🏙│ai-images

[wheat radish]

Halloween jonesy made with my LoRA 🎃

[graceful lark]

Aah this looks cool

[hard fog]

mostly because 50 steps is too undertrained

[broken topaz]

/imagemake a boy

/image

[wheat radish]

ty, used weights.gg for it

[graceful lark]

Yes I saw the watermark below

[green hamlet]

This is "Dell OptiPlex 7010/7020" Flux LoRA on Weights.gg. The results don't really change much from their original dataset images, other than additional USB ports, some PC case tweaks and colors.

[graceful lark]

😮

Give me this xdd
https://cdn.discordapp.com/attachments/1159290421676085258/1301556178110906398/cm2xelewj1dlioqb9ojrevvb3_lora_image_1.webp?ex=6724e826&is=672396a6&hm=ec8906878e0

[green hamlet]

Dell might have done something like this with their Alienware, XPS and some gaming desktop lines. Just wondering if this OptiPlex line would get the same treatment as these.

One of the images I generated even looks similar to a HP desktop PC.

[orchid obsidian]

Doc Brown 01 - Original Upscale DeepFake FLUX Dev AI Art 3840 x 2176

[alpine plover]

is this ai generated?

https://cdn.discordapp.com/attachments/1224878046251585626/1295688978728615997/cho_tbl_rblLo.gif

[graceful lark]

yep

[alpine plover]

daaaaaym

how

https://cdn.discordapp.com/attachments/1224878046251585626/1295688978728615997/cho_tbl_rblLo.gif

btw looks like https://7tv.app/emotes/01FXR14HC0000AY6WE2SPE7V9Z

[graceful lark]

So what is it called? xdd

[graceful lark]

huh? It doesn't look bad, but it's really bad

[alpine plover]

i have an idea on how to monetize this kind of art if you can show me that it's flexible in terms of themes and characters. try to recreate this but with the same goat(or whatever was on your art)

https://cdn.discordapp.com/attachments/1103016321211453471/1298006074074333226/catdance.gif

[graceful lark]

huuuuh

I don't understand what you mean?

I'm too dumb

[alpine plover]

the art is very good for a certain purpose

[graceful lark]

(btw It's a dragon)

[alpine plover]

each customized emoji like that can be sold for like 20 bucks

[graceful lark]

Я делаю часто такое

or english?

[alpine plover]

ну вот можно стримерам на твиче продавать

[graceful lark]

mhm

Но я в этом Очень Плох

[alpine plover]

у них разрешение смайлов небольшое, артефактов видно не будет

а выглядит это для смайла отлично

только с гаммой/светом поиграться

а то сильно тёмный челик

[graceful lark]

Я имею ввиду монетизация я Плох

[alpine plover]

да для начала посмотреть бы как оно с другими типовыми смайлами с твича работает

если говно выходит чё тут монетизировать

[graceful lark]

?

[alpine plover]

вот это попробуй с тем же персом воссоздать

выражение лица конкретно

[graceful lark]

Для этого мне нужно создать Отдельный LoRA

Это займёт очень много времени я думаю

[alpine plover]

что для этого нужно

куча смайлов?

[graceful lark]

Да

[alpine plover]

7tv.app можно заскрейпить

[graceful lark]

заскрейпить?

[alpine plover]

https://ru.wikipedia.org/wiki/Веб-скрейпинг

[graceful lark]

@~@

Я имею ввиду Для этого нужны разные Персонажы и Одинаковые Позы

[alpine plover]

тогда вручную оттуда набрать

[graceful lark]

Мне не сложно если ты Хочешь этого

Но это займёт много времени

[alpine plover]

если тебе интересно можно попробовать

[graceful lark]

Я не против но я должен сделать Этот концепт верно?

[alpine plover]

попробовать да

ну хотя если для каждого лору надо

то это не стоит труда

[graceful lark]

Для этого есть отдельные инструменты Например ContrlNet Но для Colab Этого не существует

Хотя я уже создавал Стикеры для тестированые

[graceful lark]

So you don't know what it's called?

[alpine plover]

nuh uh

[graceful lark]

otay

[green hamlet]

Bro just literally says nuh uh.

[shut bane]

@gaunt echo帮我把中间的谷歌logo换成Akamai

@fiery oriole

[alpine plover]

https://tenor.com/view/ez-noob-skill-issue-simply-a-difference-in-skill-gif-20780203

[snow cosmos]

some of my fave generations iv had with various ai over the last few years

i should maybe go upscale some older outputs like those

[gusty zealot]

How to do it

?

[green hamlet]

Most of these cool and fire looking AI-generated images are mainly generated with Stable Diffusion XL or Flux.1 on ComfyUI.

[gloomy thistle]

furry zelda

[gloomy thistle]

what do you think @graceful lark

[slate portal]

genera una imagen

[graceful lark]

Ooh she looks cute!

[green hamlet]

A Dell OptiPlex PC that looks like it was released two decades ago, I didn't put the OptiPlex 7010/7020 LoRA for this, sitting in the Backrooms. Wondering which Windows version is installed on this and is it usable.

Another one, which I added the LoRA to it. But this time it weights too much on this model more than the Backrooms.

[half badger]

windows 7

[tawny peak]

windows vista or xp, based off the logo design

[wild lake]

'

[mint trout]

This is the prompt

Draw a purple dog which has a purple aura around it, while its eyes are white light would come out of them. The dog [The dog is the size of a large mountain] would be behind a giant mountain, while in the distance there is an adventurer on the ground standing, looking at the giant dog. {adjectives for the dog: Purple, massive, giant} {Drawing styles: Still life} {Additional styles: Watercolor, painting, gradients} {Quality: 2K, 360p, minipixeled lights, little details}. {Description of the place: Greenish landscape, giant mountains with peaks covered by fog, clouds scattered across the sky, gray sky}.

[opal sandal]

Draw Demitra kalogeras

[opaque spear]

NFS:MW LoRA

[zinc portal]

I just used a heavier model, which is a FLUX Dev Modified by the community that works with FP16, and I tested it on my PC with 8Vram and 32Ram, and the only thing I could see is that when Low Ram is activated or when Low Ram is not activated, it takes the same time to make an artificial image, but my PC falls short with 32 Ram.

but I managed to make this DeepFake of the singer "Billie Eilish"

Billie Eilish 01 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

If I had a graphics card with more than 8Vram, as it uses 10Vram to 12Vram, and at least 48Ram, I could get the FP16 Flux Dev Model to run without problems.

[graceful lark]

@zinc portal u cat use Civitai or https://www.weights.gg/

[zinc portal]

but... it's not that it doesn't run well, it runs the same as the FLUX Dev fp8, only this Modified one that runs in FP16, it hangs 1 to 2 seconds, but it's not a problem for me.

[brittle ocean]

the resolution is fucked up u can see every pixel 💀

[zinc portal]

Bruh

Bella Ramsey 05 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

Here is the link to the Flux Model that I am testing: https://civitai.com/models/161068/stoiqo-newreality-flux-sd35-sdxl-sd15?modelVersionId=979329

[zinc portal]

Sadie Sink 01 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

[tough sphinx]

Can you draw a anime girl who is overweight and has a big stomach and big arms and big legs and is sweating a lot

[zinc portal]

Yes I can, so I'll do it in a moment.

[tough sphinx]

Thanks bro

Btw how do i do it myself

[zinc portal]

https://youtu.be/12CkOU05_eY?si=-z8L5nK_SM4jS0ta

It is in Spanish, but the important thing is the pages it gives, so just go to those links

[zinc portal]

Here is an attempt at what you asked for, I hope you like it.

[zinc portal]

People, take good care of yourselves to have a beautiful and well-cared-for body.

Elizabeth Olsen 01 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

[mint trout]

Give me prompt pleasee

[leaden oyster]

Ayo? @mint trout level 3 !!!

[steel haven]

no waifus today

[green hamlet]

Dell cheese PC.

[icy wyvern]

CupHead with minecraft effect

#🏙│ai-images CupHead with minecraft effect

[graceful lark]

Aah it's my new LoRA

[green hamlet]

If you want to do AI generation, go to #🤖│bots

[graceful lark]

Cheeeeeeeeeeeeese

https://tenor.com/view/licking-failarmy-taste-it-yummy-delicious-gif-10486140191804653605

[graceful lark]

https://civitai.com/models/910015?modelVersionId=1023171

[zinc portal]

I have just achieved in ComfyUI, Using the Flux Dev Model, and Adding Nodes and Things, to Achieve an Image Quality, worthy of an almost real photo in terms of Quality.

