AnimateDiff produces colorful garbage

[ahemberger]

Hi! I'm very new to ComfyUI as a whole and this extension specifically, and am trying to wrap my head around how it works. Currently I have what I think mimics the "simple setup" illustrated in the readme, but no matter which model I use or what I try, all I can seem to get as an output is this very colorful garbage; if I bypass the animateDiff nodes, I get reasonable image outputs. Could anyone help me understand what I'm doing wrong? Thank you!

[ahemberger]

Update, in case anyone else stumbles across this question: apparently my problem here is that I was trying to test with a single image, and animateDiff expects to be working with multiple images. 16 seems to be a somewhat-magical starting point in terms of frame count - setting the "batch size" of the empty latent image to 16 (rather than 1, as above) basically gives animateDiff 16 frames to work with, which produces results more like what I would expect.

Just a potential gotcha if you happen to be trying to slowly step from generating a single image to multiple images, as I was doing.