🙋│image-help

[real lynx]

It's that man who has glasses

And will a name in the game be known?

Or is it a character you invented?

[solar sigil]

Yeah that's a character I invented

[real lynx]

This is the character

[solar sigil]

Yes :3

[real lynx]

I'll see what I can do :3

[solar sigil]

Ty ^-^

[real lynx]

If something similar comes up I'll let you know.

[real lynx]

To see if you like it

This is the prompt for your image

That creates and generates something similar

[solar sigil]

Oh, not bad, thanks ;3

[real lynx]

Then I entered your image to make it look more similar

[solar sigil]

Definitely got similiar pose going on in the 2nd one :B

[real lynx]

This gave me a description for your image prompt

You need to touch the stars for that option to appear

[solar sigil]

Ok, ty ^^

[real lynx]

Then use character reference to make them look more similar

And also style reference

So that they have the same drawing style

Reference Content Might Help Too

[solar sigil]

Okies :3

[real lynx]

Your image is tall, take care of the dimensions

If you don't want to cut off his head

[solar sigil]

Ok, thanks for the advice :3

[night vault]

Ok, I remember you now after your re-introduction. You can follow Cami suggestion to get character and one you are happy ADD “Photo of “ in front of your character prompt to get realistic effect

[solar sigil]

Ty @real lynx & @night vault I'm almost there, thanks to your describing the picture using Ai, I'm getting pretty close to how I want it, so I appreciate it :3

[real lynx]

Put a shirt on your fighter xD

[solar sigil]

Lol it's in the description, blame the Ai ;P

[fallen obsidian]

@tall elbow here the ss.

[tall elbow]

that link it's too long

try this
https://app.leonardo.ai

[fallen obsidian]

once i click for the launch app on the top right of main website it just pop by itself

[tall elbow]

where did you click? sceenshoot

[wise canyon]

Yeah that address extension is odd
Try typing the short address manually?

[fallen obsidian]

"Get Started" also the same 😦 cant reached

[eternal urchin]

any tips on getting characters against this type of background ?

I tried the 'describe with ai' but it didn't work

[real lynx]

Hi

[real lynx]

Something like that ?

[eternal urchin]

That type of background, yes

[real lynx]

Full body miniature figure on a blue base.Digital illustration featuring an anime-style character with a bright a big blue background blue background. The character is a young female with fair skin, large blue eyes, and short white hair. She has cat ears on top of her head, adding a unique and whimsical touch. She is wearing a light gray hoodie and has a blue backpack slung over her shoulder. The overall style is clean and vibrant, with smooth lines and a minimalist color palette. The character's expression is calm and slightly smiling, giving a friendly and approachable vibe.

[eternal urchin]

Oh yeah ,switching presets fixed it

thanks

[tall elbow]

that it's not the link what i send you

https://app.leonardo.ai

[winter pawn]

Hello guys, I'm having an issue with inpainting

Why does it do this weird texture? I just wanted to fill it up with the rest of the brick wall

This is the starting image. Thx in advance for help

[wise canyon]

Hi Rene, a couple of tips: first be sure to toggle off the "outpaint" tab in the right side menu, make the edit box smaller and only place it over the black area and a lot of the wall and window, then mask the black area and prompt "window and brick wall".
depending on the gen outcome at this point, you may need to play with the "inpaint strength" slider in the right menu field to get different results too.

[winter pawn]

Hi, thank you so much for your help! However it doesn't make a big difference, even with the different inpaint strenghts. Maybe I messed something up in the settings

[manic root]

Hey there! 🌟 If you're keen to level up your skills, work on cool projects or have set learning targets 🎯, our mentori have got your back! For personalized help, just pop over to #👋│accademia and hit /mentoring. They're ready when you are!

[low bear]

I see the problem, you're using an XL model

XL models are very, very bad at inpainting

[wise canyon]

The AI tries to match what is within the edit box,
It's seeing too much silver car, so it tries to put it in the generation. Make the box smaller and only do a small section at a time

[low bear]

Try selecting a model that doesn't have "XL" behind it

[winter pawn]

I'll try that guys

@low bear @wise canyon it still doesn't work

I tried all the models and they all do that

That's DreamShaper7, AbsoluteReality and StableDiffusion2.1 do the same kind of weird stuff

[low bear]

Hmmmm that is weird

DS7 at the very least should work fine

[winter pawn]

Yeah. I must have messed up something in the settings

If it only happens to me

That's what I'm currently running

[wise canyon]

Select the smallest aspect ratio before making it smaller with the render density, then try and frame just the window, mask it the lower corner needing filling in, and try "match background" as a prompt.

[winter pawn]

Smallest aspect ratio? As in 1;1?

[wise canyon]

640x832

[winter pawn]

Super weird @wise canyon

[wise canyon]

One moment I'll do a screenshot

[winter pawn]

I tried removing the negative prompt but it does the same

I'll send you the image if you want

[undone furnace]

What do you want to do with that car image? Do you wanna add something?

[winter pawn]

(original image file)

[winter pawn]

I generated the building. But it made a weird texture around the car. So I want to fill it up with the rest of the building so it looks normal. The bricks and the window basically

[undone furnace]

Background stuff must've occurred 'cause you placed the image outside the ratio.

[winter pawn]

Interesting take. I'll try to make it fit all inside the ratio

[undone furnace]

If you wanna add a background in the blank space, just place the ratio in that space, and enter your prompt.

[subtle gulch]

Can you provide the true original image without all the areas you’ve erased?

[winter pawn]

Hello, it is the true original image. I erased parts of it in Gimp because the eraser tool works better in my opinion

[winter pawn]

That's what I was doing higher in the chat and it gave me weird artefacts

[subtle gulch]

The irregular erased parts is causing problem for the AI to generate based on what is left.

[winter pawn]

That would make sense indeed

I'll export it with a greenscreen and delete that in Leonardo and try again

[wise canyon]

I actually had the same weird results you did

[subtle gulch]

I would be easier if there is a coherent original image for the AI to base on to generate the edits.

[winter pawn]

Siloa is saying that it's because I erased it irregularly. Wait a second

[wise canyon]

Yep
So I changed to "sketch2image" and painted it in

[subtle gulch]

You can see here clearer of what was left behind

[wise canyon]

Close, but this one is a challenge

[subtle gulch]

In your "original" image after the Gimp edit

[winter pawn]

Yeah I get what you mean. I'm exporting it with a greenscreen and deleting it in Leonardo see if it works better

[winter pawn]

Viable alternative indeed

[subtle gulch]

First edit

Is it close to what you want?

[wise canyon]

Impressive Siloa 👍
I see you downloaded his original
(I was working off a screenshot so I think I had too low res for Leo to work with well)
Nice job 💯

[subtle gulch]

I figure something must be off about his image to cause so much trouble in editing.

[wise canyon]

Yes, I was scratching my head too 🤔

[subtle gulch]

I am using KinoXL btw, and there isn’t any issue once those weird parts are gone.

[subtle gulch]

When using Canvas Editor, I tend not to prompt too much so that the AI will make reference only on the existing image to generate the missing components.

The most common prompt I use on Canvas is simply “background”.

[winter pawn]

Close

I green screened it then deleted the parts into the canvas

[subtle gulch]

This is what I've got so far

[winter pawn]

This is what I've got so far hahahahha

Please explain @subtle gulch

[subtle gulch]

You're using a very low resolution version of your image

[winter pawn]

No

That's the max version

I just made it fit the canvas, 1024x1024

[subtle gulch]

oh ... please don't do that

[winter pawn]

Wdym?

You reckon I should upload the image, not resize it, and move the canvas over it multiple times instead?

[subtle gulch]

While using AI to edit, what in fact is happening is that the AI is making use of everything in the frame as reference to generate the part that is missing or masked.

[winter pawn]

I understand that

[subtle gulch]

It doesn't mean the AI can magically create everything perfectly in one go

so if there are multiple places that have issues, the AI will create random things as the reference is not enough for it to work

The most effective way to work with Canvas Editor is to edit it bit by bit

[winter pawn]

ok

[subtle gulch]

Like in here, I only masked about less than 10% of the image in compared to the whole frame

[winter pawn]

Ok

So that's why it shows green stuff?

[subtle gulch]

Mostly likely the green residue that has not be fully removed that is left on the edges

after the greenscreen edit

[winter pawn]

On the edges?

[subtle gulch]

For example, you can see the grey line on the edge here. Before I generate the right hand side of the image, I in fact erased all the grey edges first.

[subtle gulch]

Right, the green residue may not be clearly visible, especially if you shrink the image like that. It'd be safer to always erase the edges a bit before doing outpainting.

[winter pawn]

Ok, interesting

I'll try that

[subtle gulch]

If you still encounter issues on Canvas, you can drop by #👋│accademia and ask for help. The mentori there can start a private thread for you.

[winter pawn]

Thank you so much for your help

[subtle gulch]

Here's the image I've got so far for your reference. It took multiple inpaints and outpaints.

all done using only "background" as prompt

[subtle gulch]

You're welcome. Just let us know if you need any more assistance.

[wise canyon]

My most used fav Canvas prompt too 🙌

[storm jungle]

Just using "Focus mode" in cavas editor, I just used the prompt "Remove object", this was the was first result, I use "Remove object" for many things.

[proper crystal]

How and where?

[storm jungle]

@winter pawn You can see a similar post I help with, to give you other ideas that may help

Click here:
#🙋│image-help message

[storm jungle]

If you're after black and white coloring pages of Mandalas, you could try this style:

I used Leonardo Pheonix and preset style Graphic Design Vector............ I also note not to display text and words etc, but it doesn't always work due to the style of the image, but text is easy to remove afterward, but some won't have text.

