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

00014.png
00015.png
00017.png
00018.png


00019.png
00020.png
 
Last edited:
Can someone please use AI to do a Full Metal Thugshaker?
X
What the fuck is that ass on you Private Pyle?!
 
(of course an artist would say that)
BTW wouldn't a "transformative" "derivative work" be the closest to infringement, with the usual AI-made art?

Then again, modern artists (at least or especially Western ones) can be rather extremist about copyright.

I realized I may not be much of a fan of the typical modern art scene, or at least the typical modern Western one on deviantART.

They can overvalue art - like thinking works they make are like "children" - and can not take too kindly (to put it lightly) to mocking what they make. They can overthink what is or isn't art, and overvalue originality. And of course it seems they usually think of mere copying as "theft" or even worse. And it seems they're quite often fanatical and anal supporters of copyright: for example, a proposed "orphan works" part of copyright law caused a "gathering of the tribes" of artists to protest it. Also while being different isn't always bad, they can be TOO different in a freakish way. And of course there is the easy ego and elitism.

If AI makes NI*-made art a mere hobby that's not done for money, that could be a good thing, at least with deflating the ego.

*(natural intelligence)
 
Last edited:



These are good, but please remember the threadiquette: https://kiwifarms.net/threads/stable-diffusion-novelai-machine-learning-art.130730/post-14024432
Also, for some reason when looking at your images at full resolution they look pixelated, as if they were generated as smaller images then scaled up to huge 5120x2880 monstrosities afterwards.
 
Recently got a used 3090 for dabbling in AI training. For testing things out I made a Sonichu dreambooth using all of CWC's original comics unaltered as material to train with.
The results are not the best obviously as multi-painned comics are "confusing" for AI models to understand, and I only tagged them with clip (which surprisingly identified Sonichu correctly many times).
Attached some of the sample results that were generated at random during the training.
sample_13752-0.jpg

sample_13752-3.jpg

sample_14898-0.jpg

sample_20628-2.jpg

sample_22920-3.jpg

sample_26358-3.jpg

sample_28650-0.jpg

sample_28650-1.jpg

sample_28650-3.jpg

Not great by any means, but still amusing. Surprised that it was able to reproduce the colored with a crayola marker look as well as it did, which was the exact thing I was hoping for.
 
Last edited:
These are good, but please remember the threadiquette: https://kiwifarms.net/threads/stable-diffusion-novelai-machine-learning-art.130730/post-14024432
Also, for some reason when looking at your images at full resolution they look pixelated, as if they were generated as smaller images then scaled up to huge 5120x2880 monstrosities afterwards.
Sorry about that
Reading is hard
I'll make sure not to drain everyone's data caps all at once


00024-2938709917.png
00022-1991113995.png
00017-837441073.png
 
In honor of Bam Margera's latest escapades, I had Midjourney design some skateboard decks.

skateboard deck depicting bam margera as some manner of modern-day forest druid in a hood --v 5.1

BamForestDruid.png

skateboard deck depicting bam margera as a washed-up slob eating a burger off the floor --v 5.1

BamWashedUp.png

skateboard deck depicting bam margera sitting in his car and screaming at passers-by --v 5.1

BamScream.png

This one was my favorite. Unironically based and Bampilled.

Druid.png
 
I finally decided to bite the bullet and get this working last night. Really simple prompt of a Zombie apocalypse in the American civil war yielded this:
00310-569781629.jpg
Probably the best 1920x1080 image I've gotten out of this thing so far. Generating images of this size creates melted hellscapes, but it actually fits for this specific prompt.
Converting it to jpg killed the meta data, so here's the actual prompt: "Zombie apocalypse, American civil war, 1860s". There are no negative prompts and everything else is set to default.
00311-2891728864.png
Sorry, did I say trad? I meant to say tard.
 
I finally decided to bite the bullet and get this working last night. Really simple prompt of a Zombie apocalypse in the American civil war yielded this:
View attachment 5115558
Probably the best 1920x1080 image I've gotten out of this thing so far. Generating images of this size creates melted hellscapes, but it actually fits for this specific prompt.
Converting it to jpg killed the meta data, so here's the actual prompt: "Zombie apocalypse, American civil war, 1860s". There are no negative prompts and everything else is set to default.
View attachment 5115559
Sorry, did I say trad? I meant to say tard.
How long did that first pic take to render? What hardware are you using?
 
We're back! and to celebrate I come bearing kick-ass freakin' sweet robots!

Awww yeah, brother!
robot1.jpg

Coming to your house to steal your energon and your mail!
robot2.jpg

Easy to defeat if gone rouge! (Just press the glowing blue power button near the center)
robot4.jpg

Rock hard plastic robo-tits!
robot5.jpg

This apple iRobot runs on iOS (it can't access Kiwifarms on tor)
robot3.jpg
 
Last edited:
I finally got back to collaborating on some horrors (with some pleasantries mixed in) with my ol' friend SDf:

00306-2533335272-a wombat wearing a civil war union officer's uniform with rifle resting on sh...png00328-702051040-wombat wearing a civil war union officer's uniform with rifle resting on shoul...png00311-2533335277-a wombat wearing a civil war union officer's uniform with rifle resting on sh...png00313-2533335279-a wombat wearing a civil war union officer's uniform with rifle resting on sh...png00321-4268158776-a wombat wearing a civil war union officer's uniform with rifle resting on sh...png00324-702051036-wombat wearing a civil war union officer's uniform with rifle resting on shoul...png00329-702051041-wombat wearing a civil war union officer's uniform with rifle resting on shoul...png00331-702051043-wombat wearing a civil war union officer's uniform with rifle resting on shoul...png
00343-1616578812-man in civil war uniform with animal head, apocalypse background, black and w...png00344-1616578813-man in civil war uniform with animal head, apocalypse background, black and w...png00345-1616578814-man in civil war uniform with animal head, apocalypse background, black and w...png00347-1616578816-man in civil war uniform with animal head, apocalypse background, black and w...png00348-1616578817-man in civil war uniform with animal head, apocalypse background, black and w...png00350-1616578819-man in civil war uniform with animal head, apocalypse background, black and w...png
00352-3337970497-ultimate grotesquery, horror of the mind, dark palette, swirling background, ...png00358-3337970503-ultimate grotesquery, horror of the mind, dark palette, swirling background, ...png00351-3337970496-ultimate grotesquery, horror of the mind, dark palette, swirling background, ...png00353-3337970498-ultimate grotesquery, horror of the mind, dark palette, swirling background, ...png00354-3337970499-ultimate grotesquery, horror of the mind, dark palette, swirling background, ...png00356-3337970501-ultimate grotesquery, horror of the mind, dark palette, swirling background, ...png00357-3337970502-ultimate grotesquery, horror of the mind, dark palette, swirling background, ...png00359-3337970504-ultimate grotesquery, horror of the mind, dark palette, swirling background, ...png
00389-1132776295-epic astronomy scene, swirling galaxies, comets, bright, colorful, detailed, ...png00387-1132776293-epic astronomy scene, swirling galaxies, comets, bright, colorful, detailed, ...png00388-1132776294-epic astronomy scene, swirling galaxies, comets, bright, colorful, detailed, ...png00390-1132776296-epic astronomy scene, swirling galaxies, comets, bright, colorful, detailed, ...png00391-1132776297-epic astronomy scene, swirling galaxies, comets, bright, colorful, detailed, ...png00393-1132776299-epic astronomy scene, swirling galaxies, comets, bright, colorful, detailed, ...png
00412-3899263073-three-toed sloth riding a giant woolly mammoth.png00414-717678510-three-toed sloth riding on a woolly mammoth, tundra background, detailed, real...png00415-717678511-three-toed sloth riding on a woolly mammoth, tundra background, detailed, real...png00419-717678515-three-toed sloth riding on a woolly mammoth, tundra background, detailed, real...png00428-1136948652-three-toed sloth riding on a mammoth, tundra background, comet in the sky, de...png00429-1136948653-three-toed sloth riding on a mammoth, tundra background, comet in the sky, de...png00417-717678513-three-toed sloth riding on a woolly mammoth, tundra background, detailed, real...png00424-1136948648-three-toed sloth riding on a mammoth, tundra background, comet in the sky, de...png
 
Apparently there's this "new" AI program out called "Gen-2" that let's you create AI generated videos.
So far it looks like we're getting a rehash over "(insert media franchise here) as a 80s dark fantasy" only this time it's AI generated commercials about whatever.
 
Apparently there's this "new" AI program out called "Gen-2" that let's you create AI generated videos.
So far it looks like we're getting a rehash over "(insert media franchise here) as a 80s dark fantasy" only this time it's AI generated commercials about whatever.
Once a public version releases I'll give it a couple of months for when the coomers make it generate realistic porn with little to no artifacts.

The video shit people made with Stable Diffusion is pretty nice but raw text-to-video is really amazing. Chinese and Korean animators BTFO, anime industry hardened or BTFO. Don't really know, the Japs always embrace the new shit to become more efficient while Americans suddenly become ultra conservatives because the AI was supposed to replace physical labor before the things that are mostly automated or completely digitalized.
 
  • Like
Reactions: anustart76 and Puff
The video shit people made with Stable Diffusion is pretty nice but raw text-to-video is really amazing. Chinese and Korean animators BTFO, anime industry hardened or BTFO. Don't really know, the Japs always embrace the new shit to become more efficient while Americans suddenly become ultra conservatives because the AI was supposed to replace physical labor before the things that are mostly automated or completely digitalized.
It's clearly not ready to kill off all animators, but I can easily imagine one Warner Bros. Discovery intern being forced to make an entire Adult Swim show with stoner content in the style of that Pizza Nuggets commercial. WHOA
 
Yeah, the video stuff is obviously fake and wouldn't fool anyone into thinking it's real. Meanwhile, ai images indistinguishable from real photos made using consumer grade hardware have been out for at least a year now. I'm not sure how long until video catches up. Video has difficult hurdles like temporal consistency that still images don't need to clear.
 
Sorry if this is an obvious question, but does anyone know a service that would let me upload an image and "expand" it in a matching style or create a matching background? For example, going from this:

View attachment 5126183

to this:

View attachment 5126184

Is it possible?
What you're looking for is called outpainting. I only run Stable Diffusion locally so I don't know whether any of the many services support this, but I bet one does.
 
Back