🏙│ai-images

[compact pivot]

he must really like dodge-ram trucks!

[finite bramble]

Here's an image I prompted using a photo of my Loona plushie all cosy in bed.

[steel haven]

Hey, what did you use this? Just curious. Looks great.

[languid orbit]

image

[coral mango]

this is a very beginner question
which Ai model should I use for image-to-image
like changing interior desing of a room

no masking just direct transfer

[steel haven]

Nobody here knows anything. Head to civitai server any maybe someone will be able to help you out here.

...Unless you're looking for furry shit.

then stay here.

[obtuse mauve]

gemini and imagen 3 can do that i believe

[zinc portal]

Chat, I just completed a challenge this week: "Upscale low-quality images and convert them to 4K." Here's an example (ignore the image, just look at the quality):

[unreal harbor]

this shit is so ass 🗣️ 🗣️ 🔥 🔥 🔥

[slate oak]

hanuman

[graceful lark]

@keen bramble

[keen bramble]

Cursed

[hollow hollow]

Oh no, it can spell, WHY CAN IT SPELL 😭

It's getting to good 😨

[barren sonnet]

what ai is this

[zinc portal]

After trying to further improve the image scaling to achieve finer detail, I achieved the following result

Original

Upscale 4k

[hollow hollow]

Chatgpt

[lapis coyote]

mars

[full jewel]

Guys whats an AI image generator thats free

[graceful lark]

#🏙│ai-images message

[obtuse mauve]

veo 2 can spell in video lol

[wraith merlin]

How much time it took to upscale an image using Ultimate upscaler

[zinc portal]

I have a custom workbench that takes between 6 and 12 minutes to upscale a 4K image.
The time depends on the image's aspect ratio and the Ultimate SD Upscale settings.
In addition to the power of your PC.

Ultimate SD Upscale itself does not scale images, you need to have certain models and nodes connected in order to scale an image to 4k with Coherence.

If someone tries to scale an image x using only the Ultimate SD Upscale node, it will create an image scaled x2 or x4, but it will move away from your image and also will not have the desired texture.

[tawny peak]

chito & yuuri from shoujo shuumatsu ryokou

[steel haven]

you're just so helpful

[wraith merlin]

For me it took arroud 45 minutes I upscaled images 2 time on 4k resulations

[green hamlet]

Kendrick Lamar balloon.

[zinc portal]

What PC Components Do You Have?

[wraith merlin]

4060 8gb and 32gb ddr 5

[zinc portal]

It's strange, I use an RTX 3070 ti with 8 VRAM and it works super well, it does hang for 1 minute when starting the program, but it goes very quickly afterwards

[wraith merlin]

Maybe my settings were too high idk

[graceful lark]

@keen bramble Shut lol

[rustic bobcat]

มะ

[humble wasp]

/imagine prompt: beautiful chibi girl sitting in a flower field, pastel colors, highly detailed --style anime --quality 2 --ar 9:16

[green hamlet]

Please speak English, brother. This is not where you say something in Thai.

[hard fog]

bro left

[green hamlet]

So this bro can't even speak English then lol.

[brittle ocean]

😈

[storm crater]

Hi

[plucky owl]

hi

[graceful lark]

lol zabka

[vast rampart]

what

[round saffron]

r u sure

[graceful lark]

@keen bramble

[keen bramble]

boink

[green hamlet]

Marin Kitagawa in real life.

[plucky atlas]

Big, friendly mans.

Less manly friend.

[oblique citrus]

Xo

[onyx tundra]

cow

[sour nymph]

diference is clear and real

[subtle quarry]

Guess it’s jailbroken

[subtle quarry]

Here ya go ha

[livid sapphire]

Which tool are you using for upscaling?

[plucky owl]

interesting

[livid sapphire]

Great, thanks

[graceful lark]

@uncut slate muhehehehe

[uncut slate]

thats cute owo

[graceful lark]

YUS 100%

[harsh moss]

Pika-pika?

Yes, now this is kind of my style

[stoic parcel]

Hi

[graceful lark]

Have you ever seen LoRA with 1GB dataset and more than 1000+ images?

[spiral shard]

School body building

School

[graceful lark]

@spiral shard chat not for gen img

[alpine plover]

https://tenor.com/view/no-rat-nuh-uh-nope-no-no-gif-11699607390227676176

[full salmon]

hi

[plucky owl]

hi

[wraith merlin]

This image is generated by which model?

[shadow bane]

I see no reason why it wouldn't work. Also, you can extract loras from finetunes.

[wraith merlin]

You mean "flux dev fp16" ?

[clever tapir]

make an image prabowo eating fancy food in front of poor people

[green hamlet]

To generate an AI image, go to #🤖│bots and use Weights bot there.

[placid drift]

||<@&1159293140440723499>||

nomi112266

⚠️ NSFW Content Detected! ⚠️

This image has been automatically detected as NSFW content and removed.

🔍 Image Analysis Results

||
Hentai: 82.02%
Drawing: 17.82%
Porn: 0.15%
Sexy: 0.00%
Neutral: 0.00%
||

👤 Message Author

@orchid mesa
nomi112266 (929370859876012083)

💬 Message Content

my first ever ai image

Image was automatically removed · Apr 22, 2025, 7:28 AM

[cold lintel]

that was not nsfw

[distant pelican]

lmfao

[wraith merlin]

Can you please give me the workflow

[brittle ocean]

||
Hentai: 82.02%
Drawing: 17.82%
Porn: 0.15%
Sexy: 0.00%
Neutral: 0.00%
||

[cold lintel]

???

[arctic canopy]

Yo can anyone help me “i need a picture with a girl holding a paper and on the papet it needs to have this “monkeOPasf”

[plucky owl]

so make it using Ai image genaration model

[arctic canopy]

Idk wich one is goof

Good*

[brittle ocean]

Read I found it nsfw image

[cold lintel]

how is the message you replied to nsfw?

[brittle ocean]

Yes

[cold lintel]

you copied what the bot sent to another message

[brittle ocean]

I hope you're the funny one at parties 😭

[brave silo]

@brittle ocean help

[wraith merlin]

Flux dev 1 model

But for that you require a powerful GPU or you can use cloud services to use flux

[languid sinew]

probably the only best ai bodycam generate

[brittle ocean]

No

[graceful lark]

https://cdn.discordapp.com/attachments/1131350125223088284/1325282285125374054/ezgif-7-1ea5690a9a.gif

Good boi! ^^

[neat rover]

pika

[wraith merlin]

What kind of help do you need

Small Details such as ?

I'm not sure I can find it, but I'll try

[indigo holly]

Why is this channel just filled with little girls... :/

[green hamlet]

Alice Margatroid dresses as Tendou Arisu.

[indigo holly]

boy kisser

[graceful lark]

yus sir

[hard fog]

fr the last comment in 4chan

[jagged willow]

Yagami

Light yagami

[plucky owl]

kira

[jagged willow]

Kira

[dim vortex]

Is this ai or real drawing?

