👋│accademia

[hearty torrent]

Hello there! Is there a recording from yesterday?

[ruby wraith]

We hope to have something soon 👌🙏

[uncut seal]

Prompt Share: a versatile little Flow State Prompt, courtesy of @alpine hemlock , and edited a bit by me.
pixel art scene, Solarpunk style, glitch effects, ASCII particle trails, with high detail and super resolution, creative lighting, vivid volumetric colours, artistic lighting, and abstract geometric patterns

[drifting moon]

which is the best model combination for good trnsparent background generation?

[drifting moon]

preferebly that can handle text

[tough mist]

text you need phoenix or flux. though not sure about transparent capability

[subtle shell]

Hi friends. Is it possible to use Leo for product photography and videography?

I am working a project and tried many AI tools that are specific for product photography and videography but they SUCK.

[formal mountain]

Hi! For sure! I'd recommend using the /mentoring command to get help directly from our volunteer mentoring team for a project like this

[subtle shell]

Thank you very much for guidance. I'll run that command.

[next wolf]

I need a mentor, earlier a mentor was assigned but now unable to message in that group. Not sure why. Can i get a new mentor please?

[uncut seal]

what happens when you use the /mentoring command?

[sudden pivot]

If anyone needs help with Veo prompting, I've been in the Veo beta since last year.

[uneven pollen]

how do you add audio to the prompt?

[sudden pivot]

Be explicit. With vocals, explicitly say what you want characters to say, and how. It's okay with some tone stuff, but can't do accents (I think it doesn't do most other languages either, but I don't do that myself). Same with background noise. Just explicitly describe the sounds, in plain English.

[uneven pollen]

thanks! i got it going now

much appreciated

[sudden pivot]

Happy to help.

[void grail]

Yo guys, how can I avoid plastic skin in Flux Kontext ? When I want to generate human ?

[plush bison]

/missions

Why It is not working?

[uncut seal]

its currently disabled while the system is being worked on

[plush bison]

oh, thank you

[ornate spear]

/missions

[valid adder]

I am momoh sandi

[uncut seal]

two quite cool images that came out of this flowstate session: 👇

[ruby wraith]

That's an understatement!!! 2 very cool images came out!

[nimble leaf]

https://tenor.com/view/boy-math-school-asia-study-gif-18879119

[icy wolf]

so i'm a poor free person, can someone help me. I only get 150 points per day. I am trying to make

• very dark skinned african american 19 year old man
• handsome, square jawline, smooth skin,
• 4 poses: forward, back, right, left
• wearing only black tight shorts
• hyper realistic
• for unreal engine game development character (not cartoony, if you know what I mean)
• natural muscle build (not bodybuilder) thin
• hair can be anything (black color though)

I can't seem to get this. I get front and back, and diagonal, or just a face

[uncut seal]

Do you need all four perspectives in the same one image?

What prompts and models have you tried?

[icy wolf]

well let me look up the specifics. I tried to just have all 4 images in one pic, but let me double check what unreal engine wants

[uncut seal]

I cant help right now. On my way to work. But can give it a crack later today perhaps. Assuming nobody else has by then.

If you wanna let us know what youve tried so far, prompts and model, that would be helpful.

[icy wolf]

PROMPT: make a model sheet of 4 poses front back and each side view standing in A-pose of a hyper realistic detailed photograph of a 19-year-old very handsome African-American male model with very dark skin, set against a blank background, showcasing the model from head to toe wearing bare feet black tight shorts no shirt standing in A-pose, hyper realistic photography, trending on unreal engine, for hyper realistic unreal engine character creation, perfectly ambient illuminated,

and i went to the modern ai and tried phoenix and lux different versions, then back to the older ai and tried almost all of those.

oh that's right, it wouldn't let me do it

for a style sheet you need to have just the character in underwear in an A pose or T pose, and the anti-corn setting doesn't distinguish that. I guess game devs don't use this site

[cyan thorn]

Hi there i am trying to blend 2 people to make a unique person how can i achive this in leonardo ai any help

[weary isle]

Using the Nano Banana model, you could upload reference images of each character, then prompt it to create a new character from them.