Just look at that wool shirt, it just has too much quality to look like an AI

Billie Eilish 03 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

And all that quality depends on it being in 4K

[zinc portal]

Bella Ramsey 06 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

Millie Bobby Brown 02 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

[green hamlet]

1. a mossy Dell PC
2. A Dell PC catching on fire
3. A frozen Dell PC
4. Pink Dell PC

[graceful lark]

@green hamlet

[green hamlet]

Fumo.

[graceful lark]

mhmmm

[zinc portal]

Billie Eilish 02 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840.png

[umbral pawn]

Create a professional logo for a media and communication company that specializes in graphic design, editing, video camera, photography, and maintenance.

[graceful lark]

#🤖│bots message

[zinc portal]

OKEY

This is the Best DeepFake of Emma Stone

Emma Stone 02 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

[zinc portal]

DeLorean 01 - Original Upscale DeepFake FLUX Dev AI Art 3840 x 2176

[safe summit]

deepfake ?

[zinc portal]

I have a system for saving files and it's too much of a hassle to change that thing every time I save a file.

[safe summit]

understandable

[zinc portal]

You know, when you take a lot of photos or do something a thousand times, it's a pain to change the name of each thing one by one.

[safe summit]

yes

[magic swan]

how i can genrate photo

[graceful lark]

check pins

[zinc portal]

Now I got a better DeepFake, to drive you crazy, since I installed the PulidFlux node, and I got incredible results, I mix the PulidFlux + the DeepFake that I made.

original photo of PulidFLUX

original photo of PulidFLUX + DeepFake

Now it does look more like the character I want to clone, in this case, Emma Myers.

The only thing I noticed is that PulidFlux takes up more RAM and takes between 10% to 25% longer to make an initial image.

And as a final note, 2 things to say:
1.- Positive: PulidFlux is a good tool, as long as you know how to use it and have time to test images and, well, it manages to get the face of the character you want to make.
2.- Negative: Having more DeepFake options, I think PulidFlux is unnecessary, since it takes between 10% to 25% longer to create an image, and it also takes up more RAM and is difficult to install.

[zinc portal]

Conclusion, I prefer to use my original DeepFake, since if I use PolishFlux, between scaling it from 960p to 1920p and then to 3840p and it takes 30 minutes, with my original DeepFake I make 4 4k images in 30 minutes

Bella Ramsey 07 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

and to give you an idea, here are the creation times:
My Optimized DeepFake
432 x 960 - 30 - 40 seconds
10 seconds deepfake
864 - 1920 - 1:30 - 1:50 min
10 seconds deepfake
1728 - 3840 - 4 mosaics that take 1:30 - 1:50 min, which gives a total of 7 to 8 min
vs
PulidFlux
432 x 960 - 50 seconds
10 seconds deepfake
864 - 1920 - 3:50 - 4:20 min
10 seconds deepfake
1728 - 3840 - 4 mosaics that take 3:50 - 4:20 min, which gives a total of 14 to 16 min

And that's not even taking into account the time it takes for the AI ​​to process and execute the commands and nodes, which in this case, means that a single 4K image made with PolishFlux takes more than 30 minutes to complete.

And I shouldn't say this, but...
I say that it is even better to create a good shadow of the character that one wants to make Deepfake, than to use the same PulidFlux.

[leaden oyster]

Ayo? @zinc portal level 14 !!!

[cobalt tangle]

how can i get a juice wrld album cover in just lines no collor??

[graceful lark]

#######*########***#****###*#*********####%%%%%%%%%#%
#*###*+**************#####*************############%#
****+**+*********************************+**#########
****+*********+*++++**********+**+**********###******
**++++++++**********+*********%*+*+++++++***++++++++*
++++++++++++++++++==*%@%@@%%%%@%##+++++++++++=++++==+
++++++++++++++++++*#%%@%@=#@%%%@*+*+++==+=+++++++=+++
++++++++==+++++++==#%%%%#*+@@#@%#====+++=+=+=======++
=============+===*##%#%%%@%@%%+@%*==============+==++
==================+####%%%+%%%@%+#==================+
==================***%#*##*+%#%#=====================
+*****+==========:+**##**#%#+##*.+.:-================
+++**#***++====-..:.-+**#**#**#+.....:-----------====
++++==+++++===:......*=-=-+*++#:......:-=-=---------=
####%######**+==:...:=++==*#*#%*:...:==+++==---------
%%%%%%%%%%%*+#**+-==+-*===+#%%+#%#*+===-=----------=+
**#*########**=-###%*==+**#%#=+#%%#++*#*++**+++++**##
+*++++++++=##%#+##%%#.==-+*+-:%%%*+-=**#%*=**++++++++
*=++===**=#%#***#%%#%#*--+=+-+%%%%*=++*%%#==*+*+*+++*
*+**+*=++**%+-#+#*%%%%#*-=++#**%##%#%#####+-+*****+++
*++**##****%%%%%#*%%#*#+*++=##%%%%%%%%%%%#==+***+++*+
#**###***##%%%%@%%%%%%+===+*+=+%%%%%%#%@@#=-=**+=+**+
#**#*++++#*%@%%%%%@%%#-==-++==*%%%##%#%%%*+=-*+****++
*+*****+++*##%##%%%%%*--======*##%##=+#%%##*-=+******
+****###++*##%%#%#***+-=====-=+####*:-###***+*+*+++=+
+*****+++=*##%%#%###*+-------==####*#####***+********
***+++++==+**###=###=---=-----=*###******=+++++==-=:+
+++++++====+**+*-***=-:-=+:::--********+++++===+++++=

???

[zinc portal]

WTF

Billie Eilish 04 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

[tidal sedge]

Ewww juice wrld is trash

[zinc portal]

To my DeepFake Workbench, I Just Added One More Step, To Gain Between 10% To 25% More Detail On The Face.

result

Millie Bobby Brown 03 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

[safe summit]

FlUx

[dapper sequoia]

create img women

create img sad women

[opaque spear]

Recraft v3. I did not ask for that subtitle. I'm impressed and shocked.

[zinc portal]

Emma Stone 03 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

Bella Ramsey 08 - Original Upscale DeepFake FLUX Dev AI Art 1728 x 3840

[rapid saddle]

Kitchen

[steel haven]

livingroom

[frigid star]

HASSAN

AI PLESE CRATE BENTEN

[steel haven]

lol.

Stop yelling at the AI, it makes you look silly.

[graceful lark]

Boxers from @royal matrix (Flux)

[slow harbor]

Wassup Beijing

[swift sand]

recep ivedik

[green hamlet]

Crystal (Kittydog) playing sports.

[graceful lark]

Aah look cute

[royal matrix]

Yay I can finally upload :3
Here's some of my Flux Pixelwave gens from the past couple of days. No loras, just prompt generations in Forge.

[graceful lark]

Hooray!

I'm surprised that this bad picture has a good animation + squish

[green hamlet]

Crush them, melt them, cakified, exploded, squish them, inflate them.

[vestal oxide]

dubstep

[idle nebula]

https://tenor.com/view/angry-monster-cloverfield-gif-18105349

[green hamlet]

Kanye bear on each Kanye West album, generated with Flux.1 on Weights.gg.

[safe summit]

Anime image 👍

[sacred valley]

kamala supporters be like:

[leaden oyster]

Ayo? @bitter walrus level 2 !!!

[green hamlet]

#1159290752195633273

The #🏙│ai-images is for sharing AI-generated images, not sending your own YouTube links.

[terse lintel]

Okay, I just thought that an AI generated video (which is full of AI generated images) fits this category. It's on YouTube but I could have just send the source file as well... But sorry about it.

[graceful lark]

Files - Yes
Link - No
Please remove the link and send to #1159290752195633273

@terse lintel

[tawny peak]

damn the eyes look really good in this one, my model btw still working on it

[lucid seal]

i felt like putting images of mewtwo into vidu cuz hes my fav pokemon

i put this here cuz the first one was made from an image generated by sd using perchance.org

ASH KETCHUM JUMPSCARE 😱😱
also yapping mewtwo on the other two lol

[steel haven]

3k

[tawny peak]

Which one is better, No hi-res fix (left) or hi-res fix (right)?