Prompt:
Highly detailed professional graphic design flat vector art featuring a stunningly intricate and simple mandala composition, showcasing flawless symmetry with perfectly concentric circles that exude a sense of precision and harmony, rendered in bold and sleek black ink lines against a pristine full white background, adhering to a precise vector art style that accentuates clean lines, geometric shapes, leaves, circles and a minimalist aesthetic, evoking a sense of modern sophistication and elegance.

[fallen obsidian]

Hi there, just back from work, using the link that you send to me. its still can't reached 🥲

[winter pawn]

Thank you for the tip

[storm jungle]

Your welcome

[manic root]

Ready to level up your skills with a bit of one-on-one guidance? 📈 Hit up our #👋│accademia channel and just type /mentoring to get personalized support from our awesome mentori anytime!

[cold cliff]

Hello
I would need help to generate 5-10 images. I teach computer science and at the start of the school year I have a 6th grade class.
Being very fond of lessons via discord since Covid, I have integrated a reward and level system for my students. However, I am missing the logos!
The idea would be to create 5 or 10 logos in the style of prestige in call of duty or ranked ranks in video games.
I want to make 5 or 10 levels that evolve each time. The max level being the highest level, it has to be impressive 😅.
I tried a few prompts on Leonardo but was not very satisfied with the results. So I am calling for your help!

[meager rover]

I made a beautiful hammer and it's standing I want to change the way it's positioned for exp I want it to be on the ground but I want to keep the same hammer can you please tell me which image guidance tool should I use?

[cold cliff]

Hi, it would be a great pleasure! I look forward to your suggestions.

[storm jungle]

Is this sort of what you had in mind ?

[cold cliff]

@storm jungle why not, maybe work a little more on the logos and with a base that evolves for example in this idea: (not paying attention to the colors)

[cold cliff]

I like this base

[storm jungle]

The prompt I used was:

Create 5 separate stunning high detailed logos in a form of a reward ranking system, logo 1 is "level 1", logo 2 is "Level 2", logo 3 is "Level 3", logo 4 is "Level 4", logo 5 is "Level 5", each logo has its own color.

[cold cliff]

Ok thank you for the prompt

[storm jungle]

Your welcome, in prompts you can describe each item separately

You can describe each item between each logo set of rules for example:

Prompt:
Create 2 separate stunning logos, Logo 1 is yellow with black outline and in a shape of a triangle with the text "Level 1", logo 2 is green with blue outline and in a shape of a hexagon with the text "Level 2"

gives the attached result

[dim zodiac]

yo this is a COOL application @cold cliff !! Your students are lucky to have a teacher with a creative mind to keep them engaged!

be sure to do all this in the Phoenix model, as it's the only one that will handle the text well

[storm jungle]

I never had a teacher like that : (

[cold cliff]

you have to interest the new young people 😂 and then I like it when it's fun and they do too and they always talk about their prestige call of duty or warzone stuff 😂

[dim zodiac]

totally 😄

current battlepass is a lot of gold and green highlights, so if you want to be super topical use those terms as well

[storm jungle]

Hats off to you : ) Students need more teachers like you.

[cold cliff]

I like it less, I prefer something a little more worked in the logos a little more worked but thanks for the generation 🙂

[dim zodiac]

The AI went a little haywire with how many it made, but this is reminiscent of camo progression 😄

[storm jungle]

: P It does that sometimes, the ones I posted earlier where the ones that were perfect, others had mixed numbers and or overlapping each other

[dim zodiac]

AI still being AI... the randomness still kicks in

[storm jungle]

End of the day, we only need one that we want : )

[cold cliff]

I really like this style, is it possible to create something like this?

I don't care about the colors

I ask a lot of you 😅 thank you in any case for the suggestions

[storm jungle]

Do you need levels or rank number or anything, or just a logo on its own

[dim zodiac]

I think you'll have better control generating one, and then using it as image input on the next

so that they all cleanly evolve... otherwise there's always going to be a randomness element that you'll fight when trying to do them all at once

[storm jungle]

So you like stuff like this but random styles for each rank

[cold cliff]

Woaw !! I like it 😁

This is Lvl 5 😂

What prompt did you use?

[meager rover]

@night vault Can you tell me what preset did you use for this image ? and maybe what elements

[dim zodiac]

@cold cliff This would be my other thought on doing it. Find one image you like, and then use IT to generate the others

[cold cliff]

Wow wow wow 🥰 I love too

@dim zodiac @storm jungle you guys are awesome! I love it, that's what I'm looking for

[dim zodiac]

there you are... 5 ranks 😄

[cold cliff]

even better !

Yes thank you !!!

[dim zodiac]

so I started from this prompt: A ranking logo similar to Call of Duty, a small shield made of copper with a star in the center, shiny metallic highlight, grey background, Octane render, UHD

in AlbedobaseXL using RayTraced preset

and then took the resulting image as Content Reference for the next, and changed the colour and maybe one detail

[cold cliff]

I am waiting for a proposal from @thin tinsel and @storm jungle for 5 lvl and I think I create 15 lvl with 3 rank 😅😅

[dim zodiac]

typically kept Content Reference set to Mid usually, and then Low when I wanted to add crazier details

they may or may not continue to add images now... we've given you the tools - it's not really about crowdsourcing

[dim zodiac]

niiiice

Style reference?

[storm jungle]

Had to work through quite a few generations before I got a good one, then I used it as an img2img to create mroe deisgns from it, once I got 5 I put them together and generated it as one suing img2img again

[dim zodiac]

yeah so similar workflow

nice

[storm jungle]

here is brighter one

[cold cliff]

Ahhhh y’re incredible ! they are all beautiful

[storm jungle]

They are yours if you want them

[cold cliff]

how can i get them back 1 by 1? i mean separately other than cropping the image?

[cold cliff]

with great pleasure ! Thank you for tour Time

[storm jungle]

If you crop them, you can re-generate each one to get a larger one using img2img just select img2img at 0.90 weight, being max weight

[cold cliff]

is it on Leonardo ?

[storm jungle]

I used LifeLike Vision with Dynamic preset style.. Phoenix doesn't have image guidance yet, but in the future it will have.... when you use img2img, use describe with Ai to get a prompt, then generate the image... remember to set the weigh in the img2img guidance to 0.90, repeat all the steps for each image.

[storm jungle]

Very nice : )

[cold cliff]

Woaw it’s beatiful too ! But but I'm missing a little bit of color😅

[thin tinsel]

this is just to whet your appetite. 😆

[wanton belfry]

with phoenix, this is pretty convinient

[cold cliff]

I hopewait to see the dish then xD

In any case, I love the community here, thank you!

[cold cliff]

Where do you find this menu ?

[dim zodiac]

click the little Image icon beside the prompt for all image guidance options

^^^ this lil guy

[cold cliff]

ok thanks but i've this

[thin tinsel]

I think you've got what you needed from the other Mentori Zro and 3ric.

[dim zodiac]

yes, only possible in models other than Phoenix

[cold cliff]

ahhhhhh okay xD

[thin tinsel]

Just wanted to add that I didn't use Phoenix, I used GAME UI element.

[dim zodiac]

very smart! it's pretty much MADE for this 😄

[cold cliff]

I'll keep those for roles 😁

[cold cliff]

Today you are my teachers 😂

[manic root]

Eager to improve your craft or crush those learning targets? Our mentori are on standby to guide you! Just pop over to #👋│accademia and hit /mentoring to get one-on-one help. 😊🚀

[cold cliff]

a little more epic I would like it better so that the students are proud of their logos 😅

[storm jungle]

https://tenor.com/view/brasasenglishcourse-brasas-teacher-mode-on-gif-18386568

[thin tinsel]

I have no idea what 11 - 12 year old like, or what prestige call of duty looks like.
But I do know how to use Leonardo, and that's what I was bringing to the table.
As I said, the others have done a great job, and given you stuff you can work with, so I think I'll go and have amy dinner now.

[night vault]

https://app.leonardo.ai/ai-generations?imageId=60e25083-a365-4e13-9a3b-e3060015fa78

LightningXL + Anime Semi-realistic preset
Elements

• Folk Art Illustration -0.50
• 3D Sculpt -1.10
• Sparklecore 0.20

[cold cliff]

No worries, thanks again for your help and the images you were able to generate! I would like to take a little course on how to use it in the future 😉
Yours sincerely

[cold cliff]

this is perfectly 🙂

[thin tinsel]

then why did you say they wouldn't like them. 😦

[meager rover]

thanks a lot brother

[meager rover]

but I'm curious what are the effects that negative elements make

[storm jungle]

Using "Game UI" element can give you great designs depending on how you prompt them to look

[cold cliff]

I took the time to analyze them in detail and I did a survey among my children. Sorry for my hasty response

[cold cliff]

beautiful too 🙂

[storm jungle]

Yeah, just depends on the prompt description how you want them to look, you could use baby wings as a starter, then have bigger wings, it all depends what you want.

A shield on it's own, then a shield with one wing,, then a shield with two wings etc

[cold cliff]

yeah but i'm beginner n AI image generation

[night vault]

It is not well document. But what I observed, it add grains to the output. This will turn clean image into more detail, sharp edges

[storm jungle]

Some people forget that basic words or common-sense phrases will do the trick, like most who never used AI before see the big picture first and get overwhelmed, you can start with basic phrases.

