|
Vlaphor posted:"8k resolution beautiful cozy inviting Steampunk Hobbit-House Cottage in Steamy foggy Yellowstone Hotspring Geyser digital illustration polished psychadelic matte painting trending on Artstation" I decided to try some Thomas Kinkade versions. minor variations on "Thomas Kinkade painting of beautiful cozy inviting Hobbit-Hole built into Hillside in Steamy foggy Yellowstone near Hotsprings and Geysers, Lord of the rings, digital illustration, polished, matte painting, trending on Artstation"
|
# ? Sep 18, 2022 23:14 |
|
|
# ? May 30, 2024 13:53 |
|
Roger Dean art of beautiful cozy inviting Steampunk Hobbit-House Cottage in multicolored countryside from an acid trip, digital illustration polished psychedelic matte painting trending on Artstation Steps: 50, Sampler: Euler a, CFG scale: 7 edit: it gave me a literal matte painting one more. turnt up VectorSigma fucked around with this message at 04:08 on Sep 19, 2022 |
# ? Sep 19, 2022 03:56 |
|
VectorSigma posted:Roger Dean art of beautiful cozy inviting Steampunk Hobbit-House Cottage in multicolored countryside from an acid trip, digital illustration polished psychedelic matte painting trending on Artstation Transparent version
|
# ? Sep 19, 2022 05:19 |
|
It can't really handle Dr. Seus style though. Dr. Seus art of beautiful cozy inviting Hobbit-House Cottage in multicolored countryside, background forest, lorax, book illustration flat colors Jonny Nox fucked around with this message at 05:27 on Sep 19, 2022 |
# ? Sep 19, 2022 05:23 |
|
Cable Guy posted:There's probably a photoshop thread waiting in that.... here's something real quick and dirty... yeah it's too easy
|
# ? Sep 19, 2022 05:31 |
|
Uh.... Stable Diffusion... What....? Prompt was "Vladimir Putin bending over away from but facing camera" (don't ask me why...) run through the OS Hugging Face SD engine.
|
# ? Sep 19, 2022 09:05 |
|
Hadlock posted:That last one, but dos era VGA graphics And since not posting your prompt is illegal: I first cropped the image to 4:3 by hand, then I ran this command in imagemagick: code:
-colorspace rgb converts the image to linear rgb (as opposed to gamma-corrected srgb) -normalize stretches the contrast in the image -sharpen 2x0 sharpens the image -resize 320x240 resizes it to QVGA +dither turns error-diffusion dithering off -ordered-dither o8x8,64 uses a 8x8 ordered bayer dither with 64 levels for r,g and b (that is, 6 bits per channel, like vga) -colors 256 reduces the number of colors to 256 (this step would use error correcting dither it it was turned on) -posterize 64 again fixes the reduced number of colors to 64 levels per channel to fix any non-6-bit colors imagemagick selected for the palette in the previous step, no dithering this time (this step would also use error correcting dither it it was turned on) -scale 300% use nearest neighbor scaling to scale the image by 300% so the pixels are nice and visible -colorspace srgb convert the image back to (gamma corrected) srgb Wheany fucked around with this message at 10:59 on Sep 19, 2022 |
# ? Sep 19, 2022 10:56 |
|
Vlaphor posted:"watch made of human flesh, body horror, blood, stunning detail, intricately detailed, ultra realistic, ultra detailed, award winning photography, 8k, Bokeh" Oh yeah and these were really gross, thank you/gently caress you 50/50 split
|
# ? Sep 19, 2022 11:02 |
|
I was absent for a few weeks, what’s the new hot thing? Seems like everybody is using a locally installed version of Stable Diffusion? I tried it on browser months ago but couldn’t for the life of me make portraits and the process was very involved. Does it at least have a UI of sorts these days? I mostly wanna make waifus / portraits of characters to use in my tabletop rpg’s and cool vaporwave stuff
|
# ? Sep 19, 2022 14:50 |
|
I think I might have unleashed a great evil upon the world. Ruth Bader Ginsburg, Wakanda Forever, Marvel Cinematic Universe, Drew Struzan, Frank Frazetta, Greg Rutkowski, Movie Poster
|
# ? Sep 19, 2022 15:34 |
|
Deltasquid posted:I was absent for a few weeks, what’s the new hot thing? Seems like everybody is using a locally installed version of Stable Diffusion? https://github.com/AUTOMATIC1111/stable-diffusion-webui/
|
# ? Sep 19, 2022 16:11 |
|
Deltasquid posted:I was absent for a few weeks, what’s the new hot thing? Seems like everybody is using a locally installed version of Stable Diffusion? For waifus and rpg portraits midjourney is likely your best bet. The sexiest words are banned, but if you use the magical phrase "super sonico" you can make the best waifus
|
# ? Sep 19, 2022 17:32 |
|
I think the longer I look at this stuff the less I buy the hype. The AI can turn out high fidelity images quickly, but any good painting, drawing, movie, whatever, only gets there after dozens of tiny, crappy sketches. There's no nuance to the final results, nothing is being communicated past the prompt. No little details adding up to something bigger. And it needs an entire internet full of reference material to get there. It is less revolutionary than Photoshop!