[alpine plover]

left one has wider colour palette, but the eye in right one is sooo much better and more clear. incredibly hard to pick

[green hamlet]

A hippo riding a fish.

[magic radish]

google imagen 3 is very good for human types but I can't get efficiency from flux or ia's like Stable Diffusion

I think it needs to be developed more, I think it may be possible to make this picture as a lora of aia's lora, but it feels like flux is not realistic in itself.

[noble osprey]

Which application would you recommend for video-to-video with a prompt? Like Gen2 of Runway but free and easy to install on PC with GPU.

[graceful lark]

I recommend asking this at #🔍│help-ai-art

[karmic belfry]

beauty girl

[snow cosmos]

Centepede dragon hybrid in the style of a yakuza back tattoo in black and white

[leaden oyster]

Ayo? @snow cosmos level 2 !!!

[snow cosmos]

Some cool generations I has

[magic radish]

model: google imagen 3

[graceful lark]

@shadow bane my loras (not all) :

[shadow bane]

awesome

very high quality loras as i can see, ai imperfections are very rare in your images as i can see

[graceful lark]

lol are you kidding? xdd
It's just an accident
it's called Best Moments That is, I chose only the best

[shadow bane]

super awesome to be ai

@graceful lark more images for the dataset = more accurate and better lora?

[graceful lark]

aaah I don’t recommend using more than 100-200 Images for pony training

For Flux Character or Concept You can use 20 But describe it in detail

[shadow bane]

what is pony training

[graceful lark]

Pony is a Model (Pony Diffusion [SDXL])

https://civitai.com/models/257749/pony-diffusion-v6-xl

[shadow bane]

ah yes mlp model

[graceful lark]

aah yes and no pony draws Very Well all Furries, Anime, Animals, Realism

[shadow bane]

yea i've seen a ai image of a female rat character in the lora images

[graceful lark]

All pictures are Pony Model

In any case, I'm sorry if I distracted you

[shadow bane]

fuck

my pc is getting super slow suddenly

[graceful lark]

welp my laptop 24/7 Very slow xdd

[shadow bane]

welp i have an athlon 3000g lol

im saving to add a rx 580

[graceful lark]

I have an old stone laptop :3

[shadow bane]

oh

you're training in google colab?

[graceful lark]

Yes + I Use Civitai for LoRA Training

In any case, you see the result

[shadow bane]

i used colab too for the stronghold lora i made

but idk how to test it out

like, generate images with the trained safetensors file

[graceful lark]

You can use sd on colab

https://civitai.com/user/vadigr123_/articles

[shadow bane]

send link

[graceful lark]

You can read Stable Diffusion on Colab and Test

[shadow bane]

i'll keep it in mind

[graceful lark]

https://colab.research.google.com/github/R3gm/SD_diffusers_interactive/blob/main/Stable_diffusion_interactive_notebook.ipynb

[shadow bane]

tho tysm, i really needed this

[graceful lark]

yeah np np

[cosmic spoke]

kemono art style lora i made today, pretty flexible (pony-sdxl based)

[quick moss]

🦊

[green hamlet]

Luce Fumo, generated on Weights.gg. Her eyes looked big on this one, might draw her later.

[graceful lark]

But it looks very cute! I like it

[cosmic spoke]

The trick was to caption literally all of the stuff present in the dataset and then use prodigy as the optimizer, then check tensorboard for selecting an epoch (like rvc)
i trained for 1500 steps

[graceful lark]

Aah, this is the path you chose 😮

I have a dumb question
will you publish this model? 😮

[cosmic spoke]

Yes for style loras you have to caption everything, after that u have to use a slightly larger dim/alpha and way more steps than a character lora

[graceful lark]

Yes but I only use 1 tag for Style xdd

[cosmic spoke]

Nop

[graceful lark]

https://cdn.discordapp.com/attachments/1154385802936201226/1294982019478061086/mikus_cute_-cry.gif

[cosmic spoke]

Not captioning the lora have somewhat a positive quality effect in the final result but this causes the lora to not be as flexible like an one that has everything captioned

[graceful lark]

To be honest, I tried to do this (I think it was yiff v36 sd 1.5) but it didn’t work (I signed + AI Corrected)

[cosmic spoke]

like for example if u try to generate an species not present in the dataset is going to look weird

[cosmic spoke]

probably undertrained/overtrained

[graceful lark]

Mm maybe

otay Then you can show which author it is or Parse the full Dataset so that I can roughly understand how it should work (welp, I know that I’m a bad Helper)

[raven zealot]

Hello

[graceful lark]

hewwo

[cosmic spoke]

Is not hard, if you want to train an style just caption everything you see using e6 tags, for example if your image is a orange cat sitting outdoors then you have to caption it like this: activation token, cat, orange fur, sitting, feline, etc

[graceful lark]

activation token?

[green hamlet]

It might not be that hard, but I'm hard.

[cosmic spoke]

The activation word, the word u put first first in the captions

[graceful lark]

aah yeah

Trigger word I call it

[cosmic spoke]

[graceful lark]

welp I will try Although to be honest I doubt that I will succeed Good LoRA xdd

[cosmic spoke]

For style use around 50 images

[graceful lark]

50?! But is that too little?

[cosmic spoke]

My style lora is 37 images

[graceful lark]

https://tenor.com/view/loading-wtf-ruocgiff-gif-25631449

[cosmic spoke]

[cosmic spoke]

net dim was 16
net alpha 8

prodigy, both lr set to 1.0

batch size 1

1500 steps, 1 repeat (i dont train using epoch amount but for step amount, like rvc)
edit: after this i find out i should have trained for more steps

[graceful lark]

otay

ty for hewp! ❤️

[cosmic spoke]

No problem, u can ping me if you need some help

[graceful lark]

and sowwy 😞

[cosmic spoke]

You didn’t do anything bad, don’t worry

[graceful lark]

Are you guys too suspicious.. 🏳️‍🌈?

Hm??

[keen bramble]

funny enough i wanna make one of my vrc avatar

but not sure about the angles and stuff bc last time it didnt look good

[graceful lark]

I could try to do it for you Although as you can see I'm bad at this

[cosmic spoke]

dataset has to have variety, similar principes used in rvc training

tho the lora will have the vrchat art style, u can't remove it no matter the captions

[cosmic spoke]

different poses and angles

for characters try 30 images or slightly more

also character loras require less steps and lower dim/alpha

1 repeat

[keen bramble]

my shaders make it look kinda anime like

I can go to worlds that look like 2D images too

[cosmic spoke]

i'll try to make a lora of my vrchat character

[brittle ocean]

18🤑

[tawny peak]

For me, when I made sd 1.5 loras i did 10 epochs and 10 dataset repeats, and for sdxl I do 20 epochs and 10 dataset repeats and i’m able to get really great results without having to lower the weighting at all when generating

i’ll have to check what my network stuff is later as i can’t remember

[tawny peak]

Doesn't work

Though you can just use the GPT4 bot in #🤖│bots

[cosmic spoke]

i was able to remove the vrchat art style by tagging 3d_artwork in the dataset

[graceful lark]

@shadow bane cute
@sullen ingot pwease del msg and send -> #1159290752195633273

[shadow bane]

why tho

[leaden oyster]

Ayo? @shadow bane level 2 !!!

[graceful lark]

I'm sorry, I don't mean for you to delete, I mean another user

[shadow bane]

ah alr

[green hamlet]

I found a Flux LoRA model of a dog someone said anyone could use for free on Weights. So I gave it a try with autumn-themed calendar photo-type image. I'm not sure how to make light shading looking good, so I'd leave it there.

[green hamlet]

So there's an event going on with Weights. This event is called "the cake". I made the same as above, but now with a cake, indicating that Weights turned a year old.

[rigid horizon]

mushrooms growing out of the stones

[uneven path]

Oh nice images from where did u create this?

[leaden oyster]

Ayo? @uneven path level 1 !!!