The above images all I did in the prompt was:
World war game style rank award, gold shield with silver wings on a white background

You can use basic wording, ignoring grammar, and get Leonardo to enhance the prompt, start with a silver shield with gold wings, that simple, AI will do the rest, then alter the prompt that was enhanced for you.

The image attached here using GAME UI element was simply the above with a few changes:
World war game style rank award, neon purple shield with gold wings and chrome highlights on a white background, with intricate details

[cold cliff]

Do you offer Leonardo initiations ?

[storm jungle]

Settings I used "Legacy Mode"

[thin tinsel]

Thank you. I'm glad they liked some of them.

[thin tinsel]

We have the mentoring scheme where you can join up with a team of mentors, @storm jungle and @dim zodiac and me and one other are all part of a mentoring team.
If you are interested in receiving support and training for specific ideas and projects I think @dim zodiac will be the best placed to tell you about that. 😄

[dim zodiac]

Thanks Willow!

yes, we offer private threads so you can group all your help in one location, can be very useful! Plus, if it interests you - a bit of added camaraderie within the discord

[cold cliff]

@dim zodiac @thin tinsel @storm jungle @wanton belfry Okay. Thank you very much in any case to all of you who helped me with this project. I thank you very much. I discovered a community of enthusiasts and above all attentive. I greatly appreciated these exchanges and especially loved your responsiveness and the images you generated. Thank you all !!

[thin tinsel]

You're very welcome.
Tell your friends. 😆

[storm jungle]

You're welcome : )

[cold cliff]

😂

[cold cliff]

i don't have like you 😭

[dim zodiac]

this is Zro's not-so-secret secret

😉 that will take you to the V1.0 GUI with a few different options in layout

[thin tinsel]

This is what I was doing with the black and bronze images I posted initially.

[brave ether]

I would like prompts that use geometric shapes, so that the background is light and simple... Something similar to this image, but I couldn't find anything that looked good...

[low bear]

Ooooh yeah that is definitely not what the AI is trained on so it will be hard, but it might be possible...

I'm gonna give it a try and report back!

[brave ether]

I wanted to buy it to add models to a database, but I don't have a credit card... That's sad.

[low bear]

hmmm this wasn't exactly what I wanted but pretty good for a first try lol

gonna keep trying

[brave ether]

It's better than mine hahaha

[low bear]

Pretty cool image though!

[brave ether]

It looks good, but for a presentation it has too many details, I would like something cleaner.

[low bear]

Yeah true haha

[real lynx]

Albedo xl

A mesmerizing abstract geometric masterpiece featuring an intricate layout of interconnected translucent polygons. The composition utilizes a harmonious balance of varying shades of blue and white. The design evokes a sense of depth and movement, with each polygon seemingly interacting with the others, creating a dynamic and visually engaging experience.

[low bear]

getting somewhere!

this one is really neat too

Any of those match what you're looking for @brave ether?

[brave ether]

The image is very interesting, I think I can make some edits and use it, thank you very much!!! You are very creative with prompts, I have worked as a developer but I don't have that creativity hahaha

[brave ether]

That's it!!! It was perfect

[low bear]

prompt simple plain background with monocolor 2D polygons forming a gradient, white with light blue shades
model "Phoenix", on "Quality" mode
preset "Graphic Design Vector"
contrast "Low" (VERY IMPORTANT)

thats how I generated it

[quick verge]

im no helper, just decided to play, got this... to dark for sure. i cheated and put your image in "describe with ai" then ran it through phoenix with the graphic design vector preset... again im a novice, just wanted to play too 🙂

[low bear]

If you had put the contrast on low you probably would've gotten something similar to mine ;p

[quick verge]

Ruby's is 100x better, just wanted to have fun 🙂

[quick verge]

yeah, i never touch that... should 😄

[real lynx]

My prompt only uses them in these presets

[low bear]

Yeah it is great to experiment with, can change your result a lot!

[brave ether]

Any help is always welcome! You got a much better result than me haha

[cold cliff]

I love this i take 😅

[real lynx]

Your method is perfect

[quick verge]

the prompt it came out with was:
"Geometric abstract image featuring a polygonal design. The layout consists of various interconnected triangles and polygons, creating a faceted, crystalline appearance. The color palette is predominantly shades of blue, ranging from light to dark, with some areas blending into white. The overall style is modern and minimalist, with a clean and crisp aesthetic. There are no people, text, or logos present in the image."

[low bear]

Yeah thats great

[thin tinsel]

Yeah I thought they could all have numbers from the most plain looking one at 1, to the most eleborate shape at 10.

[low bear]

Hm I wonder if I can improve my !rank background images using monocolor 2D polygons I used in my prompt

[real lynx]

This is a very good tool to find prompts similar to your images

[quick verge]

amazing feature, even if you dont use its output directly. its a great starter to get a vibe... and if you just leave it itll spit out some amazingly similar things, i did a disney one that just was wow

[brave ether]

Got it. I'm not very familiar with the features.

[tired trench]

oh thank you. This could help in creating a design I can use as a mask over other creations.

[cold cliff]

Absolutely, it's a good idea.

[proper crystal]

I am trying to change the eye colour to a more brighter sparkly green but it's turning blurry or watery or just out of place for some reason

What am I doing wrong?

[low bear]

Try to use DreamShaper 7 as the model, that may help

also, your prompt should be "neon green eyes"

the AI doesn't know what to turn neon green now

[thin tinsel]

Turn render density up to two

[proper crystal]

Yeah I tried that too same results but I'll change the model!

[low bear]

Trying it out myself real quick, will get back to you

[uncut carbon]

do you need the image?

[low bear]

I figured it out I think

whoops, nevermind

[low bear]

nah I can manage without, just trying to figure out why erase is not generating anything as you say haha

[proper crystal]

@low bear Omg yeah defo better results still not quite there yet

[low bear]

Alright so, it seems the "Inpaint" mode is meant to be used only by the brush tool, which you can use after switching to the inpaint mode

You erased something in "Draw" mode and then switched to Inpaint mode, which does not work

In "Inpaint" mode, you use the "Brush" tool to draw over something which will act as a mask, and then it will inpaint something in this area

which is how I got this beautiful thing 🤣 🤣 🤣

If you just paint over the erased areas with the brush tool in "Inpaint" mode it should work @uncut carbon! :D

[uncut carbon]

yeah, no, no matter what I write, it just removes the head

[low bear]

Hahaha rip, it might help describing the entire robots

"image with two robots with LED eyes"

something like that

and if that doesn't work, you'll probably want to use normal AI Canvas for this rather than Realtime Canvas

[proper crystal]

How would I add like a fiery, neon green effect on her hair?

[proper crystal]

For this

[uncut carbon]

lmao all I got was the smaller twin towers xD

[manic root]

Need a helping hand with your learning journey? 🛠 Our mentori are ready to guide you every step of the way! Swing by #👋│accademia and type /mentoring to get personalized help whenever you need it. Let's reach those goals together! 🎓

[uncut carbon]

yep, defo going with the normal canvas

that would be the canvas editor, right?

[low bear]

Yep!

The name for that keeps changing lol

[proper crystal]

Why doesn't it show flames?

Like green flames

I am so confused

[uncut carbon]

the image stays completely the same when I click generate

all 4 of them didnt change a pixel

and I'm using up my points 😢

no matter what I type in, it doesnt change the image at all, do I need to mark where it has to change things or?

[spring trail]

canvas editor retains all images i referenced in. the past -- how to enter canvas editor with an empty canvas?

[dim zodiac]

you need to be in Inpaint mode (have the Outpaint toggle OFF) at strength 1 and use the Brush tool on the side to paint the robo faces

then it should work? I would suggest using the DreamshaperV7 model to do this

[dim zodiac]

hit the orange garbage can!

[spring trail]

that did it -- thanks

Prompt: Filipino young adult man looking directly into a bamboo mirror wearing slacks and plain polo shirt . Philippines background depicting a traditional Filipino large bedroom with bamboo walls and a thatched roof.

How to remove the legs underneath the mirror?

[dim zodiac]

easiest would be in Canvas

In inpaint mode, mask the legs, then prompt for "bamboo floor"

[dim zodiac]

nice!

[spring trail]

not bad -- i spent a lot of tokens but it looks acceptable for the preliminary image in my workflow -- next is to convert it to illustration -- thank you for the advice

[real lynx]

Hi

If you want, I'll try what you're missing.

[spring trail]

This is the closest I've gotten to my goal so far -- it's really a shame I can't get it to reflect the same colors but this prompt is probably more complex for AI to figure out

Prompt: Handsome Filipino adolescent male with a clean face no facial hair gazing intently at his own reflection in a bamboo mirror, his serious expression softened by a hint of self-contemplation. He wears slacks and a plain polo shirt, with no adornments or accessories. The Philippines background depicts a traditional Filipino large bedroom with bamboo walls and a thatched roof allowing a subtle ray of sunlight to illuminate his introspective moment.

Image Guidance: Style for the 'illustration look' and Content using the previous image above

[weary badge]

Guys, I'm trying to use canvas to expand a picture of a medieval artwork, and leonardo keeps adding robes to some character, is there a way to avoid this?

[storm jungle]

You're very welcome : )

[wise canyon]

Hi Conrad, I take it from your message history that you have removed any "robe" words from your prompt in the Canvas Editor, and have tried the neg. prompt field.
The most important aspect of Canvas to remember is that it "looks" at the image content in the edit box area as well as the prompt, when generating.
The positioning of the edit box is a critical factor so It's possible, in your case, the AI is "seeing" robes somewhere in the edit box.
There may also be other factors causing this.
We have plenty of experienced Canvas users in the community who can help, and also personal Mentori who can assist you "one on one" if you like: https://discord.com/channels/989166677390426132/1230345629776085023 and type "/mentoring" to get personalized help whenever you need it.

