

[real lynx]

What you can do is first create the character model with letters in Phoenix and then upload that image in albedo is another preset that allows you to use character image

This image was created in Phoenix

Then with the same prompt I uploaded it to image content in albedo and put a photo of myself in character

So I was able to copy the image of the letters and put my face on it.

[slate dune]

This is beautiful!

[half halo]

that is really cool

your elf like qualities have been enhanced with the powers of AI

middle earth will tremble

[humble ruin]

Hi. Does anyone know if there's a way yet to take different images of two different people and do one different image of them doing something completely different?

Sounds cheesy I know but a friend of mines wants me to do a fun image of him and his new wife riding dolphins using their pics but leonardo combines both people to hideousness i dare not show on this thread lol

[half halo]

Well you can use the ai image reference option to take the character of an image and have the prompt edit it as you like

this is where you can find it

it take some getting used to but i can recommend just trying out the feature and play with the prompt a bit

i have to admit though im not sure you can have 2 character reference images

you may need to have both people in 1 image? but im not sure, maybe someone whos tried this can chip in

[dim zodiac]

You would have to make them separately riding dolphins using character reference as Anthony suggested, and then combine the two of them in Canvas Editor

The Character Reference will only work for one individual at a time

[humble ruin]

OK. I'll try to do that. The canvas editor and I aren't the best of friends but I'll definitely give it a shot lol thanks both of you.

[half halo]

youre welcome bud, i hope it works and and would love to see the result 🙂

[real lynx]

My advice is to edit one first and then the other.

And then with some editor you put the characters together

If it is with the same prompt to preserve the background better

This is a friend

I had the AI describe its image and with that it gave me a prompt and I already put my face on it

And I stood next to his image

This tool describes the image

[half halo]

ooooh thats nifty

[real lynx]

I don't know what nifty means

[half halo]

I think nifty means a cool application of something, or a marvel, or using something in a different way to get positive results

at least thats how i use it

[real lynx]

Thanks for explaining, I speak Spanish and I didn't translate the word.

[half halo]

you are very welcome, clear communication is important 🙂

thank you for asking!

[ember current]

Hi all hope everyone well. I wondered if anyone had any prompt suggestions please to continually get a full height image of someone looking away from the camera, so I can only see them from behind, including the back of their head. I'm constantly getting results with their head turned to the side, but I don't want to see their face! Tried all sorts of pose to image, from seed, numerous prompts, negative prompts etc. Example result I continually get is like the below, but yes I dont want to see his face!

I've been using illustrative albedo, but just switched to phoenix and immediately getting better results! Sometimes just posting a help message gives you some inspiration I guess. I was using albedo for character reference. I'll keep experimenting on phoenix but any help welcomed!

[dim zodiac]

Phoenix has the strongest "prompt intelligence" so a trick you could do is generate your dude in Phoenix, and then use that as image reference to get the styling right in albedo

Also, character reference might be fighting you... It's designed to maintain facial features, so using character reference might be part of the reason you're getting forward or 50% turned faces

[ember current]

At the first attempt using this method 2 of 4 images looking good. Hair is a bit off, but I can live/work with that and its probably just because of the style. Thank you - super helpful.

[real lynx]

Hi

(((Bird's eye view camera effect aerial view ))).A cinematic shot of a man with his back turned, standing on a rocky terrain near a body of water. The man is wearing a grey hoodie and jeans. The background reveals a serene landscape with mountains, a lake, and a clear sky. The sunlight reflects off the water, creating a shimmering effect.Image is a digital illustration featuring a person with a striking punk style. The subject is seen from the back, showcasing a vibrant pink mohawk hairstyle. The individual has a light skin tone and is wearing a blue denim jacket adorned with bold, graphic designs. The centerpiece of the jacket is a large, detailed skull with pink accents and a banner beneath it. Additional patches and designs are visible on the sleeves, contributing to the edgy aesthetic. The person is also wearing dark jeans with a gold skull emblem on the back pocket. The background is a gradient of dark to light gray, enhancing the focus on the subject's distinctive style.

[real lynx]

Would something like this help you?

Lol :3

[ember current]

Hi, yes I think the 'with his back turned' is helping even further, 75% of the results are as expected. Thank you 👍

[ember current]

The background is awesome on your first set, love that

[real lynx]

Brilliant

I also put a photo of myself, but now he looks at the camera because he forced the AI to show me my features.

[unkempt kettle]

Hello, I'm new. I just did my short film using images from Leonardo, I did try Kino model at first because I love the look but surprisingly it was not as sensitive as phoenix. Is there a workaround perhaps? to get more precision in Kino

[real lynx]

Hello

You could try to get something similar by uploading the image of Phoenix with this tool so that the AI can describe your image

With premium tools you can imitate poses or make faces

Phoenix can write things

And if you have premium tools you can upload this image to albedo with reference content and it could keep those letters

[rustic tree]

@thin tinsel hey there, having a go-again at my canvas attempts. this time, tried resetting all the canvas history then just going for a mask over the hair.

did the prompt as shown but for some reason alongside it not changing the hair color, it's also changing the character's face, and ear even though they are not masked over?

[thin tinsel]

You need to erase and mask ideally.
Send me your image here. I will look at it

[marble cape]

Sp yeah anyways it’s hard to remove the whole staff from this image on my pad

For just the staff alone

Not the background just the staff

Cause it’s hard to remove the whole background and keep the staff alone

Yeah just been having a hard time

[granite horizon]

Hello

Is anybody here having trouble with Canvas EDITOR?

Namely, when you select an area to edit, write a prompt, hit ENTER and... BOOM! instead of getting changes in the selected area, you get a brand new image in a little square on top of your original image

[marble cape]

Can someone help me remove background

[granite horizon]

Let me show you

[marble cape]

Just the staff alone

[twilit fiber]

Same problem here

I'm having another problem : when it generates stuff, it completely expands/warps the selection

I'm having that problem and the one mentionned by @granite horizon ; it worked perfectly well yesterday evening, and have been doing the same actions I've been doing for months

[granite horizon]

Working well only when using Leonardo Diffusion XL. All of the other models are failing. Including my models.

