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

I did this image of Keffals with SD 1.4. He's kind of nightmarish as is, but I wanted something more stylized. I just put a Keffals photo as initial image and played with the prompt. For example: a portrait of a balding fat man with terrible hair, sweat dripping from face, by H.R. Giger. It's an amazing tool for sure.

Fucking hell man, you got me good.
Here are mine:
tmpybemcva5.png
tmp_2vyz9q6.png
tmphpgw1se3.png
tmpsn8xsqhq.png
And my personal favorite:
tmpcf73oxuz.png
Love the Chikatilo grin on this one. Cool machete, too.
 
There are several great things about this tech:

  • Mostly used by first worlders who have the money to buy the computing power
  • Requires some technical know-how, gatekeeps bottom of the barrel retards, also gatekeeps third worlders
  • No major entities control it, so development is controlled by anyone using it, and major updates are almost daily
  • I no longer have to deal with stuck-up artists who let their egos inflate their abilities
  • If said artist does not meet my specifications, I can generate 400 images and pick the best of the lot, it's very evolutionary and selective
  • I save money by using this instead of paying an artist to do something

@Null you might like this generation
 

Attachments

  • 1666060052567697.png
    1666060052567697.png
    424.2 KB · Views: 69
Last edited:
Human contact and conversation is unpredictable. Sometimes humans disagree with you (==being hitler basically)
sometimes humans say things that make you feel offended.

The fact that an AI would not be a perfect imitation of a real human is a feature not a bug. And that is why people will prefer the AI companion over a human companion. For proof of this, see twitter, blockchains and people deliberately building echo chambers. It is because they do not want to interact with people that disagree. The ultimate echo chamber is a persona AI that replaces all humans.
I feel like the only people who would really want this in a partner are the kind of anti-social, weirdo NEET types that likely wouldn't find a real partner anyways. We'd all prefer people to agree with us all the time, but one of the fun parts of building a relationship with somebody and discovering more about that person is finding out what they think, how they behave, why they think the way they do.

Learning where you and a partner agree or disagree on certain topics and discussing those similiarities and differences is one of the most enjoyable parts of exploring a relationship with someone and can't be replicated by an AI authentically in any way, regardless of what the Kevin Gibes and Keffals of the world would have you believe.
 
Coming back to this, here's a few choices.
No harm on trying to get https://grisk.itch.io/stable-diffusion-gui to run on your pc, you just install it. If it doesn't work it doesn't work. But this video shows you ways to run it online.
I got the UI version running, it seems to be working but generating images takes upwards of 4-6 minutes. That's a little longer than I expected. Is this normal or am I fucking something up?
 
  • Thunk-Provoking
Reactions: Toolbox
I got the UI version running, it seems to be working but generating images takes upwards of 4-6 minutes. That's a little longer than I expected. Is this normal or am I fucking something up?
It mostly depends on your gpu, mine never had a problem generating images in under a minute, and I'm running a laptop 2070 which while it runs at a lower wattage has more VRAM which is what is important. Though have you increased the resolution of the images generated or the iteration? That could be a problem too.
I'm confused by a few comments made here. Is there anything about this program that can damage my computer?
Unless you're grabbing it from freenotavirusprograms.info then there shouldn't be a problem. Stuff off github or the itch.io one should be fine. The only issue would be is if you're cooking your pc and it's improperly cooled. But I'm not an expert on this.
 
  • Like
Reactions: lolcow yoghurt
It mostly depends on your gpu, mine never had a problem generating images in under a minute, and I'm running a laptop 2070 which while it runs at a lower wattage has more VRAM which is what is important. Though have you increased the resolution of the images generated or the iteration? That could be a problem too.
resolution is around 512x512, my card should be performing the same as yours. I think I saw something about it calling my GPU unsupported and switching to CPU generation instead? not sure what that's about
 
resolution is around 512x512, my card should be performing the same as yours. I think I saw something about it calling my GPU unsupported and switching to CPU generation instead? not sure what that's about
It's very important to know what your gpu is concerning how well this will run. VRAM is specific to your card, it isn't your regular RAM. Also Nvidia cards are way better at this because Nvidia has been putting a lot of money into developing their cards for deep learning stuff for years. 4gb of vram on a nvidia card is the minimum for having a decently fast generation, but 8gb is what would give most people reasonable performance.
 