[uneven path]

yeah u can even say m not on any level

[lucid seal]

blue slushy cookie from cookie run generated using flux with a lora i trained on weights.gg

i think i'll avoid flux for a while, it is impressive with its ability to generate text that isnt weird gibberish, but its very janky (i like ai jankiness but this is too much for me) and there aren't a lot of decent character loras like stable diffusion
maybe with time it'll improve

[shadow bane]

illustriousXL is pretty good imo

[tawny peak]

NAI-XL, bit hard to understand, super good outputs, this is just a lowres sample tho

[lucid seal]

the closest ive ever gotten to generating an accurate mewtwo in flux
for context when you try to generate an image of a pokemon that isnt pikachu or eevee it just makes a generic creature

[tawny peak]

Does weights support lora?

[lucid seal]

yes

[tawny peak]

have you tried finding a mewtwo lora to try out?

[tawny peak]

I'm really liking this model

[worthy valve]

generate an apple

[alpine plover]

https://tenor.com/view/applesiftheywereblue-blue-apples-apple-blue-perfect-world-gif-1668985994174720221

[cosmic spoke]

LoRas of my webfishing sona and vrchat avatar
(trained with in game screenshots)

[graceful lark]

Ooh he will look very cute!

[cosmic spoke]

they look like this in their respective games

[graceful lark]

😮

[cosmic spoke]

20 images aint enough for a lora it seems, i will add more images to both datasets to see if i can improve them

[graceful lark]

Ooh welp then I'll be looking forward to New Art!

[green hamlet]

Kanye dressing as Reimu Hakurei.

[shadow bane]

mimi is the best aint she

[fair heath]

How i can do this things

[graceful lark]

Yes https://www.weights.gg/ Does its job well

[leaden oyster]

Ayo? @lucid seal level 3 !!!

[lucid seal]

more stuff featuring everyones favorite anthro psychic cat pokemon
last one is spoiler-tagged due to censored nipples (i didn't put that in the prompt it just generated that out of nowhere)

[tawny peak]

as a note for the future, just don’t send things you’d have to censor

[tidal terrace]

Is ur pfp Five night freddy

[green hamlet]

Some AI generated images of "Moo Deng the baby hippo" that I found anywhere, even with unofficial merchandises that being sold in my country, don't look like her at all. So I had an idea that I should make her into a Flux LoRA model on Weights instead. 🦛

[spark swallow]

A creative and professional image featuring the word ' Tremoa 1001 Ouled Taima ' formed using various automotive mechanical tools such as wrenches, bolts, gears, and screwdrivers. The tools are intricately arranged to shape the letters in a realistic way. --size 1:1 --sref 0 - ayoub033621 (Fast) - in process 41%

[graceful lark]

https://tenor.com/view/loading-wtf-ruocgiff-gif-25631449

[green hamlet]

What?

[green hamlet]

Moo Deng getting trapped inside the Backrooms dimension.

[green hamlet]

Please test your keyboard on Notepad, not here.

[shadow bane]

make a model of Redgear Pro Wireless Gamepad with 2.4GHz Wireless Technology, Integrated Dual Intensity Motor, Illuminated Keys for PC

[green hamlet]

This is not where you generate an image model for use.

[shadow bane]

then where to genrate?

[green hamlet]

Ping @ chatgpt4o at #🤖│bots and tell it to generate some images for you. Unfortunately, there's no option for generate an entire image model for use now, but you might wanna get some dataset images to train one instead.

[safe summit]

Kind sir would you share which lora, model and prompt was used for the rail tracks image

[small edge]

It was Ukraine and Russia border at night

[tawny peak]

no character loras lol

[fluid escarp]

DnD shenanigans

[green hamlet]

Drake trying to dress as Reimu, generated on Weights. 💀

[tawny peak]

[tardy sapphire]

#🏙│ai-images dog

[safe summit]

👍

[burnt axle]

Viral anime nature wallpaper in 4K quality, in the style of photography with a 35mm lens inspired by Isao Takahata, capturing a serene rice field at dusk, the sky a soft gradient of orange and purple, and fireflies dancing above the water; warm and nostalgic color temperature, cosmos lighting with stars twinkling in the sky, no human characters, the atmosphere is tranquil and dreamy --v 5 --stylize 1000

[green hamlet]

Ping ChatGPT4o bot at #🤖│bots and it will make an image for you.

[faint oyster]

a

[gusty comet]

BigBang

[graceful lark]

https://tenor.com/view/mashle-mss-mashle-opening-bling-bang-bang-born-creepy-nuts-gif-1188046021999379027

[fiery atlas]

What ai tool is best for creating ai model?

[tawny peak]

Onetrainer or Kohya-ss

[fiery atlas]

Fooocus and rendernet?

[green hamlet]

What AI model? Image or voice?

[tawny peak]

wdym rendernet? that a closed source program.

To my knowledge you can't make models for that

Onetrainer and Kohya-SS can both make models for fooocus because it's just a stable diffusion frontend

[fiery atlas]

Okay, thanks

[leaden oyster]

Ayo? @fiery atlas level 1 !!!

[fiery atlas]

What?

[hard fog]

pretty sure SD/flux

[fiery atlas]

You have a link?

[unborn sinew]

Burning shrimp
Generation time: 3.2 Seconds

[flat sentinel]

Yum yum

[green hamlet]

🔥🦐

[smoky cape]

Here's how the neural network processed my request to generate an image so that Mike Pollock, the voice actor for Dr. Eggman (2003-present), would look more like the character he's been voicing for a long time

[unborn sinew]

prompt: Minion on a supercar
Generation time: 3.43 seconds

[hollow basin]

Grimmace gold chain

Prompt: grimace with a gold chain

[harsh moss]

Check this out if you only have ChatGPT Plus! https://chatgpt.com/g/g-6748694a7be88191b7793d7d511f53d9-animals-generator

Hello @graceful lark

[graceful lark]

hewwo!

[graceful lark]

What should I write?

[harsh moss]

Well, first of all, do you have ChatGPT Plus? If yes, then you can write anything to generate any animal-related image

[graceful lark]

otaaaa...

[graceful lark]

tada!

[harsh moss]

Nice!

[graceful lark]

Although honestly I would recommend using SD pony xdd

[harsh moss]

Really, ok, I will give it a try, but I don’t know how to install it locally on my PC and neither I have the knowledge to do it

[graceful lark]

Hmm Hmm I think the professional at #🔍│help-ai-art I'm bad at localy bc I use colab or CivitAI for this

[unborn sinew]

gimme the prompt

I want to try it on my custom trained Model

[graceful lark]

A 3D animation-style anthropomorphic mouse with white fur, yellow ears, and a yellow tail sitting on a bench in a serene park surrounded by lush green trees and vibrant flowers. The mouse is wearing a black sweater and gray shorts with a subtle mini bulge. It is holding a cup of bubble tea, with a relaxed and content expression as it feels the cool glass against its paws. The tea is creamy and sweet, with visible tapioca pearls. The scene is illuminated by warm sunlight, casting a comforting glow over the mouse and enhancing the tranquility of the park setting.

[unborn sinew]

Thank you!

[graceful lark]

np np! good luck! ❤️

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1311720204497322105/image_1.webp?ex=6749e223&is=674890a3&hm=29df47cf093c11588e7c7feab846b3e4de72beaf477915d2b68ba76511dd19ff&=&format=webp

I love generations in 2 seconds

[graceful lark]

aah looks cool

[unborn sinew]

they go hard

Yours is more on details ig

and the mouse is a bigger

[graceful lark]

It seems I created this model for gpt4 @harsh moss I think

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1311700326981701652/image_1.webp?ex=6749cfa0&is=67487e20&hm=c83736dfe3848d24a56b9c3a9b1d276cce4981029f58abbaa797ffcbfe3ced28&=&format=webp

ayyyy

[unborn sinew]

cool cool

[graceful lark]

Although I also create mini models

[unborn sinew]

Nice nice

[graceful lark]

In any case, good luck with the project!

[unborn sinew]

Thanks! I want to upgrade the GPUs that the model uses for faster results

:D

https://media.discordapp.net/attachments/1306974853852827699/1311721393745952938/image_1.webp?ex=6749e33e&is=674891be&hm=8e6465fe18b21c8726ee149518abecd364ec3c78807ba4c4baaf0ddd45e2fe50&=&format=webp

Generation Time: 4.31 seconds
prompt: A vast, icy landscape under a twilight sky, where a colossal glacier juts out over a frozen lake. The glacier’s intricate textures and luminous blue veins are perfectly mirrored in the glass-like ice below. Overhead, the aurora borealis dances across the heavens, its vivid greens and purples echoed in the shimmering reflections on the frozen surface. Nearby, an explorer in a high-tech suit kneels, their helmet light casting a glowing circle that disrupts the pristine reflection, adding a human element to the surreal environment.

