OpenXcom Forum
OpenXcom => Offtopic => Topic started by: LuckyClover on January 28, 2023, 11:42:17 am
-
What do You think about ai generated art?
Is it more opportunity or threat?
This technology will make life easier for many people, but artists are often really affraid of it.
Another question is how would You call person that create those images-artist?
Do You have other idea for name?
In Polish language I use word "operator"-Im not sure if people would like it :)
Here is also elf (or maybe he looks more like halfling?) picture generated few days ago. I kinda like it so Ieven use it as avatar :)
-
it will just create a (bigger) copyright hell
with a tiny chance that the hell may freeze over
-
The copyright "law" can't do a thing against progress. The bloodsuckers will squirm and bite, and certainly inconvenience a lot of people, but they can't win against the technological reality. Personally, I don't think it's worth thinking about, when there are endless possibilities of using this amazing new technology for an actual benefit of humanity.
As for the word for a person operating it, I think "prompter" would fit, since it's all about writing prompts. :)
Also, definitely a halfling. ;)
-
How do you call them? What does it matter? Would you call anyone who uses a hammer (regardless of their job) "a hammerer"? (funnily enough the process of work with this AI does remind me of mining + hammering).
I know many people who are afraid of it, none of them is an actual artist. Those who lose jobs are paintbots in human form, it's not a moment to soon this job (which really became a market only little over a decade ago) is gone. Meanwhile actual artists I know grabbed this AI thing like it was a box of candies...
A book can be written on how much that does change, regarding impact on society, on art, on jobs, on our views of such things as originality and immortality, but let's just say it's a very momentous advancement.
Attached is example of my use of it in xcom modding.
-
First thing is that current version of AI art is not creative, it effective collage of millions of pictures that was used to train this AI.
You can have have "creative" mix like `X look like Y` but it can't create some unique Z, this mean true artists could always be one step ahead
because they could change style or make pictures that are very hard to AI to recreate.
For not-artists AI is blessing as it allow create custom arts that is not accessible because of skill level.
Like Dioxine show, modders could easy use AI to create content that match description and have specific style.
Another thing I see that could be great is to write books even if your language skills are low, I could image some plot, characters and AI would
fill all text and dialog that normally could be bizmo if written by me. And I could even add images that match plot and characters.
-
Those are not “AI generated”, they are AI remixed.
It looks like some bright bulbs already have started “hire an artist to make a basic set of samples, fire the artist, start auto-remixing” routine, but they are likely to be crushed quickly and thoroughly, probably with support of other publishers.
Technically, there are few differences between AI-controlled and manually controlled image editor, and they are not necessarily provable. And then, AI can be set up to do only negligible adjustments… you see the picture (sorry).
The following assumes the status quo in the powers over Internet. Things would be very different in conditions of e.g. Swiss (with approval of Russia and China) emerging as the keeper of most Internet infrastructure that matters, while holding in absolute contempt Disney, Hasbro, Salon com and Californian courts, while EU is confused by all this — but there’s no point trying to predict details of turns this sharp. :D
If AI remixes were somehow exempt — however shaped, this de facto would effectively kill both copyright (protection of authors, mainly from publishers) and IP (protection of publishers, mainly from consumers) laws. The latter obviously is not going to die anytime soon — if Disney and other dinosaurs could not man the ramparts, they would be gone already.
Thus, most likely AI exemptions are not going to happen either. If anything, it looks like a great excuse for another wave of “publishers own everything, including Pythagoras table” style crackdowns (which in turn are a tried setup for “unofficially official” censorship infrastructure, of course).
Much less likely (but not impossible, seeing how some animals are a lot more equal than the rest already (https://duckduckgo.com/?q=%2B%22anas+abdin%22+%2B%22star+trek%22&ia=web)): lawyers embrace the opportunity to metastasize some more, but are cut short by the scandal-averse bureaucrats on all sides. The publisher oligopoly strengthens its grip enough that they don’t really care what you do, because things they don’t want to go around cannot survive even on your blog anyhow. So the “creative collaborative community” and “unofficially official” press are allowed to go all out in celebration of this shocking new freedom (as long as they don’t deviate too much from the Party Line), but since they still compete with each other even when not allowed to disagree, this results in near-cyberpunk grade “I grabbed it first” jungle.
-
People are very commonly perceiving these AI image-producers as being far better at these jobs than they really are. Virtually every time that an image is presented as an example of how good AI art is, they offer one which had loads of human intervention in its design process. The ones that the AI spits out and which receive zero correction are generally very messy and full of errors.
