So we're back with this since the old one died. This is a thread for those among us who use AI generation tools to share and discuss their work because /draw/ is a space for traditional art and there is no point in us intruding on their turf.
As usual, there are some guidelines for everyone posting to follow:
1. If you're sharing any art, preferably only post your own artwork. If you really wish to share someone else's art, always provide sauce to credit the original author.
2. Be willing to explain your process and tools used to others so they too can learn how to make art like you do.
3. Criticism is to be requested by the author, not handed out willy-nilly. Some people just want to share their work, not have it be deconstructed by every armchair critic that comes across it. Be respectful of their wishes.
4. Any work of yours that is posted ITT should meet at least two of the three following criteria:
A) Your work had at least the minimal amount of your own input. This can mean manual edits to generated images, the use of inpainting/outpainting tools, the use of controlnets, and so on. This also means you made an honest effort to check for obvious mistakes such as unwanted visual artifacts or anatomical errors and fix them. Basically as long as your contribution to the piece included more than typing the prompt, it counts.
B) Your work has a specific style to it. This can mean a specific medium like oil painting, black and white ink sketch, a LORA based on a specific artist's style (or even a hybrid of two styles!) and so on. No default shiny "hyperrealistic" anime style please, the internet already has enough of those.
C) Your work made use of tools that were free/open-source when making the piece and you did not willingly give your data/money to owners of proprietary software like OpenAI
Anyway, with all covered. Happy generating!
Not to be that guy, but why not learn how to draw for real?
>>711121I did, I've been drawing since I was a kid. I even took all the extra art courses in high school because I like to make stuff with my own hands and I've owned both a graphics tablet and a sketchbook for close to 10 years now and made plenty use of both. I even studied for 2 years how to make animations, 2D and 3D assets for games at university level. However I realized after those two years that I never wanted to make art into my career because it would just drain all the creative juices out of me. Hence why I treat art as a hobby from which I don't ever wish to make money out of.
So yeah, not everyone who uses AI tools is automatically someone who doesn't know how to draw, just like every traditional artist is not Da Vinci. In fact, in my experience at least, if you are already a traditional artist (in the sense that you draw by hand) your skillset tends to translate to AI generation pretty well because you know what to look for in terms of anatomy, posing, composition, lighting and shading, color theory and so on.
>>711124Sure, but if you can already draw why even use AI?
>>711129I‘m not „shifting the goal post“, this isn‘t a debate. I‘m merely wondering why anyone would „effort post AI art“ instead of learning art. And if you tell me you already learned how to draw then I‘m wondering why you wouldn‘t just do actual art instead of using AI. If it gets your panties up in a bunch so much then nvm lol.
>AI slop
Never have I been more excited and then disappointed to find another Spurdo artist poster except he took the lazy way out.
>>711130Sorry, but I'm just tired of hearing people repeatedly yell some variation of "just pick up a pencil" at me so I kind of tend to act on reflex when I hear anything close to that these days.
>>711131>AI is always automatically slop>I took the hard way so everyone else also has to take the hard wayWhat purpose does posting something like this serve? Is the point to just make me feel bad? Genuinely curious.
>>711135>What purpose does posting something like this serve? Is the point to just make me feel bad? Genuinely curious.not him but i sort of feel a similar eery way desu knowing that soon enough as AI develops, its art can become so detailed you wouldn't ditinguish it from the real handmade deal, soon you wouldn't tell if an entire deviant art account is a human or a bot drawing and a bot responding.
i love the art you posted, but at the same time can't help but think in a doomer way about what it means for OG artists. or beyond art, if i even need to study how to draw anime characters when i know some code can make what i have in mind, gives me a "why even try" thought
what is yall favorite tool for making AI art?
>>711137Yeah, I can understand that and I'm not one of those weird accelerationists who are like "hahaha yes soon all artists will be obsolete" because those people are fucking stupid IMO since I acknowledge that all art is derivative and AI generation itself would not be possible without the all the training data made by real people through both photos and traditional artwork. I believe that there will always be a place for traditional artists because some people will always want to make art in the traditional way and some people will always want to exclusively consume traditional art, and that's a completely okay, even a good thing. It's why traditional artisan crafts like pottery have never fully disappeared despite the fact that industrial mass production allows for a single factory to be making thousands of identical pots a day. Sometimes you just want that nicely decorated handmade claypot that someone (maybe even yourself) molded with their own two hands.
