| View previous topic :: View next topic |
| Author |
Message |
|
|
|
jdnation
Joined: 15 May 2007
Posts: 2059
|
 Posted: Sat Aug 05, 2023 8:39 pm
|
 |
|
| Gem-Bug wrote: | This is only going to reinforce capitalism and the status quo. Anything anti-creative and anti-worker is essential anti-human. Folks need to stop believing things they read on twitter.  |
If anything, AI will even the field of capital for more people. More indy creators will be able to compete with the big guys, just as middleware and other engines do for the indy videogames community versus the big publishers. There is even the possibility of competitive pricing and lower consumer prices thanks to lower budgets and the potential for more output if only because it's faster.
It does not have to be 'anti-human' in any way. Sure some disruption and change will always sort people into winners or losers in the outcome, but that's the natural cycle of adaptability. Maybe time is up where some jobs will no longer be viable or possible at high salaries, just as video killed the radio star, and YouTube and Streaming killed cable, and cell phones killed landlines and smartphones and tablets ate some of the PC market, so too maybe AI will reduce the salaries of a number of jobs, but that also creates an opening for new skills and industries, tool creation, and more subsidiarity in the market for a cottage industry middle class. Sure, the top earning and celebrity types are the most at risk in this instance, but once upon a time many industries went through similar changes, like the commercial airline industry, where it was practically a luxury to be employed, but that's life.
I'm optimistic that the playing field can be more leveled, that animators can have it easier, and that as Japan's population ages and there are fewer people to employ, anime can still be made with a smaller team aided by new generation of tools. That maybe the next 'Makoto Shinkai' can be discovered, that perhaps great storytellers who may not be great artists or animators can do something decent and get noticed. We can even begin to see more talent rise up in other places in the world where animation is more niche, like South America and Africa, India etc. that can compete with the Disney/Pixar and top Anime, giving us more diversity in story, culture and product.
There are legitimate long term concerns about AI, but it needn't be any more worrisome than any other disruptive technology, all which had their doom fortellers. And in some ways, they were right, but in several other ways they never foresaw, there were many benefits.
|
| Back to top |
|
|
|
Gem-Bug
Joined: 10 Nov 2018
Posts: 1282
|
|
Posted: Sat Aug 05, 2023 9:12 pm
|
|
|
| Megumi Chisato wrote: | | How? |
I mean by taking things they read on twitter with a grain of salt?
|
| Back to top |
|
|
|
MarshalBanana
Joined: 31 Aug 2014
Posts: 5469
|
|
Posted: Sun Aug 06, 2023 12:44 pm
|
|
|
| jdnation wrote: | | True. But this same argument was also leveraged at Photoshop. How are artists supposed to hone their skills when they can use the 'Undo' button? And did outsourcing in-betweening to Korea destroy the Japanese key animator's ability to work? |
Doing In-betweens is what what started animation industries in south east Asian countries, and studios in Japan still do some of the in-betweens. Photoshop is still basically the same principal when it comes to line art, painting is different, someone who has only ever painted on Photoshop would be less skilled than someone who paints with acrylics or watercolours.
More than anything else, the view of in-between as nothing but grunt work is a real sticking point to me, because it undervalues the actual artistry. Watch this at 0.2 and see how the in-betweens work as more than identical drawings that are slightly different, they are inconsistent, they morph into weird shapes, are even completely out of proportion in a few frames. All of which works well in motion, but an AI won't figure that out, all it can give is a generic in-between.
| jdnation wrote: | | Only if that's how you chose to read what I said. Don't concept artists use reference materials and photography in their work? How is leveraging AI any different? If the director wants some quick sketches one fast, the concept artist can reduce his personal 'grunt work' by leveraging AI to create an angle of a set already trained on the library of visuals and cinsistent set decoration for the film, background art that needs to remain consistent and the draw in only the new layer of details and characters that the director wants. |
I apologise if I misunderstood you. I did once know a professional concept artist, and one of the things he said is when he is in meetings he will sketch character designs down as they are talking about them, that way he already has a rough idea when he leaves. Artists can produce sketches very fast, faster than it would take to setup the instructions for an AI program, so the only reason you would want AI for concept art is if you don't have any artists to do it. They've done AI background in a few Anime, and even after being touched up they still looked bad, all they did was take a job away from an actual artist.
| jdnation wrote: | | There is no reason to fear skafreak51's scenario of too much garbage flooding the market any more than that human artists will be made obsolete. The market demand for quality and originality will filter out what's poor and uninteresting. If large studios destroy themselves with false hopes of AI, then the Indy houses will eat their lunch with a better understanding of the market. |
If a studio does decide to replace staff with AI, the final result might be poorly received, but there is no guarantee that it won't be financially successful.