|
# ? Sep 19, 2022 17:32 |
|
FunkyAl posted:I think the longer I look at this stuff the less I buy the hype. The AI can turn out high fidelity images quickly, but any good painting, drawing, movie, whatever, only gets there after dozens of tiny, crappy sketches. There's no nuance to the final results, nothing is being communicated past the prompt. No little details adding up to something bigger. And it needs an entire internet full of reference material to get there. This isn't for serious art appreciators. what this stuff is good for is when you need bulk art. like if you are making a monster manual for D&D and you need a picture of a wolf and an ogre and a goblin, whatever. You just need a bunch of art and it's ok if it's not a masterpiece. that's what this is good for
|
# ? Sep 19, 2022 17:37 |
|
Rutibex posted:This isn't for serious art appreciators. what this stuff is good for is when you need bulk art. like if you are making a monster manual for D&D and you need a picture of a wolf and an ogre and a goblin, whatever. You just need a bunch of art and it's ok if it's not a masterpiece. that's what this is good for Why wouldn't you just draw a goblin though? Using your imagination is the fun part.
|
# ? Sep 19, 2022 17:41 |
|
FunkyAl posted:Why wouldn't you just draw a goblin though? Using your imagination is the fun part. lol because I can't
|
# ? Sep 19, 2022 17:42 |
|
FunkyAl posted:Why wouldn't you just draw a goblin though? Using your imagination is the fun part. Because you lack the artistic training and want a unique portrait for your D&D character that isn't just a google image result?
|
# ? Sep 19, 2022 17:43 |
|
FunkyAl posted:Why wouldn't you just draw a goblin though? Using your imagination is the fun part. I’m better at writing than drawing, so using AI goes faster. Sometimes I also want a very specific mood like asking it to give me a CGI rendition of Columbo shooting a goblin, but not sufficiently for me to spend more than 15 minutes on it
|
# ? Sep 19, 2022 17:47 |
|
Personally I think while what we have now is cool but not as world changing as some people think, what's more impressive is how far the technology has come in what, less two years? Hell IIRC the original paper theorizing these models is less than a decade old. If things keep improving this poo poo is going to be scary good in like, five years time.
|
# ? Sep 19, 2022 17:47 |
|
Yeah, its important to realize that is that all of this AI art stuff only recent became easily accessible to the public. Now that it is out there and easy to experiment with, there is going to be much more brain power dedicated to improving it. I think the next big leap is being able to use it for video with temporal stability. Its also a tool. Its not quite there yet, but pretty soon I think it will make independent projects much more feasible, and they won't require organizing a huge team or the creator working themselves to death.
|
# ? Sep 19, 2022 18:01 |
|
FunkyAl posted:I think the longer I look at this stuff the less I buy the hype. The AI can turn out high fidelity images quickly, but any good painting, drawing, movie, whatever, only gets there after dozens of tiny, crappy sketches. There's no nuance to the final results, nothing is being communicated past the prompt. No little details adding up to something bigger. And it needs an entire internet full of reference material to get there. With img2img in stable diffusion there is a lot more room for composing a more complex scene in pieces. It lets you block out some basic scene structure and colors that the final output will be based on. Then when it outputs an image that is almost what you want you can scrub out the random third arm that someone in the scene grew and send it back through the machine for a cleaner final output. Just mashing out a prompt to get a cool image can work depending on your needs, but the potential to do a whole lot more than that is here.