[green hamlet]

Look up the file name.

[green hamlet]

Don't spam.

[prisma ruin]

Cute

[hushed gulch]

Големы под 140 чистого выжигают каждый.

представь ебало морфа

[modest rain]

#🏙│ai-images hi

[plucky owl]

w tf

[gritty glen]

h

[hushed gulch]

брррратан

[jolly kettle]

um the server is English only

[graceful lark]

#🌏│русский, #🌍│español , #🌏│한국어

[harsh moss]

In pencil sketch form

[tawny peak]

Very cool

[graceful lark]

https://tenor.com/view/protogen-vrchat-hello-hi-jumping-gif-18406743932972249866

[quaint tiger]

XD

[graceful lark]

https://cdn.discordapp.com/attachments/1287170529760903251/1337623331926704200/FURRYCLUB.gif

[graceful lark]

Pony!

[wheat radish]

this server is English only, and this channel is only for ai images, not screenshot of game things

@hushed gulch

[solar raven]

Why i dont have image perms

@uncut slate could i have image perms pls

[uncut slate]

you can but you need to chat more

[solar raven]

Chat with who bruh the chats are dead

[green hamlet]

To get an image permission here, make sure you talk a bit in chat until your name turns light green.

[solar raven]

Aight

[uncut slate]

you can chat with chatgpt here https://discord.com/channels/1159260121998827560/1202754985255764060

[solar raven]

No thanks

Anyways thanks for info

[uncut slate]

no problem

[graceful lark]

To get permission you must have level 5 to get level 5 you need to talk here

https://tenor.com/view/level-5-pic-perms-sea-of-umjee-umjee-gif-15630646039760648385

[plucky owl]

nope

[green hamlet]

Wild.

[plucky owl]

where is that bot

[graceful lark]

#🤖│bots?

[harsh moss]

😯

😡 🤣

😭

[graceful lark]

ooh @harsh moss!
I think this is violating the Server rules! Can you delete these photos?

start

[graceful lark]

but here he look so cute!

[harsh moss]

yeah

[graceful lark]

hes face just so silly xdd

like me xdd

[harsh moss]

Upscaled to 4K

[graceful lark]

Pov: When you were a child, when did you go to the dentist:

[solar raven]

im oe letter away

[mossy maple]

wait how

[graceful lark]

Just like blood and other "scary" stuff

[thick vale]

sick colors and lines. what model produced this? GPT?

[quaint tiger]

its thanks to my fursona lora! though the base model is FifthMixIllustrious_bananaDaiquiri

[thick vale]

mm i didnt think about that. thats a pretty brilliant way to scrape a style. use MJ/CPT to generate the style you want -> train a local on thousands of iterations of this style. new tech

(this is my intuitive assumption on how it was trained)

[shadow bane]

Err I don't think he should be drinking that

[granite fossil]

model?

[tame atlas]

dev

[quaint tiger]

i have a bunch of different arts of my character, i just train ai on those images i have of him to get a model that can somewhat generate him, then when i get almost perfect images, i photoshop them to perfection to match my character, which then again are used as data for a newer better improved model c:

[final vapor]

this was genned in voyages.

[loud veldt]

create image of 2 year old baby boy

/image create image of 2 year old baby boy

#🌁│image-creations create image of 2 year old baby boy

[white tiger]

Very cute

[pallid thicket]

Leavanny x Mega Banette x Lopunny

[pallid thicket]

Froslass x Mega
Lucario

[cold lintel]

by chatting and reaching level 10

or something idk

[graceful lark]

10? nah ig like 5!~

https://tenor.com/view/level-5-pic-perms-sea-of-umjee-umjee-gif-15630646039760648385

[cold lintel]

im level 7 and i dont have pic perms

[graceful lark]

@echo goblet, @median plume, @eternal bay , @void dew 🐹

[eternal bay]

?

[serene storm]

why the hell would you ping mods

what

some arent even mods 💀

[shadow bane]

Its level 10, I'm pretty sure

I couldn't post images until I got to level 10

[serene storm]

yeah

[whole wagon]

how you make a pictures

[green hamlet]

I use Adobe Photoshop.

[whole wagon]

you can make ai pictures there ?

how much is the cost per picture

[green hamlet]

Read pinned messages here.

[full latch]

Is there any section for ai song ?

[green hamlet]

No.

[full latch]

@pure dagger

[full latch]

Any section for ai song ?

[green hamlet]

Please stop asking the same question. There's no channel for uploading AI generated songs here.

[full latch]

I'm giving suggestion to him 👀, you shouldn't be frustrated

[green hamlet]

To suggest a feature in this server, at least check #1159516963014451302. Asking an admin to add channel in this specific channel won't gonna work.

[full latch]

Thanks for guiding, I'm new I didn't know

[pure dagger]

Asking an admin isn't wrong either

[pure dagger]

Most people misuse the channel for promoting their own YouTube or Social Media accounts, that's why we removed it

[harsh moss]

Aww...

[pure dagger]

Adorable

[calm dock]

How do you make these

[pallid thicket]

ComfyUI, downloaded a bunch of checkpoints and 2,808 lora's, but 99% go unused most of the time, some checkpoints understand numerous pokemon thankfully so no heavy reliance on Lora's

[final vapor]

for ai users who need more than one lora.

[final vapor]

which ai?

[warm rampart]

looks cool

[pallid thicket]

Nah, gotta have them for preservation's sake, plus I like collecting styles, meaning 1 is never enough noob, boost those rookie numbers up

[final vapor]

i hope you don't mind waiting 10 years.

[green hamlet]

"Free" Weights users when AI image generator:

[green hamlet]

I usually use more than two LoRA models to combine two entities into one. Arona becomes Tung Tung Tung Sahur.

[harsh moss]

I used ImageFX from Google

[fading wagon]

mine

[pallid thicket]

My stuff generates quite quickly depending on what I'm doing with it and how much of it I'm doing. Lowest end is just a few minutes through all the nodes I got in one circuit.

[quaint tiger]

I dont agree with that when you make more detailed images <u< especially with custom characters

[graceful lark]

yus stupid loras

[tall turret]

/chien

[final vapor]

if you have premium than so be it. i am about to try premium sometime soon.

[pallid thicket]

Fuck that, I run this shit on my PC, I don't pay a dime cause I like to keep my money for things that actually matter

[final vapor]

if you have powerful enough pc specs then sure. my card is a gtx 1650 so ai is basically a no go.

[pallid thicket]

I got a 4070 gtx with 64 gigs of ram and 8 gigs worth of Vram, I can handle quite a bit

[final vapor]

as a 1650 user ai is out of the question.

[shadow bane]

4GB VRAM is enough for some AI things.