So it can do smaller booba. Coomers in the itty bitty titty committee will be happy I guess...

Moar and a technical update: I am using a Nvidia GTX 970M (3 GB VRAM) from 8 years ago and am consistently getting 0.60 - 0.63 it/s, so it's reasonably fast enough to tinker with (cannot really do anything else graphics intensive concurrently though, which is expected).

Captain Picard spooky. Looks very plasticy and action-figure-like. @make_it_so as a Sith lord I guess?
1666131277033.png

@Feline Supremacist vodka:
1666131415579.png

Obama and Putin shoe shopping:
1666131952550.png
1666131981636.png
(Last one is confusing as all get out).
 
Last edited:
It's very important to know what your gpu is concerning how well this will run. VRAM is specific to your card, it isn't your regular RAM. Also Nvidia cards are way better at this because Nvidia has been putting a lot of money into developing their cards for deep learning stuff for years. 4gb of vram on a nvidia card is the minimum for having a decently fast generation, but 8gb is what would give most people reasonable performance.
Mine has 8gb of VRAM as well, but it's not Nvidia. I think I'll try one of the other methods for getting it to run. Thanks for the information.
 
  • Informative
Reactions: Toolbox
Pulling this one out because it's a modern masterpiece.
View attachment 3748270


Why would you spoiler this, then set it as your avatar, you butt?



Relevant pages from "Boku no Tsuma wa Kanjou ga Nai | My Wife Has No Emotion"


If you're that level of tech retarded, just pay NovelAI.


What are they going to spend that money on now? Custom models?
Wow, I never expected to see My Wife Has No Emotion here. For anyone who hasn't read it, it's a crazy good manga that just looks like Chobits on the surface and in reality delves into Deus Ex levels of genre deconstruction.
As this thread is starting to evolve into 'is AI bad actually' I would like to say that I, for one, welcome our new robot overlords. If humanity is going to die off, we may as well be comfortable.
People have always retreated from conflict and into their own fantasies, long before computer generated anime art or robot waifus. If anything, it'll reduce the amount of harm those kinds of people do to others by attempting to force those fantasies onto the real world.
 
  • Feels
Reactions: naaaaiiiiillllll!!!
as someone who has zero artistic ability, this and that thing that transformed your nonsensical doodles into really landscapes/formations, I am loving the shit out of this. I’ve spent hours generating weird hellscapes, images that would make Junji Ito reconsider his life choices, and food. These various models create some ridiculous looking hamburgers.
 
Mine has 8gb of VRAM as well, but it's not Nvidia. I think I'll try one of the other methods for getting it to run. Thanks for the information.
From what I've heard amd can be a pain in the ass with this stuff, yes your longer generation time can be normal for AMD. Nvidia cards from the past few years are just more efficient in deep learning versus AMD from the same period.
 
Mine has 8gb of VRAM as well, but it's not Nvidia. I think I'll try one of the other methods for getting it to run. Thanks for the information.
Most versions of Stable Diffusion require a Nvidia card as the actual processing runs on CUDA cores, which AMD cards don't have. There is this guide for getting it to run on AMD cards but I have no idea how good the result is. Probably still a lot slower if I had to guess.
 
To anyone worried about this killing art, this is literally the equivalent of a fleshlight to an actual coochie. Women haven't died out yet.
Did artists disappear when the camera showed up? No.

The main concern of artists is that their art is being used in models without their consent, but Rembrandt wasn't complaining when his students became as good as him. While God made the landscape, man captured it with his own depictions, so it's hard to argue against it as theft.
 
Back