|
# ? Sep 19, 2022 18:17 |
|
FunkyAl posted:Why wouldn't you just draw a goblin though? Using your imagination is the fun part. I can't loving draw anything that isn't based on angles or straight lines or can't be done with a drafting machine (mechanical drawing is all I can do). I have tried over the last fifty years to get better at other kinds of art drawing than mechanical, but I continue to be stuck at elementary school scribbles. AI art is a god send to me.
|
# ? Sep 19, 2022 18:26 |
|
Mr Luxury Yacht posted:Personally I think while what we have now is cool but not as world changing as some people think, what's more impressive is how far the technology has come in what, less two years? Hell IIRC the original paper theorizing these models is less than a decade old. Ah yes, the Disco Stu fallacy
|
# ? Sep 19, 2022 18:27 |
|
AI emojis gonna be off the chain. Just send a unique version of your expression every time instead of the same old eggplant.
|
# ? Sep 19, 2022 18:29 |
|
the queen of england at a tattoo shop getting a tattoo of herself on her lower back
|
# ? Sep 19, 2022 18:41 |
|
Rahu posted:With img2img in stable diffusion there is a lot more room for composing a more complex scene in pieces. It lets you block out some basic scene structure and colors that the final output will be based on. Then when it outputs an image that is almost what you want you can scrub out the random third arm that someone in the scene grew and send it back through the machine for a cleaner final output. you can iterate like this using midjourney too. if you layer your original seed over top of the AI output you can selectively delete features and send it through the AI for another pass. people saying this isn't impressive are really underestimating this. I had basically zero art powers, but I could throw things togeather in Photoshop. these AIs give me so much original material to work with my art power level has grown exponentially. the AIs are like transforming into an art super Saiyan, it multiplies your power 50x
|
# ? Sep 19, 2022 18:59 |
|
Handen posted:
Rutibex posted:you can iterate like this using midjourney too. if you layer your original seed over top of the AI output you can selectively delete features and send it through the AI for another pass. How do you get the seed?
|
# ? Sep 19, 2022 19:03 |
|
Humbug Scoolbus posted:
You have to start with an image. Put the image URL at the front of your prompt and it will use the image instead of a random seed
|
# ? Sep 19, 2022 19:05 |
|
Rutibex posted:You have to start with an image. Put the image URL at the front of your prompt and it will use the image instead of a random seed Ah, I thought you were talking seed number like SD uses
|
# ? Sep 19, 2022 19:08 |
|
you can do that with MJ too, but the only way I know how to do that is to specify the seed in advance
|
# ? Sep 19, 2022 19:10 |
|
FunkyAl posted:I think the longer I look at this stuff the less I buy the hype. The AI can turn out high fidelity images quickly, but any good painting, drawing, movie, whatever, only gets there after dozens of tiny, crappy sketches. There's no nuance to the final results, nothing is being communicated past the prompt. No little details adding up to something bigger. And it needs an entire internet full of reference material to get there. The guy who lived in my house before me was a commercial artist for...vox.com? Verge.com? Some v named blog. Vice? The one that interviewed Obama Anyways his job is basically to draw the art you see at the top of your article that might summarize part of the article visually. And for bigger articles he might draw 2-5 more drawings for the day's headline article There is no soul in what he draws, it's just commercial art, he's like living clip art Anyways his career is probably doomed now once someone packages this up as an 8gb desktop app and sells it for $2.99 on the app store
|
# ? Sep 19, 2022 20:03 |
|
FunkyAl posted:Why wouldn't you just draw a goblin though? Using your imagination is the fun part.