Here is an example of an AI-generated image which had a large contribution of work from a human:
https://www.deviantart.com/shad-brooks/art/PALADIN-Stable-Diffusion-AI-conversion-936296172 (https://www.deviantart.com/shad-brooks/art/PALADIN-Stable-Diffusion-AI-conversion-936296172)
The human who worked to generate this image has detailed his design process in a YouTube video:
https://www.youtube.com/watch?v=7PszF9Upan8 (https://www.youtube.com/watch?v=7PszF9Upan8)
To demonstrate the quality of AI art without heavy influence from a human, just look at AI art and notice how frequently it is filled with flaws. But I gave stable diffusion a prompt: "a prince fighting a dragon, and the dragon is breathing fire" and of the results given, this was the only one that was even a match to the prompt:
https://imgur.com/a/tRRwuy3 (https://imgur.com/a/tRRwuy3)
I did not edit the image any further. This was the second prompt offered, as the first one was deemed possibly too difficult for the AI to parse (as none of the offered images were a match). I did not perform any more tweaks to the prompt, or ask for it to generate more attempts. This is what "AI art" actually looks like.
To give an excellent example of why human-made art really stands apart from AI-generated art, I would like to offer an excellent example of high-detail art generated by a human:
https://www.reddit.com/r/drawing/comments/ztbtw9/a_muse_in_the_warzone/ (https://www.reddit.com/r/drawing/comments/ztbtw9/a_muse_in_the_warzone/)
Note the combination of themes in the image and how they make sense together. Note the fine details and how none of them are weirdly proportioned, warped, or shifting into structured unrelated to their position in the image. The hallmarks of AI-generated images are not present here. The evidence of human intervention is all over it. It doesn't look like AI-generated imagery at a glance, and it only looks more human the deeper you peer. And this is just one of many examples I could give.
It's not just that AI fail the Turing test at every turn, but even humans pretending to be AI will usually fail the reverse Turing test as well. The gap is far greater than many grifters pretend.
AI art is not going to make human art obsolete; far to the contrary, AI image generation will expand art production and allow our society to tap into the creativity of artists lacking painting or other "artistic" talents, artists whose art otherwise would have never seen the light of day.
Thank you for coming to my Ted Talk. I would like to end my 94% serious rant with a relevant bit of Star Trek: The Next Generation. I tried but failed to find a concise clip, but in the episode titled 'When the Bough Breaks', a child is offered a tool which enables him to carve wood into any shape he can imagine. In fact within that episode are many bits of wisdom about human brain development and how we can be mentally healthier. Good episode, loosely related to the topic.
-
To demonstrate the quality of AI art without heavy influence from a human, just look at AI art and notice how frequently it is filled with flaws. But I gave stable diffusion a prompt: "a prince fighting a dragon, and the dragon is breathing fire" and of the results given, this was the only one that was even a match to the prompt:
I think you did use weaker AI generator that other used. I do not thing any your interaction could archive looks like:
https://www.youtube.com/watch?v=RdwCE8ScPk4
But still most what you said is correct, even example I give where most images look like "real" you can spot flaws that no resolvable human would put in.
-
The images in the linked video clearly are heavily influenced by a human-designed source material. The consistency of the world demonstrates this: the humans being beneath the androids, the similar architectural styles of each android from many angles, the consistent cyborg elder, the same look and feel of the setting, the logistical consistencies such as rubble being adjacent to ruined structures, corpses next to murderers, just the way things are juxtaposed together. It has human design written all over it. And not only is it clearly influenced by a human, but most importantly the human design work is almost entirely what makes it so exciting. All the AI did was the busy work of making more images, views, and angles. What the AI can do is akin to a team of animators who replicate cartoon characters designed by the lead animator.
AI art isn't going to reduce the amount of work available. People will lose jobs over it for two reasons:
1.) Most importantly: Expanding the amount of people capable of being professional artists by more than the amount of expansion of the field of professional art will inevitably result in some of the existing people falling out of the field. Mostly this will be people who do busy work and don't contribute much by way of creativity such as base-level animators or illustrators--however with any change there is going to be some significant amount of actual upheaval in the community just as we have seen every time large changes happen rapidly.
2.) Most critically: Large and greedy corporations will take advantage of every opportunity to make people poorer. In many cases they aren't even doing it to save money directly so much as to disempower the people and smaller organizations from competing with them. Consider what Levi's is doing with their "diversity" program in which they use AI instead of paying brown people. Obviously they're downgrading the quality of their modeling, and the cost of modeling compared to their income is insignificant. So it isn't about the money, clearly, it has to be about something else. People say don't ascribe to malice that which can be ascribed to incompetence, but it just seems too convenient to me that these plots always seem to hurt the same people. I think it's blatant racism and/or classism. Either way, it's corporatism at the core, not AI. The AI didn't make Levi's, the AI didn't corrupt Levi's, and most importantly, Levi's has been one of many anti-human corporations all along. The AI didn't create, cause, or even exacerbate the problem. It was always humans, and the AI is just a scapegoat like always.
