AI art in compos
category: general [glöplog]
I feel this whole AI text to image charade is such a sad sad misuse of computers, code and people. Of hadware, software and wetware. Basically nothing more than a further trivialization of already painfully trivial culture. More of the same. But behold how expensive it is. What a waste of resources. What a wasted opportunity to create something truly original. What, this instead? So this is what all your pictures, conviniently tagged were for? I’m so dissapointed. What a weak primitive mind it was that set this idiocy to motion. Then again, if this whole ‘art’ side-job serves to enhance the general AI project, then I guess it’s OK. But to see so many people indulge in this shit, obsessing over it is so sad. Friends too.
@farfar: sounds like you want to replace "freestyle graphics" with a "digital painting" competition. Because "freestyle" already suggests that it's neutral to the technique.
Separating like that might make sense. But it's still very unclear how to enforce that separation in a meaningful way.
Separating like that might make sense. But it's still very unclear how to enforce that separation in a meaningful way.
I believe 'AI will replace artists' as a general argument is a bit exaggerating. We are talking about 2D digital image generators based on industry aesthetics (excessively ornamental post production effects that are already available in graphics editors anyway and used to the death by creators) with otherwise major limitations in pretty much every aspect. Art is about material, concept, technique, space, time, senses and so on. I don't see Midjourney or Dall-E creating installations, or sculpture, or brush stroke qualities. Sure it may produce results that are stimulating to certain peoples' eyes but that's about how deep it goes.
That said, AI image generators sure are fun as toys. For artists are yet another tool that some may use, and others will ignore. Also may serve as a base for interesting philosophical talks.
With regards to whether Midjourney or Dall-E or others will play a role, and to what extend, in certain industries that use 'art' for (or as part of) their high-volume production/consumption products (and thus affect jobs), this remains to be seen. But this has always been happening and it even is very well recorded historically since the industrial revolution.
You may read about the Arts and Crafts movement story (https://en.wikipedia.org/wiki/Arts_and_Crafts_movement), which is an even better example of technology claiming to replace craftsmen than the photography replacing figurative painting one.
That said, AI image generators sure are fun as toys. For artists are yet another tool that some may use, and others will ignore. Also may serve as a base for interesting philosophical talks.
With regards to whether Midjourney or Dall-E or others will play a role, and to what extend, in certain industries that use 'art' for (or as part of) their high-volume production/consumption products (and thus affect jobs), this remains to be seen. But this has always been happening and it even is very well recorded historically since the industrial revolution.
You may read about the Arts and Crafts movement story (https://en.wikipedia.org/wiki/Arts_and_Crafts_movement), which is an even better example of technology claiming to replace craftsmen than the photography replacing figurative painting one.
Not really no - you don't have to digitally paint something to create a composition for example. That's neutral to the technique as you say.
I think the freestyle graphics compo should be for artworks that have been created by artists themselves, meaning that the vast majority of the creative decisions have been made by the artists themselves. (so I guess I subscribe to the metoikos viewpoint earlier in the thread, and what darya said) I don't know if that's naive or backwards or too difficult or whatever.
If we want to incorporate AI artworks into scene compos, I think they should be in their own compo category as I said initially.
I think the freestyle graphics compo should be for artworks that have been created by artists themselves, meaning that the vast majority of the creative decisions have been made by the artists themselves. (so I guess I subscribe to the metoikos viewpoint earlier in the thread, and what darya said) I don't know if that's naive or backwards or too difficult or whatever.
If we want to incorporate AI artworks into scene compos, I think they should be in their own compo category as I said initially.
sorry, that was @kusma obviously :)
farfar: In that case, I'm back to thinking you just don't like AI stuff and want to arbitrarily exclude that. As I said, the point of the "freestyle" graphics compo is to allow all techniques. Drawing a line at AI (which as I've pointed out before is fuzzy, so it doesn't really work) is just arbitrary. You might as well not like lens flares, and exclude that.