[elder bear]

Hello everyone,

I am asking if there is a way to generate a consistent character and keep the color of the T-shirt the same. I used a seed, but the color still changes. Does anyone have any idea how to achieve the same color? Is there a way we can use a hex color value?

Thanks!

[feral bough]

Hello, could you give us a bit more context?
It seems you're experiencing a failed generation. This is a bug/problem that has been reported by a lot of people the past few days. I believe the team is still working to fix it.

[fickle lintel]

From 7 days, i am facing the problem. The team say "boy" word is the problem but other user used my prompt and they successfully generated the image.

[feral bough]

Yes this is not a moderation filter issue as that would send you a different error message. So this has nothing to do with your prompt most likely.

[fickle lintel]

They messaged me as they are not real human.

Are they?

[manic root]

Eager to boost your skills or conquer challenging projects? Our mentors have got your back! 🚀 For personalized guidance, just pop over to #👋│accademia and send a /mentoring request. Let's level up together!

[feral bough]

The one with the M icon next to the message is a human, the one with the black icon is a bot

[meager rover]

Maybe use the content or character reference tool and add the green flames or use pheonix and enhance it with other models using image guidance tool since the quality of phoenix is awful

[proper crystal]

WHAT HAPPEND HERE 😭

Nvm I fixed it

[proper crystal]

Wdym?

[meager rover]

Ignore the canvas and use that image to make the image you want

[proper crystal]

You might wanna elaborate

[meager rover]

Ignore the canvas completely go to image generation and use the image guidance tools try(Image to image ,content reference or character reference) pref on high and on the prompt write eyes with green flame or something like that

[proper crystal]

With the same preset that I've used before right?

I can't do content or character reference as I am a free user

[wanton belfry]

canvas is generally to fix details like shirt color or hairstyle or messed up hands

everyone still probably generates in Img Generation

[proper crystal]

Didn't work

:/

[meager rover]

image guidance strengh?

[proper crystal]

0.30

Should it be higher?

[meager rover]

Can you give the image I will try but maybe not now

[proper crystal]

I'll DM you it

[night vault]

can you elaborate what did not work?

[wanton belfry]

you want something like this?

[proper crystal]

Defo something like that

[wanton belfry]

sadly even though its easily done with Depth, however it will take alot of work to make it work with img2img

without proper image it comes out something like this

[proper crystal]

So how'd I go about it?

[wanton belfry]

find a very suitable image of girl with firey hair with composition you want

img2img takes alot of colors, composition from image

depth pastes detail over it

[proper crystal]

So I can't use the image I generated?

[wanton belfry]

I used this to make this

[wanton belfry]

maybe you can, if its in right colors

you see even if you ask for green flames, its still gonna do yellow because its yellow in original image

so if the composition of your own image is of your liking then it can be done

[night vault]

I dont get what vally asked yet

[wanton belfry]

he wants a girl with Green firey hair

he doesn't have premium so limited to img2img

we're finding ways to get it done

[night vault]

ok

i see it linked back like 15hr ago

[proper crystal]

So it's alot harder if the image I want to reference doesn't have flames to begin with?

Coz I honestly thought you can just add it on with prompts or something

[wanton belfry]

to a extent, but if it doesn't have proper noise present in the picture...it'll not materialize something from nothing. You need something there on hair which it can change

this is a limiter on img2img

Content allows you change colors/add effects, Depth pastes a few things over the noise with even minimal noise there

[proper crystal]

Yikes free users don't have much to work with

[wanton belfry]

there is still one option, let me check

[proper crystal]

Okayyy

[wanton belfry]

this is depth from base SD, Dreamshaper

I think its not premium

if quality is not satisfactory you can run img2img again on it in anime XL

[proper crystal]

Where to find?

[wanton belfry]

After running the same image as posted above in Anime XL with same prompt

[wanton belfry]

go into legacy mode, go to models and select Dreamshaper 7

[proper crystal]

Is that it? What's next?

[wanton belfry]

use depth, put in your prompt, once you get decent looking image take it and run it in Anime XL to uplift quality

[proper crystal]

I am so lost

[wanton belfry]

to summarize, just use Depth2img of DS7 which is not premium to get the image and if quality is not good enough then use whichever best output you got into Anime XL

I know its a very roundabout manner of doing things but other than having a perfect image for img2img this is alternative

[proper crystal]

I am not seeing a depth2img anywhere

[wanton belfry]

in DS7?

[proper crystal]

I have DS7 but there is no depth2img

Depth2img is a premium feature

[proper crystal]

OH WAIT

I am blind

[wanton belfry]

No the format for legacy and new mode are different so obviously people get confused

I get it

also do not put depth to 1.00 power, keep around .50

otherwise you'll get garbage output

[proper crystal]

So for the depth2image, I put this image for reference?

[wanton belfry]

sounds like a plan

[proper crystal]

And then for the prompt do I use the same one orrr?

[wanton belfry]

there is one more issue, since DS7 is previous gen, output is going to be pretty sad looking

that bioluminescent prompt, of course

I got this

[proper crystal]

Ohhh right right

My generation failed

And it took my tokens

This is terrible

[wanton belfry]

told ya, its previous gen....the quality is pretty outdated

[proper crystal]

Is that the only way?

[wanton belfry]

that's why you re-run it with anime xl

[wanton belfry]

only 2 ways are, either this or get a image with right color and composition or you can even alter that image in some software to get it there

[proper crystal]

It does absolutely nothing like the original image except it making it look goofy and it just wastes my tokens

[proper crystal]

Image with right color and composition meaning?

[wanton belfry]

for example you want green flame hair, you need some noise that'll let AI understand where to put firey hair, green color should be adquaetly present in color composition too

with going back into depth, you've given additional noise for it to work with

[proper crystal]

What's noise?

[wanton belfry]

lets say there is a white background and image of character in between, AI sees that character as noise

it changes it to provide different output

while staying true to original composition

now if that character doesn't have long hair, AI can't easily put in long hair...it has to compensate with something to stay true to composition, so some other part is taken as hair or its ignored in prompt altogether

this is why it was ignoring your prompt because it doesn't have right noise details to work with to make firey hair

[proper crystal]

Ohh okay I think I understand so it needs to have firey hair in the first place in order for it to work properly?

So that the AI can use that?

[wanton belfry]

more or less yes, this is why I went back to DS7 outdated depth, to get noise and re-run into Anime XL, now hair has fire

once it has right sort of noise in place to play with, it can interpret it as fire

[proper crystal]

So the DS7 depth helps adding the fire and re-running it into Anime XL enchances the quality?

[wanton belfry]

yes, but the problem is DS7 is old gen, not only quality but the whole thing is outdated as heck...alot of outfits will go bad too

[proper crystal]

Will there be a DS8 soon or something like that?

[wanton belfry]

because if you go Anime XL to DS7 depth and back to anime XL you'll have very different image

[wanton belfry]

DS8 is already out, not many people care much about base SD now though

especially after Lightning models, whose gonna care about it

just look at this quality....

[proper crystal]

What model beats it?

[wanton belfry]

none yet

Kino is on slight better side, since it can also do realistic

[proper crystal]

There's loads to learn lol

[wanton belfry]

All models have their uses, Most popular are Kino, Anime XL, Albedo

out of which anime XL is the only lightning model

there is also phoenix

oh wait phoenix

[wanton belfry]

Oh I keep forgetting there is phoenix

[wanton belfry]

bloody heck, if you can spare tokens, Phoenix is capable of doing it

however no img2img

it has to made from scratch with prompt

[livid inlet]

Phoenix is only 15 tokens on small setting 🔥

[proper crystal]

Oh yeah phoneix

[wanton belfry]

I feel a bit dumb, I keep forgetting there is phoenix now

even though you can't do img2img to get phoenix output, its not like you cant use image gotten from phoenix and img2img it in Anime XL

[proper crystal]

Yeahhh just phoenix uses alot of tokens

I don't use it much

[livid inlet]

I call that the Kentox Method now lol

[wanton belfry]

I already named this sort of practice "Cross modeling" Long back

once it sees noise in right place, it can play with it

While Phoenix lacks for making great art in terms of quality, its very prompt relevant and very good get basic structure of what you want

you can cross model to get the quality+

this is the third method

[proper crystal]

Sooo I can use phoenix first and then use anime XL after?

[wanton belfry]

yea, but no real need if the output is already decent

if its bad then maybe

or if its realistic that you want to convert into anime then again

[proper crystal]

If I do convert it into anime, will it keep the details?

[wanton belfry]

you can compare the outputs...phoenix on left and anime XL of the same on right

oh waits that's depth though

sorry not img2img

https://tenor.com/view/oops-yikes-awkward-jennifer-lawrence-gif-9504720997818108080

[low girder]

@real lynx help for create logos

[elder bear]

I need help please

[long coral]

I need help modifying an image please!!

[dim zodiac]

The more detail you can provide on your prompt for original and new images, the more we can help out! My suggestion would be to be as descriptive about the colour as possible, use the same terminology in both prompts

[dim zodiac]

Heya Menda! What are you trying to achieve?

[long coral]

I need a cool background, the one on the left has to have darker hair and brown eyes and the one on the right has to have blue eyes and a shorter beard. And fix the hand on the shoulder hahaha

[elder bear]

She is wearing a plain light green T-shirt.
But, I always get different light green color.