[cyan thorn]

Ahh I see

Thank you so much

I was literally lost lol

[vapid sable]

Hey guys what are good resources to learn how to prompt better?

[uncut seal]

Well, until the new 3rd party models came along and changed everything, this is how I prompted everything #🙋│image-help message

It works fine on those models too, just when Image Ref gets involved you'll wanna give it instructions rather than the 'shopping list' format prompt. 🤔

[ruby wraith]

https://discord.gg/leonardo-ai?event=1433876615435256020

[limber stone]

Prompt 10: The Final Complaint
Final wide shot, melancholic dusk lighting. The original local cat, a living memory of a forgotten covenant, sits alone on the sharp, cold edge of a cement curb. It looks silently down the modern, sterile street, its posture conveying a deep, silent grievance. ––ar 16:9 ––style raw

[steep patio]

Good day, night, morning, evening, or whatever it is for you. Please help me create a full-length image of a girl. Nothing erotic, nothing 18+, just a girl IN CLOTHES! Here is my prompt: “A full-body photograph of a shy Slavic woman with delicate features, pale skin, and long blonde hair, dressed in simple everyday clothes, modest and attractive.” I will sincerely pray for whoever helps me.

[weary isle]

Hi there, it's often difficult to get a full body shot when using a 16:9 aspect ratio,
the trick is to specify some type of footwear, in order to force the AI to generate a full body image.
Your prompt modified:
A full-body photograph of a shy Slavic woman. She has a modest and attractive appearance, with delicate features, pale skin, and long blonde hair, dressed in simple everyday clothes and wearing brown leather lace-up boots.

@steep patio You can take it further, by using that prompt and loading your first close-up image as a character reference:

[steep patio]

You're my hero! (tricks) Thank you so much! I'll keep working! 🙂

[weary isle]

You're welcome, Keep at it 😉

[uncut seal]

Another tip would be to specify what the char.'s standing on. Even if its pointless like 'standing on floor'.

But yes, aspect ratio also has its preferences, so full body would be easier in say portrait mode. 👍

[steep patio]

Hello everyone, I need help making my hand look normal. Here's my prompt: “hand touches waist, fix right arm, natural shape, no distortions.”

I tried different promts, but my hand still remains crooked. Here is the original photo. prompt:A photorealistic portrait of a Russian woman aged 20-25 with light hair,
standing directly facing the camera (face straight ahead), full-length in the frame. She is holding her phone with the camera turned on neatly
holding a phone with the camera turned on in one hand, standing in a relaxed, comfortable pose,
with a playful, slightly smiling expression on her face. She has a proportional, slender figure (neither thin nor full)
emphasized by a summer dress and white lace-up sneakers that accentuate her silhouette.
She is standing on Red Square in Moscow on a warm, cloudy day with no one around. Her skin is rendered with soft, pale
shades, a natural professional photography style, high detail,
realistic skin texture, soft shadows, and subtle contour lighting.
Composition: subject in the center, camera at eye level, portrait framing 35-50,
shallow depth of field creating a refined intimate atmosphere reminiscent of
a high-quality professional portrait.

[uncut seal]

Basic advice: dont ask for things you dont want. The AI doesnt know what 'No distortions' means. And since you mention them, it may make some for you.

Similarly, just saying 'fix right arm' likely doesnt work either.

I think you like the second image, but just want her unoccupied hand on her waist, right? In that case, omni edit that image with a simple prompt "one hand holding phone, the other on waist/hip"

[pallid gorge]

@digital walrus hi Roberto any update when rank command and stuff will become active again or are we using other system now to increase rank and stuff? Thanks in advance

[digital walrus]

hello Stef! We will dedicate some dev time for that task this week or the next, I know we all miss that, sorry for the delay

[lavish plinth]

You convinced me.

[formal mountain]

it's always time for emotional support noodles

[uncut seal]

I made a Comic with Nano Banana Pro. I hope this makes some people chuckle.

and i also made a guide on how to create a comic of your own. Here ya go: https://discord.com/channels/989166677390426132/1444639481813798954

[dark sage]

Where can I find myself some demon babes like that?
Perferably ones that I won't have to share the pizza with.

[uncut seal]

Instagram? 🤷‍♂️

ooh wait. you mean page 12..

