|
if someone were to merge the standard stable diffusion model with a leaked model at 50/50. It might produce interesting results, like better hands and give a more realistic style.
|
# ? Oct 8, 2022 01:36 |
|
|
# ? May 30, 2024 07:53 |
|
Mindblast posted:Lmfao. The timing is amazing(a little too amazing). And it seems some enterprising weeb burned a zero day for this? I got no confirmation but if true that is some dedication. God bless our troops
|
# ? Oct 8, 2022 07:29 |
|
Rutibex posted:Avengers Assemble! You got to hand it to the Captain for staying sober through all that poo poo.
|
# ? Oct 8, 2022 10:20 |
|
axolotl farmer posted:You got to hand it to the Captain for staying sober through all that poo poo. He didn't have a choice vodka might as well be water to him it took an entire bar to get him shitfaced. Do you know how much that costs to stock?
|
# ? Oct 8, 2022 11:52 |
|
Mr Bean as a Good Guy Chucky Doll Rutibex fucked around with this message at 13:36 on Oct 8, 2022 |
# ? Oct 8, 2022 13:34 |
|
Rutibex posted:Mr Bean as a Good Guy Chucky Doll Mr Bean at a rave Mr Bean on trial for war crimes for fucks sake fucked around with this message at 20:45 on Oct 8, 2022 |
# ? Oct 8, 2022 20:33 |
Do not like. Fits the October horror theme perfectly. I expect it will only get worse.
|
|
# ? Oct 8, 2022 21:46 |
look all i know is i saw it on retweeted and figure it belongs here https://twitter.com/urban_meadows/status/1578392347028643845
|
|
# ? Oct 8, 2022 22:43 |
|
Lol there are no AI artists. There are artists who use AI but anyone calling themselves an AI artist isn't an artist. Taking credit for what your prompt generates is like taking credit for your poo poo being black after drinking Guiness.
|
# ? Oct 8, 2022 23:19 |
|
|
# ? Oct 8, 2022 23:21 |
|
Same but people who use fancy photoshop brushes and/or filters.
|
# ? Oct 8, 2022 23:44 |
|
The Sausages posted:Lol there are no AI artists. There are artists who use AI but anyone calling themselves an AI artist isn't an artist. Taking credit for what your prompt generates is like taking credit for your poo poo being black after drinking Guiness. Source your quotes
|
# ? Oct 9, 2022 00:25 |
|
Anything created, captured, or presented as art is art and the person doing so is an artist. Final exam time! What is art? ____________ There, you just completed Art 101. That will be $25,000.
|
# ? Oct 9, 2022 00:36 |
|
The Sausages posted:Lol there are no AI artists. There are artists who use AI but anyone calling themselves an AI artist isn't an artist. Taking credit for what your prompt generates is like taking credit for your poo poo being black after drinking Guiness. You're only a real artist if you make your own paints by hunting beetles, grinding them up, and you make your own brushes by trapping foxes in Russia or whatever. edit: Nope, still not the artist in that situation. It's the bugs and flowers used to make the paints and the fox for the brush that is the real artist. Chainclaw fucked around with this message at 00:50 on Oct 9, 2022 |
# ? Oct 9, 2022 00:47 |
|
I have aphantasia so I've been using AI to generate concepts (b/c I lack the ability to envision things in my mind) and then draw things based on them, which is pretty much being an AI Artist -> -> I do not like admitting this b/c many people are so against the idea of anything ai-generated being "art" that they don't even consider drawings derivative of the ai-generated images to be art but I see aphantasia as a major disability in art and this lets me overcome it pretty easily. Both of these images are too complex for me to have deliberately put together on my own without some kind of reference material allowing me to imagine the scene but I also in both cases tried to do something to differentiate it from the AI-generated originals. deep dish peat moss fucked around with this message at 01:13 on Oct 9, 2022 |
# ? Oct 9, 2022 01:02 |
|
surgeons operating on pumpkins Rutibex fucked around with this message at 02:07 on Oct 9, 2022 |
# ? Oct 9, 2022 02:05 |
|
I think I might have broken it somehow. Santa claus as a samurai, cyberpunk, realistic image, Steps: 50, Sampler: DPM2 a, CFG scale: 15.5, Seed: 318045479, Face restoration: GFPGAN, Size: 1280x768, Model hash: 7460a6fa, Denoising strength: 0.7 gave me this madness.