[hard fog]

having more vram is more helping than having more system ram (32 GB is usually enough)

desktop 4070 has 12 GB tho

so it is usually enough

[pure dagger]

That's is marvellous

[rich moss]

How to make ai images

[green hamlet]

Read pinned messages here.

[bronze abyss]

The plus option that adds media is not enabled. How can I share images in thr group?

[lofty folio]

flux 1 pro

[echo coral]

Wait what

That looks like a real image

From a distance

[lofty folio]

that is from flux 1 pro

[gritty shadow]

Princess Boo + Mario lora's

[echo coral]

That's wild

[hard fog]

looks like reskinned Peach

[echo coral]

shine bright like a diamond

shine bright like a diamond

[languid osprey]

What app or website do you use to make these?

[languid osprey]

Fire

[echo coral]

Fire

[gritty shadow]

Stable diffusion

Local

[split bough]

k

k

k

k

k

k

k

[fading wagon]

black hole,red hypergiant,blue giant and white dwarf

[tawny peak]

quite an upgrade in clarity

and some pretty damn good hands

[warm rampart]

😂

[plucky atlas]

Remember to stay active!

And drink plenty of water, now that the weather's heating up again!

[shadow bane]

I think I'm doing it wrong

[west quiver]

One message removed from a suspended account.

[quaint tiger]

Just some random stuff i have made o3o

[warm rampart]

[harsh moss]

Hello!

[hexed frost]

...

Damnn

[echo coral]

Holy ai dump

[wraith merlin]

Which model is this

[wraith merlin]

It's looking so real

[warped basin]

I made this album cover for my song today. Not sure what you guys think of it.

[warm rampart]

some ai generated stock images from our customer project

[warm rampart]

your work is awesome dude, we like it a lot keep going ❤️

[vernal sorrel]

is that 4o new image generation? or flux/mj/other models?

[warm rampart]

its with flux

[wise shard]

Just bored

[magic radish]

prompt: GTA 6風のスタイル、ロマンチックな雰囲気。メインキャラクターはトルコ人の若い男性と日本人の若い女性。二人は親密に並び立ち、都会の夜景を背景にしている。背景には大阪の街並み（ネオン、看板、狭い路地など）。小道具としてバクラヴァと寿司がテーブルに置かれている。画面の端に控えめに映るトルコ人女性（髪を下ろした）と日本人男性（クールな表情）、脇役として小さく描かれる。全体の色調はネオンピンクとパープル。画面には架空の日本語テキストコードも含める。GTA 6ではないが、その雰囲気を再現。

I created v4 vasryosyon by mistake.

[west quiver]

One message removed from a suspended account.

[gaunt anvil]

How to use

how to use

[magic radish]

lora ?

open ai?

[west quiver]

One message removed from a suspended account.

[magic radish]

No need to spend her credit

I know a site where you can produce unlimited open ai images with original api

[west quiver]

One message removed from a suspended account.

[magic radish]

florafauna.ai sign up with temp gmail from this site and when you run out of credits, sign up again

but here's the important note

Use gmail with a dotted (.) and copy the box to download it and paste it into a text document, it will be like jason, there is a url there, copy it and download it from the browser

@west quiver

but you can sign up with temp gmail and the dot (.)

[fiery ore]

reate a video showing an anime style adorable furry character

[harsh moss]

Would you want this?

Used this as a reference image:

To make this image:

[naive wave]

Okay but how is this both soft and fire at the same time?? The freckles, the eyes, the vibe literal perfection 🔥

[west quiver]

One message removed from a suspended account.

[livid oracle]

сиськи

boobs

[livid oracle]

hello go meet

[naive wave]

Wow, the detail in this prompt is amazing the emotion, the rain, the city lights, everything comes together perfectly. Really well done!

[graceful lark]

https://cdn.discordapp.com/attachments/1097590033319465040/1276198495379914783/me_when_your_ears.gif

[brisk drum]

Thanks pony diffusion doesn't really like when I use detailed prompts so I mostly use my own model to get stuff made

[reef wren]

What was the prompt?

[topaz acorn]

@plucky owl nice

[plucky owl]

I made it through sets of prompts.

[cold lintel]

@warm rampart no advertising

[echo coral]

These are nice

[oblique relic]

how do i post here?

[naive wave]

ngl this turned out super clean. the lighting and fabric movement just hit right

[reef wren]

Ohh but no doubt it's really good 👍🏻

[green hamlet]

To post an image here, talk a bit until your name turns light green.

[round cloak]

one thing i noticce in most ai image gen is if the promt about nature, it will create realistic output like this one, but when you try to promt to generate human, ai will give you a cartoonish human, like fully covered by make up

[oblique relic]

anyone know the best image model for realism ?

so far ive been trying gpt 4o and i produced images indistinguashable from reality

however it took alot of prompts and editing

[harsh moss]

Man, I love the new 4o image generator in ChatGPT. It generated a one page comic dub I want

This is the story:

Scene 1: Living Room (Evening)

Boy (smiling, munching on snacks): "Man, this cartoon never gets old!"

(The CRT TV suddenly turns to static.)

Boy (frowning): "Huh? What's wrong with this thing?"

(The screen flickers back to the cartoon, but within seconds, it turns static again.)

Boy (sighing, getting up): "Oh, come on!"

(The boy walks up to the TV, lightly knocking the side of it.)

Boy: "Hey, work with me here."

(He knocks it again, a bit harder. Still static.)

Boy (frustrated): "Why now?"

(He gives the TV one more knock, but suddenly pulls his hands back.)

Boy (wincing, looking at his hands): "Ow! What the..."

(A small cut appears on both his palms, a thin line of blood forming.)

Boy (rubbing his hands): "Weird... I didn’t hit it that hard."

(He yawns, glancing at the wall clock.)

Boy: "Guess I'll just take a shower and get some sleep."

Scene 2: Bathroom (Night)

(The boy finishes his shower and goes to bed. Fade out.)

Scene 3: Bedroom (3:12 AM)

(The digital clock shows "3:12 AM." The boy stirs in bed, rubbing his palms.)

Boy (half-asleep): "Ugh... why do my hands feel so weird?"

(He sits up and turns on the bedside lamp. His eyes widen as he stares at his hands.)

Boy (voice trembling): "W-what the... fur?"

(His right hand is now covered in sleek fur, fingers reshaping into paw-like structures. The left hand follows.)

Boy (panicked): "What’s happening to me?!"

(He gets up, stumbling. The view shifts to his back. A tail slowly pushes against his pants, straining the fabric until it rips, revealing a feline tail.)

Boy (breathing heavily): "This can’t be real..."

(He looks down at his feet, which are lengthening and morphing into digitigrade, cat-like paws.)

Boy: "No, no, no..."

(His hair changes color and texture as cat ears emerge atop his head. His face gradually elongates, nose and mouth forming a feline snout.)

Boy (touching his face): "I’m... a cat?"

(The final scene shows the boy’s full body: a slim, anthropomorphic cat, trembling in shock.)