[thin tinsel]

@granite horizon and @twilit fiber I am going to report it via the intercom I recommend you both do clear cache and cookies and if there is sitll a problem report it to

[marble cape]

I guess it’s back to editor then

[thin tinsel]

The problem I am having is that it regenerates a new image in the generation window regardless of what it masked

[twilit fiber]

I've reported it as well

I guess the more people act, the better

[thin tinsel]

[twilit fiber]

It's actually really random

But highly consistent with the base image

[thin tinsel]

Seems to depend on what strength inpaint is on

[twilit fiber]

I my case, I didn't change any setting

Either since yesterday, or for those tests

[granite horizon]

Funny

[thin tinsel]

I tried to just do a simple outpaint, and it gave me something very random

[granite horizon]

And it's same either on focus or infinite modes

Again. Just Leonardo Dffusion XL working OK

Top on the list. All of the other models are failing.

[marble cape]

Can just someone just do this once for me

[stable carbon]

@granite horizon @twilit fiber @thin tinsel thanks for pooling these infos. i suggest not wasting too many more tokens while this issue is being looked into and resolved. I'm getting higher up people to have a look as well. Thanks as well for informing the Intercom Team on teh website itself.

[granite horizon]

Thank YOU!

[thin tinsel]

@granite horizon and @twilit fiber I will am sending the information to Frap who will send on our behalf

[marble cape]

Yeah I’m trying it’s not working for me

[twilit fiber]

Thanks 🙂

[granite horizon]

I was playing a little more with the "bug" and found that it doesn't seem to affect non-human or non-skin-related objects; i.e. in the image I was playing with, I was able to modify the baseball bat and the mask effectively. Unfortunately I didn't take screenshots. Or maybe it was luck.

I hope that helps with the incident research, LOL

[stable carbon]

and this started today? it wasnt this way yesterday..?

[granite horizon]

Last night it was ok

I'm in central time

[stable carbon]

ok. well thats very good to know. so something changed/was changed since yesterday. sounds like an unintentional side effect of an update to me. 🤷‍♂️

[twilit fiber]

Ah, someone's replying to my ticket

[twilit fiber]

It worked fine until 2am for me this night, and when I tried it again 10 hours later, it was broken

[granite horizon]

Forget what I said about non-human LOL

HAHAHAHAHAHA

Those were the last tokens I spent on the bug. Just wanted to make sure. 😆

[granite horizon]

USually it's how it happens... cof cof Microsoft cof cof

[stable carbon]

that is a beautiful elefycle 👌 ✨

[karmic pulsar]

Beautiful

[proven kiln]

I am having issues with a fine tuned model - Absolute Reality v1.6

creating some really weird atifacts of faces. Started happening as of late last night

[thin tinsel]

please share

[marble cape]

im havign a hard time here

It’s not working

Guys please

Guys

[real lynx]

I liked the little eyes

It got a little deformed xD

@marble cape

I really liked this one

It's in more realistic mode

[marble cape]

maybe you dress that up for halloween

[real lynx]

Yes I thought about it :3

[marble cape]

oh really lol

[real lynx]

From what I know about canvas, it will take longer to repair so I ask for your patience and wait ♥️ thanks

[marble cape]

ok but im still trying to understand

sorry

[real lynx]

I'm glad you understand, thanks again.

[marble cape]

i understand but i dont how to go about removinng it since i didnt create that and someone gaved it to me

[real lynx]

I would have liked to have just one eye like that in my image

[marble cape]

no the staff

[marble cape]

as you can see ive tried removing it

[real lynx]

If you want I can delete it

[marble cape]

can you do it on your end if you can that will be great

my next projcect was trying to generate exactly like this:

and this

[marble cape]

remove the circle

[real lynx]

Good luck with the rest, you can achieve things if you set your mind to it.

[marble cape]

and that blue things

[real lynx]

There is less work left

[marble cape]

oh

[real lynx]

Please erase the circle 👍

Now he plays for the Argentine national team

[marble cape]

you cant remove the blue thing?

[real lynx]

I'm going to watch the game

👋

[marble cape]

oh

ok hope they win

[real lynx]

Let's hope that it doesn't matter so much, it's because I have visitors at home and I have to attend to them there.

[marble cape]

I wonder that staff can have a more yugioh trading card game art style effect

[real lynx]

I loved it

[loud vessel]

Hey all, what settings do i use to make logos in Leonardo please?

such as Preset, style etc

[real lynx]

Hello the Phoenix preset

[loud vessel]

Hey there, the phoenix preset i believe does not offer transparency?

[marble cape]

Is it working

[marble cape]

Yeah so I forgot how to remove it how do you remove it again?

For the circle. it’s just erasing right?

[granite horizon]

Yeah! My best elefcycle so far

[loud vessel]

hi i am trying to make a more improved version of this logo, however it seems that the logo keeps getting cut off when its produced an image, i try to unzoom it but it becomes very messy, is there a way to have produce a full image without any cut off please?

[hollow breach]

Have you tried changing the image ratio?

[loud vessel]

No what option is that? i dont see it

[hollow breach]

Click on the cog

Then pick the right image dimensions. You need something that is taller and less wide than your current setting

[frank leaf]

Cami, I was not able to get help, and had to step away from the project for a few weeks. I'm looking to try again. Here's my original post again. I hope it's Ok to re-post, since the problem was not solved....

Hey everybody! I'm having trouble creating a photorealistic headshot of a little girl that I'm trying to make into a character. I have an example image that I use as a character reference for the face. But, I want her to have a different hairstyle. I've tried 'massaging' my prompt to get her to look as close as possible to what I'm looking for. However, Leonardo always gives her cornrows in the picture, and the ponytails don't look anything like they should. I've attached examples of the face reference, and what hairstyle I want her to have. I'm using legacy 'Photoreal' with image guidance in the mid to high range. Here's an example of the prompt I'm using: "Photorealistic headshot of the character referenced. She should have a hairstyle of two loose ponytails. Her hair is not very curly and lends itself well to keeping in short ponytails. The ponytails are loose and are held in place with black rubber bands. ((Short ponytails)) Her skin is dark. She is ten years old. The background of the image should be light in color."

[sullen light]

Hi to everyone, I want to learn more about presents and which choose ed when. is there a guideline tutorial? Thanks to everyone

[marble cape]

Some remove this circle real quick

[feral bough]

Just erase the circle with the eraser tool, put it on Outpaint, prompt 'blue background', and click generate

[marble cape]

But editor is not working

I’m just trying to get it into this background

[marble cape]

I’m just trying to remove the whole back so I can add the spear staff on my creatures

Creature

Think I got

Well not really

Well not really

[sonic linden]

Hello guys,

I'm trying to use Leonardo AI to add a "sub image" (which cannot be generated by Leonardo because it doesn't know what this sub image is) to an image that has been generated with Leo.