[sly carbon]

this is awesome, I love it 🙂

[icy wolf]

Help me: you get what I'mt trying to do? this prompt just won't deliver the right pic

young 19 year old african man with deep black obsidian Nilotic skin, high cheekbones square jawline hazel eyes, in a classic D&D cleric holy armor and white flowing robes with runic X holy symbols on them, deep hypnotizing hazel eyes oiled Babylonian beard and long Afar tribe style coiled hair, he is a high level cleric of the archbishopric, he has his arm extended outward casting mass heal over a huge triage tent full of wounded warriors with a battle still raging in the background, the warriors all lay on cots with bandages and partially clothed in their armor, lower level cleric tend to them too, a soft emissive golden glow flows over all the warriors and settles over their bodies coming from his outstretched hand, each cot with a warrior on it has soft golden glow around the warrior, closed eyes, he has a solemn calm look eyes closed on his face praying to his god for healing, high fantasy setting, style of Tithi Luadthong, 8k resolution photography, hyperrealistic photography, action pose, he wears rich glistening armor with embossed holy symbols on it, profile casting holy magical mass heal over wounded warriors on cots,

[sly carbon]

what results are you getting? how do they differ from what you are looking for?

[digital walrus]

lovely!

[icy wolf]

can someone help me? i just can't get the prompt to generate the pic i want.

a video game arcade, machines are lined up on either side of the isle carpet in the middle of the machines in between, but in back of the machines you spot rows of hamster wheels with hamsters furiously running inside the hamster wheels powering the machines, surreal dystopian atmosphere, arcade carpet, half of the machines are blinking because the power is sporadic

[icy wolf]

i'm sorry, i gave up. i do well in ai competitions, but sometimes it just doesn't give me anything useable

[dark sage]

Sometimes you just need to simplify your prompt a bit more so it's easier for the AI to understand.

[pallid gorge]

Any update on this Roberto?

[digital walrus]

not yet, we are still working on that thing, but this is a priority for us as well, as soon as we are able to we will relaunch it

[graceful badger]

i need help. do anyone know how to take a floor plan, switch the perspective, and make it "3D"? and it needs to be accurate.

[uncut seal]

do you have a floor plan to experiment with?

[graceful badger]

yes

[uncut seal]

i meant, can you post an example of one, so we can try and help you get your 3d thing from it..?

[graceful badger]

like this. and would be even better if the furniture, kitchen and stuff were accurate too..

[uncut seal]

okay.. but how would you see all of it in 3D with all the walls in there..? 🤔

[graceful badger]

oh, what you do is screenshot one of the rooms, then you ask it to recreate it (somehow)

say you want isometric, you get something like this..

or make the entryway like this

seedream and nano are probably the best models for this. at least they were the closest ones to what i were after..

[uncut seal]

"depict this exact room from a first person perspective. you're standing in the lower left corner of the room. detailed, realistic"

i imagine your floor plan is too complicated, bc that worked fine for me. sure its not perfect, but its pretty ok

[graceful badger]

which model is this?

[uncut seal]

banana Pro

[graceful badger]

ok. i try that. tx 😄

[graceful badger]

i added the grill 🥩

[graceful badger]

futuristic look 🛸

[uncut seal]

Found a neat little prompt and changed it up a bit

[void grail]

Hello guys, if I would like to create very realistic female who wears for example a turtleneck. Which model I need to use for the better output ? I would like to avoid plastic skin all of that. I have only the turtleneck in ref image

[uncut seal]

do you hav epaid membership or only free user tools?

[void grail]

Paid

The 30 dollars plan

[uncut seal]

for my understanding, do you have the clothing you want to apply as reference images?
or just looking for a model to accurately understand what you want?

[void grail]

Yes I have the clothing in flat lay that I put in reference

And I would like to generate a model who wears it. But every model looks like fake imo

Or I have a prompt issue ahah

[uncut seal]

so you have the clothing as reference, but you're letting the AI create the human itself, yes?

[void grail]

Exactly

[uncut seal]

well it is likely one of the more expensive options, but i suggest nano banana pro.
what prompt have you tried?