[manic root]

Hey there! 🌟 Need a helping hand to level up your skills, crush that big project, or hit your learning targets? Our awesome mentors have got your back! Swing by #👋│accademia and type /mentoring to get personalized support on your journey. Let's grow together! 💡

[dim zodiac]

Ok this sounds like a Canvas project. I would generate a background for them in Albedobase separately, just an environmental scene. Then I would remove the white background from this image, and later it over the background in canvas. Finally I would use inpaint mode in canvas to prompt for those changes

[dim zodiac]

Yes so be even more specific "light emerald green" "light seafoam green". Hex codes I've not seen consistent results with

[night vault]

Seafoam https://cdn.leonardo.ai/users/b8b7cb49-52d0-41e7-9766-6d1c4c5ea529/generations/9c04bdef-9353-4efc-8670-30a58e40ee55/Default_honkai_impact_Lois_van_baarle_merged_with_Chris_Sander_0.jpg

[elder bear]

Thanks so much @dim zodiac
Any idea how to generate the character in side, back, top view?
I tried many times to generate it with seed in side view, I failed.

[night vault]

just prompt for "side profile" at the front of prompt for better chance to get side view. also seed number does not matter much in this case.

[elder bear]

"side profile" it works, thanks a lot @night vault
But the side is not 100%, any adjustment?

Regarding seed, do u mean useless?
What I know it links to the image I generate and want to work with in different poses, situations.

[night vault]

yes, the seed works but only if the change is not significant. you can change shirt color, hair style, pant to skirt, etc

if there is significant changes like perspective change, front to top view, major change posture, zoom level change, those are likely not benefits from seed

[elder bear]

I noticed when using "side profile", the body generated in side view, nut face still looking to the screen "front profile".
Is there any way to fix?

[night vault]

does "looking to the left/right" help?

[elder bear]

Not exactly, the results are as her face looking forward to screen

[elder bear]

But I tried before creating Character Sheet, Leo generate back, side perfectly

[thin tinsel]

You can use STYLE REFERENCE, once you have an image with the green you like and then use that an the style reference image. Also use the specific colour name in the prompt as 3RIC suggested.

@elder bear You have various options with image input.
Sketch to image, pose to image, content reference etc etc to get the composition you want for your image. If you share in more detail what you need to accomplish, then we can provide more detailed guidance.
Hi, I'm Willow, one of the Mentori for the Leonardo Discord server.

[elder bear]

Welcome @thin tinsel
Thanks for your help
Will try and update
Thanks all

[dim zodiac]

cyber monkey temple?

[storm jungle]

Monkey business

[pure geyser]

I want this image looks real fire and thunder for digital download which model preset should I use to create this type of image @muted canyon @spiral axle @dim zodiac @storm jungle

[subtle gulch]

Do you mean you want to edit this image to make it more realistic?

[pure geyser]

To generate this image with prompt

Model preset?

[subtle gulch]

AnimeXL can create effects like that

With the Semi-realism preset you can get something similar I believe

[pure geyser]

See this guy's image so realistic digital artwork (he use mid journey)

I want to generate like this realistic (editing part I can do myself)

[pure geyser]

Which model preset is best

[spiral axle]

@pure geyser you need help ?

Vision is for my taste best for fantasy realism

But u can do it well with PhotoReal

[storm jungle]

What @spiral axle said, LifeLike Vision, with either "Dynamic" or "llustration"

This example was illustration style

This one using dynamic

Dynamic would probably work best, but you can see how both styles turn out.

The prompt I used for both is below, just change it to your liking:

Highly detailed and dynamic photorealistic featuring an anime character with intense facial features and a fierce expression. The character has glowing yellow eyes, sharp teeth, and a wide, menacing grin. Their skin is a deep, fiery orange, and they are surrounded by swirling flames and embers. Black, tribal-like markings are visible on their face and body, adding to the intense and powerful appearance. The background is filled with vibrant, fiery colors, enhancing the overall dramatic and energetic atmosphere of the image.

[subtle gulch]

I don’t think any models in Leo can generate an image up to that level of quality from MJ. MJ’s model has been extensively buffed with copyrighted materials to reach that level of image quality.

[spiral axle]

realistic

[manic root]

Hey, if you're eager to level up your skills or take on some cool projects 🚀, our mentors are ready to back you up! Just pop over to #👋│accademia and type /mentoring to get personalized help.

[subtle gulch]

You can still get something decent with Leonardo if you know how to manipulate the elements and make use of the universal upscaler.
(Quite a lot of MJ art creators I know in fact use Leo to upscale images)

This one is created with AnimeXL.

[hardy sentinel]

Great image, tropical

[spiral axle]

but yeah upscaled still very realistic 😄

[hardy sentinel]

Sehr schoen. Upscaled bekommst du dann halt mehr Detail

[wanton belfry]

everything can be done with Anime XL

[spiral axle]

😮 @wanton belfry have u tried combining Style with the old Legacy Mode ?

[wanton belfry]

yes

here's fire

[spiral axle]

what styles 😄 ?

[wanton belfry]

It was bit before, I can't remmember the pics

this is without any image ref

I usually go for semi-realistic or Anime Illustration

those are both fancy

[spiral axle]

that new ui

[wanton belfry]

I can get good stuff in old mode too, however its very varied

[spiral axle]

depending on what style I change (same Prompt ... just changed Digimon to Monster Hunter)

[subtle gulch]

I’m still using the mobile app more than the new ui though, even for editing.

[wanton belfry]

I would use style more if it textured only a certain thing, not the entire pic

[wanton belfry]

never used Mobile App, does it still have all the features?

like 16:9 ratio

[subtle gulch]

No universal upscaler
Old UI (legacy mode) only
Canvas editor only in Focus mode

[wanton belfry]

I see you cant change resolution either then because its only in new mode

honestly speaking this is the only thing that made me keep using it

[subtle gulch]

Mobile app is only in old mode

[wanton belfry]

this looks like diffusion

[spiral axle]

Anime XL ... its just Dark Souls combined with Digimon

[wanton belfry]

go bit, get 16:9 make wallpapers

[spiral axle]

like this 😄 ?

[wanton belfry]

yea but its mostly empty, need more stuff populating it

you can even directly do content ref on a wallpaper and change it...its too good, very fun

[wanton belfry]

very Dark Souls like...

I can hear hans zimmer playing behind the scene

[sleek gorge]

Looking for image around soho

[spiral axle]

soho ?

[proper crystal]

How is it inappropiate????

Wtf leonardo

[night vault]

and smaller gen for same token cost

[night vault]

negative element can 👀

[night vault]

yeah I was frustrated too today 🫂

[subtle gulch]

That is what I meant…

[subtle gulch]

Try removing the word “hair” or “burning”

The AI for some reason thinks it’s an act of violence 🤦🏽‍♂️

[night vault]

ai has no agenda. their filter model just sucks.

[subtle gulch]

Really? 🤔

[proper crystal]

The AI is a snowflake fr

[supple junco]

How do I remove a filter from a persons face? I have the 12 dollars subscription

[night vault]

same 12 per 4, 15 per 5 (1 controlnet)

[feral bough]

'Hair burning' is associated with violent imagery by the filter which is against the TOS. In your case it's not against the TOS, however the filters are not so refined that they can detect the context, the part of the image you're wanting to change, etc. It's an unfortunate by-effect of having to tighten the filters to prevent the generation of illegal content. The devs are always working on making the filter fairer & smarter to prevent OK prompts like yours from being blocked.

[spare basin]

Hi! I'm trying to create a prompt for my D&D character. Generally the description is working out fine. I'm trying to affect the camera angle. The character is a dwarf, so I want the camera to be up higher a little bit looking down. I've tried adding a few things to the end of my prompt like

The composition of the shot should be as if the viewer is a much taller person than the subject.
and
He is viewed from a slightly elevated angle, making him appear more compact and sturdy.
But neither of these seem to have any effect on the composition at all. Any suggestions?

[real lynx]

Fisheye effect camera aerial view.

I'm also a dwarf and it works well 😅

[twilit fiber]

Hiyall 🙂 When trying to expand images via the Canvas, I always get results like this, where I can't get it to see the edges of the head, or hair locks. Any idea on how to do ? I've tried things, and "top of the head, hairline, scalp" usually gets some results... but not consistently at all, and not everywhere.

[sinful basin]

Hello all, any pointers to ensure there are no typos/spelling mistakes if I want to model to use a specific text in the image?

[subtle gulch]

It’d be more consistent if you include the whole head of the character in the work frame as well while trying to expand from the edges. It’s hard for the AI to understand what to expand if you do not provide enough reference.

You can see how hard it is if you were asked to imagine and complete the rest of the image but were shown just the tip of the character’s hair.

[subtle gulch]

It’s kind of hard to generate 100% accurate text if it is too long.

Breaking that down into smaller chunks and lines may help.

[thin tinsel]

you can just let it happen, then repair it afterwards.

[storm jungle]

You could playa round with my text guide here, if this helps

#1270660361829220384 message

[thin tinsel]

Hi, I'm Willow, one of the Leonardo discord mentors.
I shared my process for text here: #1268274778112786442 message

[twilit fiber]

I will test out all that ; I try to include as much from the image as possible in the frame, so it draws data from it, but in my first exemple (the blondish hair), that's exactly what I did, and it still messed up (repeatedly)

I'll try to use shorter inputs too... And Yes, I could repair, which is what I do all the time, but hair is always such a pain in the *ss 😛

And thanks to @thin tinsel and @storm jungle for the links

[storm jungle]

👌

[manic root]

Ready to enhance your capabilities or dive deep into a new project? Our expert mentors are eager to help! 🌟 For personal guidance, swing by #👋│accademia and enter /mentoring. They've got you covered!