A basic example is:

• Leonardo generates an image of a guy holding a sheet of paper
• My sub-image is e.g. a precise comic book cover.

Attached images:
leo-image.jpg => the original leo generated image with some kind of "placeholder" in which I want the sub-image (which can't be leo generated) to be placed
sub-image.jpg => the image that can't be generated and needs to be placed
expected-result.jpg => what I want (here it's butchered with Photoshop)

I'm pretty sure it can be done all within Leonardo : can someone point me to the right direction? All my attempt at using the Canvas Editor are going nowhere.

Thank you for your help!

[vocal portal]

Did you get help?

[marble cape]

I figured it out

[sullen light]

Hi, good afternoon, I need an help; I would like to know how to use preset. Is there a guide? Thanks

[marble cape]

What you guys think this as staff too for my creature?

[dim zodiac]

Can you explain what you mean a bit more?

[sullen light]

i want to know about presets and in which situations i have to use them

[sullen light]

i mean: leonardo, anime, cinematic kino, concept art ... etc etc

[dim zodiac]

Do you want that in your mentoring thread so it's all in one location?

[sullen light]

ye thanls

yes

thanks

[broken elm]

How do I invite people to my "mentoring" channel?

[dim zodiac]

Only other mentors are able to join, who are you hoping to invite?

They are designed to be private threads, so unfortunately cannot invite other users

[broken elm]

really? that really sucks

also I have a huge problem , I lost my progress in canvas editor, im panicking

I wanted to invite in @hardy sentinel

[dim zodiac]

He's a mentor, I've pinged him in your thread

[broken elm]

thank god I had made a save of my canvas image

[real lynx]

Steven The mentor can enter your private thread

[livid inlet]

https://discord.gg/cKfy2qvf?event=1292075029584150569

[eager bridge]

Image guidance problem. It’s broken.here is my screen captures

[low bear]

What happens if you set the strength higher?

Does it produce a completely different image?

[hardy sentinel]

If you still did not clearing your browsing history. It should be there.

[hollow sky]

Im having problems putting characters either in the background, left side or right side. here is the prompt "Visualize a dual-layered world split into two realms. On the top layer, in the background, the faceless figure, wearing a cloak covering their head, appears drained and expressionless, surrounded by a barren landscape with heavy clouds weighing down. Beneath the surface, within the lower realm, vines wrapped tightly around their body represent the physical symptoms—fatigue, aches, and sleep disturbances—caused by depression, illustrating its effects on the physical body as well as the mind. 3D and Hyper Realistic"

can someone please guide me?

[leaden jacinth]

https://cdn.leonardo.ai/users/46484726-47e0-47c2-8d2b-6ecb58d855f2/generations/86333864-6163-4ae5-bfee-f24bfb75881e/Leonardo_Phoenix_A_captivating_tshirt_design_featuring_an_owl_3.jpg

https://cdn.leonardo.ai/users/46484726-47e0-47c2-8d2b-6ecb58d855f2/generations/86333864-6163-4ae5-bfee-f24bfb75881e/Leonardo_Phoenix_A_captivating_tshirt_design_featuring_an_owl_2.jpg

https://cdn.leonardo.ai/users/46484726-47e0-47c2-8d2b-6ecb58d855f2/generations/eef877db-1080-4284-a377-6248cd5fb461/Leonardo_Phoenix_A_luxurious_and_captivating_scene_of_exquisit_3.jpg

https://cdn.leonardo.ai/users/46484726-47e0-47c2-8d2b-6ecb58d855f2/generations/eef877db-1080-4284-a377-6248cd5fb461/Leonardo_Phoenix_A_luxurious_and_captivating_scene_of_exquisit_2.jpg

https://cdn.leonardo.ai/users/46484726-47e0-47c2-8d2b-6ecb58d855f2/generations/037fc41d-e4c4-461a-a68b-f9c02bf9276f/Leonardo_Phoenix_A_photo_of_a_glamorous_golden_statue_of_a_han_3.jpg

[sudden yoke]

Hi, can anyone please explain Contrast?

[open echo]

Yello 😄
That's a very good question.

So the contrast will work a bit like in photography: It's the contrary between the dark colors and bright colors.
A high contrast will have a more vivid difference (really dark shadows and very well lit face) and a low contrast will have more of a "gray" to it: everything is lit up evenly.

[sudden yoke]

@open echo I see. Thank you for the answer. I understand contrast in images, but I was just curious if it meant something different in AI art generation. It seemed strange to have something that's typically a post-process in the prompting stage.

[open echo]

Well, in image generation the image can be delivered with all adjustments already done
(Of course I'm a perfectionist so I take it out to my photo editing program anyway... 😅 )

[sudden yoke]

I have another question. What's the best way to get hand-drawn looking images in Leonardo? In Midjourney, there's Niji. I've tried using the Leonardo Anime XL model, but the results were surprisingly more CG-looking.

[open echo]

Funnily enough, I had great results in different models (Kino, AlbedoXL, Anime)
The exemple here was done in Kino, with the important part being, it starts with "Style of Edgar P. Jacob"
(A belgian comic book artist)

It really depends on the prompt. I found that putting too much words in a prompt will "confuse" the AI... like putting too much ingredient in a salad, you may not know what comes out.

Or, alternatively, you can focus the results (limit the pool of possibilities) by using Elements.

[sudden yoke]

@open echo That looks fantastic! 😍 I've been playing around with different Elements, but I was thinking the models would be more robust.

[open echo]

I played a bunch with classic comic book styles, and you'd be surprised how it's easy to get a good prompt.

I think the biggest issue I've seen people have when they start with Leonardo.Ai (not saying you're doing it, havent seen your prompts :D), is to overcomplicate

[sudden yoke]

Wow, and it's with Kino. Okay, I haven't done much with that one. My understanding of it was that it's the Cinematic one?

[open echo]

Yes, but it also does great vintage 80's anime as well

Pretty much anything that could have appeared on screen.

[sudden yoke]

I think you're right about confusing the AI. As soon as it sees words that trigger cinematic/photographic/or CGI, the results cease to look hand-drawn

[sudden yoke]

How cool!!!! Looks great!