[void grail]

I generate the prompt through ChatGPT

But I think it’s not the good way

Like this

Ultra-high-resolution editorial fashion photo cropped from chin to mid-thigh,
showing only the torso of a sophisticated Italian man wearing a medium grey melange tailored coat.

The coat features a clean mandarin collar slightly open,
structured shoulders,
front buttons partially closed,
tailored silhouette emphasizing chest and waist.

The fabric displays a refined woven texture composed of light grey and mid-grey fibers,
resulting in a balanced, warm medium grey tone
(no black, no charcoal, no blue undertone).

The model’s arms are relaxed, one hand subtly adjusting the front of the coat,
the other resting near the pocket, natural and elegant posture.

Backdrop: softly blurred neutral-toned urban or studio background,
keeping full focus on the coat.

Lighting: smooth, diffused natural daylight from the front-left,
producing gentle highlights on the fabric texture
and subtle shadows along the seams without darkening the grey tone.

Camera: 70mm lens, f/4 for full garment clarity and depth;
slight organic film grain for a refined Vogue-style editorial finish.

No face, no accessories, no logos, no text.

[void grail]

Something like this

[uncut seal]

be with you in just a min. need to finish my food.. wlel, maybe 5min

[void grail]

Arf Im at the gym bro ahah. It will difficult I think

Can we open a chat in academia ?

[uncut seal]

sure can. use /mentoring
then ping me there with @ if you want

[void grail]

Awesome thanks !

[uncut seal]

in my opinion, that prompt is needlessly long.

here is what i did. Model: Nano Banana Pro > Dynamic

i used that random coat as reference image.
then prompt: "depict a woman with orange wavy hair wearing this coat. she is striking a sporty pose. she is wearing zebra-print sweatpants and orange boots. blurry highstreet background"

some other small tips:

• if you're using the clothing as reference image, you dont need to describe it in detail again.
• dont ask the prompt for things you don't want to see. "no black, no charcoal,.." just focus on what you DO want to see

[tiny garnet]

What happened to the Nanobanana?

[weary isle]

That's weird, looks like the old UI
no FLUX.2, Ideogram, or Seedream either?
maybe try a different browser?

[forest reef]

GPT 1.5 vs Nano pro

[void grail]

Yo guys which IA you recommend to generate some realistic videos of for example Dolce Vita Italian vibe, boat etc ?

[sly carbon]

probably the most realistic one would be Nano Banana

[uncut seal]

Chapter Two is here:

and a reminder, if you want to make your own comic on leonardo, here is a guide: https://discord.com/channels/989166677390426132/1444639481813798954

Looking forward to reactions 😁 ✨

[digital walrus]

I truly love this

the consistency you achieve is really good!

[limpid forum]

I love this! I gotta know who owns Chibi Thom's soul!!

[tardy leaf]

I attempted to locate the little rabbit's presence in the first chapter. 🤣🤣
-# 🏷️ DemonDuel

－－－
⛩️ Chapter One：#👋│accademia message

[formal mountain]

this was basically how you could mutate in the mana game that got rebranded as ffiii on gameboy 😂

all my people were mutants 😂

[tough mist]

https://tenor.com/view/dungeon-meshi-delicious-in-dungeon-marcille-dungeon-meshi-marcille-marcille-dungeon-meshi-gif-4938728210725164459

[uncut seal]

🐥

[formal mountain]

awwwwwwwww

[uncut seal]

and reminder, we have a guide thread for comics 😉
https://discord.com/channels/989166677390426132/1444639481813798954

[nocturne sky]

Material Study: The Eye of God. 🏺

[formal mountain]

this is a very lovely image. What model in th eLeonardo toolkit would you suggest for users wishing to try this out?

[uncut seal]

@fathom cloud are you still lurking?

[dark sage]

So, I heard you like mudkips

[uncut seal]

hmm, not what i envisioned, but maybe someone wants to make this cooler. 🤔

[dark sage]

It's pretty cool just the way it is, I say

[dark sage]

I don't know if I'd say these are cooler than yours. They're certainly different

i took this one in a different direction.

[sick ocean]

Hi

[old jungle]

What is the purpose of this channel? Thanks.

