It doesn’t really feel like “art” in the making. When I’ve used AI to create an image, it doesn’t feel different from using search terms and tags on an imagebooru, or trying to find a piece of clip art for a presentation.
I think there might be fruit for exploration in digital collage, training ones on models in creative ways… I’m not really seeing anyone using these tools to really “do art” though. I’m seeing lots of anime girls, porn, ShrimpJesus Facebook slop, hamfisted political comics, and occasionally an “artist” crowing over like a generic image of a tiger. I’d like to see better, but I’m not.
Also - if you like making art, I don’t understand the appeal of taking out “process.” You type some keywords, you adjust them if you don’t like what you see.
This might be more personal preference, but something that I’ve come to enjoy working with paint is that you have to wait for it to dry. That it splatters and doesn’t always go where you want it. That the image you have in your head will not ultimately be the image you get on the canvas. That sometimes it’s a process of weeks of dialogue between you and the canvas.
A lot of AI art enthusiasts do seem fixated on product, not process. I don’t know if you are really an “artist” if there isn’t some element of “process” that you are involved with.
One of my favourite small indie bands trained and used AI to make music videos for their most recent album. They were very upfront about it basically saying it was the only way they could have made the type of video they wanted as they simply wouldn’t have had the resources to do it another way.
I still don’t like AI art and I don’t feel great about it but this is the closest to a legitimate application of AI art I’ve come across.
I think I’ve encountered one person that could reasonably be said to be using AI generators to make art, in that a discord server I’m in used to have a guy that made something of a hobby out of trying to get chatbots and coding AIs to make “shaders” (I don’t know exactly what this implies, since what he posted weren’t recognizable as shaded images but some kind of abstract patterns or shapes). He was always talking about tweaking some technical aspects of various models, that I didn’t really get the terminology of to understand, and seemed to spend a lot of time messing around with them to only occasionally get something he found interesting enough to share. It wasn’t how people typically use “AI art” generators though, for sure.
That video looks like designing a video game asset to me, or putting together a map for W40k match. Part of that might be some personal bias because I don’t really have an appreciation for depictional fantasy art.
One thing I’ve noticed with AI art which I think is holding the genre back as a whole is a focus on realism, especially photorealism, or at least staying vary far away from abstraction. Other styles are largely stolen - like, I’m sure it’ll be happy to make a Basquiat of Ashoka Tano or a Klimt “The Kiss” of Anakin and Padme. But that’s the options as far as moving away from photorealism - the style of artists that already have their stuff in museums.
This also seems to tie into the “AI helps people who don’t have the time/ability to develop drawing skills” argument - that the only way many folks can even conceive of making “good art” is getting really good at drawing photorealistic faces.
It doesn’t really feel like “art” in the making. When I’ve used AI to create an image, it doesn’t feel different from using search terms and tags on an imagebooru, or trying to find a piece of clip art for a presentation.
I think there might be fruit for exploration in digital collage, training ones on models in creative ways… I’m not really seeing anyone using these tools to really “do art” though. I’m seeing lots of anime girls, porn, ShrimpJesus Facebook slop, hamfisted political comics, and occasionally an “artist” crowing over like a generic image of a tiger. I’d like to see better, but I’m not.
Also - if you like making art, I don’t understand the appeal of taking out “process.” You type some keywords, you adjust them if you don’t like what you see.
This might be more personal preference, but something that I’ve come to enjoy working with paint is that you have to wait for it to dry. That it splatters and doesn’t always go where you want it. That the image you have in your head will not ultimately be the image you get on the canvas. That sometimes it’s a process of weeks of dialogue between you and the canvas.
A lot of AI art enthusiasts do seem fixated on product, not process. I don’t know if you are really an “artist” if there isn’t some element of “process” that you are involved with.
One of my favourite small indie bands trained and used AI to make music videos for their most recent album. They were very upfront about it basically saying it was the only way they could have made the type of video they wanted as they simply wouldn’t have had the resources to do it another way.
I still don’t like AI art and I don’t feel great about it but this is the closest to a legitimate application of AI art I’ve come across.
I think I’ve encountered one person that could reasonably be said to be using AI generators to make art, in that a discord server I’m in used to have a guy that made something of a hobby out of trying to get chatbots and coding AIs to make “shaders” (I don’t know exactly what this implies, since what he posted weren’t recognizable as shaded images but some kind of abstract patterns or shapes). He was always talking about tweaking some technical aspects of various models, that I didn’t really get the terminology of to understand, and seemed to spend a lot of time messing around with them to only occasionally get something he found interesting enough to share. It wasn’t how people typically use “AI art” generators though, for sure.
This sounds like shaders used in graphics pipeline, which is something you code. So probably using a LLM.
I’m curious how you feel about something like this https://www.reddit.com/r/StableDiffusion/comments/19d8uwo/inpainting_is_a_powerful_tool_project_time_lapse/
That video looks like designing a video game asset to me, or putting together a map for W40k match. Part of that might be some personal bias because I don’t really have an appreciation for depictional fantasy art.
One thing I’ve noticed with AI art which I think is holding the genre back as a whole is a focus on realism, especially photorealism, or at least staying vary far away from abstraction. Other styles are largely stolen - like, I’m sure it’ll be happy to make a Basquiat of Ashoka Tano or a Klimt “The Kiss” of Anakin and Padme. But that’s the options as far as moving away from photorealism - the style of artists that already have their stuff in museums.
This also seems to tie into the “AI helps people who don’t have the time/ability to develop drawing skills” argument - that the only way many folks can even conceive of making “good art” is getting really good at drawing photorealistic faces.