[graceful lark]

New LoRA ~ new testing

[graceful lark]

Good morning!

[graceful lark]

LEEEGO

[green hamlet]

Red Angry Birds plush floating in an ocean.

[prisma ruin]

Me & Natsuki - A.I Generated on Weight's!

Another more I generated image I try 2× attempt getting rid for this a prompt, and I did it..!!

Look Compare for images!

• First Smile like Me Goofy & Monika her mouth just lips and shadow.
• Second Not Smile cause unhappy & Monika her lips are too huge lustly!

[graceful lark]

@sullen ingot -> #1159290752195633273

[harsh moss]

Hey @graceful lark , I have updated my GPT: https://chatgpt.com/g/g-6748fb30b0f0819196e9a21e8a9caa3b-animals-generator-exaggerated, to be more exaggerated

[leaden oyster]

Ayo? @graceful lark level 54 !!!

[graceful lark]

huh huh

[harsh moss]

I have two versions:

1. Non-Exaggerated: https://chatgpt.com/g/g-6748694a7be88191b7793d7d511f53d9-animals-generator
2. Extreme Exaggerated: https://chatgpt.com/g/g-6748fb30b0f0819196e9a21e8a9caa3b-animals-generator-exaggerated

[graceful lark]

[harsh moss]

It's cursed, right?

[graceful lark]

Aah, did you want it to look like a Furry or something? 😮

[harsh moss]

No, no, what I'm saying the GPT itself creates "cursed images", right? Like the image you shown me on the right

[graceful lark]

Yes, on the right it looks cursed

[harsh moss]

@graceful lark, how do you even create these images? What do you use?

[graceful lark]

Stable diffusion!

All I'm sending here are all my Stable Diffusion Pony Models

[harsh moss]

Yeah, but what kind of software / website you're using to generate these?

[green hamlet]

He might either uses Forge or ComfyUI to do these.

These are the most popular Stable Diffusion UI forks that use 1.5/XL/Flux models to generate. Other than this, I'm not sure about other ones.

[hard fog]

he might be looking for some cloud sites, since he only wants to spend money on those services instead of saving for gpu upgrade to do locally

[graceful lark]

https://civitai.com/ or colab + u can use localy

[unborn sinew]

First public Video of my text to video Model.

Quality is not very good but I will fix that in the future.

[leaden oyster]

Ayo? @analog tree level 1 !!!

[analog tree]

require('@tensorflow/tfjs-node'); //running tfjs on laptop cpu
const tf = require('@tensorflow/tfjs'); //require in Tensorflow JS lib
const Users = require('../../mongodb/models/users')
const Posts = require('../../mongodb/models/posts')

function knn(features, labels, predictionPoint, k) {
features = tf.tensor(features);
labels = tf.tensor(labels);
const {mean, variance} = tf.moments(features, 0);
const scaledPrediction = predictionPoint.sub(mean).div(variance.pow(0.5)) //Scaled Prediction const for standardization
const apple = features.sub(mean)
/* Step 0 - Standarization of features for /
.div(variance.pow(0.5))
/ Step 1 - Find Distance between features & prediction point features /
.sub(scaledPrediction) //Broadcasting operation
.pow(2) //Elementwise operation to square each element
.sum(1) //Sums along x(1)-axis
.sqrt() //Elementwise operation to take power of .5 = sqrt of each element
/ Step 2 - Sort from lowest distance to greatest distance /
.expandDims(1) //We expand dimensions of distances tensor across the x-axis to get the shape of [4,1], the same as the labels distance
.concat(labels, 1) //We concatenate labels to distances across the x-axis so they are linked by the same indices in one tensor
.unstack() //We unstack our 1 tensor into 1 Vanilla JS array containing multiple tensors
/ After unstacking our tensor, we are dealing with vanilla JS array and we can ONLY use vanilla JS methods from this point on /
.sort((a,b) => a.get(0) > b.get(0) ? 1 : -1) //Sorting function tp sprt tensors in order of least to greatest distance
/ Step 3 - Average the label value of the top k records */
.slice(0, k)//Get Top k records

return apple
}
module.exports = knn

[hard fog]

no one here uses tensorflow and js

[green hamlet]

Please #1159290752195633273 your YouTube channel at #1159290752195633273.

#🏙│ai-images is for upload AI generated image, not for promote your video or account.

Kendrick Lamar standing in front of someone's house.

[graceful lark]

@safe summit new lora:

[safe summit]

wow that was very cool👍

[shadow bane]

Crie um rato

Make a fish

[graceful lark]

cuti2

[unborn sinew]

Real

[graceful lark]

https://tenor.com/view/peashooter-pvz-plants-vs-zombies-plants-zombies-gif-6998445874810870689

[graceful lark]

I'm too thoughtful to understand you

[leaden oyster]

Ayo? @unborn sinew level 5 !!!

[unborn sinew]

No comment.

[latent halo]

jett

[graceful lark]

@shadow bane

[alpine plover]

lmao

can you do neco arc?

https://tenor.com/view/cat-cat-with-tie-gif-11241827778194923856

[graceful lark]

mimi neko?

[alpine plover]

this

[graceful lark]

yeah but I mean the character or what? 😮

I'm confused

[alpine plover]

like use this lora on neco arc

to get a sad one

sad cute little racist cat

[graceful lark]

aaah

[alpine plover]

i'm just stating facts

[graceful lark]

u mean this? or other (wait)

[alpine plover]

yea yea this

ty

very cool

damn bro do you have some kind of tutorial on how to create these loras?

[graceful lark]

or this?

[graceful lark]

yeah wait

I only have how to create LoRA via colab
https://civitai.com/user/vadigr123_/articles

[alpine plover]

ty, will check it out on vacation

https://cdn.discordapp.com/attachments/1103016321211453471/1298006074074333226/catdance.gif

[graceful lark]

Although you can do it localy But you will need to ask other helpers

[alpine plover]

no problem, i'll figure it out

[graceful lark]

otay!

Good luck with that

@alpine plover I hope you don't look like that when you do it

[alpine plover]

don't underestimate me

https://tenor.com/view/bernardson-silly-goober-silly-cat-gif-3512342629753561110

[graceful lark]

I just wanted to send this picture xdd

[alpine plover]

send them more

[graceful lark]

u just chill neco?

[shadow bane]

i'm forever grateful now.

[coarse leaf]

so colorful I'd have to say

[jolly jay]

hentai lebanese

[alpine plover]

borga

https://tenor.com/view/cat-kitty-tired-sleepy-relaxed-gif-7257672

[jolly jay]

https://tenor.com/fITBLLf71Er.gif

[keen bramble]

average workflow

result

[alpine plover]

daaaaamn

want to export and upload here?

[keen bramble]

playing around with my OC

[keen bramble]

such a clusterfuck it will be hard to explain

[alpine plover]

i mean you dont need to explain anything, just export json and upload it here

[keen bramble]

arcane reference

[lucid barn]

@graceful lark can u create smth for me?

[graceful lark]

aaah otay?

I'll try but it depends what you want

[lucid barn]

can u do a villager or pomni?

smth like that yeah

[graceful lark]

otay?

but what should a pomni or villager do on art?

[lucid barn]

that little up thingy u send here, but both do it

[graceful lark]

ahh

[glass marsh]

yo

can anyone help me generate an ai image

i have this ai image already but i need it to be bigger using an ai tool but idk how to do that

[lucid barn]

ask poyo

[glass marsh]

where can i find him

[lucid barn]

he's right here

[glass marsh]

poyo

[glass marsh]

poyo

can u plz help me

[graceful lark]

huuuh huhh?!?

mb u can use #🔍│help-ai-art?

@lucid barn i saw

[glass marsh]

but theres noone talking in there

[lucid barn]

saw what?

[lucid barn]

oh

my deleted message

[graceful lark]

sowwy I'm a little busy right now so ask this question at #🔍│help-ai-art later I'll look at otay?