Just because AI generated art exists doesn't mean we need to all suddenly stop making all other art or that traditional art is obsolete. That's nonsense. For many artists the process of learning and making is the part they like the most. For me? All forms of artistic expression are valid because intent matters more than how much work you put into making a piece. But that's just my opinion.
>>711140I use Krita's Stable Diffusion Add-on with local ComfyUI instance run through Stability Matrix for generation, in/outpainting and Krita's own native tools and interface for manual drawing. Much more user-friendly than what I used to use (Automatic1111's Stable Diffusion WebUI).
based bread
I hate AI slop but I will never ever ever ever ever stop jacking off to it
>>711128The introduction of power-looms into England probably reduced by one-half the labour required to weave a given quantity of yarn into cloth. The hand-loom weavers, as a matter of fact, continued to require the same time as before; but for all that, the product of one hour of their labour represented after the change only half an hour’s social labour, and consequently fell to one-half its former value.
https://www.marxists.org/archive/marx/works/1867-c1/ch01.htmamazing how "marxists" on a "marxist" website are continually "confused" as to why people would use labor-saving technology
>b-but muh environmentburgers refuse to transition to solar/wind/nuclear, so it's their fault AI destroys their shitty 100 year old power grid
>>711135>>AI is always automatically slopLiterally yes. Generative AI can't be anything other than slop by virtue of how it works on a technical level. It creates an amalgamation of a library of inputs that homogenizes everything. Instead of the details being decided by an artist for a purpose, it's all just an echo of what other people have done before.
Moreover, the fact that it's created by people and for people who know that most of the "effort" is the computer mashing data together means that you aren't consuming the "art" in the same way that people generally understand "art" to be consumed, where the artist is communicating ideas, feelings, impressions through conscious choices. The output is instead reduced to something that looks "good enough" at a glance but usually falls apart if you try to engage with it more seriously. It's ultimately disposable, something to pop up on your screen and elicit a surface level response, and be scrolled past. It can't be more than that really, because there isn't anything more to it.
It's hard to imagine something that can more accurately be described as "slop." That doesn't necessarily make it "bad" but in terms of the production and consumption involved, that is what it is.
>>711274doki doki if it was good
oh god that game is soon to be 9 years old
I read the full thread and I have some questions but please don't think I'm trying to be discouraging or anything: if you can draw, why do you use AI?
I can draw myself and I thus I don't feel like using AI at all. I dont know. I just don't get it.
Having said that I just wanna establish that I don't consider AI to be le evil or anything. In fact, what I consider to be evil is that western companies released these technologies in spite of the environmental damage. This is entirely on the companies, not on the users (because if technology becomes available can you really blame the people for actually using it?). Also the copyright concerns are a meme to me because copyright shouldn't even exist in the first place and most artists are porkyminded shitheads.
And finally, I think AI art is actually art. The process of selection and training imo does say something about the person and their taste and even if you can make the argument that it isn't and will never be anything other than low tier art, its still art.
Anyway: why do you do AI art if you can already draw? I'm legit really curious.
OH as a final note, I think that the artists that are being replaced by AI slop were all just manually doing soulless slop only good for a quick dopamine hit anyway, no artist doing something above slop are in any way threatened lmao.
>>711256>Instead of the details being decided by an artist for a purpose, it's all just an echo of what other people have done before.does he know that humans just echo each other as well
>ooooooooooooh but the hvman sovl makes le original stuffspooky
>>711278>the environmental damagewouldn't be a problem if we weren't still using fossil fuels
>>711285I mean, AI isn't just an upgrade of drawing/painting, but more of a sidegrade. With AI you can very quickly do stuff that would normally take a lot of time and work but you lose a lot of control. For example in a normal drawing you can give every line an specific weight to convey an specific feeling or mood and you can be much more deliberate and selective about texture and really just about everything.