|
# ? Oct 9, 2022 03:49 |
|
Just found this randomly. Holy poo poo. https://www.youtube.com/watch?v=oPbtZ-PZusw
|
# ? Oct 9, 2022 07:32 |
|
deep dish peat moss posted:I have aphantasia so I've been using AI to generate concepts (b/c I lack the ability to envision things in my mind) and then draw things based on them, which is pretty much being an AI Artist Thats awesome, you're an artist, and I mean no disrespect to people who are using this technology to explore artistic expression. This tech blurs the line between patrons and artists and we should mock people who can't tell the difference when they try claiming they did it all for themselves. You're an artist, Greg Rutkowski is an artist, some cryptobro typing GR's name into a prompt to mass produce nfts doesn't need to be respected as an artist even if they qualify on sophistry-adjacent technicalities. There are only artists, there are no AI artists worth recognizing as such (yet), the AI is both the artist and the medium. Gonna have to rewatch Zima Blue and smoke some weed and think poo poo over some more
|
# ? Oct 9, 2022 07:55 |
|
KakerMix posted:Source your quotes they're not wrong, at best you're a muse to the AI
|
# ? Oct 9, 2022 10:56 |
|
The Sausages posted:There are only artists, there are no AI artists worth recognizing as such (yet), the AI is both the artist and the medium. yeah, I'm gonna go ahead and say Midjourney is an artist. a loving fantastic artist
|
# ? Oct 9, 2022 10:59 |
|
Large Testicles posted:they're not wrong, at best you're a muse to the AI "Artist" is a meaningless term, if you guys want to change defense now to "ok it IS art but YOU aren't an artist " then by all means have any it. AI is making art and causing all sorts of drama getting people all upset and passionate which is the most "art" thing there is.
|
# ? Oct 9, 2022 14:04 |
|
a steampunk alchemist tiger, character concept, in his cozy steampunk laboratory
|
# ? Oct 9, 2022 14:52 |
|
this AI music vid got posted in the wyd thread: https://www.youtube.com/watch?v=wb4RGmWO5Es
|
# ? Oct 9, 2022 15:35 |
Somewhere in the ancient past one caveman paused in making their hand and mud print antelope cave wall art , saw another cave person pick up a burnt stick, and wandered over to grunt something at them because they didn't have language yet to communicate "It just doesn't feel like it has the same soul..."
|
|
# ? Oct 9, 2022 16:14 |
|
TheGoonspiracist posted:Just found this randomly. Holy poo poo. perfect example of how AI can enhance in the hands of an artist edit: actually, the video was made in 2010? it's all hand made? double edit: actually the original video without the effects were made in 2010, the update is AI generated ymgve fucked around with this message at 18:34 on Oct 9, 2022 |
# ? Oct 9, 2022 17:14 |
|
I let my twins come up with prompts the other night and this is what we got. As you can see, they are obsessed with pugs, pirates, Pokémon and Lego.
|
# ? Oct 9, 2022 17:27 |
|
|
# ? Oct 9, 2022 19:24 |
|
The Sausages posted:Lol there are no AI artists. There are artists who use AI but anyone calling themselves an AI artist isn't an artist. Taking credit for what your prompt generates is like taking credit for your poo poo being black after drinking Guiness. I scrawled this doodle with like crayons or something. The narrative idea I had in mind was a dragon perched on a hill above a tunnel, intently staring at the footpath below. After two hours of farting around with stable diffusion img2img and a blazing 3090 warming my feet I had this. As per tradition I poured a word salad of gibberish incantations into the prompt thing. With the latest Automatic1111 build the token limit has been deleted so you can go hog wild. Still not sure about the AND operator. "Stable Diffusion posted:(epic:1.2), (whimsical:0.5), (digital fantasy illustration:1.1), (by Brian Froud:1.2), (by Tim Hildebrandt:0.9), (by Greg Rutkowski:0.8), (by Richard Corben:0.4), (by Thomas Kinkade:0.4), (by Stanislaw Beksinski:0.4), Dragon with golden scales and folded wings lurking on a grassy hill above a (dark brick-lined tunnel:1.2), the dragon is staring at the tunnel, glowing eyes, misty forest background, muddy path foreground. (Cinematic lighting:1.2), haze, muddy path, pine trees, distant mountains, intricate detail, (moss overgrown cracked flagstones :1.1), (reeds:1.4) It took a few generations of selectively photoshop compositing from masked lucky shots from various prompt iterations. Like, I liked this style of tunnel stones from this one, I liked this bit of dragon wings or scales from a few others. Oo, nice misty forest background. Obviously since my input sketch was so crude and I'm no artist and I have no idea how dragon anatomy works I ended up with sort of a nonsensical dragon design where the two wings aren't remotely alike. I guess I liked the one resembling a bat-wing but I wasn't arsed to try and coax symmetry out of it. The dragon body is completely undefined, just I guess a long squiggly golden worm? The tail looks flat. It was much worse initially though! Because my input sketch had a poorly defined body form it took a few iterations of body horror and completely disconnected individual elements before the "idea" of the dragon was solidified enough that it stayed coherent. I got all kinds of neat designs for various bits along the way. A bejeweled crown on the dragon's head. Brian Froud esque creatures emanating from the rock pile lower right. Often time the dragon had really fantastic facial designs but which didn't carry the narrative I was intending. Eventually after about 2 hours had passed I was happy enough with the design and called it a day. I don't at the end of this session think of myself as the "artist" so much as maybe a creative director since nobody else was present to volunteer creative decisions or overrule mine. There also was nobody here but me who wanted this thing to exist so in that sense I was also a producer. I thought of Stable Diffusion as my unpaid, uncredited artisan laborer patiently rendering and re-rendering variations of ideas: Give me 20 takes on this inpaint or just redo the whole image with various prompt tweaks. All that thankless work just turns into waste and occasional saved scraps pasted into the basis for the next iteration. Is there no artist present? Then whence this piece? So is there an artist present? I only volunteered a napkin doodle. The only other candidate is SD, unless you get into paranormal territory and suggest SD is invoking or embodying the spirit and talents of dead and alive actual artists to some degree. So did Tim Hildebrandt or Brian Froud paint this? I suggest that those giants could have done far better, and given my humble narrative prompt a much more clever take.