Boy (whispering): "What did that TV do to me...?"

I reimagined a story based of this video:

[harsh moss]

Bruh, I don’t know why, I’m so obsessed with the new ChatGPT 4o image generator, and as a plus, I recently upgraded to ChatGPT Plus, just for that.

[harsh moss]

Bruh, why are these so good? Made with ChatGPT Plus

[harsh moss]

Used ChatGPT to modify the style of an image. First picture, reference image. Second picture, resulting image

[shadow bane]

I wish I could afford to pay for it x.x

[fading wagon]

by chatGPT

[steel steeple]

well it was fun while it lasted lmao

[graceful lark]

you can use google colab

but without gui

[harsh moss]

This is unreal…

This is AI generated you know… by the new 4o image generation in ChatGPT

[harsh moss]

The Shibuya Crossing in Tokyo after an apocalypse

After an earthquake

[bronze cedar]

a dog

[harsh moss]

UwU… I like kissing boys…

[tawny peak]

found a good looking config

[graceful lark]

👍

[tawny peak]

some kemono friends characters worked well with it

[harsh moss]

So stylish

[granite fossil]

What the heck? You got a prompt??

Used Gemini to generate an image for a story I’ve been developing using ChatGPT.

[harsh moss]

Could you give me that story, if you can? Because I can generate a comic-styled page for you to depict your story

[harsh moss]

This one I used image-to-text, still using ChatGPT Plus

[granite fossil]

Got any examples?

[harsh moss]

Sure, just give me a minute

[round crater]

i think i'm cooked, i aint got no idea what to make

[marsh cape]

how do i make an image

[exotic warren]

How do I make an image here ?

[cold lintel]

do /image in #🤖│bots

[unreal olive]

strimanze O_o

[lost thistle]

thsi is cute i like chibi what ya do (going afk shopping ping me i'll read later)

[tawny peak]

?

[lost thistle]

how did you get it to be like that

[tawny peak]

prompting

[lost thistle]

you could be a bit more specfic but w/e also your profile is more of a interjection than pronoun =p

[tawny peak]

I just used chibi, chibi only to get that look. Also so what if it's like that.

[hard fog]

nice sticker material

[tawny peak]

thx

[lost thistle]

i need to make some like that

[rustic ingot]

why not learn to wake up and not be insufferable on discord

[flat falcon]

hi

[graceful lark]

hewwo!

[vivid mango]

i told chatgpt to recreate this image with it's own style

[green ridge]

Hello everyone, does anyone know which AI preserves the quality of a person's image? I mean, when generating an image with different effects, the person's face should remain as it was originally. However, both GPT and Grok distort it during generation. Which AI doesn't do this? Even if it's paid, it doesn't matter, please tell me about all the ones you know via private messages. Well, you can write here too, but it would be preferable if you could message me privately, thank you

[magic radish]

Model: Google imagen 4

[magic radish]

Prompt: ```text
A wide image taken with a phone of a glass whiteboard, in a room overlooking the Bay Bridge. The field of view shows a woman writing, sporting a tshirt wiith a large OpenAI logo. The handwriting looks natural and a bit messy, and we see the photographer's reflection.

The text reads:

(left)
"Transfer between Modalities:

Suppose we directly model
p(text, pixels, sound) [equation]
with one big autoregressive transformer.

Pros:

• image generation augmented with vast world knowledge
• next-level text rendering
• native in-context learning
• unified post-training stack

Cons:

• varying bit-rate across modalities
• compute not adaptive"

(Right)
"Fixes:

• model compressed representations
• compose autoregressive prior with a powerful decoder"

On the bottom right of the board, she draws a diagram:
"tokens -> [transformer] -> [diffusion] -> pixels"```

[naive wave]

this one goes hard, the colors are wild in the best way. love how the eyes pop and that sword pose feels ready for action

[naive wave]

this is adorable the expression and lighting together just sell it. feels like a quiet moment but still full of life

[harsh moss]

Lounging…

[harsh moss]

Cool! And made your character realistic…

[quaint tiger]

Ewwww i dont like this xD

Not a sleight at you, it just looks

It makes me uncomfy xD

[quaint tiger]

I been improving at making loras of my sona, this must be like model 6000+ i have made at this point xD

[quaint tiger]

I think the word im looking for is that it looks uncanny

[quaint tiger]

This is so beautiful and adorable ♡

[harsh moss]

@quaint tiger, what do you think of my image, that has the message, “Lounging…”?

It’s a first person view of a furry grey wolf drinking coffee in a cafe during a rainy day

[quaint tiger]

That one is actually cool :3

[haughty zinc]

How used

[shadow bane]

_>

[green hamlet]

URL links are banned in this channel to avoid spamming and promoting.

[shadow bane]

it was short mp3 clip but I understand XD

[magic radish]

Model: Google imagen 4

[magic radish]

Model: Google imagen 4

[magic radish]

There were youtubers who said that it doesn't support Turkish.

[versed badger]

🐁

[crimson osprey]

/imagine Full body professional portrait of a young European woman, 22-24 years old, Slavic features, piercing blue eyes, long blonde hair with natural waves, wearing a stylish mini skirt and elegant blouse, standing in natural confident pose, soft professional lighting, high fashion photography style, clean background, commercial photography aesthetic, full body shot, 8k resolution, photorealistic

[graceful lark]

-# bro what

[graceful lark]

@keen bramble your mom took your phone and PC

oh bruh x3

[tawny peak]

smol and normal adachi rei

[tawny peak]

DAAAAMN!!! (literally)

[small edge]

Hello, wanna be friends?

And should I label my ai images into real life location names?

So you can know

You all

[rocky tangle]

If I was going to sign up (pay) for ONE AI platform to use, what would you guys suggest? My initial interest would probably be for image generation / video generation.

[shadow bane]

I'd wait and buy one of the new Intel GPUs with 24 or 48GB vram for super cheap.

[finite bramble]

Blinky Bill is being forced to play Russian Roulette with the dingo mafia, Extraordinary!

[umbral geyser]

thumnail for youtube grow a garden roblox

[magic radish]

model: google imagen 4

[round cloak]

this gen pretty looks so real, and it's like the same with what i am experimenting in savro site

[graceful lark]

i use ChatGPT

[sage lake]

/imagine uma carro voando

[magic radish]

Model: google imagen 4
Style: Studio Ghibli

[magic radish]

guys who made tumrail with youtube
making imagen 4

[magic radish]

only the dead-end sign is faulty

[magic radish]

I found the right prompt
image generated with google imagen 4

[tulip light]

guys need help

damn why cant i share pics

[harsh moss]

Man... I found a new text-to-image AI generator online that can generate images up to 2K, for free (well, the first time you sign up, it gives you 3 trial uses, that doesn't cost any credits, then it will use your actual credits after. On first sign up, it gives you 600 credits

BETTER THAN MIDJOURNEY!