[glass marsh]

u cant just do it? im tryna get this tattoo tmrw and i need to get the pic ready tnite

😞

[graceful lark]

pweease

30m

[glass marsh]

ok but plz dont forget about me bro

[graceful lark]

otay! describe in detail how I can help, I’ll try to do it

[lucid barn]

don't forget abt me too 😞

[graceful lark]

@~@

otay otay

waait pweease

[lucid barn]

we will

[graceful lark]

@lucid barn

[tawny peak]

[mental ice]

Damn

Why did you change the image generation? This is frustrating!

[tawny peak]

Wdym change?

[mental ice]

Well, before you created a specific model of real artists and the image generated in Weights had realism and unique details. Now if I generate something completely generic and boring it will appear.

WEIGHTS, GO BACK TO WHAT YOU WERE YESTERDAY.

[tawny peak]

I don't control that so not sure how to help you.

[mental ice]

It's just to generate greater visibility. Maybe they'll fix this soon.

[green hamlet]

This character name is Binary. Generated on Weights.

A side note. ||This character came from an NSFW artist, unfortunately. I had to manually edit some freaky stuff out from him||, so I could train and publish this model entirely SFW on Weights.

[merry spruce]

#🔍│help-ai-art

[green hamlet]

Binary holding a hamburger.
https://generated-images.weights.gg/cm4cmv7nl1i0bmk18kz5dk7yq%2Flora_image_1.webp

He even holds a double stacked of cheese and beef.

[vivid mango]

i made dante dmc on weights

https://www.weights.gg/loras/cm4dlpbxm0bnko818eej7v46y

[safe summit]

flux loras on SD 3.5 large

[jade vine]

an absulute potato

[green hamlet]

https://generated-images.weights.gg/cm4ezvf140fa6ke18h5hfjmwk%2Flora_image_1.webp

Moo Deng inflatable balloon. 🔥

[gentle minnow]

Mad dog of Shimano as an anime girl#

[green hamlet]

This is not where you generate image, damn.

Ping the ChatGPT4o bot at #🤖│bots and it'll generate an image for you light purple name.

[frigid star]

Hi

[green hamlet]

Is there anything we can help? You may start a conversation at #🧬│ai-chat for chat or get some help at AI image there at #🧬│ai-chat.

[primal kiln]

hawk tuah

[green hamlet]

https://tenor.com/view/skibidi-toilet-spiderman-skibidi-gif-11268458470528183028

[daring jewel]

what thats actually good

what website is this?

that your using?

[unborn sinew]

Prompt?

No clue what my AI model did

[unborn sinew]

Well thats all for now.

[zinc portal]

Marty McFly 04 - Original Upscale DeepFake FLUX Dev AI Art 3840 x 2176

Doc Brown 06 - Original Upscale DeepFake FLUX Dev AI Art 3840 x 2176

[green hamlet]

I didn't make this. Someone used my model on Weights to do this. https://generated-images.weights.gg/cm4hdtcfg01a5mj18drnc6vnl%2Flora_image_1.webp

I made this one. https://generated-images.weights.gg/cm4i4q6jq0ipxm718ni9spz9u%2Flora_image_1.webp

[safe summit]

https://civitai.com/posts/10136526?returnUrl=%2Fmodels%2F796382%3FmodelVersionId%3D1026423

[safe summit]

guys i have 1.2m (18gb all jpg) images i want to train a flux or sd3.5 model how can i do it on the cloud

and how can i run clip on them ?

[tawny peak]

rent a gpu

[unborn sinew]

Why is this AI almost the same name as mine 💀

[shadow bane]

oh yea i've seen it on weights.gg lol

and actually it's shivertermal's model lolz if you're not joking

[green hamlet]

Yeah, that's the same model used for those images.

[safe summit]

?

[unborn sinew]

I named my model CivilityAI and this website is named CivitAi

Crazyyxx

[safe summit]

lol

CivilityAI is a finetune of which model ?

[unborn sinew]

Its completely custom trained

[safe summit]

there is always a base model

[unborn sinew]

[safe summit]

companies like openai and stability are spending millions to make base models how did u do it? bro u are a genius

[unborn sinew]

Prompt: "A sleek car driving down a scenic mountain road during sunset. The car’s polished surface reflects the warm orange and pink hues of the sky. The road winds through lush green hills, with occasional glimpses of a sparkling blue lake below. The sound of the engine hums smoothly as the car glides past, and the tires grip the asphalt effortlessly. The atmosphere is dynamic yet peaceful, showcasing the thrill of the drive."

Generation time: Around 10 mins

prompt: "A busy city street during a sunny afternoon, filled with people walking in various directions. Some are dressed casually, carrying shopping bags, while others are in business attire, focused on their phones or briefcases. A child holds a parent's hand, skipping cheerfully, while an elderly couple strolls leisurely, enjoying the day. The scene is alive with movement, chatter, and occasional laughter, with the backdrop of tall buildings and bustling storefronts."

Generation time: Around 8 mins

[unborn sinew]

Bruh why is discord failing to load the videos

[graceful lark]

my silly lora

[worthy flare]

@graceful lark goldfish

[graceful lark]

https://tenor.com/view/fish-gif-7422720

Otay

[wind gate]

Anyone know what's the best AI that I can use to make these fanmade movie trailer thumbnails.

Preferably something free.

[graceful lark]

flux is free if you use localy :3

[unborn sinew]

Flux can do Texts?

When I tried it it couldnt do it

[graceful lark]

ahh wait

video can create lumalabs.ai
flux can make a nice "cover" (thumbnail) for Yt Or initial image for luma

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1316753044893073438/image_1.webp?ex=675c3155&is=675adfd5&hm=7e7c9c3c1605e3ded2ac83325e69f08908949265c23c729e8b7bb2457393e77f&=&format=webp

[unborn sinew]

Looks cool!

https://media.discordapp.net/attachments/1306974853852827699/1316753808411594792/image_1.webp?ex=675c320c&is=675ae08c&hm=e1c08682e37fec252f5bbe96d40b83866906806a6efeb5735738c1e09a36d7f5&=&format=webp

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1316768186074664970/image_1.webp?ex=675c3f6f&is=675aedef&hm=96374eafc8dc7cca42de498a7f007b711472ccb97c60d112589d977a76ad9a1a&=&format=webp

https://media.discordapp.net/attachments/1306974853852827699/1316769150160601098/image_1.webp?ex=675c4055&is=675aeed5&hm=1e7debe4639830adfe6869eab0432108ae408254931c926fbafb516f97b203a0&=&format=webp

[graceful lark]

@hushed wasp -> #1159290752195633273

[zinc portal]

The following images are sensible.
I ask the admin not to delete them, they were made out of simple curiosity to see the potential of Grok 2 (AI From page X (Twitter))

It should be noted that the images are scaled x2 with FLUX.
Original image of Grok 2 is 960 x 720

Donald Trump and Paul McCartney 01

Mao Zedong and Michael Jackson 01

Obama 01

[lucid seal]

i made a lora of mewtwo, specifically the one from one of the pokemon movies, the generated images look a bit inaccurate but you can edit and draw over them if you'd like
https://www.weights.gg/loras/cm495h40b0uoknj180g0ial2k

also try out my other mewtwo lora, based off of its appearance in smash bros ultimate (generates 3d render styled images)
https://www.weights.gg/loras/cm3nt2le3003ev7918c5ah9hu

[unborn sinew]

https://media.discordapp.net/attachments/1201176761422061678/1317114026089517216/image_1.webp?ex=675d8186&is=675c3006&hm=c39bfa96bd7f10a97b18b01b5ca19f69b35a88fac66c5e2e203c00bad544aa9c&=&format=webp&width=960&height=480

https://media.discordapp.net/attachments/1306974909347663972/1317115416396431380/image_1.webp?ex=675d82d2&is=675c3152&hm=ec705766f1812bb938c12403d0e194398a5051508d305c8658bc7e3a041f070d&=&format=webp&width=960&height=480

https://media.discordapp.net/attachments/1306974945963671614/1317117434691584040/image_1.webp?ex=675d84b3&is=675c3333&hm=dde89fab759c7e7ed08ca5a0e42c2969aa9a0ae0eead8b50343828e9e6b98b23&=&format=webp

https://media.discordapp.net/attachments/1306974945963671614/1317120111882010696/image_1.webp?ex=675d8731&is=675c35b1&hm=1a547a4b0ea233c10eb8a4e2a46c00d0aad690ce330eaca5ffa0421838b5e2c4&=&format=webp&width=960&height=480

[unborn sinew]

Prompt: "A sleek, futuristic supercar with a glossy metallic finish racing down an empty highway at dusk. The road stretches endlessly into the horizon, flanked by open landscapes and glowing streetlights, with a dramatic sunset casting vibrant colors across the sky.