[night vault]

leave editing area about 25% max for compare to whole canvas view

[subtle gulch]

To avoid such inconvenience, while you’re generating images, you may include some keywords that can help preventing unnecessary cropping of your character, like “full body”, “close cropped”, “wide angle”, etc.

Another way to prevent it is by using a more appropriate dimension ratio.

[robust iris]

Hello

Which model preset and style use for this image

[eternal kayak]

looks like anime + semirealism

[robust iris]

Yes But i am unable to Got the same result for series of images

now i use the same style model everything to generate the image that little girl is eating ice cream but i am unable to generate

[odd meteor]

do you know if there is a special preset or setting for creating interier image?

[robust iris]

i am new to everything i am learning

[foggy zephyr]

ho to create un image

[vestal lake]

hi guys how can i combine these images, with no writing? thanks

[real lynx]

Hello, you want a similar image and to be able to write about it

[vestal lake]

nah i need to the images mixed together with the camera film roll

[real lynx]

Oki

First fail I had not understood

I'm going to try again

[real lynx]

Something like that ?

[sacred yoke]

hey guys! Quick question in case anyone knows this.
I am using Leonardo.ai to generate images for a flipbook animation I am doing right now and I am having trouble controlling the camera angle.
I thought it would be obvious that when I say e.g. Top View of .... bla blah i would get a top down camera angle. When I say side view of people running in a street i would get a camera on the side angle but it refuses to obey my commands and i end up wasting a lot of credits to get something remotely decent out of it.
Do u know if there is a way to force camera angles in Leonard via my text prompt?
Here's an example of a bad prompt i did today:
a man in a hoodie running in absolute terror in a downhill city street, camera from the opposite sidewalk view

at this point I was frustrated that I couldnt direct the camera with traditional terms that my english broke...
However I don't know how Leonardo would interpret the camera placement but at least a man running in a downhill city street means that the man would be running downhill... not upwards.
And yet this is what it gave me

what i actually was trying to achieve was a side view like this:

[vestal lake]

these are perfect, the big one is the one ,what was your prompt ,thank for the help

[real lynx]

Digitally created image with a mystical and spiritual theme. The layout features a central figure in a meditative pose, surrounded by a glowing aura and intricate mandala designs. The background is a gradient of dark blue and orange hues, giving a cosmic feel. Around the central figure, there are six individuals, three on each side, sitting in a similar meditative pose. They have dark skin and are wearing traditional orange robes. Above the central figure, there is a bright, radiant sun-like symbol. The text 'Thought Meditation' is at the top in a white, serif font, followed by a paragraph of text in a similar style. At the bottom, the text 'Heart & Crown' is prominently displayed in a bold, white font, with additional text below it in a smaller size. The overall composition emphasizes a sense of harmony, spirituality, and enlightenment.a spiral of film strips emerges from a bright, radiant light at the center, displaying various images of people and landscapes. The background showcases a dynamic sky with clouds and a gradient of warm colors, enhancing the mystical atmosphere of the scene.

In Phoenix

[real lynx]

To mix the contents use the AI description in images of the two images

And I only added the part of the cinema images that you wanted to the prompt

To the other description

spiral of film strips emerges from a bright, radiant light at the center, displaying various images of people and landscapes. The background showcases a dynamic sky with clouds and a gradient of warm colors, enhancing the mystical atmosphere of the scene.

[vestal lake]

ah you can use 2 images ,i never thought of that

[real lynx]

No

[real lynx]

First I went for the description of the image you wanted

Then I went for the description of the second image

And take the description of this is the part that talked about the films that you needed in the first

Copy that description and paste it to the description of the first image

[vestal lake]

ok now i get it ,thanks i didn't think of that as well lol

[real lynx]

The descriptions of each image were obtained with this tool

Fan of mixing prompts and images :3

[vestal lake]

yeah i use that all the time but never thought of using 2 image description thanks this should step my game up a bit 😉

[real lynx]

@calm acorn

[calm acorn]

Where i can find it ?

[real lynx]

Hello, tell me a little more about what you are looking for.

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

And welcome

[real lynx]

Does something like this work for you?

I need you to explain a little more what you are looking for to help you.

A captivating illustration of a trendy streetwear design inspired by brutalist architecture. The main character is wearing a hoodie that features bold geometric shapes and stark lines, resembling the style of brutalist buildings. The background showcases a cityscape with imposing, blocky structures, while the character stands confidently in the center, embodying the fusion of urban fashion and architectural aesthetics.

[calm acorn]

No, but like this

[sacred yoke]

Hi Camila sorry to jump in like this. Just in case u got a few seconds and in case u missed it, I posted a question here:
#🙋│image-help message
You seem like the right person to ask 🙂

after u help Clover of course!

[real lynx]

Oki I'm going to try it myself

Dark-themed, poster-style image with a gothic and horror aesthetic. The layout features a central figure of a skeletal warrior with a detailed, menacing skull, wearing dark armor with red and purple lighting effects. The background is a textured, dark surface with subtle patterns. The top of the image prominently displays the word 'KING' in large, white serif font. Below the skeletal figure, there is a rectangular section with the word 'SLAYER' in a distressed, white font, accompanied by various smaller text elements and logos in white, including 'Post Human,' 'Survival Horror,' and several brand logos. The overall color scheme is dominated by dark tones with striking red and purple highlights.

[sacred yoke]

I should mention that I am using Anime (General) for this since the whole Flipbook is Japanese 90s Anime Style

[real lynx]

👍

Question: I wanted to know if you have premium tools

[sacred yoke]

No sadly I am doing everything on the free plan

[real lynx]

Ok I'll have a coffee and try again :3

Tends to go up the hill :3

[sacred yoke]

yeah i noticed that too i dunno why hahha

[sacred yoke]

and no sideview either

[real lynx]

But with premium tools I was able to force it to go down

[sacred yoke]

how did u achieve it?

some of these are pretty damn close, all i would need would be city area instead of a mountain and a man with a gray hoodie on running for his life!

😄

for reference this is what i'm making:
https://streamable.com/krvdpw

[real lynx]

Use these two premium tools

One to imitate my features and another to imitate the movement

[sacred yoke]

ah right i saw those. was bummed they were premium 😄

[real lynx]

From this image

[manic root]

Need a helping hand to sharpen your skills or conquer tricky projects? 🎯 Our mentori are on standby to guide you! For personalized help, just pop over to #👋│accademia and fire away with /mentoring. 🌟

[real lynx]

But we are going to try to do it with free tools

First a coffee and I'll be back

My character was very tired xD

[sacred yoke]

when its time for a selfie, its time for a selfie

[sacred yoke]

all of this was so far made with pictures i generated from Leonardo and then animated in Kaiber

The song will be released next thursday so hopefully i will manage to make Leonardo obey me by then 😄

[real lynx]

You can use my premium tools at your disposal if you want

[sacred yoke]

Aww thanks a lot! I might take up on that offer if I really get stuck at some point.
I'll try to learn how to use it better in the meantime.

[real lynx]

That seems perfect to me

[calm acorn]

Is this need upload image sample ?

[real lynx]

If it would help me a lot

[real lynx]

Vibrant, animated style image featuring a young woman running on a trail through lush, green hills under a bright blue sky with scattered white clouds. The subject has light skin and short, wavy blonde hair that flows as she runs. She is wearing a white t-shirt, red and black shorts, and red running shoes. The background showcases a picturesque landscape with rolling hills, a small village nestled in the valley, and distant mountains. The overall scene is bright and dynamic, capturing the energy and movement of the runner against the serene, natural backdrop.

Free tools

I think they are all going down :3

You wanted a man

[sacred yoke]

What is this sorcery? How are they all downhill AND sideview?
U didn't even mention these in the prompt.
Can u use this magic to create a man wearing a gray hoodie and instead of a mountain its an urban env (a very steep street in the city) (daytime)

And he is not jogging. he is running for his life (scared)

😄

[real lynx]

I'll try

Do you have an image to keep in mind

[sacred yoke]

its basically a blend of these two here

the man and the city would be the first one

but the camera and angle would be the second one

so what I mean is I would take the character from the first picture and the vibe of the city from the first one
And instead of the camera showing his back and instead of him going up some stairs, the camera would be on the side and the guy would be running down the stairs/sidewalk

also please use 16:9 in case u get it right

i would just use yours straight away 😄

thanks so much for trying I know im asking a lot. if u don't get it its totally ok!

[real lynx]

I haven't tried anything yet But I'm going to try now I'm traveling to work

[real lynx]

I didn't ask you what kind of clothes he wears

I think you wanted it to be seen from the side

[real lynx]

I had to go down not up :3

[wanton belfry]

derp~

[real lynx]

He wants to go down some stairs, your view is perfect :3

[wanton belfry]

no stairs in this one but soon

with stairs its hard to get sideview...it keeps going behind...gotta try different method

here it is

hmmm the BG looks atrocious in that

much better

[elder bear]

Hey guys!
I am using in my project:
Style Ref only, strength is Medium,
Illustrative Albedo,

And using all these negative keywords:
deformed features, deformed face, deformed body, deformed limbs, deformed hands, deformed arms, deformed legs, deformed feet, extra limbs, extra head, extra body parts, extra fingers, extra toes, gross, disgusting, unnatural look, crossed eyes, two heads, weird colors, morbid themes, fused fingers, fused toes, fused body, missing limb, collapsed structures, collapsed body, anthropomorphic creature, text, Deformed eyes, Asymmetrical eyes, Crossed eyes, Sunken eyes, Cloudy eyes, Misshapen eyes, Undersized eyes, Misaligned eyes, Bleeding eyes, Eyes with extra pupils, Eyes with multiple irises, Hollow eyes, Misshapen hands, Asymmetrical hands, Swollen hands, Shrunken hands, Gnarled hands, Twisted hands, Malformed fingers, Missing fingers, Webbed fingers, Bent fingers, Crooked fingers, Clubbed fingers, Atrophied hands,

BUT, it does not work :(, I got really bad illustrations in some parts specially eyes, hands and fingers.