Give me a prompt, I don't have any ideas bruh...

BRUH!

I did NO upscaling, whatsover

Bruh, I just realized, I did not prompt it to generate an image of Nick Wilde from Zootopia, but it looks like Nick Wilde

[cold lintel]

thats probably what they trained it with

[harsh moss]

Another one

[magic radish]

google imagen 4

[finite bramble]

Looks like Blinky's caused enough trouble for Mayor Pelican one time too many...

[subtle osprey]

/imagine A cute young boy named Gleb wearing a simple t-shirt, with messy curly hair, holding a Samsung phone and staring at the screen with heart-shaped reflections in his eyes. His expression is dreamy and lovestruck, as if he's looking at something adorable or exciting. The background is slightly blurred to emphasize his emotions. The art style is semi-realistic with soft lighting and warm colors to enhance the romantic vibe

[quaint tiger]

reminds me of something

[harsh moss]

@subtle osprey

[heady trail]

That sucks i cant send images

[naive wave]

Is Dreamina AI a good tool to generate cartoon like photo?

[harsh moss]

Yes, if you want you can try it yourself by going to the link I provided in the image, signing up for free account, using Gmail, or your own email, then you’ll get 600 credits, and for Standard (1K) resolution, it consumes 10 credits only

[harsh moss]

Bruh

[magic radish]

Did you just learn about this site?

I knew the Chinese AI was shit.
I say Turkish and the Arab draws it

they don't care about artificial intelligence

It's called drawing Arabs. It's annoying.

@harsh moss new model of flux is out

but not open source from the site

[harsh moss]

Yes, I just found out about it, two days ago

[harsh moss]

It’s not Chinese AI, this AI is literally made by CapCut

[magic radish]

CapCut > TikTok > Chinese

THAT'S HOW I KNOW

[harsh moss]

Ok, ok, no need to snap. I just thought it wasn’t Chinese

[magic radish]

I'm not angry at you, I'm angry at the Chinese.

Do we look like Arabs to you?

[sly badge]

how you make these photos

like with what ai

[cold lintel]

please stop talking about this, users are free to use whatever ai they want to

[harsh moss]

Sorry for the late reply, I made with Dreamina AI

[cold lintel]

no one calls chinese ai "drawing arabs"

[harsh moss]

The link is in this picture

[harsh moss]

The link

You get up to 600 credits daily for free, once you sign up

[sly badge]

Ooh okk thank youuu

[harsh moss]

If you sign up for the first time, they let you try generate pictures with the highest resolution and latest model for 3 times, as they said, 3 trial uses, then after that, it will consume your credits. For the Standard (1K) resolution, it consumes 10 credits only with the latest model, for the High (2K) model, it consumes double the credits

So 20 credits

They also have editing tools for your pictures there, to try it, click Create on Canvas on the home page and upload your picture. Click the icon on the left of the canvas size, usually 1024x1024, to make it scale within your image dimensions. Then, there are Retouch, Upscale and Background Remover. Retouch and Upscale are both cost free, and doesn’t consume any credits, except Background Remover, it will consume 20 credits

Bruh, I just found out that Microsoft’s Bing Image Creator just introduced a text to video generator in preview stage

[magic radish]

I did not judge

I gave my opinion.

and I'm not angry.

Am I being misunderstood?

[harsh moss]

Bruh, Microsoft’s Video creator is so bad… I just tried generating one video, they said it will be ready, in a few hours?!

[magic radish]

Did Microsoft get into the video AI business too?

[quaint tiger]

if im allowed to be honest, i have literally no clue what you mean with drawing arabs or what it has to do with arabs o-o

[cold lintel]

you shoulda seen what else he sent that i deleted

[quaint tiger]

I honestly have no clue what it is even about o.o.... arent we here to just use and enjoy AI? Why are different countries/races being pulled into that type of conversation?

I am hella confused

[cold lintel]

yeah someone was complaining about how long it was taking to generate a video and he said "This is what happens to those who support (insert country thats fighting a war)"

[quaint tiger]

Wtf

Why cant people just leave politics out of stuff like this x.x

[cold lintel]

yep

[night kestrel]

Craziness

[night kestrel]

Query - I only got 120 credits instead of 600

Did I miss something?

[sly badge]

alrrr thank youuu for letting me know

[night kestrel]

Also, I can't seem to actually use the image generator for Dreamina? The site works well, but every time I click image generator, my screen freezes, and the website is forced to quit

[naive wave]

look nice, what AI prompt did you use to generate this photo?

[quaint tiger]

You shud be able to throw it into png info, also i used the new boykisser loras i have made and uploaded to civitai c:

Im working, so i dont have access to the prompt myself atm

[magic radish]

an example of an open source artificial intelligence made by a Chinese

there are a lot of images in stock related to turkey

Before the update, he looked like a Saudi Arabian.

because of the hat

[quaint tiger]

Thats just how the model has been trained, if you dont like it, dont use it.

Its pointless to drag politics in this.

Theres people that might like using it for other stuff, no need to shit on it cause its not accurate or good enough on certain stuff.

[cold lintel]

#🏙│ai-images message and you replied to an image of a fox which has nothing to do with turkish/saudi arabian men

if you really want pictures of turkish men i recommend grindr

[quaint tiger]

And if i may be honest, i dont like ai generation of real looking people anyway, that stuff is just weird imo.

No judgement to those that do like it, it just most certainly aint my cup of tea.
I like generating wallpapers, characters, nature stuff, etc.

[magic radish]

It has nothing to do with potlitika

incomplete train

[magic radish]

For example, flux is not Chinese, but it can't draw.

even looks very plastic

Recraft v3, for example, does not draw Turkish

Seedream 3.0, for example, does not draw turkeys

This has nothing to do with China.

I don't like the models.

can it be fixed with lora yes

the important thing is that the loa does not leak

[hard fog]

so you think it's kind of political influence? you can go research it

[cold lintel]

no one said anything about turkey either

[magic radish]

No, I didn't say that.

[magic radish]

I'm saying the model is not drawing.

we're probably having trouble understanding because of the translation.

I'm also translating from deepl.

[harsh moss]

Bruh

Using Google Imagen 4

Using Dreamina AI

[magic radish]

there's also an ultra model of imagen 4.

I didn't see

google

always draws a blurred background

[graceful lark]

wow!

[quaint tiger]

Pretty gud right? Doesnt look AI to me tbh x3

[graceful lark]

yeah so good!

my lora.. look not good x3

[shadow bane]

what are some free good Ai image generators that can generate high quality detailed pictures?
for example if i want multiple angles of a duffle bag and for it to be detailed and high res.

[wraith merlin]

Hey, how are you, I think you was the helper in Ai hub but you account got hacked?

[graceful lark]

Flux + Upscaler ig

[covert dome]

they don't have bot here man

😐

[cold lintel]

/image in #🤖│bots

[magic radish]

Yes, it was me. I gave up the helper role.

