I had some time to try three of the current AI art generating platforms. Confirmed some initial impressions, formed some thoughts of both it's utility, and threat to arts professionals.
Hey Max, interesting thoughts on Midjourney. To be honest, I'm in the horrified camp. I was following this photographer on Instagram and suddenly he started posting everyday: yup he's a "concept artist" now (using AI). It pissed me off and I unfollowed him. If, as I have heard and read many times, attention is the most sought after commodity and now virtually anybody can generate and post "concept art" within a minute, does it not change the game (in a bad way) for the artists who toiled hours to generate a piece?
Hi Daniel! Well to get something actually decent it takes a bit more than a minute, but a lot less labor than doing it ourselves for sure. I see it very likely being used as we do already, to generate stuff that doesn't have to be too specific in its parameters, just eye catching. And often it will be a starting point that people then paint over to bring the last mile. Saving them a lot of time for the gram likes.
But I think if the plan was to build a career mainly on attention then yeah, humans are going to be unable to keep up with humans using AI image generation in some ways. Though people will still always be far more impressed when it's all "done by hand".
However AI is pretty crap at doing very controlled specific things, so stuff like my job as an animation designer is pretty safe for the moment. And with something like comics it might work for an writer who wants to be able to just go and do it for cheep rather than working with a human artist, and is ok with getting something sort of like what they had in mind? In the end for them having a human do their take on what they wrote, and having an AI do it, is probably pretty similar feeling.
But it won't take notes, and they will need to do their own page layouts, and all the other things. The AI will ONLY spit out images, any time it tries to do a comic page or graphic design it's pretty loose at best and has no intention or thought behind the work. It's not laying out panels for readability or flow. So the comics I've seen done via AI are interesting, but pretty standard at best for page design, and that was from a professional AD so someone with a design background.
Seems the page won't display longer text so broke this up:
AI also still screws up some fundamental stuff a lot, and can be pretty unpredictable and inconsistent. So while I can definitely see it becoming a tool in professional work I don't see it replacing us in professional work at this stage.
And the thing is it'll probably be easier to build tools into it so that a human designer can use it as a tool better, make revisions and direct it as it's generating more. Than it will be to give the AI enough of a 'mind' and true general AI to do all the thinking that goes on behind good art. I don't think people really realize how far we are from that level of processing power. Folks like Elon make it sound like it's around the corner but I seriously doubt it.
It'll take basically billions of CPU to model what Billions of Neurons can do. Globally there are something like 800k to a couple million CPU/GPU dedicated to crypto, because there's money they think in that. In all there's something like just over 2 billion computers in the world at this time. And THAT? isn't enough to truly. emulate a mind.
Do you see anyone commandeering two or tree times the worlds current processing power, to dedicate it to generating concept art? Might be technically possible but no one is going to do that. So computing tech will need several orders of magnitude improvements in performance before actual General AI is possible, and then I don't think art is the first thing they will use that for.
Hi again Max, in your pod you mention how you have more ideas for projects than you'll ever have time to work on (I suffer from this too, I even thought about hiring you at one point) and AI might be helpful. It's an interesting thought. I've heard in an interview how it can enable people with disabilities to create work, and I've also seen a 3D artist (Bryn Jones) use it to create a texture or color scheme for his work. All very cool.
I agree with you that people are more impressed by "done by hand" work.
I've not considered or given thought t how much processing power is needed to emulate a mind or to improve Midjourney. However, I think it's telling that an artist like Dave Mckean is worried (I'm listening to his with Palle Schmidt now and I've also listened to a longer one with the Living The Line guys). Now one could argue that he's worried because Midjourney is good at making images that look like his work but one could also say that he's an artist who's at the forefront of using new tools, so ...
In any case, you've moved back to drawing on paper so ;)
Hey Max, interesting thoughts on Midjourney. To be honest, I'm in the horrified camp. I was following this photographer on Instagram and suddenly he started posting everyday: yup he's a "concept artist" now (using AI). It pissed me off and I unfollowed him. If, as I have heard and read many times, attention is the most sought after commodity and now virtually anybody can generate and post "concept art" within a minute, does it not change the game (in a bad way) for the artists who toiled hours to generate a piece?
Hi Daniel! Well to get something actually decent it takes a bit more than a minute, but a lot less labor than doing it ourselves for sure. I see it very likely being used as we do already, to generate stuff that doesn't have to be too specific in its parameters, just eye catching. And often it will be a starting point that people then paint over to bring the last mile. Saving them a lot of time for the gram likes.
But I think if the plan was to build a career mainly on attention then yeah, humans are going to be unable to keep up with humans using AI image generation in some ways. Though people will still always be far more impressed when it's all "done by hand".
However AI is pretty crap at doing very controlled specific things, so stuff like my job as an animation designer is pretty safe for the moment. And with something like comics it might work for an writer who wants to be able to just go and do it for cheep rather than working with a human artist, and is ok with getting something sort of like what they had in mind? In the end for them having a human do their take on what they wrote, and having an AI do it, is probably pretty similar feeling.
But it won't take notes, and they will need to do their own page layouts, and all the other things. The AI will ONLY spit out images, any time it tries to do a comic page or graphic design it's pretty loose at best and has no intention or thought behind the work. It's not laying out panels for readability or flow. So the comics I've seen done via AI are interesting, but pretty standard at best for page design, and that was from a professional AD so someone with a design background.
Seems the page won't display longer text so broke this up:
AI also still screws up some fundamental stuff a lot, and can be pretty unpredictable and inconsistent. So while I can definitely see it becoming a tool in professional work I don't see it replacing us in professional work at this stage.
And the thing is it'll probably be easier to build tools into it so that a human designer can use it as a tool better, make revisions and direct it as it's generating more. Than it will be to give the AI enough of a 'mind' and true general AI to do all the thinking that goes on behind good art. I don't think people really realize how far we are from that level of processing power. Folks like Elon make it sound like it's around the corner but I seriously doubt it.
It'll take basically billions of CPU to model what Billions of Neurons can do. Globally there are something like 800k to a couple million CPU/GPU dedicated to crypto, because there's money they think in that. In all there's something like just over 2 billion computers in the world at this time. And THAT? isn't enough to truly. emulate a mind.
Do you see anyone commandeering two or tree times the worlds current processing power, to dedicate it to generating concept art? Might be technically possible but no one is going to do that. So computing tech will need several orders of magnitude improvements in performance before actual General AI is possible, and then I don't think art is the first thing they will use that for.
Hi again Max, in your pod you mention how you have more ideas for projects than you'll ever have time to work on (I suffer from this too, I even thought about hiring you at one point) and AI might be helpful. It's an interesting thought. I've heard in an interview how it can enable people with disabilities to create work, and I've also seen a 3D artist (Bryn Jones) use it to create a texture or color scheme for his work. All very cool.
I agree with you that people are more impressed by "done by hand" work.
I've not considered or given thought t how much processing power is needed to emulate a mind or to improve Midjourney. However, I think it's telling that an artist like Dave Mckean is worried (I'm listening to his with Palle Schmidt now and I've also listened to a longer one with the Living The Line guys). Now one could argue that he's worried because Midjourney is good at making images that look like his work but one could also say that he's an artist who's at the forefront of using new tools, so ...
In any case, you've moved back to drawing on paper so ;)
I love the idea of distributing zines in neighbourhood libraries!
Though my real dream is building and decorating a mini neighbourhood library :P