edit: If you have anything else you want to say, i would recommend not being as long as before, because I can only rationally replying so much of what you say.
|
| Back to top |
|
|
|
|
jdnation
Joined: 15 May 2007
Posts: 2059
|
|
Posted: Sun Aug 06, 2023 4:22 pm
|
|
|
| MarshalBanana wrote: | Doing In-betweens is what what started animation industries in south east Asian countries, and studios in Japan still do some of the in-betweens. Photoshop is still basically the same principal when it comes to line art, painting is different, someone who has only ever painted on Photoshop would be less skilled than someone who paints with acrylics or watercolours.
More than anything else, the view of in-between as nothing but grunt work is a real sticking point to me, because it undervalues the actual artistry. Watch this at 0.2 and see how the in-betweens work as more than identical drawings that are slightly different, they are inconsistent, they morph into weird shapes, are even completely out of proportion in a few frames. All of which works well in motion, but an AI won't figure that out, all it can give is a generic in-between. |
My charecterization of in-betweening as 'grunt-work', was not to downplay the skill involved the job, but the time consumption and expense. I predict that AI will function akin to the 'Fill' tool in digital painting versur painting every cell by hand as in the old days. Just as one has to check and correct in-betweens now, AI can speedily redraw things based on corrections, immediately, akin to real-time rendering on the fly. As a rule studios and schools should still train students and employees on the old methods, which many art schools already do now, despite digital tools to develop skills. This can be leveraged for smaller projects - 15-30 sec commercials, experimental short films, conceptual pitches etc., while AI is used for larger projects on tighter deadlines and budgets.
| Quote: | | I apologise if I misunderstood you. I did once know a professional concept artist, and one of the things he said is when he is in meetings he will sketch character designs down as they are talking about them, that way he already has a rough idea when he leaves. Artists can produce sketches very fast, faster than it would take to setup the instructions for an AI program, so the only reason you would want AI for concept art is if you don't have any artists to do it. They've done AI background in a few Anime, and even after being touched up they still looked bad, all they did was take a job away from an actual artist. |
You'll find no argument from me. AI is still in its infancy. Until the day that a producer or director can literally talk verbally to an AI and have it function like an actual human artist, at best it will only be a technical tool that will be useful at scale, and even then can only produce something generic that will at most be usefull as a base. The strenghts may be at initially figuring out things such as perspective and lighting for reference, or if trained on a specific visual style, to help maintain a visual consistency that the art director wants. But it will always require an artist to finalize it, either by redrawing or tweaking the code or both. I can especially see it useful in gaming where large open worlds may require automated content beyond any team's ability to handcraft everything to scope.
| Quote: | | If a studio does decide to replace staff with AI, the final result might be poorly received, but there is no guarantee that it won't be financially successful. |
True. But that has always been there, there is plenty of low-brow click bait content thanks to the internet and social media that makes money. But people will still show up for the good stuff, it's just that at the price ratio, audiences now have higher expectations from a $20 movie ticket than a more amateur free YouTube video running ads. So the big studios can no longer just coast. The little guys grow more competitive, and even many of the little guys care about the art more than just maximizing revenue. Crowd funding also gives the enthusiast audience a say it what they want. So someone will always step up to fill a vacuum.
|
| Back to top |
|
|
|
|
noriah
Joined: 12 Apr 2015
Posts: 23
|
|
Posted: Mon Aug 07, 2023 3:26 am
|
|
|
|
Having software do the in-betweens automatically is a software category/feature that existed before "AI image generation." Digipaint and vector animation are the technologies, and software that's been used a lot include Flash, toonboom and CACANi
I suspect that the workflows that these software use to interpolate frames is more precise that what currently exists in AI tools
|
| Back to top |
|
|
|
|
TheKillerAngel
Joined: 02 Mar 2018
Posts: 85
|
|
Posted: Mon Aug 07, 2023 9:12 am
|
|
|
| SciasSlash wrote: | | ai is mostly focused on automating art and voice acting, there's relatively little work being done on automating things like heavy labor. |
I disagree with the statement, as the past century has seen significant automation in industries such as agriculture through machinery and robotics. However, the continued advancement of these physical processes is often constrained by limitations in materials science, energy sources, and associated technologies. In contrast, software automation, including areas like voice acting and art, faces fewer such constraints, allowing for more rapid progress in its current state.
|
| Back to top |
|
|
|
|