Any help plz

[real lynx]

Hello, that is your negative prompt.

Tell me what you would like to create

To help you

[elder bear]

Thanks so much,
The prompt :

A cheerful family of four enjoying a meal together at a wooden dining table. The dad, wearing a blue shirt, is enjoying the meal and engaged in the moment. The mom, in a pink shirt and a long purple skirt, is also enjoying the meal and engaged in the moment. The son, in an orange shirt, is enjoying the meal and engaged in the moment. The daughter, in a blue shirt with a cute design, is enjoying the meal and engaged in the moment. The wooden table is set with plates of food, including salads, bread, and drinks (juice). The scene captures the joy of family time during meals in a bright, welcoming setting. The scene should be bright, colorful, and engaging, suitable for children aged 4 to 8 years old.

[storm jungle]

Style reference is mainly used to grab the same colours etc from another image, if that is what you're going for.

[elder bear]

Yes, but the style ref is perfect

[real lynx]

Where is the family eating, is it a restaurant or at home?

[elder bear]

home

[real lynx]

Do you want them of any particular ethnicity?

[elder bear]

No, general.

[real lynx]

A breathtakingly detailed illustration of an affectionate Cue family, warmly aglow in a serene, natural-lit dining area, where a loving couple, with gentle smiles, shares a delectable meal with their two young children, a bright-eyed boy with a mop of messy brown hair and a rosy-cheeked girl with a big bow in her curly blonde locks, both beaming with excitement. The family gathers around a beautifully set, richly polished wooden dining table, adorned with vibrant, hand-painted ceramic plates piled high with a sumptuous feast, and candles that cast a soft, golden glow. The walls of their tastefully decorated home, filled with plush furnishings and lush greenery, are painted a soothing creamy white, and the air is filled with the aroma of freshly baked bread and roasting vegetables. Every detail, from the delicate lace tablecloth to the intricately carved wooden chairs, exudes warmth, comfort, and happiness, transporting the viewer to a sense of peace and tranquility, all rendered in hyper-detailed, ultra-high resolution, with exquisite textures and subtle, nuanced color gradations.

Phoenix ilustración

[elder bear]

Thanks for your time, but when use Phoenix, I can not use Style Ref.

I use Albedo in every prompt, to be sure I got the same style in all images

[real lynx]

I understand I didn't like how it looked in albedo that's why I told you to use Phoenix

[sacred yoke]

This one is really nice as well! The character wears a gray hoodie. I would definitely use it if he did. What prompt did u use on your pics. You managed to get side view on some.

[sacred yoke]

Also this one looks very good despite not being side view.

[night vault]

what is strength for style ref, max will likely distort the output

Also I generally not recommend putting negative that long given that I have not seen any significant improvement from new models

[fierce flame]

Amy chance of the Texture generator getting a axis reference so we can specify top bottom sides etc and training the model on that besides that it is amazing

[elder bear]

Mid

[thin tinsel]

put it as a suggestion. I don't think anyone will pick it up from this channel.
type a / then the word suggestion.

[thin tinsel]

My Advice is that you

1. use Albedo with your style reference on whatever strength (makes no difference to the quality of your image).
2. Use Cami's Prompt
3. Use Cami's image set as Content Ref on low.
4. Get rid of ALL your negative prompt words.
   Manage your expectations in appreciating that doing an image with multiple people is a known issue for AI, and you should not expect perfection.

[fierce flame]

Thanks mate just did that

[thin tinsel]

That's great. They are pretty good at looking at the suggestions! (and even implimenting them) 🙂

[fierce flame]

Yeah hope they do for 3d work I use stereokit as my augmented reality engine and when using claude for code assistance as its a code only engine I used a 3d axis indicator on objects whe n I sent in images improved its accuracy alot

[elder bear]

I removed all negative prompt words
Still getting ugly, strange fingers and eyes.
I spent a lot of token to fix.

[thin tinsel]

You followed all four steps of my advice?

[thin tinsel]

Really interesting to hear about the different ways it's being used.

[elder bear]

Sorry, I did not notice Content Ref
It worked better now

So, providing a photo as Content will fix these un wanted or error parts from happening.

[manic root]

Eager to enhance your expertise or take on bigger challenges? Get personalized help from our mentors whenever you need it! Just pop over to #👋│accademia and use the /mentoring command. 🌟

[thin tinsel]

Yes providing an image for content reference will help the AI a lot.
What I often do is create an image with Phoenix because Phoenix is so responsive, then use that as my content reference.
And I really like using style ref too. Albedo is my favourite, to work with image reference images.

Here is mine using @real lynx 's prompt and image + my own style ref image on mid.

@elder bear We run a mentoring programme, if you would like to work with a mentor on any of your projects, just let us know.

[elder bear]

Is it paid or free?

[thin tinsel]

It's completely free.
The way it works is that you will be paired up with a few of us from the mentoring team.
We put you and your mentors in a channel on this discord and then whenever you have any questions you can ask them.
So it's a bit like this image-help channel, but more personal to help you with whatever you need. 🙂

[elder bear]

Please

[thin tinsel]

That's great! 😄
/mentoring

[elder bear]

write this?
/mentoring

[thin tinsel]

Yes. Type that in and then you will have a form to put some information

You don't need to write a lot. You can say if there is anything specific you'd like help with, or you can just say that you're new and would like to learn more.

[elder bear]

Thanks so much Mr. @thin tinsel

[twilit fiber]

Hi again. I have been trying various methods based on your feedback, and I have to admit, it goes nowhere. When there are bits of matter/texture, the ai wants to do something with it, refusing to fill with whatever else. In this exemple, I tried expanding, using a Mask, Erasing... using words, negative prompt, I cannot get rid of the flow of hair... I know how to get better framing in general, but sometimes, the perfect image isn't perfect on all details !

[solemn shale]

What model and settings should I follow to have the artstyle of this image?

like, I know I'm not going to have the same artstyle. But a similar one could be nice

[storm jungle]

Is this what your trying to do

[twilit fiber]

Yes, more like it !

[storm jungle]

This is the area I masked, dont go too far wider than you need...... I used in the positive prompt only "Remove object" it took 3 goes, and it cleared

I use "Remove Object" alot, no matter what it is

[twilit fiber]

🧐

[storm jungle]

There will be times it may tske a few extra regenerations, but it will get there

[twilit fiber]

How do you adjust the Guidance Scale ?

[thin tinsel]

You can watch my video here, and try the technique I use.
https://youtu.be/adiCB8Esm6U?si=T6hYxAqRd7VKJ07d&t=752

[storm jungle]

Click on the guidance icon, and some options will appear

[thin tinsel]

@twilit fiber if you would like 1:1 support and training 🚀 Our mentori squad is all set to back you up! Just pop over to ⁠👋│accademia and summon assistance with /mentoring. Let’s level up together! 🌟 It's free!

[twilit fiber]

I don't have tokens anymore, so I will have to postpone my attempts. Thank you, I will first try the Remove Object method, and then your Erase one from the video.

Thank you.

[storm jungle]

All the best, we will be here later when you're ready

Remember you don't need to over mask your objects.

[twilit fiber]

What is the difference between Accademia and here ? ^^'

[twilit fiber]

To be fair, I have tried many sizes of overlay !

[storm jungle]

I always mask just slightly over the edge, it's not always a must, but I find it works better to mask as less as possible..... if your maskiing objects that have shadows, you need to mask them too.

[solemn shale]

I guess I have to train my model at the moment

[night vault]

if you know artist name, that may help with older model like dreamshaper 5

[thin tinsel]

Just use the 'Describe with AI' option with any model in the new UI.

[thin tinsel]

If you would like to take up the offer to receive 1:1 mentoring then just type /mentoring and complete the form.
I don't really know about the accademia channel.

[surreal tusk]

I need help with expanding my images

[dim zodiac]

Heya! We have a few guides on this

https://discord.com/channels/989166677390426132/1101117155401015398

There's a guide on the basics behind it... And I'll post another that shows the updated Canvas version, as that guide was Canvas 1.0

https://discord.com/channels/989166677390426132/1126679661741609120

@surreal tusk if you want more interactive help we can open a mentoring thread for you

[subtle gulch]

In #👋│accademia you can request a private thread from there with a mentore so your discussion will not be interrupted

[twilit fiber]

Ohhkay, that's cool thanks

[thin tinsel]

This is the discord server for Leonardo.ai.
Please do not promote third-party applications unless it is needed where Leonardo meets it's limits.

Leonado's CANVAS EDITOR is for expanding your images, and is actually very good at it!

[dim zodiac]

Thanks Willow

@pure geyser she's correct, which is why I had linked several guides on how to do it in Leo. I'll be deleting these links

[thin tinsel]

Thanks @dim zodiac 🎉

I love Canvas Editor it's one of my favourite things in existence.
The other well-known tool, I can take or leave.