I'm just hanging out

[harsh moss]

The link is here, @shadow bane

You asked, what are some good AI image generators. These are the ones

Upscaling and retouching your images are free there

[harsh moss]

Hmmm…

[harsh moss]

Hmmm…

[harsh moss]

Bruh…

[graceful lark]

OWWW! CUTE!

[green hamlet]

Sometimes, using a translator can mislead an English speaker to think about something else too far. Even if I use my entire brain to type in English, they still don't understand me lol.

Anyways, make sure to refrain about politics and ethnic topics anywhere in this server because not everyone would find them fun to talk about.

[wraith merlin]

I'm glad to see you back

Welcome

[naive wave]

love the simple animated photo, poor kitty cat, wanna see more of this

[hard fog]

I actually love this fluffy one

[finite bramble]

Don't remember James taking a wrong turn in Greenpatch.
Wait... NO!

[calm carbon]

You do not have Manage Server permission to use that command.

Please wait 43 more second(s) before reusing the clean command.

Please wait 35 more second(s) before reusing the clean command.

[naive wave]

Yo, why does my guy look like he’s rethinking all his life choices while she’s out here chillin’ like it’s Saturday morning cartoons! 😅

[finite bramble]

James is supposed to be a ladies' man, but he keeps attracting the wrong king of ladies.

Just to clarify, Daisy Dingo is also a whore.

[thick rose]

hello

[graceful lark]

OOOW! >////<

SO CUTE!~

[fresh mulch]

can someone help me begin generating ai images to help me learn anime art

[vital olive]

How to generate those images?

[vital olive]

?

[shadow bane]

ca someoone help me

[quaint tiger]

you need a proper strong pc to run this stuff

then i recommend a program called Stability matrix

which you can use to install a bunch of AI stuff

[rocky tangle]

Cool, which image AI program did you use to produce those images after installing Stability Matrix?

[shadow bane]

idk about stability matrix, but I just used SwarmUI

[rocky tangle]

Thanks, added to my list to check out

[small edge]

Guys

Say one random country

I dare ya

Or city

Anywhere

[green hamlet]

For normal conversation, go to #✦│chat.

[green hamlet]

For any help about Stable Diffusion or any other AI program, go to #✨│ai-help or #1192011222023950368 and explain about your issue there.

[fading wagon]

by gpt-4o

[naive wave]

that cute ai dog gave me mixed emotions it made me smile, but also feel a bit sad. It reminded me of my dog, Debit, who passed away three months ago

I still miss him a lot.

[quaint tiger]

i personally use forge stable diffusion

[rocky tangle]

Thanks for the feedback!

[small edge]

Feel free to guess

[quaint tiger]

Guess what?

[cold lintel]

chicken butt

[small edge]

The cities

[finite bramble]

Alright, who violently murdered Blinky Bill?!

[graceful lark]

Original and silly gpt:

[tawny peak]

hires fix is amazing with how much quality you can get

[magic radish]

Imagen-4-Ultra-Exp

Imagen-4 For the purpose of comparison, this is not ultra

[past hatch]

yuuuuuukaaaaa

[naive wave]

this is really a high quality image, i have question, what file formats should I use to preserve high-quality details and consistency across generated images? i would love to read answers

[quaint tiger]

Start using adetailer as well, and boom

[tawny peak]

Yeah I have used it before, personally for me it just increased gen time and didn't make it look much better, though that was like a year back so it could've changed by then

[quaint tiger]

oh i cant live without it

it makes my stuff SO MUCH better

[tawny peak]

Let me try it and ill see if it's better

[graceful lark]

cute

[graceful lark]

xd

[quaint tiger]

Certainly not perfect, but i liked the style x3

[shadow bane]

that's adorable♥

[naive wave]

this is adorable, love the color combination id like to ask what prompt u used to achieve this output

[quaint tiger]

It was a very simple prompt containing keywords like:
Nolineart, verctor, vector style, vector art,

As for the colors, that comes mostly from the lora that i made of my fursona, this is my character afterall, though i do think i also used a vector art lora that made the colors a little more pastel

[compact fjord]

Hello, I want to make a video of Lionel Messi greeting my father on his birthday. I already have the audio and video that I would like to use. Does anyone know of any free tools that I could use to make the deep fake?

[quaint tiger]

the only member that i remember doing deepfakes is Pikachu

so they would be your best bet

i think atleast.

[quaint tiger]

Very gud :3

[graceful lark]

thanks ^^

[rocky tangle]

Are you working on making a comic / story book or something?

[graceful lark]

welp Stable Diffusion can create but need so many time for good result + photoshop ig

[graceful lark]

woah 😮

[grim marten]

This is quite good.

[naive wave]

this is really impressive! what AI photo tool have u been using, mind sharing?

[grim marten]

ComfyUI feels pretty good.

[harsh moss]

I used ChatGPT’s newest image generator, which is the GPT 4o image generator. But you can only access it with a plus or pro subscription

[harsh moss]

:3

[harsh moss]

Another version:

This is Boykisser’s girlfriend

[hard fog]

I like this

[echo rover]

TF IS THIS FURRY AH SERVER

0

[harsh moss]

Maybe kind of… it started all because of me

[naive wave]

oh, is it easy to use or it has lots of functions to remember or to use?

[naive wave]

like this one too, was it generated or enhanced by chatgpt too ?

[harsh moss]

It was completely generated by ChatGPT

[harsh moss]

I just gave it the 2D picture, which is this ^, and then I gave it this prompt:
Render this character in a modern, high-quality 3D animation style.
Just like that

[quaint tiger]

O.o

[quaint tiger]

Why do you believe this? O3o

[green hamlet]

[past hatch]

[rich saffron]

this is totally ai

trust

[graceful lark]

https://tenor.com/view/rat-rodent-vermintide-vermintide2-skaven-gif-20147931

[graceful lark]

@rich saffron you sure its ai? 😮

[night kestrel]

Is it possible to remove the dreamina ai logo from the image after generating?

[rich saffron]

no

its a commission

[graceful lark]

welp ig better del

bc here only ai

[graceful lark]

yeah ig

[night kestrel]

I've just been taking screenshots of the completed product, do you have a method of just downloading it directly without the logo then?

[graceful lark]

HTML code or just "online remove watermark"

ooh - legs x3

[night kestrel]

Thanks! 🙏

[graceful lark]

yw?

[bright pagoda]

does anyone have realistic ai images of girl selfies will pay for expert

[graceful lark]

Uuuuugh ChatGPT? X3

[rich saffron]

NO

chatgpt has horrible ai images

like im not even an ai person and i know its bad

[hard fog]

because ppl have done ghibli AI

[rich saffron]

guh

grandma ai

[hard fog]

yep, grandma in ghibli style

[graceful lark]

??!