But much as they tried to scapegoat corporations (and not the people who run them), they are now trying to scapegoat the computers. And in both cases, they will continue to use the corporations and the computers to further their self-serving ends at cost to the quality of life of everyone else. And just as the scapegoating made the corporations stronger, this new scapegoating is just going to make the computers stronger.
Unless...
Unless we can finally take a stand against the actual threat: excessive wealth, belief in meritocracy, and the idea that extremely rich people actually earned their wealth or have any positive value to society.
-
Can you stick a banana to the wall with a duct tape? Can you draw a square and paint it black? Would you ever be able to nail your real life manparts to the Red Square? Do you think AI can do this? Do we even understand what is "this" until we see it? Are those who call themself artists doing "this", or do they just craftsmen? Anyway, I just asked AI to nail it manparts the Red Square. After it failed to do The Black Square. I'm attaching the result now. Yes, today is a good day to change the avatar.
So, I really dont know why do people bother so much about it: AI-generated art is just like a super brush. A support for bigger, grander pieces of art made by human mind. This elf of yours is not an art, its just a picture of an elf. The meaning of its look, the idea behind the whole picture is what makes it an art. more precisely, the idea which raises in the minds of the beholders.
So, its not artists who'll lose their jobs, but rather some craftsmen will lose their jobs, but that's it
With all of this in mind, lets just call people who anyhow produces interesting pieces of art artists, and people who will lose their jobs craftsmen as they should be called since forever.
-
With all of this in mind, lets just call people who anyhow produces interesting pieces of art artists, and people who will lose their jobs craftsmen as they should be called since forever.
Yes!
It's the difference between an artist and an artisan. Back in the Renaissance, artisans were in high demand because it took so much time, effort, and resources to learn to do a high quality job. But these days machine artisans produce FAR better work than a human artisan ever could. The number of artists in the world keeps going up as people become more educated and industries become more automated, but the number of human artisans is going down just as it has been for some two centuries.
-
I like how it uses big phallic buildings as analogy for the manparts and the street-lamps and other people as nails.
Also the background could just be an image on a wall and it is very well possible that his testicles are nailed to it from behind.
-
Just had this generated and wanted to share it somewhere.
-
I like how it uses big phallic buildings as analogy for the manparts and the street-lamps and other people as nails.
Now it is suddenly deep.
Nice dogs, by the way.
-
Speaking as an artist, I think that diffusion mapping is here to stay. I think there are certainly a lot of issues that need to be addressed, such as copyright and deepfakes, but that is a level of hell that I don't want to get into.
The big issue I have with anti diffusion model arguments is that it's not "art" as the computer is unable to perceive emotion and this transfer it to art. I think that this is a cry of artists who suddenly feel their existence and talents are no longer "special". Which is incorrect, as models will still rely in other artists to submit work to build training models.
The other common criticism I hear is that it's just "copy and paste" and not giving credit to original artists. While that is certainly true in some cases, that is model dependent. I could go out and take pictures or create art myself, catalog it, create a model with it, and then create art based on my own artwork.
For many other cases, diffusion models are strongly different from their source materials, and I would consider them to be inpressions. For many of us we started our artwork being influenced by other artists, and I consider diffusion models to be somewhat of that ground of "learning", albiet that they are just "starting out" and are quite capable of producing extremely realistic photography and artwork.
-
Markov models were there since the early 20th century, when Markov himself used 1-gram model to simulate text writing and physical processes. It was already obvious, that by implementing an N-gram model, one will be able to simulate human reasoning or any physical process. So Markov claimed that God doesn't exist, since all nature's processes could ultimately be understood and implemented, which got him excommunicated.
But it took more than a century to properly implement N-gram models, and, believe me, people tried hard ever since computers were invented. There were 2-gram models, 3-gram models, etc... but natural language conversations are many sentences long (i.e. more than 100-gram model is needed), and one needs a lot of training data to for a blind and deaf brain-in-a-vat to understand humans. In fact it required scrapping the entire Internet.
As of now AI is just a tool. You need a lot of expertise to do anything actually useful with it.
1. Understand how foundational models work.
2. Train LoRA.
3. Prompt the model in a consistent way.
4. Detect errors somehow. Model can generate bad art, and unless you're an artist, you wont notice what is precisely wrong.
And it will still look bland and generic, unless you train LoRA on something you personally designed in an unique way.
But in the end it can save you a bit of time or allow to produce photo-realistic content if you really know what you're doing.
Yet in observable future with well trained multimodal modal models it will be a real challenge for humanity to handle, since it could work without anyone guiding it precisely. I.e. a language model will do planning and initial composition, then image model will do a sketch, feeding result back into the language and spatial models, which will fix inconsistencies like 6 fingers.
These image models already have some basic interfacing with language models - they even use the same word embedding scheme. And further multimodal integration is driven by the need in robotics to navigate the complex real world environment.
I'm personally not good at competing even with other people, so guess I have already lost to the AI which will be there in 100 years. So I will just continue working on my projects.