You should have written the last paragraph first, would have gotten less downvotes.
I know this is a dirty thing to say, but this feels like one of the few actual genuine use cases for AI in games.
Animations are extremely finicky and requires a lot of manual tweaking and adjusting to get right, it doesn’t surprise me that they struggled to make a procedural animation system because if they could solve that, they’d actually solve a well known industry challenge.
If you were building spore today, you could train a little local model that purely creates reasonably good animations for arbitrary creature designs. It wouldn’t be perfect, but for a game like Spore it would be good enough.
slimerancher@lemmy.world 4 days ago
PhoenixDog@lemmy.world 4 days ago
I know this is a dirty thing to say, but this feels like one of the few actual genuine use cases for AI in games
And everyone just immediately stopped reading your comment.
Orphanator@lemmy.world 3 days ago
No, that’s a problem we’ve already solved, with math.
You can make a rig like that yourself, just by following some advanced tutorials.
It feels like you’re saying “this is hard, let’s throw AI at it”.
That’s not the fun way for humans to do things.
FooBarrington@lemmy.world 3 days ago
Can you link what you’re talking about? Because I’m not aware of any animation system that’s purely mathematically derived AND that can generate aesthetically pleasing animations for arbitrary body shapes.
There are certain techniques like Inverse Kinematics that might vaguely fit your description, but that’s a tiny piece of the puzzle - it might get you 5-10% of the way, but given arbitrary body shapes it’s gonna look horrible in most cases, and it doesn’t give you actual animations since you’d still need to purposefully move the creature’s extremities.