[trim elbow]

I need help. I need something basic. I want the wolf to have the same expressions and pose (eyes, mouth, hand, eyerows and hat) of the second picture

In the style of the wolf pic. I don't know if you understand what i mean

[uncut seal]

all of the mentoring threads and a lot of other threads like games are linked to this channel. and the main room here is a bit for showing off and a bit for spreading things around, like cool prompts, I'd say.

[uncut seal]

sounds like something nano banana should be able to do. have you tried?

[trim elbow]

Yes but it didn't work at all that's why I'm here

[uncut seal]

wanna share the prompt you used? then i can try and adjust it to what I'd try myself

[weary owl]

xd

[trim elbow]

I got the same pic

The problem are the eyebrows are not arched

Like in the emoji

[trim elbow]

It's in Spanish it's okm

?

[weary owl]

If you like the result, but need to make changes, you can continue to modify it here

[trim elbow]

If you can make the other eyebrow to be bowed like the other one I think will get it!

[trim elbow]

What was the prompt bro? I liked it

[weary owl]

Prompt: use the image 1 and create a emoji like image 2, expression
Nano Banana Pro

[trim elbow]

Thanks bro

[weary owl]

you´r welcome buddy

[trim elbow]

hey guys need help again an easy request maybe but i struggle and nano banana pro sucks I just want the wolf style to have the expression of the Jerry one

something easy

(without the coffee tho)

I'm getting this shut in return. So now you understand why I'm requesting for help

[tough mist]

you can hint “awkward smile” from jerry that can help model pick up subtle meaning

also try seedream and gpt - they can be cost effective (maybe you need to roll a lot to get one with right style) for similar capability with nano

[trim elbow]

Gpt made a worst work

[tough mist]

there are certain styles some model is better than other. for flat 2d cartoon with clear border like the wolf, seedream or nano are better than gpt

[regal elm]

[regal elm]

complicada la cara y el gesto de ese raton

[trim elbow]

no se parece

[trim elbow]

creo que lo conseguí

[regal elm]

Pasa el prompt

[weary isle]

@wise bloom Hi there, please note: self promotion and job soliciting is not allowed in the server, thank you

[dark sage]

This gave me an idea...
I went back and got your isolated island prompt and modified it to put the island in a teacup.

[uncut seal]

are the labels accurate?

[dark sage]

I had to look it up to be sure...
Vancouver is actually right next to Portland.
But other than that, yeah. Pretty close.

my first atte3mpt was a bit messy

[uncut seal]

no use crying over spilt tea

[dark sage]

That isn't tea, it's my tears.

[formal mountain]

it was sensible of you to cry over those tea leaves.

[dark sage]

The tears steeped faster than expected, and the salt helped the tea taste less bitter!

[tired glen]

does anyone have a good model or any suggestions on how to create good 2d maps to use for a dungeon in D&D, i am going to need many of these and i cant get it to come out right

Even getting it to be completely top down seems difficult its always at an angle

[tired glen]

finally got some where using some image references and a lot of tries

https://cdn.leonardo.ai/users/4c09927f-a365-48fb-9e9d-7bde90775857/generations/1f125f46-6866-6e10-b6cb-d319bce79668/nano-banana-2_A_vast_ancient_Forgotten_Fortress_stronghold_with_a_unique_layout_featuring_15_d-0.jpg

[tired glen]

https://cdn.leonardo.ai/users/4c09927f-a365-48fb-9e9d-7bde90775857/generations/1f1260ea-919b-6d30-a26a-8a832afd1298/nano-banana-2_Image_is_a_detailed_top-down_style_dungeon_map_featuring_a_grid_layout_with_10-1-0.jpg

got there in the end

[formal mountain]

glad to hear it! was there a particular prompt that worked?