Generation time: Around 3 mins

Not the best result

Its alright

https://media.discordapp.net/attachments/1306974945963671614/1317134096337600572/image_1.webp?ex=675d9437&is=675c42b7&hm=c7e6a5076259c4acbbfc5b3eaf818cb4d80e8e18fb891bcf058b4eefcad76f01&=&format=webp

Generation time: 2.5 seconds

[unborn sinew]

cool

[unborn sinew]

https://media.discordapp.net/attachments/1306974945963671614/1317149194921836624/image_1.webp?ex=675da247&is=675c50c7&hm=b5ee46e7bff5dd78403b5052207ba6de4d11169c68ee6819abaf0fa7568371e5&=&format=webp

[graceful lark]

aaah forgot to add that this is spongebob style*

Although I think it's clear

[unborn sinew]

hmmm

https://media.discordapp.net/attachments/1306974945963671614/1317150126669697044/image_1.webp?ex=675da325&is=675c51a5&hm=500e43c43792b1db9dce8a9bc45a2c51f85b76ceb47b4d0695ac473098cdb2b8&=&format=webp

holy spongebob

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1317151023437185204/image_1.webp?ex=675da3fb&is=675c527b&hm=697f70dfff3304600efe4ae865fce6b6e327e278e1e34cd9492f01282f9ced41&=&format=webp

Prompt: Burning shrimp
Generation time: 2.3 seconds

https://media.discordapp.net/attachments/1306974853852827699/1317152195627581591/image_1.webp?ex=675da512&is=675c5392&hm=af4dd9488a79f0e3eaf1d1a7628f9c40f05be236daa9c80f5a2e368f9a1103cf&=&format=webp

[final thunder]

merry xmas from pikachu and me ^^

[unborn sinew]

D:

Pikatchu :D

[final thunder]

yeah

[shadow bane]

Do I need a special role to post images here?

[final thunder]

no?

[shadow bane]

there's no upload file button or anything

[unborn sinew]

oh yea

you need to reach a level

[final thunder]

ah

[green hamlet]

Talk a bit, and you should be able to send image here.

[unborn sinew]

yea

[shadow bane]

oh okay thanks ^^

[final thunder]

i can post pics

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1317152550004199454/image_1.webp?ex=675da567&is=675c53e7&hm=9dc8c9a5a498587e03ba8ab05fccf2012416b6ec27f54bf000d46db8674f1210&=&format=webp

I got it a nice room :D

[final thunder]

OO very niceu

[unborn sinew]

1 min

[final thunder]

yeah?

[unborn sinew]

I will add pikatchu

[final thunder]

can i add u on here sky?

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1317152885510901780/image_1.webp?ex=675da5b7&is=675c5437&hm=2e847da4257f4a4352b64626fe5810909e3db2e7fc567cabdc31a86edaf4436a&=&format=webp

[green hamlet]

The Grinch dressing as Arona from Blue Archive.

[unborn sinew]

sure ig

[final thunder]

there 🙂

[unborn sinew]

Accepted

[final thunder]

ty

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1317153355122085969/image_1.webp?ex=675da627&is=675c54a7&hm=4e85e995c5c81ef95271dbd324f6d3d1b9d391d2668970aaf0a0f758cb9fd416&=&format=webp

[shadow bane]

I can post stuff now yay

[unborn sinew]

yippie

[final thunder]

there ^^

[unborn sinew]

not bad

https://media.discordapp.net/attachments/1306974853852827699/1317154922122444861/image_1.webp?ex=675da79c&is=675c561c&hm=5376b92b9a096c4d108c0a7bb7c4e36a730d4abdd19256726868883cf1497538&=&format=webp

[graceful lark]

Thank you! Merry xmas to you too! ❤️

[unborn sinew]

not bad

[graceful lark]

https://civitai.com/models/1033924/spongebob-squarepants-or-style-pony-and-illustrious

[unborn sinew]

pay to generate Images?

wtf

[graceful lark]

mmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm yes and no

You use local currency to use their server to create (you also have the choice to donate to buy buzz)

[night vortex]

hi

[unborn sinew]

hi

[unborn sinew]

Whats the prompt? :D

[shadow bane]

Well technically it's my own flux finetune so... idk what it would make on the normal flux model

This image is a high quality anime screenshot of a furry baby with long pastel purple hair and iridescent purple eyes. Her fur is purple with pink stripes. She is wearing a cat's bell collar. Sparkling purple background, sprite, character portrait. thick white eyebrows, looking at viewer, mischievous expression, She is sitting in midair. purple feetpaws, adorable, cute. one eye closed, single fang, 4k high definition, intricate details. Realistic scene, ultra detailed, lens flare. The mood of the image is one of innocence and cuteness. The image is a perfect rendition of a furry character. The character has small childlike proportions. She has a small fluffy tail to her right side. The image is sharp and in focus with no blurring. The image has thick black outlines. She is holding a popsicle and licking it in a catlike manner. Speech bubble with the text "Flux is nice, like this popsicle♥"

[unborn sinew]

I have my own custom trained model and I wanted to it xd

[shadow bane]

1920 x 1088 100 steps, flux guidance scale 6, seed is 1908857447

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1317272493349666826/image_1.webp?ex=675e151c&is=675cc39c&hm=f3ec95b9db857f3abaa18c295ad3357f1d1c16b674018e8bee2242e5fbf179fe&=&format=webp

[shadow bane]

XDD

close

[unborn sinew]

yea...

My main work is currently on text2video LOL

https://media.discordapp.net/attachments/1306974853852827699/1317154922122444861/image_1.webp?ex=675da79c&is=675c561c&hm=5376b92b9a096c4d108c0a7bb7c4e36a730d4abdd19256726868883cf1497538&=&format=webp

This image is insane ngl

[shadow bane]

it looks amazing o.o

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1313140283331838025/image_1_watermarked.png?ex=675d8d30&is=675c3bb0&hm=e00787d10295805e347548ac72532b5427c9f5060fa7dae7bcaa3529462d8492&=&format=webp&quality=lossless&width=550&height=275
https://media.discordapp.net/attachments/1306974853852827699/1311858744803721236/image_1.webp?ex=675d80e9&is=675c2f69&hm=02bdfd80d13b06a50eb37d87e4cd259b1b98400426d0d8c9287c65021a6a0eda&=&format=webp

[unborn sinew]

Thanks!

Feel free to share more of your art work ;D

I would like to see more tbh

https://tenor.com/view/crunch-cat-luna-eat-gif-3997701764820700221

[shadow bane]

idk what caused the weird color noise in this one

[unborn sinew]

well thats something "special"

[shadow bane]

o.o what

[unborn sinew]

It looks like a furry LOL

so its special

[shadow bane]

oh these are supposed to be furries o.o possibly

XD

[unborn sinew]

I see

I do realistic Images or fun stuff lmao

[shadow bane]

so far I've made a furry model and a chibi model... I guess the chibi one would be more useful but whatever

[shadow bane]

wow

[unborn sinew]

Flux can prob beat these results lmaoo

idk what this shit is

biggest fail ever

[shadow bane]

XD

[unborn sinew]

I like this Image because of the reflections

[shadow bane]

flux is surprisingly good at reflections

[unborn sinew]

Well when I tried flux...

It failed

idk how it is these days

[cosmic spoke]

is dis the artist that starts with p and ends with ton

[unborn sinew]

dam

[shadow bane]

o.o why would you know that?

[cosmic spoke]

i know every furry artist

[shadow bane]

well.... yes, maybe

[cosmic spoke]

[shadow bane]

it can't do NSFW o.o don't worry

I just think his cats are adorable

[unborn sinew]

[shadow bane]

don't judge me o.o it only makes fur

[unborn sinew]

It looks cool

[graceful lark]

Have you tried PonyDifftusion?

[shadow bane]

meh... i prefer having somewhat unique images

[cosmic spoke]

pony is overcook af

[graceful lark]

huh wdym?

[cosmic spoke]

overtrained

[shadow bane]

the p...ton guy would charge like...13,000yen for a commission and would take like over 3 weeks to make stuff. I mean his art is worth it, but then again I can make a cheap ripoff in a few seconds for free.