[surreal tusk]

he sent me a recommendation and i just replied back 😭

[thin tinsel]

Yes, I see what happened.
Did you view the links to the guides you were sent earlier.

Let me know if you still need any help

[swift forum]

any1 got a idea how to make a cartoon profile pic like that

[copper warren]

How closely do you want to mirror this style? What aspects are you trying to capture?

[night vault]

for outline that will need “die-cut sticker”

[swift forum]

mostly the cartoonish style so outline big eyes wide smile expressiveness mostly

[swift forum]

in the prompt?

[night vault]

hmm, I need to test my own prompt, something like this on animeXL or albedoXL

die-cut sticker, cartoon, vector art, a man with brown fluffy hair in yellow shirt, big circular eyes, pink solid background

[livid inlet]

I used describe with A.I to get the basic look then tweaked it a little to get the white outline. All you'd need to do is change the details of the person to suit

[swift forum]

damn thats sich

thanks man

[real lynx]

@rich turtle

Hi

Here you can post images

[rich turtle]

hi sorry not well versed in discord

[real lynx]

Send me your idea

And we begin

There is probably something similar to what you are looking for on Google.

And then we adopt it to your idea and I explain how it is created.

[manic root]

Hey there! 🌟 Wanting to level up your abilities, conquer big projects, or hit some learning milestones? Our mentori stand ready to help out. Just swing by #👋│accademia and type /mentoring to get tailored support whenever you need it.

[rich turtle]

ok, I'm trying to create images of cartooned hot rods similar to this but keep getting this.

And this is one of the better ones. most are pretty disfigured and rough

[real lynx]

Ahh

I had thought of something else...

A whimsical cartoon scene of an assembly line, where a team of animated workers are diligently creating vibrant and sleek hot rods. The vehicles are a mix of classic and modern designs, with bright, bold colors and elaborate custom paint jobs. The background shows a bustling cityscape, with neon signs and a retro-futuristic vibe. The overall atmosphere is lively and energetic, filled with creativity and excitement.

[rich turtle]

Thats a pretty good prompt but I'm just looking for something clean, crisp and not cluttered.

[real lynx]

I thought about this style

I'll try again now with the style of your image

[rich turtle]

those are pretty good. Just a bit too much going on. More minimalistic is more of what i'm looking for

ok thanks.

[real lynx]

Looking for something like a minimalist logo?

Or do you want the car?

Stylized, cartoonish illustration of a classic blue car with exaggerated features. The car has a boxy shape with a prominent front grille and large, round headlights that are brightly illuminated. The body of the car is painted in a vibrant blue color with black racing stripes running along the hood. The wheels are oversized with shiny, chrome rims, and the car appears to be lowered, giving it a sporty stance. The license plate reads 'T_Bucket_Tees' in black text on a white background. The overall style of the illustration is bold and dynamic, with clean lines and a polished finish.

[rich turtle]

Just the car. and more, for lack of a better term, funky looking,

[real lynx]

Remember to give me all the possible instructions, I don't know anything about cars :3

And your opinion is good, what is missing

[rich turtle]

ok, let's see. I'm looking for a more lowered look with oval wheels and tires not rounded, a slight leaning forward to give the impression of speeding down a road, and a squattier look to the body of the vehicle. If that makes sense?

[real lynx]

I'm going for the new idea

[rich turtle]

Nice, a little bit too European styling for what I'm after. I'm looking more 1970 and older American muscle car and hot rod.

I took your above prompt and changed a few things and this is what I got. Not what I was looking for but pretty good.

[real lynx]

😅

[rich turtle]

😀 Just a taaaaaad bit too Disney

[real lynx]

These are premium tools that imitate the positions of the car

That's why the last one looks so similar to the first image.

[rich turtle]

The first one was done in Leonardo. The second on was done in a program beginning with the letter I.

[real lynx]

I have to go to a course to study, I have to start preparing to leave.

I work at night today is going to be a long day

But if I have the strength at night I will try to create more images for you

[rich turtle]

The last one is pretty close to what I'm looking for, but it needs to be les realistic and more cartoonish.

Really appreciate the help. You have a great rest of your day.

[real lynx]

Thanks 😅I like to help

I'm going to ask on the Spanish channel if there are any car fans

To help me with the concepts and thus put together a nice prompt

I feel that since I don't know about cars I can't generalize the idea.

With premium tools I can imitate and come up with the solution to make it as you say realistic and csrtoo drawing style

[meager rover]

I added sparklecore at 30

It's the only time I get a copletly different image using sparklecore

[rich turtle]

👍

[storm jungle]

My 2 most fav cars in the world nice, GTHO Phrase 3, and an almighty Mustang

[real lynx]

Hello good evening

I think you know a lot more about cars than I do :3

And that you could help him better

Far better

[storm jungle]

I been workin on it actually, because I like this style too, and even thought about this not long ago.... I am trying to exaggerate the wheels but it's not being nice : P

I might need some more time with this one

I just need to get it lower and squashier <-- is that a word : P

[real lynx]

This one came out with big wheels xD

[storm jungle]

I like the first one i did, I just need to get it lower, and the wheels need to be squashed and being the focal point, and possibly a hint of a chibi look on the whole image

[real lynx]

I copied the first image but he still wants it more cartoon style

[storm jungle]

Yeah I know the axact look, this style is quite popular in Australian car magazines, I am not sure where he is from, but the car with the 351 is Australian, Ford Falcon GTHO Phrase 3, it may be available in the US in a similar style possibly different lights but also known as a Falcon.

[storm jungle]

Gonna need more time with this one, unless someone else has an idea that I can't think of.

Although a popular style, but it's not an everyday one, not in the eyes of Ai anyways.

[wise canyon]

I love the dynamics you created in these 3, very cool images! 💙

[storm jungle]

Thanks, i'm trying to get the front wheel squashed, to give it an exaggerated aggressive look

[wise canyon]

Yeah, I wonder if "caricature of car" would help??
Just a thought 🤔

It might try to add a face though

[manic root]

Looking to level up your skills or need help with a tricky project? 🛠️ Our mentors can guide you! For personalized help, just go to #👋│accademia and enter /mentoring. Let's grow together! 🌱

[storm jungle]

Add a note with a slight chibi and caricature look

[wise canyon]

That looks cool!

[storm jungle]

it's not every car though, some are like my earlier ones, but if I can get one or two then we are on the right track

It needs to squash the front wheels, like its made of rubber, it likes making them perfectly round : P

[hollow breach]

Guys, how do you deal with two different characters. for instance, if the prompt states a short blond hair man and a long dark hair woman. how do you prevent to have two blonds, 2 women, 2 long hair or other combination . Phoenix can probably deal with it but is there a way to limit the "leaking" from one character to the other?

[storm jungle]

You can try naming the characters, a girl named Joan, and a guy name Fred, then reference your description to those names, Fred is wearing a hat, Joan has a purple jacket, whatever you call them by name or alien, try to reference it to exactly that

Or.. The person on the left has blah blah, and the person on the right has blah blah.

[hollow breach]

I can try that technique.

[thin tinsel]

You can use the names of actual lesser-known celebrities, so it doesn't actually look like them but would still carry their main features such as hair colour.
You can also find an image from the internet and use that as content reference or image to image on a low-weight.
Given all the techniques you can use, this is still rather down to luck and in a batch of four images, you might still have some that don't work out.

[summer iron]

how can I get baby yoda blend in the picture? 🤣 🙈

[thin tinsel]

Do you have a better quality image of Grogu?

[summer iron]

Yes

because I the pictures on canvas.. and ask the prompt too blend.. 😅

[thin tinsel]

Ok lovely. Are we doing this in Canvas Editor?

[summer iron]

Yes, I am trying too make it work

[thin tinsel]

Step 1, I would remove the background from baby-yoda first
Go into canvas, erase as much as you can around him
Then take the image back into Leo with img-2-img, on highest weight + transparency on. (Use original prompt if you have it).

Let me know when you've done that

[summer iron]

Yes we will do it 🙂

[thin tinsel]

@hollow breach

[hollow breach]

Thanks I will try this too

[thin tinsel]

@summer iron
You're welcome to use one of mine if you like. I used your Flower-girl image as style reference on low, so that that grogu would be the same colour tones and hues, and so the light-source would come from the same direction.
That way it would blend together more. 🫶

[summer iron]

Oh thank you I will use that 🙂

we are still figuring it out lol

[thin tinsel]

So if you use one of mine we are ready for the step 2.
You need to draw a shadow for Grogu on the grass.
You may find my video useful: https://youtu.be/_NEhsc3bcAA?si=kQyy3Lv_WAK2jP5X

[summer iron]

Im doing something wrong... 😅

[thin tinsel]

no you're not, it's fine.

[summer iron]

doesn't wanna blend.. 😅 .. should I putt the baby yoda beside the image?

[thin tinsel]

give a moment to try it myself

Ok, go to the sketch tool.
In the colour type in #E1E4E8

Then take the transparent slide down so the blob is under the green.

Then zoomed in as far as you can, with a small brush,
Very gently and carfully colour over the top of Grogu

Let me know when you're ready for the next step

[summer iron]

Let me tr this.. 😅

[thin tinsel]

@summer iron
Make sure you draw in one stroke and don't go over your own lines.
You can use the eraser with sketch only selected to tidy up around the edge. but a little outer glow will give him a nice ethereal look.

[summer iron]

Yes thank you 🙂

[summer iron]

oei... looks really dark now... 😅