How an AI Anime Is Created

It's weird how consistent the 'ai art style' is for each medium. You really only need a glance at the right frame in these to know immediately. (https://www.aventos.dev/our-work) Not looking for errors or anything, just style. The ramen one is markedly subtler about it than the others at least. I always assumed this was due to people using 'default' styling by not specifying anything in particular, but these people claim to have a pretty in-depth process that presumably would nudge away from that. When I look at old pictures of dall-e 2 or midjourney v3, the hands are all mangled and stuff, but they're not in the 'ai style'.

The end result[1] looks quite impressive! The animation feels like it's quite high quality--very smooth. However, some things seem off: bad lip sync, minor animation errors, discontinuous scene-to-scene feel. Given that these are supposed to be their best examples, meant as advertisements, I would guess these errors are hard to fix using only the models.

So my verdict is that AI-only anime is still not there yet (although it would be cool to have a world where one dedicated and talented artist could make a high quality anime). But at the same time, anime studios will probably start incorporating AI into their workflow, getting high-quality-looking scenes for very cheap.

[1] https://www.aventos.dev/our-work

I must admit, the short sample for “How Asians Pay the Bill” was kinda fun :)

I have seen a few samples of AI anime, and it is impressive how far the technology has come; the only thing that really stands out are the cuts every few seconds, because video generation currently can't produce anything longer. However it's definitely enough to tell a story, particularly one that's nearly 100% fast-paced fighting scenes.

> [we did this so anime production companies would] pay the employees(artists) better

I mean, has pay gone up in other fields where AI has taken over a lot of jobs?

In any case, I think the reason pay is awful in anime is the same reason as in gamedev: if people get into it for passion, the companies can pay less and still have employees.

> One, fans here actually get to make anime in the US. Instead of needing to fly to Japan, speak Japanese, and preferably be Japanese to break into the industry.

Does this come off as wildly tone deaf to anybody else? "We love your product and what you do, but could we get it without associating with or paying you in any way?"

> ...Smaller and faster means more financially feasible so we can give better wages for artists and enable a new industry to thrive in the US.

I've been in Asia this week, speaking to artists and creators. The traditional systems do create challenges, but this just sounds like a U.S. AI-based company giving itself carte blanche to steal the cultural output of another country, wrap it up in and American style business, and eventually push existing artists or long standing studios out of the space.

No, I don't think I want a smaller team of artists and animation professionals doing art and animation. That feels like a very sloppy outlook to me.

[Edit: While I'm sure the process is interesting for some, the introduction especially just leaves me wondering who thought this writing sounded good? Maybe the target market for this is AI investors or people who aren't actually interested in the artistic part of things, and that's why I find it such a turn-off.

E.g. "We could license good music," implying their music is bad? Maybe it would be less jarring if they didn't specifically call out the importance of writing?]

Pretty cool overview of the process even if I would rather claw my eyes out than view the result.

This is pretty cool. I could see a future where these kinds of techniques become the norm, and the role of "animator" becomes less about hand drawing animations and more stylistic and taste choices, with I guess the occasional involvement by hand to get the exact result the animator wants. Still, the boundary between art and slop is super fine here.

1. Theft

  - Akira Yoshida, Marvel