Or use Flux idk

[jaunty dagger]

Ask

[rich pollen]

love the vibe switch! goth on the left, chill on the right. both look cool 🔥

[graceful lark]

https://c.tenor.com/3cQ0OnhQVb8AAAAd/tenor.gif

[naive wave]

i love their aura and vibes might use it for a inspiration for ai influencer i am using savro although i tried comfyui and other i just find savro easy but the quality is not so perfect but it's fine all goods on my need

[graceful lark]

Thanks!
ig i try make new "list" ai gens

[lapis whale]

isnt boykisser a boykisser

[old matrix]

Nice

[gilded gulch]

Axel rose 18 años

[rich pollen]

okay this actually looks kinda real, good stuff.

[small edge]

Seoul

[graceful lark]

https://tenor.com/view/raised-eyebrow-eyebrow-raised-eyebrow-emoji-silly-cat-wtf-gif-8109444744753568214

[tight sky]

Wait this is Boykisser chat room?

[graceful lark]

uuuugh kinda? x3
here for everyone ai arts

[tight sky]

:3

[tight sky]

hehehe, I'm going to download every image
here

[graceful lark]

woah? 😮

for wha

why

[tight sky]

cool Wallpaper

[graceful lark]

i see 😮

make sense xd

[tight sky]

[graceful lark]

im rn honestly also do Wallpaper x3

with ai ofc

@rustic ingot

and @cold lintel @alpine plover

i recc using "Search"

That could have been very nice...

[naive wave]

this is freaking hot, what prompt did u use for this?

[limpid igloo]

Hi, I’m not technical at all, but I’m currently searching for a model I can run locally to simulate model collapse, any suggestions?

[magic radish]

I didn't save the prompt

I wanted to do that drawing style a lot but it didn't happen, I'm talking about the drawing style palette
ok men it happened by chance

“Young Japanese woman in bikini with big boobs and tattoo of Mustafa Kemal Atatürk and young Turkish man in front of a vending machine with romantic background in an alley in Osaka”

this prompt 100% must be missing not what I remember
I think I wrote something close to this
the artificial intelligence was saying that I couldn't draw a sugreek, I'll force it out by trying and trying.

I did it through Poe

safe feature is one click less in api

[rich pollen]

damn that lightning one tho

[compact pivot]

Am I invited to the poyo party??!

[naive wave]

oh thankss🙏

[graceful lark]

https://64.media.tumblr.com/7e85c2576b37928b6f7c8ada54afa3c6/8992d32d515e5799-bb/s1280x1920/d1c11500d169a03e87b088abe1f10b0cdf39dbd3.gif

why not

[naive wave]

what really caught me here was luffy with gojo senpaiii😍 not being biased but love them bothh, anyways ur output were superb!

[quick plinth]

what's the best local ai image generation framework right now? is it still automatic1111 or forge or what?
is flux.1 still the best local model?

[pallid thicket]

I personally use ComfyUI, it has served me well

[quick plinth]

how do you run it? I typically run things like this out of docker containers but I've run into some issues

[pallid thicket]

I run it on my main storage device, I don't use external ones cause it might fuck things up if I download something and it doesn't go to my main drive

It helps a lot having 4Tb of storage though lol

[compact pivot]

oh boy i cant wait to try some poyo punch!!

[graceful lark]

https://tenor.com/view/kirby-sombrero-mexico-party-dance-gif-11374426840349412739

[rich pollen]

the cat just chillin under the tree like it’s got no worries at all lol

[graceful lark]

WOAH! so good!

[naive wave]

i wish i was a cat, no problems just meow meow only under the threee

[rich pollen]

same 😭 cats living my dream haha

[compact pivot]

do NOT drink the poyo juice

[tall dawn]

/imagine

[stark dagger]

An ultra-busty young Japanese woman in a bikini with a tattoo of Mustafa Kemal Atatürk's signature, leaning against a vending machine in a narrow alley in Osaka, Japan. Next to her is a young Turkish man. They are both looking at the camera with a romantic, dreamy atmosphere in the background. The vending machine has soft, colorful lighting, enhancing the intimate and cinematic vibe.

prompt: Young Japanese woman in bikini with big boobs and tattoo of Mustafa Kemal Atatürk and young Turkish man in front of a vending machine with romantic background in an alley in Osaka
model: google imagen 4 ultra

[fierce sapphire]

What am i reading

[compact pivot]

hello guys yes this is my Al generated image

[primal totem]

damn that looks real

[plucky atlas]

Same idea, different execution.

[magic radish]

prompt: A photorealistic, cinematic, full-body shot of a young, romantic couple on a private balcony overlooking the Osaka skyline at twilight. A beautiful Japanese woman with long black hair, wearing an elegant dark-colored bikini, is being tenderly embraced from behind by a handsome Turkish man. He rests his chin gently on her shoulder, and they both gaze out at the city lights with serene, content expressions. The mood is deeply romantic and peaceful, with warm, soft lighting from a nearby lantern enhancing their affectionate moment. Ultra-realistic style, high detail, soft shadows.

model: google imagen 4 ultra

[uncut slate]

This is showing too much and they are too close. Stop posting these

[rich pollen]

thanks for sharing ur prompt. that was a good output or result @magic radish !<3

[charred thicket]

/imagine

[cosmic scarab]

Chill out dude

[rich pollen]

character design is cleannn

[rich pollen]

hello!!

anyone tried tools other than Savro for creating AI influencers or characters?

[pallid thicket]

I myself use ComfyUI as it allows total freedom and is just the most convenient to use

[outer chasm]

Does anyone have a flux dev checkpoint they can recommend that interacts well with a charcter lora I made, I want lifelike amateur realism that picks up skin textures, clear backgrounds, etc. Ive tried so many checkpoints that achieve this but screw up the face.

[jolly pier]

Looks so real! What AI tool are you using?

[final vapor]

genned using voyages.

[modest wraith]

What's required to be able post images here? I don't have the option

[tawny peak]

Chat W/ People in #🧬│ai-chat until your name turns green

[rich pollen]

oh cooool! u should try as well pykaso and savro, they work so weell. comfyui is a good tool! i already tried it

☺️

show some result haha if it work well for u

[naive wave]

was it really fanastic? or u just forgot letter t? btw love this version bro ahha

[halcyon sapphire]

That was the ai I think

I was using flux

[naive wave]

anyway bro, what other ai tool that generates realistic model u know? been exploring savro but would love to ask if u know aside from it :>

[halcyon sapphire]

I use grok, Microsoft bing copilot, or Diffusers Image Outpaint, just those add realism to my eye I feel

@naive wave

[vernal inlet]

@wheat radish

omnigen

[wheat radish]

seems cool

[vernal inlet]

it can do realistic stuff, image editing, in-context generation and other stuff

chroma

[undone oriole]

flux kontext just lunched open sourced 👽 🔥

[wary vortex]

Where are you guys generating ai images? Can help share links ?