Anyway that post kind of answers my question: if you're not interested in what I previously said and you don't particularly like the process of drawing/painting then there's no reason not to use AI instead other than maybe libminded moralism. But to be fair, most artists aren't really doing stuff you can exclusively do drawing manually. Check the majority of twitter artists and they're really doing hyper-rendered porn slop that does not have any emotion "or soul" and that was obviously drawn only to give a quick dopamine hit. You see those drawings and they already look as if they were made by a robot, so why not actually get a robot to do that?
>>711283Yeah. Have the chinese already made a superior eco friendly generative AI shit? asking earnestly.
>>711278Thank you for the good faith question, it's appreciated.
>if you can draw, why do you use AI?It's just what I feel like doing, why does it need to be any deeper than that? It's a hobby, not my job. People already had this debate two decades ago when digital art was first becoming mainstream and painters and pen and paper artists were lamenting the fact that digital art "cheapened art" because it made the process of drawing so much simpler and made it so mistakes didn't matter anymore because you could always just press the undo button.
Also to the environmental impact… I do believe that has actually long since been disproven in the sense that AI is not some uniquely polluting thing like crypto or NFTs (where the impact is down to the fact that things like cryptomining requires you to constantly have your computer running and doing complex calculations, ie. lots of electricity usage), but it just comes with the nature of most big proprietary (mostly american) AI models (specifically LLMs) being run in massive datacenters that eat up a fuckton of electricity and thus the energy requirements mean more production of energy is needed to meet demand, which produces emissions because the majority of energy production is still done with fossil fuels. It's why there's a constant drive to optimize models to drive down costs even as they are constantly trying to also make them bigger and more complex, increasing the energy requirements.
By contrast when I generate an image locally off my computer, I am literally producing less emissions in the same time than if I'd spent that time running one of the latest AAA game titles because I can only use as much energy as my graphics card (the component doing the heavy lifting in generations) can handle and often only as much as the model needs, (which for the ones I use is a minimum of 8GB of VRAM) and the generation process for an image with 1536x1536 resolution takes like half a minute as opposed to a game having to constantly be rendering new frames onto the screen every second as long as its running.
>>711290>It's just what I feel like doing, why does it need to be any deeper than that?Well yeah. Pretty satisfying answer unironically lmao people can just do things.
As for the other stuff: I didn't know any of that! if damage to the environment is entirely dependent on electricity use and therefore things that are more taxing to a computer (and therefore use more electricity and are worse for the environment), then why is there such a panic regaring AI art and environmental damage and not the same amount of panic for crypto or even for high end games?
The people panicking over generative stuff and damage to the environment should be campaining for more 2D games that can run in a toaster, less bloated operative systems and for less (or no) crypto. Why aren't they?
Kinda starting to believe that the whole "AI art destroys the planet" thing is just manual slop-artists (I call them "porn rendermaxxers" because that really defines everything about them lol) feeling threatened.
>>711287>Yeah. Have the chinese already made a superior eco friendly generative AI shit? asking earnestly.no, but they make 95% of solar panels on the planet.
reminder luddites also seethed in the 1800s when photography came out
>noooooooooooooo you can't photograph, you're supposed to paint everything
but art just pivoted to less realistic styles instead of disappearing
then in the 90s when photoshop came out, luddites also seethed. they were like
>noooooooooooooo you're supposed to use ink and paper and all this other stuff and not do it digitally. digital is fake. it's unfair you get to just ctrl+z when you make a mistake. you have no real talent
and it's funny because in this latter case they didn't care about
>muh environment
they didn't care about
>muh trees for paper
or
>muh ecology of ink and paint sources
they just seethed at technology.
Now, with AI, they say
>ohhhhhhh you destroy muh environment reeeeeeeeee
but they don't really care about that. what's that? natural resources are consoomed by generative AI? yes. human artists also consoom natural resources when they make art. they sit in air conditioned homes and use the bathroom and eat refigerated food. I can't wait until the studies come out and show that not only is the AI faster but it consooms fewer resources than a human artist would over the several hours they would take to make a similar piece of art.
then there's
>muh plagiarism
petty bourgeois defense of intellectual property
>muh AI is only as good as the training data
so is the human brain
>AI just copies what humans have already done
so do humans. they learn from other humans and draw inspiration from past works
but what really kills me is the boycott attitude. they think they can stop or reverse this trend by seething about it and boycotting it. they're like knights and samurais disgusted by the introduction of gunpowder. They find the leveled playing field an insult to their nobility.