You've got me excited to try Kino!

[open echo]

It's a surprising one 😛

And it doesn't even need the "Quality" generation. Fast will give you a little more of those "imperfections"

[sudden yoke]

I had been using Vision for most of my painterly generations, but haven't gotten great hand-drawn results from it

[open echo]

From my experiences, Vision is great at sci-fi/fantasy/high artistic photorealism ... like, not a selfie with a phone camera... but a full on fantasy scene with mythical creatures and magical auras and stuff....
(I understand myself... that's the important part 😅 )

For drawn: AlbedoXl, Anime and Kino are great

But Albedo is usually my goto

[sudden yoke]

Ah. I like that aesthetic (sci-fi/fantasy). I really like Moebius

So you go "Fast" generation mode on all three of those models? @open echo

[open echo]

Normally only for Kino

But when I'm just messing around, I go with Fast for AlbedoXl too 😄

Vision is usually better with Quality

Exemple: fixed seed, same "robot face" prompt on Vision XL but second one with the "Old School Comic" Element

[sudden yoke]

Whoa! That hand-drawn is so good!!!!!!

@open echo How much of the elements do you usually apply?

[open echo]

To tell the truth... even after all this time I usually forget they exist... then take too long to do something that would have taken 10 seconds with an Element 😄

[open echo]

And if you want to really go crazy with the tools.... 😉

Using the CGI image as Content Reference (Mid) and applying the same prompt with the Old School Comic Element

[sudden yoke]

So if the Element is at 0.7, it's 70% the Element and 30% the underlying model? Or is the model the main 'brain' for the generation and the element helps denoise when it can?

[sudden yoke]

You've really done some good work with this character. It seems like the Old School Comic Element goes full black on the shadows mostly

[open echo]

I kept it as High Contrast, so yeah, it's really cutting these dark and light details with a knife 😄

[real lynx]

I thought you wanted a good night sketch

[sudden yoke]

It looks incredible.

[real lynx]

You can combine elements and you can also put a texture of a sketch so that the AI imitates that style

[sudden yoke]

@real lynx If you had told me you actually drew that, I'd have believed you! It looks so realistic!

[open echo]

@sudden yoke And just to go full-circle to the original question with the race car, you can use these technique to apply the style to an existing image
Just keep the prompt simple since the Content Reference and the Element will be doing the heavy lifting in that case

[open echo]

I knew there'd be a Cami-Bot on it's way 😛

[real lynx]

That is the description of the image created by Tuxman :3

[sudden yoke]

That is the comic book element? It looks gorgeous!

[sudden yoke]

I really like this style you've made, Cami. Do you find the coloring book Element helps keep things looking hand-drawn?

[sudden yoke]

What were you using as Content Reference? Another generation?

[open echo]

I have a love-hate relationship with it.... after all my researches on 60-70's european comic book artist, and refining the prompt and the model.......... they dropped the Element 2 weeks later 🤣

[open echo]

Yes, you original CGI race car image

[real lynx]

Camí bot

[sudden yoke]

Which model/element were you using before that? Toon/Anime?

[sudden yoke]

Pretty eyes

[open echo]

None! 😄
(That's why it hurts haha)

[real lynx]

Improved

[sudden yoke]

Yes, they've created about 30 new elements or something since I started using Leonardo!

[sudden yoke]

I like how the face is more rendered. It gives the sketch focus. I love the loose linework

[open echo]

Exactly! It really helped hone in on specific styles and simplify the prompts

[sudden yoke]

I hadn't used Old School Comic Book before this evening: thank you!

[real lynx]

We have an element that you will love about vectors

[sudden yoke]

Oh yeah? Let me look for it!

@real lynx Is that the Simple Flat Illustration?

[sudden yoke]

Woooooooowwww! 👍👍👍

[hollow sky]

Does anyone know how to add 2 characters into one image?

[sudden yoke]

One way: you can use a Content Reference on the High setting. The black and white image is the Content Reference. The resulting image has two people in the same poses as the Content Reference.

[hollow sky]

thank you so much! I will try that.

[hardy sentinel]

Please ask your specific questions here @sullen light

[eager bridge]

No it does not. It does not change much. It looks like its stuck in one position. Of more like it stuck off position.

[eager bridge]

@low bear

[manic root]

Hey Leonardians, join us in today's Halloween Tales challenge in #1290802900540588052! 🎃🦇 Engage each day as the story evolves with fresh chapter themes. Support your peers by voting with the Leonardo emoji , and don’t miss out on winning some major prize tokens! Have a blast!

[broken elm]

@feral bough I finished my first FINAL image I got 2k credits left, I THINK, like 90% sure it's SFW?

[karmic pulsar]

Lit

[marble cape]

Is editor working?

[vernal gulch]

I can't see where to do this. Where is the setting for combining elements please?

[thin tinsel]

next the prompt the elements are there

[vernal gulch]

Thank you! It's not visible in the app but I can see it in the web version.

[thin tinsel]

oh ok I see. I don't use the app.

[vernal gulch]

Thanks for your help 😀

[real lynx]

Android apk does not have all the updated tools

The application is very new

Little by little it is also growing

[hollow sky]

Anyone know how to generate the same character, with same clothes, but different poses? I have tried using the same prompt but still comes out with different clothing and hair style as well. I am using the seed and max character reference. Any help, please? 🙂

[real lynx]

The easiest thing to do is go to Phoenix and use comic book style panels to generate the same characters with different scenarios.

Phoenix may include signs

Prompt

A 3-panel gag comic dialogue with a different scenario:Character 1: Blonde woman pointy ears chibi minecraft style 3d render background Minecraft world

Panel 1:
Character 1: "Remember if you need help you can activate the /mentoring command "

Panel 2:
Character 1: looking at a massive list "This will open an exclusive and private channel for you and the mentors of all communities. "

Panel 3:
Character 1: *Standing with a giant sign held by her hands that says welcome * "My name is Camila mentor of the Spanish channel welcome "

[hard sand]

So I love the way that anime semi-realisim looks but is there a way to make it so the eyes of the woman is less anime?

Trying to make her but her eyes are to anime and she isnt as realisitic as the girl with ginger hair

or is there a way to get Leonardo phoenix to look like the images above? I try style and it doesn't do a good job.