[undone oriole]

probably comfyui

[vernal inlet]

locally

[wary vortex]

How ?

[rich pollen]

what a thug haha

[shadow bane]

turn the character into a wizard

(flux kontext) XD

Sailor Moon style

[north yacht]

#🏙│ai-images me as an Indian girl

[jolly pier]

There are AI tools that generate images.

[bronze oracle]

Well Of Course Love

[supple scarab]

are most images here generated on local models like sd1.5

[quaint tiger]

i personally use illustrious models

[shadow bane]

I use Flux :3

[shadow bane]

convert to a retro pixel art style

flux kontext is amazing XD

[shadow bane]

oh.

[umbral escarp]

cool

[pure dagger]

@lunar mortar This channel is for AI creations only, let's stick to the topic 😁

[shadow bane]

replace the background with a bed

[zinc mantle]

When i was using remix.ai ; what i could salvage before they closed it without announcing it to anyone

[pure dagger]

cUTE!

[graceful lark]

YEAH! CUTE!

[green hamlet]

AI Hub by Weights-chan. The model was trained on PixAI.art, using "NoobAI-XL (NAI-XL)". With added the latest AI Hub logo, the logo can now be generated accurately.

[vernal inlet]

flux kontext

[tough moon]

dis the right one?

[shadow bane]

Character by my model, background by Flux Kontext

[shadow bane]

the detail is insane XD 1536 x 1792 no upscaling.

[harsh moss]

I just found out that I think Microsoft’s Copilot app uses an image generator similar to GPT4o, because when generating an image, it displays that blurred image, and gradually go from top to bottom as it generates

[harsh moss]

I knew it

I don’t even need ChatGPT Plus to use this!

If I knew this from before, I would’ve installed it already

[naive wave]

ohh thanks man, been into savro but i'll try what u recommend bruh,,,thankss

[harsh moss]

THIS IS THE MOST EXPENSIVE THING I’VE EVER SEEN ON THE INTERNET!

This is literally made with Copilot. Don’t get me wrong, it’s literally Copilot

[vernal inlet]

fake site probably

like 1000s of them popping up any time a new model comes out

[harsh moss]

It’s not a fake site, it’s the official site

[celest relic]

Hi all

[harsh moss]

It’s for industrial use, that’s why

For large-scale companies

[shadow bane]

pffft I can give you a discount of only $500/month to use mine... jk

why would a company pay that much to use the dev version, it'd be more cost effective to buy a 3090 and use it locally

it's sloooow but perfectly usable

[shadow bane]

generation time 12 minutes on a 4090... 3 minutes for the character 9 minutes for the background

[graceful lark]

oow cute!

[tawny peak]

yeesh 12 minutes on a 4090? I'm sticking with my 30 second gen times on a 3060

1.5 - 2 minutes if I upscale

[shadow bane]

well it's using the full fp16 Flux models for both images and no upscaling

also 60-75steps

at 1536 x 1792

[tawny peak]

Oh man really 60 - 75?

Seems unreasonable

[shadow bane]

higher resolutions look better with more steps

for me anyway

[tawny peak]

I don't use flux, I've been sticking with Illustrious (SDXL) but I use 20 steps at 1024 x 1280 or sometimes 1280 x 1536 and yeah it looks fine

Plus I can basically get the "Kontext" part of it using IP-Adapters

though idk, maybe 60 -75 steps is reasonable for flux because of how different it is to SDXL

[shadow bane]

in my experience, 20-40 is fine usually but it has more of a chance for something to be deformed
60 is usually great
above that it starts to lose detail but become more refined
and at a certain point it's just overkill

[tawny peak]

I think the deformation could be caused by the high base resolution since Flux is trained on 1024 x 1024 same as SDXL

[shadow bane]

idk maybe it's because I trained my model on large complex images

i have some 3000 x 3000 or larger images in the training data XD and I can generate up to 2500 x 2500 ish

[tawny peak]

Geez, I think when the model is trained it downscales the images to fit the model

[shadow bane]

flux is only supposed to work up to 2 megapixels

i used bucketing so I don't know what that does

but it seemed to help

[tawny peak]

it keeps the aspect ratio while still downscaling the image

[harsh moss]

Used the Flux Kontext Pro online using the Playground from Black Forest Labs

[shadow bane]

lol

[harsh moss]

Made with Flux Kontext Pro

[harsh moss]

What prompt do you use?

I just used the Flux Kontext Pro online in their AI playground, I didn’t run it locally

[shadow bane]

for my model, I used a high quality illustration of a qzr furry character with messy medium length brown hair and brown eyes. he has messy brown fur with white patterns. He is wearing an adorable outlaw cowboy outfit and cowboy hat. He is leaning against an old western bar. Full body, feetpaws. Serious expression, holding a beer glass. Simple black background. Realistic shading, intricately detailed, fanart, commission, furaffinity. Thick outlines.
and then I just asked Flux Kontext to replace the background

[strong ridge]

you're telling me i can manufacture artificial furry images from my 4090

i play minecraft, and you're telling me i can manufacture artificial furry images from my 4090

I read that right

[fading wagon]

by gpt-4o

A cute and fluffy maltese,sleeping in christmas theme rooms,under the christmas tree,around is gift box,outside windows is blue starfield and snowy ground,japanese anime art style

[tawny peak]

90s triple baka

[shadow bane]

Using a 4090 to make high definition furries. Thousands of years of human evolution led to this moment.

[plucky atlas]

I wasn't aware Flux Kontext Pro was available for download. 🤔

[rich pollen]

yo that 90s aesthetic is so clean

[harsh moss]

Using Flux Kontext Pro

I asked it to remove everything that is outside the sliding door

[harsh moss]

Cool! I also coloured it, using Flux Kontext Pro

[plucky atlas]

Funky!

Though her shirt seems ominously flesh-coloured.

[harsh moss]

Yeah, I should try to change it

Nevermind, it still doesn’t know how to differentiate between clothing and skin

[harsh moss]

Using Flux Kontext Pro

[rich pollen]

yoo this is cuteee! can i ask @harsh moss ? what sites or software did u use for this? any tips for giving prompts? i only used savro and comfyuii.

[harsh moss]

You can search in Google Black Forest Labs playground, clicking the first link and signing up, then go to the Edit Tab, upload your image, and describe what you want to change

And make sure to click the 3 vertical dots, set the Batch Size to 1 by sliding the bar all the way to the left so that you don’t consume as many credits. Cause one image costs 4 credits and for first time users, they give you about 200 credits. And if you want more, you can create a temporary email and sign up with that every time your credits finish

For me, I have multiple Google accounts so I don’t have to worry

[rich pollen]

oh. got it. thank youuuu! 🥰

[harsh moss]

I ran the image of my living room through Flux Kontext Pro, and said to remove everything in the room except the floor tiles, curtains and the things that are attached or on the wall