|
# ? Sep 19, 2022 20:22 |
|
FunkyAl posted:And it needs an entire internet full of reference material to get there. so just like a human
|
# ? Sep 19, 2022 20:24 |
|
Also midjourney at least is capable of deeply surprising levels of abstraction and theme-setting from non-prompts (which might stem from it being a skinny model, since dalle2 and stablediffusion are both terrible at this). Look at all this crazy poo poo it drew me based on nothing but a speech from Shakespeare's Henry V. https://twitter.com/shambibble/status/1562970374953783297 It's possible to describe your setting in pure prose knowing zero prompt tricks and MJ will make a decent go of it. Elotana fucked around with this message at 20:31 on Sep 19, 2022 |
# ? Sep 19, 2022 20:28 |
|
Humbug Scoolbus posted:I can't loving draw anything that isn't based on angles or straight lines or can't be done with a drafting machine (mechanical drawing is all I can do). I have tried over the last fifty years to get better at other kinds of art drawing than mechanical, but I continue to be stuck at elementary school scribbles. AI art is a god send to me. The best life drawing instructor I ever had showed us how to draw the figure in all straight lines, he made us do just the horizontal and vertical measurements of everything before we went into the diagonal vectors of the body. I'm positive you could apply your mechanical drawing sense to other things, and even if you can't id bet your scribbles are more interesting. pixaal posted:Because you lack the artistic training and want a unique portrait for your D&D character that isn't just a google image result? How do you get to Carnegie hall? Anyway I'm getting sidelined, here is my point I want to shout until my lungs are hoarse and explode and I am dead: it does not matter if your drawing is bad. Your drawing is an artifact of your thinking. The thinking is the fun part, and the creative part. It isn't something you have to do, it's something you get to do. Much of what you like about art was organic or accidental, occurred during the process, is viewed as a mistake by the artist, just like the most interesting thing about the ai art is that it appears to be rendering things differently than we know how to. So yes, it is good for editorial illustration and horrible corporate products and all that. But if you personally are using it to draw for you, you are robbing yourself of your own thoughts.
|
# ? Sep 19, 2022 20:46 |
|
FunkyAl posted:The best life drawing instructor I ever had showed us how to draw the figure in all straight lines, he made us do just the horizontal and vertical measurements of everything before we went into the diagonal vectors of the body. I'm positive you could apply your mechanical drawing sense to other things, and even if you can't id bet your scribbles are more interesting. My brain doesn't have the ability to retain or recall pictures or images (at all -- this is a diagnosed clinical condition which also has the fun side effect of complete face-blindless), which means regardless of how much I may practice, I can't translate my thoughts to an image in a physical space. I've learned through muscle memory to draw very simple, cartoonish objects and characters, but I have to draw them identically every time. For example, I can draw a cute little dog, but can't really change his stance or the viewing angle. I am not robbing myself of my thoughts, because I can't see the mental image of a giant koala bear crushing the Space Needle in my mind anyway. But I can get good at describing it to the AI -- quality prompting is very much a skill you can train -- and together we can create something amazing. So while I appreciate the rewards that many people get from handcrafting their own art, I personally see these AI tools as a way to make art that I simply never could otherwise. WhiteHowler fucked around with this message at 21:22 on Sep 19, 2022 |
# ? Sep 19, 2022 21:02 |
|
heIL yeeh
|
# ? Sep 19, 2022 21:05 |
|
FunkyAl posted:
I find using Art AI to be fun and creative process all the same. I made bad MS paint style art in GIMP sometimes it's a shockingly similar process for me where I rarely know what the hell I'm drawing until it's done. It's okay people like different things and enjoy expressing themselves in different ways from you.
|
# ? Sep 19, 2022 21:09 |
|
DALL-E Discord posted:We heard your feedback: You can now upload and edit photos with faces!
|
# ? Sep 19, 2022 21:17 |
|
|
# ? May 30, 2024 13:53 |
|
I'm a person who has made a ton of "art" over the years, some of my emotes and dumb memes circulate around the internet still, which always makes me smile. I don't like being possessive of stuff I make, it's why I liked doing avatar making threads and the like. This is just another way to make stuff and share stuff, for me. I got really good at prompting really fast by experimenting and pinching bits here and there and then sharing those out. now I see people posting those prompts around and editing them and it gives me the same feeling of seeing something I made pop up on some random stream or website. for me it's about the process and the sharing of techniques for a brand new thing
|
# ? Sep 19, 2022 21:50 |