[hardy sentinel]

An elephant is safe for work

[eager bridge]

Now Image guidance it is workin normally! Why it was broken 2 days? Can anyone tell that?

[real lynx]

Hello good day I would use some element

If you want to keep the anime you could combine this element with some of the anime

[hard sand]

thanks ill try that

[ancient notch]

/Create A sunny backyard.
Characters: Lily, a small girl with a worried expression, looking up at a tall tree. Max, a confident-looking boy, standing next to her.
Dialogue: Max: "Come on, Lily! It's fun up there!"

[feral bough]

Bug

[feral bough]

Try running the image through the Anime Upscaler or the Ultra Upscaler, this will change the eyes

[eager bridge]

🪲

[desert saddle]

crear logo para compositor musical contemporaneo EI

[real lynx]

Hola

Tu logo se crea solo en la web no en discord

Cómo consejo te doy usa el preset Phoenix

Que es nuestro especialista en logos y letras

Lo que quieras que escriba también ponelo entre comillas

Si tenes más dudas o necesitas ayuda no dudes en preguntar

[hollow sky]

Thank you for your help! Is there a way to do it with real people, not comic book style? I apologize if I did not specify.

[real lynx]

With premium tools you have image character

[hollow sky]

I am using character reference on high strength including the seed, am I missing something else?

[real lynx]

This is a modified image of my photos

Also my profile picture

[real lynx]

Here I imitated your image

[hollow sky]

thank you so much, will try that.

[manic root]

Hey Leonardians! Dive into today's Halloween Tales challenge 🎃 🦇 in #1290802900540588052. Engage daily as the story grows with new themes, and cheer on participants by voting with the Leonardo emoji . Great prizes await! Enjoy!

[livid rune]

the top two pictures are really nice!

[feral bough]

Exactly your taste🥸

[livid rune]

indeed!

[delicate lily]

Look how pixelated the result is

Tried it on PC and the results are almost the same quality

[vocal portal]

I see

[delicate lily]

Tried both desktop version web on Android and nothing

[vocal portal]

Did you clear your cache and have you tried a different browser?

[delicate lily]

Try the image yourself and see the results

[vocal portal]

can you use the other upscaler as opposed to instant refine?

[delicate lily]

Yes I've tried

[vocal portal]

give me a second

[delicate lily]

Instant refine literally destroy the picture itself, it's not worth it, that's why I only use upscaler and live canvas

[vocal portal]

that's why I said the other one

[delicate lily]

Even using it still destroy some details

Because how pixelated the results are on Android

[vocal portal]

I'm upscaling it now.

[delicate lily]

I was hoping for the Leonardo android to feature live canvas hoping the problem will be fixed

They don't even have the upscaler 😭 🙏

Which now left me only using a pc

[vocal portal]

It doesn't really look pixelated to me. Are you on the same device you normally use?

[delicate lily]

Send screenshot

Literally I wanna see

[vocal portal]

and it's a little slow today. I'll send the image when it's done spinning

[delicate lily]

I'm on my usual device android Samsung a52s 5g

Goated phone btw

@vocal portal u use android or iOS?

[vocal portal]

iOS but I'm on a pc most of the time

[delicate lily]

Can I see the results on pc?

Please send and I'll show you

[vocal portal]

yes

see above

that is cinematic on creativity 3

this is on ultra upscaler

universal upscaler General creativity 5

[prime merlin]

anyone know how to get characters like this? they are always too beautiful or not grunge enough

[real lynx]

😅

[prime merlin]

perfect HAHAHa

I got this one

[prime merlin]

HAHAHHA

WOW

how did you do it

the bottom one looks straight from a video game

Is there a way to keep the style of the reference?

chat chat chat look

[real lynx]

These are premium tools that I used

[prime merlin]

I think i am using character reference right now but I want some more originality but if i go too low it just spits out beautiful women

haha

[real lynx]

Thanks to these tools you can transform your photos, for example, into a doll.

[prime merlin]

Wow looks like it was taken of that website where they sell garage kit models to paint

[real lynx]

I also described your image so that it gives me a similar prompt of your image

[prime merlin]

Ok let me try describe

[real lynx]

Use the pass of a Himiko Toga doll

[prime merlin]

wow she is a crazy character

[manic root]

Hey Leonardians! 🎃 Remember to join today's Halloween Tales challenge in #1290802900540588052! Each day brings a new chapter theme, so jump into the storytelling excitement. 🦇 Show support by voting for your favorite entries with the Leonardo emoji . Great prizes are up for grabs! Have a blast!

[manic root]

For sharing personal information. Please avoid sharing personal details. Contact a moderator if you think this was a mistake.
-# If you believe this was an error, please contact a moderator.

[vernal gulch]

I'd like to create the following but I am having very little success.
Any tips?
a cartoon character with the head of a young Asian male and the body of an ant
Thank you! 😀

[lapis pasture]

anyone can help me with that?

[dim zodiac]

answered on your original post

[silver shell]

How to create a image

[open echo]

Leonardo.Ai is in fact a website with a full suite of visual tools. Simply go to http://app.leonardo.ai and you can start generating images for free 😄

And here's a quick 7 minutes tutorial to get you started: https://youtu.be/Rukln4nr_Z0?si=8ZctAfZZJqvpMVvl

[marble cape]

hey all soo how do i match the same color from the creature to the staff or might as well make the creature more black color?

please let me kow

[sudden yoke]

Use your image as a Content Reference set to High. I also used your image as a Character Reference

A surreal humanoid figure with an elongated, slender body and a single large, oval-shaped eye for a face, holding a spear-like object. Its head is shaped like an inverted cone, with the top piece of the head floating above, giving an eerie, levitating effect. The figure stands in a dark, moody atmosphere with a gradient background of black transitioning to deep blue. The environment feels mysterious and abstract, with minimalist details. The image is inspired by surrealist art, with elements of dark fantasy. The lighting is soft, creating subtle shadows that enhance the eerie feel. Ultra-realistic render, 4K resolution, cinematic lighting.

[real lynx]

It turned out very well

[marble cape]

No no

[frank leaf]

I apologize that I'm posting this again. But, I have received no help. And really need some. 😦

[hollow breach]

A lot of the mentors are not around at that time. Most are in Europe and North America, so it can be hard to get an answer. You can try to type /mentoring

[frank leaf]

Thanks, John. I'll give that a try.

