Stable Diffusion, NovelAI, Machine Learning Art - AI art generation discussion and image dump

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
After (what's the opposite of doomscrolling?) browsing through @ap_hamham on xitter, I wanted to see how Bing could do. I don't think GOTTE has gotte to worry.
1.jpg2.jpg3.jpg4.jpg5.jpg6.jpg7.jpg8.jpg9.jpg
0.jpg
 
  • Feels
  • Like
  • Winner
Reactions: I'm a Silly, TapewormSalesman, indomitable snowman and 7 others
The Mass Shooter Ron Soye said:
Stability AI, Makers of Stable Diffusion, Could Be in Huge Trouble (archive)
Stability AI CEO Emad Mostaque Resigns Amid Chaos at the AI Image Generator Company (archive)

Unprofitable company, CEO stepped down after researchers and execs left.

This company could be nuked from orbit and open sores image AI would be fine, but I'd prefer if they stuck around a little longer and burned more cash releasing stuff that the plebs can use.
Click to expand...
Forbes was rather damning in their coverage despite a slightly sympathetic tone, sounds like the facts are just too brutal:

So you say "Open Source image AI would be fine". I'm interested why you think that. It sounds like we're not even sure if version 3 of their model will be released which would be a big loss if it isn't, no? If Stability AI go down, what are our options for Open Source model generation? And how well could they really compete with the big proprietary versions? I'm not doom posting - I genuinely am not well up on the state of things but it seems to me that losing Stability AI would be a big deal.
 
  • Like
  • Agree
Reactions: woowie queen, Roland TB-303 and Vecr
Overly Serious said:
So you say "Open Source image AI would be fine". I'm interested why you think that. It sounds like we're not even sure if version 3 of their model will be released which would be a big loss if it isn't, no? If Stability AI go down, what are our options for Open Source model generation? And how well could they really compete with the big proprietary versions? I'm not doom posting - I genuinely am not well up on the state of things but it seems to me that losing Stability AI would be a big deal.
Click to expand...
Everything up to SDXL is released, I believe Stable Cascade is released, and they have promised to release Stable Diffusion 3. I have a feeling the models (they say it ranges from 800M to 8B parameters) would leak if the company went under immediately, but the company will stick around in the short term from what I gather reading the Forbes article. So there's a lot to play with.

I think the question is whether or not there will be anything after SD3. But quality seems to be plateauing and they are refining how it works with Cascade. The big trained models are what could take up to millions of dollars of cloud compute time to train, but the tools that use the models could be coded by anyone.

I make no promises about video or audio models which they are also working on. I'm sure those could benefit from many more years in the oven, and they are lagging behind ClosedAI's Sora.
 
  • Like
  • Thunk-Provoking
Reactions: Overly Serious, Mr.Miyagi and Vecr
  • Winner
  • Like
  • Feels
Reactions: I'm a Silly, anustart76, woowie queen and 2 others
ducktales4gameboy said:
ComfyUI question: Anyone have a good workflow for testing the same prompt vs a bunch of checkpoints, preferably with some compositing to combine and annotate the results?
Click to expand...
Ok, back at my computer, I've slimmed my workflow a bit to give an idea of what I'm doing. Lots of custom nodes, I think ComfyUI Manager will find them all.
The Save node will shove the extra data into the extra_metadata exif field.
 

Attachments

  • Informative
Reactions: ducktales4gameboy
  • Agree
  • Feels
Reactions: The Mass Shooter Ron Soye, woowie queen, anustart76 and 1 other person
even DeviantArt at it's peak did not contain the depths of depraved horrors I've witnessed by browsing CivitAI without any filters.

HORROR.png
horror (1).png
horror (2).png
horror (3).png
horror (4).png
horror (5).png
horror (6).png
horror (7).png
horror (8).png
horror (9).png
horror (10).png
horror (11).png
horror (12).png
horror (13).png
horror (14).png
horror (15).png
horror (16).png
horror (17).png

it legitimately has to be the most degenerate website currently on the internet. All of these models are only from the past month.
 
Last edited by a moderator:
  • Horrifying
  • Islamic Content
  • Informative
Reactions: Instant_Pot_User, Bungdit Din, Dededon't and 15 others
Nabol O'tyrin said:
even DeviantArt at it's peak did not contain the depths of depraved horrors I've witnessed by browsing CivitAI without any filters.

it legitimately has to be the most degenerate website currently on the internet. All of these models are only from the past month.
Click to expand...
It is a pretty funny contrast with the featured stuff on the front page, which is always squeaky clean. To their credit, CivitAI has done a pretty good job of setting up content controls and blurring to hide that shit if you don't want to see it.

Part of it is that PonyAI got very popular in the last couple months, and it is easily the best anime/porn/cartoon model for SDXL so far. Before that all the coomers were on SD1.5 still, and now they are switching over en masse along with all their degeneracy.
 
  • Agree
  • Like
Reactions: PureP, woowie queen and Train Operator
indomitable snowman said:
That toilet one cracked me up, but yeah I feel dirty going to civitai. Even the filtered home page is full of softcore crap. Just because your extremely busty cat eared waifu is wearing clothes, doesn't make it SFW or anything I want to look at.
Click to expand...
I disagree, the head is supposed to be coming out of the bowl, fucking philistines.
 
  • Feels
Reactions: Slav Power, woowie queen and Train Operator
You must log in or register to reply here.
Back