|
# ? Oct 9, 2022 19:41 |
|
So I retrained Ghibli Studio on Waifu Diffusion 3 with even more images Real life stuff (denoising strength around 4.2) Anime screencaps (denoising strength around 3) Model Download partial labelled training data (didn't use, I was planning on making a fine tune, but at this point it seems almost uncessary.) Prompt is studio_ghibli_anime_style style
|
# ? Oct 9, 2022 20:31 |
|
frumpykvetchbot posted:So is there an artist present? I only volunteered a napkin doodle. The only other candidate is SD, unless you get into paranormal territory and suggest SD is invoking or embodying the spirit and talents of dead and alive actual artists to some degree. So did Tim Hildebrandt or Brian Froud paint this? I suggest that those giants could have done far better, and given my humble narrative prompt a much more clever take. Sure those painters might have been able to do better but we're just scoring by top historic influence, I'd say it was painted by Adolf Hitler
|
# ? Oct 9, 2022 20:36 |
|
frumpykvetchbot posted:I scrawled this doodle with like crayons or something. The narrative idea I had in mind was a dragon perched on a hill above a tunnel, intently staring at the footpath below. I mean that's the same workflow as making something with photoshop using existing photographs, except instead of using a search engine through the stock library, you've got more control over details.
|
# ? Oct 9, 2022 20:43 |
|
You guys can argue about what a "real artist" is till you're blue in the face. I'll keep making more goth chicks with pumpkins.
|
# ? Oct 9, 2022 21:38 |
|
Tunicate posted:I mean that's the same workflow as making something with photoshop using existing photographs, except instead of using a search engine through the stock library, you've got more control over details. Yes but that has SOUL, you see, now, what is soul you ask? well....
|
# ? Oct 9, 2022 21:40 |
|
IShallRiseAgain posted:So I retrained Ghibli Studio on Waifu Diffusion 3 with even more images Thank you for posting this. These results are awesome!
|
# ? Oct 9, 2022 21:44 |
|
Messing around with combinations of artist names in Midjourney and I hit on this one with cool results: by Alex Ross, John Berkey, Genndy Tartakovsky, Roger Dean, Alex Grey, Alan Moore, Kurzgesagt, Vincent Di Fate, Michael Whelan, C. Foss, Boris Vallejo Whimsical Canyon of Dreams Mojave Canyon of Dreams Futuristic spaceport docking bay "In the depths of the crystalline caverns an ancient computer awakens", by Alex Ross, John Berkey, Genndy Tartakovsky, Roger Dean, Alex Grey, Alan Moore, Kurzgesagt, Vincent Di Fate, Michael Whelan, C. Foss, Boris Vallejo Ancient Computer in the Depths of the Crystalline Caverns, by Alex Ross, John Berkey, Genndy Tartakovsky, Roger Dean, Alex Grey, Alan Moore, Kurzgesagt, Vincent Di Fate, Michael Whelan, C. Foss, Boris Vallejo e: These pics are all after an Upscale -> Remaster deep dish peat moss fucked around with this message at 22:00 on Oct 9, 2022 |
# ? Oct 9, 2022 21:50 |
|
a render of Santa Claus playing an electric guitar at a heavy metal concert in a snow storm, white hair, white beard, red clothing, rotund body, handsome face, symmetrical face, highly detailed, dark lighting, foreboding, Detailed and Intricate, Digital Art, CGI, Canon50, Beautiful Lighting, Cold Color Palette, Angry, Evil, Artstation, by Greg Rutkowski, by wlop, Unreal Engine, Cycles Render, Corona Render, Detailed Render, Colorful some of these were done using image to image
|
# ? Oct 9, 2022 22:11 |
|
IShallRiseAgain posted:So I retrained Ghibli Studio on Waifu Diffusion 3 with even more images Didn't want to let this go past without comment, this is very cool.
|
# ? Oct 9, 2022 23:10 |
|
I did a little experiment, results weren't that great, but I can see the potential in the future. https://www.youtube.com/watch?v=t9Qim_xKT_I
|
# ? Oct 10, 2022 00:18 |
|
|
# ? May 30, 2024 07:53 |
If you followed the Voldy guide to set up, how do you make it do the update/git pull?
|
|
# ? Oct 10, 2022 00:42 |