[exotic tinsel]

im lookinf for the first guidance to generate 1st image here. thanks for advice

[hollow breach]

Leonardo.Ai is in fact a website with a full suite of visual tools. Simply go to http://app.leonardo.ai and you can start generating images for free 😄

And here's a quick 7 minutes tutorial to get you started: https://youtu.be/Rukln4nr_Z0?si=8ZctAfZZJqvpMVvl

[sudden yoke]

Hi Brian, I was able to use the Realtime Canvas to get the hairstyle you want. I just painted with the background colors until Leonardo understood what I was trying to make. I downloaded the result, then added it as a Content Reference (& Image Reference). I also added the Face desired image as a Character Reference and as a Style Reference. Finally, I added the Digital Painting Elements. If the face needs to be exact, you could use a Face Swapping AI.

A hand-painted portrait of a young girl with dark brown skin, styled with two puffball ponytails tied with black hair ties, standing against a soft, greenish watercolor background. She wears a striped t-shirt under denim overalls, with soft brushstrokes and warm lighting highlighting her serene expression, inspired by children's book illustrations and contemporary character art.

[thin tinsel]

Hi Kevin,
We have a mentoring thread open to help Brian with this project. Can you access it, incase there are any follow up question?

[sudden yoke]

@thin tinsel I cannot see it. I'm happy to help with follow-up questions

[spiral axle]

Max Character Reference

[thin tinsel]

@sudden yoke we have a guide that one of our mentors wrote about Focus mode.

I'll just find it

https://discordapp.com/channels/989166677390426132/1280845243561803838

It's a really good guide for anyone to who wants to get familiar with the different features. 🙂

[sudden yoke]

Jackpot! 🎰🏆 Thank you 🙂

[thin tinsel]

Focus mode is tip number 2.
I absolutely love it, since I realised how to use it

[sudden yoke]

@thin tinsel I'm going to try it out. I've always relied on Photoshop for finishing images, but it would be great to learn these new AI tools

[thin tinsel]

Canvas Editor is better than Photoshop in its ability to ahere to the style of the original image.
The only thing I miss from canvas is the ability to Skew

[sudden yoke]

You're a Skewer? 🍢 I hope they add the feature to the Canvas Editor and don't make it to obskewer.

[thin tinsel]

LOL, it's just when you need to add perspective to something. Like adding text to a box for example.

[sudden yoke]

That makes sense. A perspective warp would be incredible! I'm in the Canvas Editor now...following the tips and videos you've shared 🙏.

[thin tinsel]

Excellent. Do share your results if you're happy with them, and ask if you need any help with it.

[sudden yoke]

I'm not sure you want to see images of me and Keanu Reeves doing karate together. I've had some things I need to bring to life though.

[thin tinsel]

Oh I'm quite sure you can't say that and not share it. 😆

[sudden yoke]

HAHAHAHAAHHAHAHA

What a rabbit hole that would be! 😄

@thin tinsel Do you give yourself assignments like that for learning Leonardo?

[thin tinsel]

Um 🤔 Not really just for the sake of learning. Mainly started off just fixing issues like our dual-trunked Elephant friend.
So I think out of necessity.
But mainly I use it because I really enjoy it. I'm a traditional Artist (acrylic paint, watercolour etc), I'm a digital artist (Photoshop, illustrator), and now an AI artist, so the Canvas Editor feeds into the joy I get from painting and photoshopping. 😍

[warm forge]

Hi. Can someone please help in making a lava burning mermaid similar to this, but with flames?

[night vault]

you can try fiery elements in XL models

[warm forge]

It doesn’t get the tail, and liquid lava and flames

[night vault]

if it get too much flame then let’s turn that off, maybe try words like flume blaze

[sudden yoke]

@warm forge Maybe something like this?

I drew some crude flame shapes in the Realtime Canvas with a very low Creativity Strength. After using Instant Refine, I downloaded this image. I then brought both images into the Canvas Editor, erased the face and unwanted details from the Realtime Canvas-generated Image, then used a mask brush to create an image that 'patched' the two images together. In the second screenshot, the images were on top of each other, but I separated them to show how the right-most image was created.

soft illustration of a lava-burning mermaid with flames, one eye is covered by hair, eyes are orange, looking at the camera, the style of loish, sketch

[hard sand]

Ello's I want to have this style of image so that any image I make has this type of style but I cant seem to get it unless I do style referance which doesnt help. I also wanted to do character ref to have the same person but doing different things but it just has her upclose.

[real lynx]

Lying on a rock .ultra detailed cute chibi mermaid Of fire with cute top and long hair, full body, lots of iridiscent bubbles, cute fishes, dynamic pose, light particles, stuningly detailed underwater dungeon, five fingers on each hand, ultra detailed illustration by hayao miyazaki and Loish, wlop, ni no kuni, zoomed, hit definition, contrasting colors, colorful holographic, 32k resolution, best quality, volumetric lighting, best quality, masterpiece, very high saturation

Be careful with this element, it can burn everything if you use too much force.

[sudden yoke]

@hard sand
Have you tried the Leonardo ChatGPT? It's great for getting good prompts!
https://chatgpt.com/g/g-FwjLmXklT-leonardo-ai-prompt-maker

Here's an example that I used with Kino XL...

A beautiful woman with delicate features, wearing a vintage lace gown with soft folds, sitting by a wooden table in a cozy room filled with books and flowers. The room is lit by soft, natural sunlight streaming through the window, casting warm, gentle shadows. Her hair is styled in elegant curls, and she has a calm and serene expression. The atmosphere is peaceful and intimate, with soft candlelight illuminating the space. The overall aesthetic is classical and romantic, inspired by 19th-century paintings and Victorian fashion, with highly detailed textures and soft-focus lighting. Cinematic 85mm, soft bokeh, and high-resolution 4k render, with photorealistic lighting and subtle depth of field.

Negative prompt:

mutated hands, mutated fingers, extra limbs, deformed body, deformed hands, deformed fingers, missing limbs, extra fingers, disfigured, fused fingers, abnormal hands, distorted fingers, unnatural limb positioning, broken limbs, poorly drawn hands, blurry hands, malformed hands, extra arms, extra legs, unrealistic proportions, double limbs, unnatural bending, misshapen body parts, incomplete limbs, awkward anatomy, incorrect anatomy, disproportionate body parts, bad anatomy, unnatural hands, excessive fingers, extra body parts, overly long fingers, hand deformities, body deformities, extra joints, glitch, incorrect human anatomy

@hard sand I used your images as references

[hard sand]

I didn't know about ChatGPT for Leonardo thanks!