I don't think you can come up with a sane explanation of what "created by artists themselves", and that's exactly the point I've been trying to make. There's a continuum of techniques that range from "manually colored every pixel" to "computer drew some images and I chose the one I liked". You have so far offered no explanation of where on that continuum you want to draw the line.
Didn't you just say that creating a composition could be enough right before?
If I understood what metoikos was trying to say correctly, that was actually calling for a digital painting compo rather than a freestyle graphics compo.
As far as I understand Darya's argument, that's really just speaking against the "the computer makes all of the decisions" end of the spectrum, but I think her position has the same problem as your: How on earth do we define that?
Sounds more to me like you don't want it to be
Quote:
I think the freestyle graphics compo should be for artworks that have been created by artists themselves,
I don't think you can come up with a sane explanation of what "created by artists themselves", and that's exactly the point I've been trying to make. There's a continuum of techniques that range from "manually colored every pixel" to "computer drew some images and I chose the one I liked". You have so far offered no explanation of where on that continuum you want to draw the line.
Quote:
meaning that the vast majority of the creative decisions have been made by the artists themselves
Didn't you just say that creating a composition could be enough right before?
Quote:
so I guess I subscribe to the metoikos viewpoint earlier in the thread, and what darya said
If I understood what metoikos was trying to say correctly, that was actually calling for a digital painting compo rather than a freestyle graphics compo.
As far as I understand Darya's argument, that's really just speaking against the "the computer makes all of the decisions" end of the spectrum, but I think her position has the same problem as your: How on earth do we define that?
Quote:
I don't know if that's naive or backwards or too difficult or whatever.
Sounds more to me like you don't want it to be
@kusma: hmm.. well you've challenged me a few times whether I just don't like AI or not. I don't have a problem with the technology itself - it's a tool. I do have a problem with AI using existing art as raw material without the original creator's knowledge or constent (the issue that gargaj outlined) - but that's a dislike of the current wild west state of AI where midjourney and others are even profiting from it.
What I meant was you don't have to practice digital painting in order to create a composition. You can use 3d, make a collage, paint or draw something physically, paint with math like IQ does - I just meant I don't have a desire to alter freestyle to a digital painting compo. I think it works fine the way it is / has been working up until now. But AI (for me) changes the game a bit.
In terms of compos I think they should be a competition of skill between people... it's about people and their creativity, and what they can do with the tools at their disposal. What they can do with the tools at their disposal.
Then we can resume the discussion about content aware fills and other procedural methods, and where the line should be drawn between "manually placing every pixel" and "AI generating everything for you" .. I do agree that it's hard to define an objective rule for this - in the same way it's hard to define objective rules for preselecting music. We still manage.
Question! What do you think about this? -> if it's okay to enter an AI image in a freestyle compo, where the extent of your own creative actions has been to enter a query of what you wanted the AI to produce - then is it also okay to enter a compo with something you searched for on pinterest or google images? (or artstation?) You more or less performed the same action in both cases.
Quote:
Didn't you just say that creating a composition could be enough right before?
What I meant was you don't have to practice digital painting in order to create a composition. You can use 3d, make a collage, paint or draw something physically, paint with math like IQ does - I just meant I don't have a desire to alter freestyle to a digital painting compo. I think it works fine the way it is / has been working up until now. But AI (for me) changes the game a bit.
In terms of compos I think they should be a competition of skill between people... it's about people and their creativity, and what they can do with the tools at their disposal. What they can do with the tools at their disposal.
Then we can resume the discussion about content aware fills and other procedural methods, and where the line should be drawn between "manually placing every pixel" and "AI generating everything for you" .. I do agree that it's hard to define an objective rule for this - in the same way it's hard to define objective rules for preselecting music. We still manage.