>muh talent
good for you that you spent all that time cultivating a talent. no really. but guess what. so did hand loom weavers who were replaced by power looms. This happens to other workers all the time.
and besides… do you really want art to be limited by the ability to do something? What about a person with no arms who wants to make art? What about a person with shaky hands or PTSD or parkinson's? This enables the disabled. But the disabled by definition aren't "talented" because they don't have "ability." Why restrict art to those who won a lottery and were gifted by nature with a brain and body capable of doing things?
>but i practiced hard! it's not fair!
some people aren't as lucky as you. they practiced even harder but couldn't accumulate talent because they didn't win the body brain lottery.
also people ignore that you can just mix trad art with AI art just like people mixed trad/digital art before. Draw something on paper, scan it, add some details, use AI for finishing touches.
Or the reverse order
>>711292>Why is there such a panic regaring AI art and environmental damage and not the same amount of panic for crypto or even for high end games? Because AI is the hot new thing being actively pushed by porky as a perceived cost-saving measure while crypto is already old news by now in the public discourse. That and it's perceived as directly threatening to the livelihood of what we would consider the mass of low-to-mid level artists that used to just about get by drawing people's DnD characters and furry OCs for a decent amount of cash because those people didn't really have an alternative besides trying to learn how to draw themselves or befriend an artist who was willing to draw an illustration for free. Those who've already reached a high-enough level of skill in their field don't really care one way or another because they know they can make stuff that's better than a newly hired intern being told to pump out assets with AI, or they make it a natural part of their workflow just like they've done with every other tool that's come along over the years.
Now those people can just run a program off their computer and get an okay-ish result if they know what they're doing without having to pay anyone a cent (or if they don't have a powerful enough computer, buy a subscription to one of the hundreds of image generation services) I suspect that the direct threat to self-interest is a bigger factor than the environmental impact for many who are actively crusading against AI (at least when it's not just an influencer or video essayist jumping on a bandwagon because hating on AI is trendy and an easy way to score social capital), the environment is a convenient excuse to join the hate brigade so they can feel a sense of moral righteousness when they head out to witchhunt anyone they think might be using AI. Even in situations where they admit they can't be 100% certain that someone used AI to make something, they will still say "we just know, okay?" and when they inevitably get false positives, they blame AI for making them paranoid.
>>711279>does he know that humans just echo each other as wellThey choose when and how to do it, not just copying a pattern based on a statistical model.
>>711278If you have the know how and skill to build furniture from scratch, then why buy furniture made in factories?
>>711299>>Now those people can just run a program off their computer and get an okay-ish result if they know what they're doing without having to pay anyone a cent (or if they don't have a powerful enough computer, buy a subscription to one of the hundreds of image generation services)you can just make tons of shit on huggingface for free
>>711302>They choosespooky
>not just copying a pattern based on a statistical model.this is literally what the brain does. the idea that it's a "choice" is just an illusion.
>>711450Nobody said humans invent things from scratch. The point is that people iterate and change based on a conscious design, not just eternally echoing what has already been done. Generative AI is not capable of producing the kind of iterations you posted because they require further developing something, not just repeating what already exists.
>unironically being a religious retard who treats making AI slop as some cardinal sin that taints your soul or whatever
jfc just call it slop and move on with your life
>>712268where's the nipple
I heard novelai is pretty good and you get 30 free generations per email adress
>>713949I can tell from experience that it's pretty decent at following prompts these days. They've also got a tool now to put in a character reference so if there's a character with very little to no existing fan art (like say your super unique OC), you just take a decent reference pic (meaning something like a clear frontal view of the full body with a simple background) from the source and put it into the reference tool and boom, it's able to recreate it. First pic is the reference for Toki (From the Ghibli film "Princess Mononoke") that I used, she has less than ten pieces of fanart in existence from what I could find, so normally not anywhere close enough for any model to inherently know what she looks like. The second is the NSFW piece I was able to make because of that reference that I otherwise wouldn't have been able to.