[thin tinsel]

I also had a go.
AlbedoBase XL | Edge to Image 0.70 | Elements Fiery Flames 0.20

Prompt: A fierce mermaid sitting on a rock on a bed of lava with her tail to the side and flames all around. Her hair is like flowing lava and her tail pattern has lines of running lava.

[sudden yoke]

@thin tinsel This is beautiful! How'd you keep the Disney-style face? I find that the models tend towards anime-faces

[thin tinsel]

I used the Illustrative Albedo preset for this one.
I can only assume it associates mermaid with the little mermaid

[sudden yoke]

That's nice: I'll try that preset. I didn't come near this style in my attempts! They were all anime faces.

[thin tinsel]

I got others in the batch that weren't quite so disney

This was my actual favourite, but there's just one small problem

[sudden yoke]

I was getting a strong Loish-vibe from the original

[thin tinsel]

Oh, I'm not familiar. Who's Loish?

[sudden yoke]

She's one of my favorite digital painters!
https://www.instagram.com/loisvb/

[thin tinsel]

Nice. I'll check her out

[thin tinsel]

Do you do that kind of illustration as well?

[sudden yoke]

@thin tinsel https://www.instagram.com/nomadic_sun/

They are (mostly) digital paintings + Leonardo/Midjourney/Firefly. Some are plain digital paintings

Do you?

From your video streams, I can hear your passion for fashion illustration!

[thin tinsel]

Thanks. These are great. I really like this style.

[sudden yoke]

Thank you! I'm very excited about the nexus of digital painting and AI

[thin tinsel]

No, I never really got into it myself. I was taking an interest in that kind of digital artwork shortly before I got into AI image generation.

[sudden yoke]

Your knack for the Canvas Editor makes me think that you'd be good at it! A lot of it is knowing digital image editing tools

[thin tinsel]

Yes indeed. Also passionate about other design professions, such as Graphic Design and Product Design.

[sudden yoke]

@thin tinsel Hahahaha! It's good to have those related-but-different passions. They will carry you through life!

I'm also an animator and musician...all things are being influenced by AI now!

@thin tinsel Are you using Canva?
https://www.canva.com/newsroom/news/leonardo-ai/

[thin tinsel]

Oh wow, that sounds great. Do you have any animations I can see. 🙂

[sudden yoke]

Speaking of Disney, I tried doing a digital painting/relighting/animation of a shot from 101 Dalmatians (with an AI-generated background):
https://youtu.be/eloIYEuv_8I?si=gcZRI7wgLW-GFd3a

@thin tinsel Here's a Cat Genie
https://www.instagram.com/reel/C-vF5vmpdm3/

The music is AI-generated

I also do some 3D animation in Blender
https://www.instagram.com/reel/CgDQskZAGrV/

[thin tinsel]

It's not letting me open it. Do I need to be logged in?

[thin tinsel]

Oh yeah I saw the bot on your YouTube channel, and was going to ask if that was Blender

[sudden yoke]

I changed the link. It should work now!

(the Cat Genie)

Maybe not

[thin tinsel]

No cat Genie unfortunagely, it just loads a blank screen. 😟
Is it also available on your YT channel?

[sudden yoke]

https://www.youtube.com/shorts/CLj47yMSJfc

The cat is conjuring the mysteries of the Universe! 🪐🔮🐱

Is any of your Graphic or Product Design online?

@thin tinsel The background was Leonardo, I believe!

[thin tinsel]

nice. 🙂

[thin tinsel]

Which one did you use?

[sudden yoke]

https://suno.com/
Have you tried it yet?

[thin tinsel]

Yeah! 🎉 😄
Lots of Suno fans around here.

[sudden yoke]

Oh! If you hadn't I was going to surprise you with this! 😄
https://suno.com/song/422ed71d-3e71-4d64-bb84-bdf419056a7e

The lyrics assumed you are in London

[manic root]

Hey Leonardians, join the thrills of today's Halloween Tales challenge! 🎃🦇 Jump into #1290802900540588052 and be part of the storytelling with daily chapter themes. Support your peers by voting with the Leonardo emoji . Great prizes up for grabs! Enjoy!

[thin tinsel]

Aww how lovely. I love it.
Did you literally just make that? That was speedy. 😆
What a great little foot tapper.

[sudden yoke]

I enjoyed it too! So many people are afraid of AI (and rightfully so), but it's the greatest magic we can do right now!

Lots of beautiful things can come from it as is seen on this server

[thin tinsel]

Agreed.
Thanks for sharing all your creations. 🙂
I'm off to sleep now. 👋

[sudden yoke]

Sleep well! Thanks for your help today!

[civic sapphire]

anyone know how to get images to look this natural

im talking about the girl pulling guys shirt up

[hard sand]

Not sure why it wont have her reading a book

everytime it just her and no book or she doesnt even look at the book

[thin tinsel]

Put this in the Mentoring thread, and then Cami will see it when she finished work. 🤗

[plucky meadow]

Ok

[stable carbon]

or you could post the prompt you're using here and anyone could try to help 😉

[open echo]

Just to reiterate what my friends asked: what have you tried so far? 😁

[frank leaf]

Kevin, thanks! That's really impressive! I've never used Realtime Canvas before. I'll have to give that a try!

May I use the image you generated as a reference for the images I'm creating for a book?

[sudden yoke]

Thank you. Yes you may! 🙂

[frank leaf]

Thanks for the follow-up Willow! Kevin replied to me. His answer was very helpful!

[frank leaf]

Thank you so much, Kevin! Would you also be interested to know that I recently published a book with illustrations I made using (largely) Leonardo? I'm trying to find Leonardo's physical mailing address, because I want to send them a copy.

[frank leaf]

DerD4nnY, thanks! That's really cool!

[sudden yoke]

That's awesome! Congratulations. 🎉

@frank leaf I'd love to see what you've made. I know Leonardo AI is based in Sydney, but I'm not sure where (somewhere in Neutral Bay?)

[manic root]

For the protection of our server, the sharing of links is limited based on server rank.
-# If you believe this was an error, please contact a moderator.