[tired glen]

yes, i used AI to describe the promt from an image and went from there

@formal mountain

[tired glen]

practice makes perfect, much time later

https://cdn.leonardo.ai/users/4c09927f-a365-48fb-9e9d-7bde90775857/generations/1f12f97d-74db-6fc0-8e28-49bb1089e91c/nano-banana-2_make_sure_the_crystal_room_connects_to_the_rest_of_the_hallways-0.jpg

https://cdn.leonardo.ai/users/4c09927f-a365-48fb-9e9d-7bde90775857/generations/1f12f984-f39d-6280-b4f5-deeb3aef6496/nano-banana-2_This_is_great_make_a_variation_on_this_with_different_terrain_and_different_biom-0.jpg

https://cdn.leonardo.ai/users/4c09927f-a365-48fb-9e9d-7bde90775857/generations/1f12f994-874e-6c40-954c-09a4b8dc844d/nano-banana-2_This_is_great_make_a_variation_on_this_with_different_terrain_and_different_biom-0.jpg

and its totally modular, loving it

variation on the 2nd one

https://cdn.leonardo.ai/users/4c09927f-a365-48fb-9e9d-7bde90775857/generations/1f12f99b-aae6-6890-b7a9-836eb3e09ddc/nano-banana-2_This_is_great_make_a_variation_on_this_with_different_terrain_and_different_biom-0.jpg

[formal mountain]

fantastic! any tricks you'd like to share?

[tired glen]

thats fake

jasonlitt got his account hacked

[formal mountain]

thanks for the heads up!

[tired glen]

Well, its easy to keep plowing away to get what you want.
You run out of tokens fast, especially with the 3rd party ones, they are extremely expensive.
What i have found is to get what you want its better to reverse engineer images you like into prompts
Then look at those prompts, see what they have in common and try to experiment with adjusting them.
It is still a exhaustive process and expensive (imo there should be a sort of experimentation setting on low quality for low cost)
But in the end you will get what you want.

Having a sort of very low quality option for each of these models that you can use, kind of like the flow state option.
So just to see what about the prompt may deliver, like a rough estimate for a low cost.
That would be very helpful

I spend 160 tokens a time now just to see if im going in the right direction, its very expensive

[formal mountain]

yikes yeah, it's true. gone are the days of doing 4 images per generation lol

[tired glen]

Oh yes those days are gone indeed 😄

good old legacy days

[formal mountain]

2 tokens an image 🤣

[tired glen]

https://tenor.com/vQ7Ro6KVzpL.gif

I suppose you could add another token set, something like Sketch tokens

so you can do it on like the lowest quality setting

so you can just RNG it, sometimes it just takes a lot of tries

[formal mountain]

ooh yeah that would be nice. worth suggesting

[tired glen]

consider this the suggestion then 😄

❤️ Pookie

[formal mountain]

passed along 🫡

could also do something like a different mode, kind of like realtime gen, but that works with the shinier models 🤔

[tired glen]

Yes i think that would work.
A sort of low cost option to get the general prompt right.
Then spending the tokens to refine it.
Would result in a lot better end result in the long term i think.
And more user friendly.

[tired glen]

@sly carbon

above is a scam

[sly carbon]

thank you dealing with it now 🙂

[tired glen]

@sly carbon

[dark sage]

Gone 🫡

[tired glen]

@dark sage another one

@uncut seal I see you are here

[uncut seal]

done. those appear like 20 times a day here (and a lot more are caught by the bot automatically). you needn't waste your time on each 😅
this channel isnt frequented toooo much luckily, and we're refining the bot to be able to hopefully deal with these more effectively too.
Thanks, but dont feel the need to point them out every time 🙂

[tired glen]

Gotcha i just check it every few hours

ill just leave them be then

[uncut seal]

yea, i usually check in every few hours too. if those 2-3 were the only ones you saw, then i'm doing pretty well on removing them 😂

[weary owl]

https://tenor.com/view/iceage-possum-peaceout-gif-2272177960667492692

[formal mountain]

you can feel free to let me know if you see these hanging around. dm or ping or whatever. only if you want, you don't need to feel like you have to, but your input is much appreciated 🙇‍♀️

[tired glen]

@formal mountain

[formal mountain]

heya, thank you!

[uncut seal]

Be sure to join the active #1496501436014526606
Win prizes and discover what one of Leonardo's newest tools is capable of

Here's an example. Spot all the easter eggs? 😬
Original 👈 / 👉 Upscale (Creative Strong)

[tough mist]

I fell like even strong is not enough

[uncut seal]

it doesnt display well confined to discord size. theres a lot of extra detail zoomed in

[tough mist]

I zoom it above and spot dog face

[cyan dust]

Your message could not be posted here. If you believe this was in error, please contact a moderator.