Question! What do you think about this? -> if it's okay to enter an AI image in a freestyle compo, where the extent of your own creative actions has been to enter a query of what you wanted the AI to produce - then is it also okay to enter a compo with something you searched for on pinterest or google images? (or artstation?) You more or less performed the same action in both cases.
Just my 2 cents.
Obviously midjourney and others are suffering from the problem of "ethically sourcing" the material. The technology itself is pretty impressive and we should embrace it in some form or another.
An idea I had a couple months ago when first stumbling upon midjourney was to make a themed compo with some restrictions and have people come up with the best (or funniest) promt and judge the result.
The idea would not be to make it a super serious compo but to start integrating the technology (and the awareness of it - which I'm sure not everybody is aware of) into our events and the community because - let's face it - it'll be there and it's going to change stuff. If it'll change it for the better or worst - we'll see.
I remember the outcries with people using dx in 64ks and yelling that "that's cheating etc.). So yeah, we overcame those times and I'm confident that we'll find a way to cope with this new technology as well.
\o/
Obviously midjourney and others are suffering from the problem of "ethically sourcing" the material. The technology itself is pretty impressive and we should embrace it in some form or another.
An idea I had a couple months ago when first stumbling upon midjourney was to make a themed compo with some restrictions and have people come up with the best (or funniest) promt and judge the result.
The idea would not be to make it a super serious compo but to start integrating the technology (and the awareness of it - which I'm sure not everybody is aware of) into our events and the community because - let's face it - it'll be there and it's going to change stuff. If it'll change it for the better or worst - we'll see.
I remember the outcries with people using dx in 64ks and yelling that "that's cheating etc.). So yeah, we overcame those times and I'm confident that we'll find a way to cope with this new technology as well.
\o/
Quote:
I remember the outcries with people using dx in 64ks and yelling that "that's cheating etc.).
I genuinely hate the short-sighted comparisons between a new technology that expands on human creativity and a new technology that reduces the need for it.
Quote:
Quote:I remember the outcries with people using dx in 64ks and yelling that "that's cheating etc.).
I genuinely hate the short-sighted comparisons between a new technology that expands on human creativity and a new technology that reduces the need for it.
Well I see potential in new technology for lesser-skilled people - which technology generally is supposed to augment. IMO at least.
Quote:
Question! What do you think about this? -> if it's okay to enter an AI image in a freestyle compo, where the extent of your own creative actions has been to enter a query of what you wanted the AI to produce - then is it also okay to enter a compo with something you searched for on pinterest or google images? (or artstation?) You more or less performed the same action in both cases.
@farfar: That one is easy; no, that's not OK. That's clear-cut copyright violation. And there's other common compo-rules against this (previously released material, mult be self-made etc). That's the big difference here.
Does the world need to figure out where the boundaries should lie with copyright and AI? Sure. Is compo rules the right place for that? I don't think so.
But also, you keep on ignoring the point that I keep on pointing out: There's no reasonable way of defining compo rules that does what you want it to.
Compo rules needs to be something that both the organizers and attendees can read and understand what means, and I still after all this discussion have not been able to get anything out of you that resembles a definition... If you come up with one that makes sense, I'll reverse my position, but I doubt that's going to happen.
@kusma yeah, I appreciate that we don't seem to be getting anywhere :)
Quote:
Well I see potential in new technology for lesser-skilled people - which technology generally is supposed to augment. IMO at least.
There's a difference between "augment" and "replace".
Quote:
There's no reasonable way of defining compo rules that does what you want it to.
Sure there is: "Fully AI generated pictures are not allowed."
If people wanna paint over one? Sure, I might even be agreeable with that, and the steps will let the audience decide anyway.
@Gargaj That's a valid point but one may argue that practice of illustration using a computer has mostly reduced human creativity anyway.
But to narrow the talk to demoscene compos; would it be compo-valid if someone used midjourney to generate some pics but not to use them as they are; rather as ingredients for creating a compo-pic?