[frank leaf]

The Botticelli won't let me share the link. What do I do?

[sudden yoke]

@frank leaf I added you as a friend. You should be able to DM any links

[frank leaf]

Ok. I'll try to do that. New to this whole Discord thing...

[frank leaf]

Does anyone know how I increase my server rank, so I can post links? I've asked the Botticelli, but it's completely silent. 😦

[night vault]

A lot has changed recently. There used to be self-promotion channel now I dont see it. The bot that help ranking up is gone too and Botticelli is not stepping up for that role yet.

[frank leaf]

Ugh. 😦

[night vault]

you need to ask homie around here. I was busy with my irl and return and a lot has change without clear announcement 👀

[frank leaf]

Ok. Who's homie?

[frank leaf]

Ok. Who's homie?

[night vault]

i mean usual faces around here

[night vault]

you can ask @copper warren about the bot

[frank leaf]

Ok. Well, thanks for the tip! I'll have to check back in later. I gotta go. Thanks!! 🙂

[frank leaf]

pookah says she doesn't take DM's anymore. 😦

Sorry, gotta go. I'll check in later...

[manic root]

Hey Leonardians, it's time for our Halloween Tales challenge! 🎃🦇 Check out #1290802900540588052 and join us as we unveil a new chapter theme each day. Dive into the story, contribute your piece, and don't forget to cheer on your fellow members by voting with the Leonardo emoji . Amazing prizes await! Let's make it spooktacular!

[copper warren]

Heya! My apologies for the less than friendly message in my bio. DMs were turning into a way for people’s messages to be lost. But you are always free to ping me here. How can I help you?

[copper warren]

Oh! Actually I saw your message, it showed up in Botticelli’s mailbag. I hope the reply was helpful.

[half halo]

looks like someone's experiencing the downside of success

soon there will be ai paparazi at your house

😄

[nimble bay]

Missions

[errant forge]

Who’s going to take me on a tour here

[half halo]

I am not sure about a tour but if you have questions you can post them here. You can also use the /mentoring command to get some one on one help from a skilled community member

[steep elm]

Hello! Thank you for accepting. I'm trying to use Leonardo to make images like pic related, but every time Leonardo is hellbent in making faces as realistic and digitalized as possible. Can anyone help?

[hardy sentinel]

welcome owerbart

[steep elm]

Hello Steven P

[hardy sentinel]

I am sorry, I can not help you

[steep elm]

Can you pinpoint to a tutorial or something?

can anyone help?

[half halo]

hmmm, i think you may want to try using the style preference option

thats where you can find it

under image guidance

but im not sure what the chances of success are

let me know if it gets you closer to what you are looking for

[steep elm]

it gets a little bit closer!

Still not great, I want something even more simplistic.

i'm trying to use Chatgpt to adjust the prompts.

[half halo]

You can set how close you want it to be to your style, make sure thats all the way up

try stuff like simple charcoal sketch in the prompt

and you can also use the negative prompt option to tell it what you dont want to see

and remember the beginning of the prompt weights heavier than the end of it

[steep elm]

its getting better, you the GOAT

[half halo]

Thank you 🙂 im glad i could help. Good luck and i would love to see the final result!

[iron spade]

hey i require some bit of assistance, i got a 2d map that i want to be changed a bit but the ai tool Gives me A COMPLETE different image of a landscape ( don't get me wrong it looks cool ) but it's 100% wrong from what i am trying to achieve

this is the map i want to be enhanced this is what i got

[half halo]

hmmm like owerbart i think you should use the style reference

and use the negative prompt and put in something like 3d

and on the normal prompt, 2d overland map

try those and see what happens

[iron spade]

what should i type on the prompt ?

[half halo]

I would suggest, 2d overland campaign map, detailed sketching, hand drawn

something like that

and refine from there

[iron spade]

and the model

[half halo]

thts a tricky one, i think you should try them

[iron spade]

is there something i should use specificly ?

oh ok

[half halo]

its hard to say which one would be best

concept art might work

but its a trial and error tbh, all these models are so different

[manic root]

Hey Leonardians! Dive into today's Halloween Tales challenge 🎃 🦇 over at #1290802900540588052! Every day brings a new chapter and theme, so join the adventure and make your mark. Remember, you can vote for your favorite stories with the Leonardo emoji . There are major prizes up for grabs! Enjoy!

[half halo]

If you dont get what you want with one model, try another, so often i have wasted many tokens adn then switched models and got what i wanted right away

[iron spade]

what about elements ?

[half halo]

i dont have a lot of experience using that yet

im also still learning 😄

but i like to help where i can

[bold wedge]

i'm certain Leonardo tutorials are on youtube

[iron spade]

i appreciate it

[iron spade]

i will keep trying i appreciate it anyway ❤️

[half halo]

you are very welcome! 🙂

[open echo]

So you can use Image Guidance features like « Depth to Image », « Edge to Image » and « Content Reference ».

Keep in mind the challenges with this image is that first, it is very wide. Leo does 3:1 ratio maximum, so some details might be skewed or missing. Second, it’s a very detailed map with very low contrast between Colors, so again, some details might be lost.

But here’s a quick exemple of what I got using Edge to Image and a very basic prompt. We see the sea on the left and the river on the right, but a lot of other details have been interpreted as rivers instead of roads.
My guess is some adjustments to the prompt could help guide the generation in the right direction, or enhancing the Colors of the map in a photo editing application to heighten the contrast.

[open echo]

Looking for something in that vein?

[nimble bay]

Anyone here knows what this means?

never seen it b4

comes up when I use Edit with AI

[open echo]

I just used Edit with AI and it worked correctly… so I couldn’t reproduce it :S

[nimble bay]

So

I used image to image and style reference to get this on Phoenix

[open echo]

I had image to image, I’ll add a style

[nimble bay]

After I turned of both guidances and wanted AI to use this seed but without helmet

get that ms and never seen it b4

nor do I get it what it means XD

[open echo]

It seems to be a bug with how Style Reference interacts with Edit with AI

You can send a bug report on Intercom and I’ll do the same ( better if more people report bugs than just one 😜)

[nimble bay]

Cool. Will do.

[nimble bay]

you sending a bug report to yourself? Busy I see...

message botton wont work for me now either lol. wil try agin later

button

[open echo]

Only if they hire me as a dev.. for now it’s somebody else’s problem 😜 (more qualified than me haha)