[unborn sinew]

well thats the power of AI LOL

and its unqiue every time

no matter what you do

[shadow bane]

I even bought his e-book and used it as part of training... I guess I paid for it so you can't say I'm stealing o.o

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1317285151952211988/image_1.webp?ex=675e20e6&is=675ccf66&hm=6d600a8001b26a472449e70c82b9d188a474764d33e846e94b168ffd98c1815e&=&format=webp

dang what a rip-off

[shadow bane]

cute XD

i can prompt it to make ponies too in the same style

[unborn sinew]

Well it only took 3 seconds to generate

so its alright

[shadow bane]

an extra ear o.o

[unborn sinew]

https://media.discordapp.net/attachments/1306974853852827699/1317286900591104080/image_1.webp?ex=675e2287&is=675cd107&hm=0fd9a5fcaba8c2a1c4c6b9068ecf48ba7800dc9d6e233dfce1686700dbf303b4&=&format=webp

[unborn sinew]

dang

[shadow bane]

nintendo 64 style

[unborn sinew]

https://media.discordapp.net/attachments/1308178673983295538/1317287328166969475/image_1.webp?ex=675e22ed&is=675cd16d&hm=d883ff8325938de15cabeaf5802fe0cdc5389028f02bc263bb50a0d39003ee4a&=&format=webp

[shadow bane]

those are very good

[unborn sinew]

thats v1 of my model.

I have v2 which is mind blowing

https://media.discordapp.net/attachments/1308178673983295538/1317287367064948776/AurWcGEPnqlZWD51QV0IDRahRa7vHtsaggqp4af8xwpRbVYWqwOgTEAAF3pB7d7mVDUO7rOOU7zynSeWM0UZPpyDBPEleiQHYnQvPuCb3XSWuqYfTrTnBohGUEhijEQwKzYRVwOfdIECToeFS5XsQZuwZGtS8ugKgAASCJBIQNRZ9WYgMaOngQS0AA7EDjLdzFt15VC7xVxsDKm9eAAAA.png?ex=675e22f6&is=675cd176&hm=dd239e83ef17402416d2c7296a487146a4b7716e7b9a8038ea8291c26cbb61c4&=&format=webp&quality=lossless

https://media.discordapp.net/attachments/1308178673983295538/1317287605322121286/B6ltVLe9ecOr6JBsBuu7QKBcsoPSQhM5hv70Fh05tf7ZNqEuJ9OERoOvlzIGhAkdPkHK7USQA0tdjnsrq0GuSWoK0S4ZpgVfAWQ9uV82N9gmeul8BHazs7tOzF3hdUrFlse6EzL07cMBU7389pUZ0tknSH77iQh5xeVoy44oNMA4YSadwSOdd8wK6ZHIEEqJeAf71VAAAA.png?ex=675e232f&is=675cd1af&hm=09fc62bb38a44597b5c84704fd7fd80d6f52b7fbaf5d5e0c7682dfd2a882ca37&=&format=webp&quality=lossless&width=550&height=309
https://media.discordapp.net/attachments/1308178673983295538/1317287148105498654/zefOH3qOpboQCnbFAjUzNfq8dkdhcTkjnVbtdcj5xWdvvnSsBtLMj8eTi0M7YScY7tP79Fghn68Zy1EE3MetykxKmR4UhcWlEi5NvmCAE7gfBzBxrUgdyZ174BgVneylD9HKmHPCBoWgAAA.png?ex=675e22c2&is=675cd142&hm=91ef3500e766945444c08d5ee8ef57a49bd38a1e552d014200f8accd74ca9db8&=&format=webp&quality=lossless

[shadow bane]

XD

[unborn sinew]

But thats not public yet

https://media.discordapp.net/attachments/1308178673983295538/1317288117350301766/image_1.webp?ex=675e23a9&is=675cd229&hm=fdd90b560a164cfc57ab87a125ceb8c478f3705f4b25d15ca0d7e991ac45ab83&=&format=webp

v1
Generation time: 3.4 seconds

[shadow bane]

for that speed, it's incredible

[unborn sinew]

ikr

https://media.discordapp.net/attachments/1308178673983295538/1317288911051030549/image_1.webp?ex=675e2466&is=675cd2e6&hm=422c56d21046edc1f556d088b9ac86ada1fd24bdadbc8ce6b6f64a5781223217&=&format=webp

Will I get copyright claimed for that?

@rustic ingot ???

[shadow bane]

I'm using full size 23gb flux models at fp16 precision, some images take a few minutes depending on resolution / steps

lol

[unborn sinew]

dang

flux seems to be slow lol

[unborn sinew]

https://media.discordapp.net/attachments/1308178673983295538/1317289927205064824/image_1.webp?ex=675e2558&is=675cd3d8&hm=612a2b2d72dcead7781351fcf2b39b7ac1eab5f3e37bced01f70e5613f9cf7e3&=&format=webp

Sorry if I send so many images of my Model. I can understand if its annoying over time xd

[shadow bane]

If I try to generate stuff rapidly, the quality definitely decreases

this is 8 sec

[unborn sinew]

lemme seee

https://media.discordapp.net/attachments/1308178673983295538/1317290401891356772/image_1.webp?ex=675e25c9&is=675cd449&hm=5e4c4c68756f237e6c80e95189532629777fa6a3aeed2cf707a54653d365046a&=&format=webp

3.21 seconds

Its not that bad

Its still good asf

[shadow bane]

only 10 steps

compared to like...

[unborn sinew]

mine does 4?

and HD mode does 7

[shadow bane]

maybe

I should use another base model for training x.x

[unborn sinew]

https://media.discordapp.net/attachments/1308178673983295538/1317291520465965086/image_1.webp?ex=675e26d4&is=675cd554&hm=a413ac52e478693f3c1ec9a97ce352da1ed33dad34df5c7aaf6ede0f7cb1e581&=&format=webp

[unborn sinew]

https://media.discordapp.net/attachments/1308178673983295538/1317291878110199889/image_1.webp?ex=675e2729&is=675cd5a9&hm=0b781068ff234b3ceed5334a76920bb1f71d6dfe3e96ede6771736cf947f4fa8&=&format=webp

[unborn sinew]

2.70 seconds

I feel like a poser 😭

[shadow bane]

i wish my outputs looked so good with so few steps x.x

[unborn sinew]

xd

https://media.discordapp.net/attachments/1308178673983295538/1317292644631838811/image_1.webp?ex=675e27e0&is=675cd660&hm=343e8e8daf45573b81a86514dcd66be495b4c5f9d5ee970f11af422001e87d39&=&format=webp

2.21 seconds

[shadow bane]

2.84 seconds but it's 512 x 512

[unborn sinew]

but thats cute

I like it

[shadow bane]

XD

[hard fog]

https://tenor.com/view/vtuber-hololive-ホロライブ-hakui-koyori-koyori-gif-25999227

[shadow bane]

:3

okay I'm done...for now >.< sowwy for the cat spam

[unborn sinew]

It was fun

Prompt: A volcanic eruption in the distance, with fiery lava shooting up into the sky in spectacular bursts. The lava flows down the sides of the volcano, lighting up the surrounding landscape in fiery shades of red, orange, and yellow. The ground around the eruption is covered in dark, charred earth, and the sky is filled with thick, swirling smoke and ash. The intense glow of the lava contrasts with the dark atmosphere, casting shadows and creating dramatic reflections on the wet, cracked surface of the ground.
Aspect Ratio: 16:9
Quality Mode: HD
Generation Time: 4.55 seconds
https://media.discordapp.net/attachments/1308178673983295538/1317293976096866315/image_1.webp?ex=675e291e&is=675cd79e&hm=41640a6177f229b5a2cca4e71a17a3002d51d67d50c8de73fded700607555659&=&format=webp

[graceful lark]

https://files.shapes.inc/f2f13a47.png

[unborn sinew]

Dang....

[tawny peak]

Wow, haven't seen this channel this active in a while

[shadow bane]

my fault >.<

[unborn sinew]

Grrrr

Why are you doing this!

[shadow bane]

meow

[unborn sinew]

Woof

[zinc portal]

Merry Christmas, Happy December to everyone in Chat.

[unborn sinew]

Noice

Flux dev