But to narrow the talk to demoscene compos; would it be compo-valid if someone used midjourney to generate some pics but not to use them as they are; rather as ingredients for creating a compo-pic?
Gargaj: You still need to define what "AI generated" means, that's not a clear-cut term at all.
Quote:
would it be compo-valid if someone used midjourney to generate some pics but not to use them as they are; rather as ingredients for creating a compo-pic?
I'd be okay with that, because from that perspective it's not that different from stock textures, photos, etc. In that case it really does become just a tool - though of course the "ethical sourcing" questions still remain. From the scene perspective I'm more worried about the cases where the effort is reduced to "type in prompt, wait, rightclick, save, submit to compo".
Or what if someone generates an image, and horizontally flips it, or croos it slightly... Where should the line go?
Ah ok, so I got my answer before asking.
Then, perhaps compos need a new approach. Perhaps similar to 'traditional-ish' call for artists, where the organizers of exhibitions set the theme and the rules that the artists have to follow. So, regardless media/materials used, the participant needs to address other, deeper values, before committing to the actual result.
Isn't Evoke that sets a further rule to pixel-art compo by setting a certain palette the participants need to use? Perhaps this needs to be taken further.
Then, perhaps compos need a new approach. Perhaps similar to 'traditional-ish' call for artists, where the organizers of exhibitions set the theme and the rules that the artists have to follow. So, regardless media/materials used, the participant needs to address other, deeper values, before committing to the actual result.
Isn't Evoke that sets a further rule to pixel-art compo by setting a certain palette the participants need to use? Perhaps this needs to be taken further.
Quote:
Gargaj: You still need to define what "AI generated" means, that's not a clear-cut term at all.
That's an unnecessary nitpick: we're not trying to file a legal suit here, but inform the reader and potential competitor what the general understanding and direction is - at some point you have to rely on the moron in a hurry, because otherwise 99% of the compo rules are not well defined. Edge cases and gray areas are always (and always have been) left up to the discretion of the organizer.
Quote:
Or what if someone generates an image, and horizontally flips it, or croos it slightly... Where should the line go?
Here's a non-AI example: I distinctly remember lug00ber and I disqualifying a tune from one Solskogen that was an edit of Noisia's Block Control - from the perspective of fair use, it may even have been legit (none of us are lawyers, obviously), but it wasn't in the spirit of the competition.
Quote:
That's an unnecessary nitpick
I don't think it is, and I've already explained why: Compo rules needs to be something that both the organizers and attendees can read and understand what means
If not, we end up wasting both the organizers and the contestants time. I've already spent enough time arguing with contestants about compo rules over much clearer rules than that.
What you suggested is not clear at all.
Quote:
Here's a non-AI example
That's neither relevant, nor answering the question. We need the rules to be understandable, so people can create entries without being paranoid of breaking the rules.
WAnd as you point out, we all have catch-all rules that already allow us to disqualify blatant stuff. Do we really need to be paranoid about what might not even turn out to be a problem, and make the rules needlessly convoluted in the process?
It's clear enough for 95% of the target audience, both organizers and attendees.
Sure, pull out some numbers out of your ass. I'm going to drop this debate now, because this is turning more toxic than I'm willing to deal with.
Quote:
That's neither relevant, nor answering the question.
It does both though - both scenarios are situations where it may pass the strict "textualist" reading of the compo rules, but are not in the spirit of the competition and organizers are welcome to decide against allowing the entry in their own discretion, which is what happens with entries that either deliberately or involuntarily skirt the rules. This is how it's always been.
Competition rules have never been a letter of law, but a description of mutual understanding between the audience and organizers how competitions will be run, with the same mutual understanding there are both gray areas, room for negotiation and good faith - otherwise wild compos practically wouldn't exist. I firmly believe (or rather, want to believe, for my own sanity) that with a situation like this, the overwhelming majority of organizers and attendees would read a line like "Fully AI generated pictures are not allowed", or variation thereof, and would come to the same definition of what that means.