2 ms·
The problem would still be temporal stability. You'd have to train on series of frames, but then I'm not sure diffusion models are well suited for that?
by operator-name 4y ago
The problem would still be temporal stability. You'd have to train on series of frames, but then I'm not sure diffusion models are well suited for that?