After seeing how someone used Seedance 2.0 to improve a famously bad anime scene (check the post here), it got me thinking: if in the near future you can just feed a rough storyboard or even a CBR file to an AI and get a fully animated episode, what’s the point of the traditional animation pipeline?
Either the industry adopts these tools en masse, or we’ll have a situation where the “fan-made” AI version of a show drops online before the official one is even finished. And if studios do use AI, how will the final product be any different from the countless fan remasters flooding the web? Feels like the whole definition of “official” animation is about to get very blurry.
about the link specifically… I couldn’t identify what was wrong with the original version? Because I can think a lot of famously badly animated scenes, and the only problem I could see with the original scene is that they just stand around talking and doing nothing.
The AI version, while I did identify that a small number of shots were objectively better, I also noticed characters being replaced with a completely different art style and other random changes that imply a different authoral intent.
Also, the commenters in the thread did not seem to agree the new one was better. As tech experiment it’s not a bad comparison, but in terms of ways to blow $50, OP wasted their money, when literally anyone could tell you 5 anime scenes they wished were drawn better.
I think slop studios will use this for 90% of animations soon, and higher end studios might use future iteration of this to boost to 60fps and/or reduce the number of keyframes drawn by hand, which could in theory lead to a higher quality output at the same-or-reduced cost (with the same amount of labor, or in less time).