This means that technically you could make Alunya with it and I'm confident she'd come out looking pretty decent. Haven't tried yet myself but I'd be willing to give it a try if people would like that.
secks
>>718632requesting Karl Marx, Gilles Deleuze, Guy Debord and Mark Fisher in Disco Elysium style
>>718632Could it create Starsector or other game portraits?
>>719412Sora 2 isn't available in my country so i get what i can get, run it again see if you get something cooler
i asked it to give another hand to this neoclassical Kali it fucking got wrong the first time
AND IT GOT IT WRONG THE 2ND TIME LMAO
>>719403Please sit on my face!
>>719292>>714942god why are men with firm bellies so inticing to women?
>>719635uoh secks
>>719675also segs
>>711119> it tries to mash multiple different gun models together when all I want is a glock.can't you train a separate model on guns and merge it with your pic gen model
(I have no idea how anything works)
>>719939AI sucks for making guns
3D gun nerds have to rely on other gun nerds constantly pointing out stuff we missed because finding reference for this stuff is usually such a pain.
>>719939It only works up to a degree in that it will get the overall silhouette right, but it will still screw up individual parts (such as the fire selector on an AKM), and all SDXL-based LoRAs that are available publicly are specialized for making one specific kind of gun. Flux Loras are better at following prompts and generally just more accurate, so if you have a LoRA that makes most common types of guns in general look better and specify that you want a 1911, you'll get about the result of what I made with
>>718632, where the mistakes are so minor that you can easily edit them out by hand.
>>720076This is also still true though, if you want your gun to look right, you still gotta know guns yourself to be able to both spot and edit out mistakes. It's why I recently visited a museum that had a display section full of old weapons all the way from muskets to the first half of the 20th century.
>>720086i'm a huge gun nerd, but out of memory you can only remember so much.
which is what blueprints and references are for.
one misshape and you'll have every commenter and their mom saying how the angle is 30° off
>>720078what are your text prompts like to get the specific results you want
>>711117>always provide sauce to credit the original author.Absolute state of artist faggot protecting their private property.
>>720171Depends on the model, what infrastructure it's working off of, and what you're looking for. With solo pictures like the Alunya one, it's easier because you only gotta worry about the details of one character. With multiple characters, you gotta stop and think about formatting to avoid having details bleed in between characters (something I struggled with early on). The more specific you are with your prompts and the higher the CFG (Classifier-Free Guidance, basically how closely the model should follow the prompt), the less randomness the model will introduce when it's in the sampling phase of its generation process. However, especially with SDXL-based models, there is the issue that if you crank the CFG too high, combined with the wrong kind of sampler, it'll just result in your image having eye-bleedingly bright contrast, which you then have to fix by adjusting the color values, brightness, and so on in Krita's native filters.
Just as an example, here's the prompt I used for the Alunya piece.
Style Prompt (This is a Krita thing, it's mainly for if you have a style preset that you've set up for a specific workflow): expert shading, masterpiece, best quality, amazing quality, absurdres, very aesthetic, 5toes, 5fingers, {prompt} (←- This is where the rest of the prompt starts from)
Common Text Prompt (basically the one that applies to the whole picture, this is where I usually put stuff that's in the background):
sfw, abstract, disco elysium, expressionist, harsh brushstrokes, painterly, traditional media, outdoors, urban, cityscape, park bench, trash can, dirt road, grass, tree, motion lines, sunlight, white wooden sign, metal fence, english text written in big bold red letters reading "NO LEANING ON THE SIGN!" and "(FINE UP TO 100$)" written underneath in smaller text,
Regional Prompt (Basically what you apply to, like an individual layer or layer group): BREAK (←- This is a way to separate sections of the full prompt manually to avoid bleeding) 1girl, solo, full body, catgirl, toned female, biceps, black cat ears, pink inner ear, black cat tail, short hair, black hair, bob cut, red eyes, eyelashes, thighs, hips, breasts, choppy bangs, thick eyebrows, pointy hair, matte skin, teeth, fang, smug, open mouth, smirk, standing, facing viewer, leaning on white wooden sign, leaning forward, standing on one leg, looking at viewer, black dress, pleated skirt, short sleeves, red belt, black boots, pointy red and black bandanna around neck,
Negative: Empty (Models based on Flux-architecture don't use negative prompts anymore)
Unique IPs: 42