[Feature Request/Offer] Easy endless looping

Nobody has claimed this yet.

I'm experimenting with the 201 frames right now as discussed in the attached Reddit thread. There's a bit of work to do with the rendering even of that, but it's looking close and I wanted to raise my hand that I'm looking at it. Tick box for "Closed Loop" would restrict the Time slider to 6.7 seconds, but actually render and trim for 201 (is the current aim anyway). I've played a little with just vibing sliding context window and multi-pass solutions to hopefully get loops of relatively-arbitrary lengths, but that needs work. I'm open to suggestions here, too, as beyond the 201 frames possible Hunyuan trained "sweet spot" suggestion, I really only have AnimateDiff Evolved Context Window Options and generally pretty "mid" understanding of context in image diffusion to work with, but have always wanted a working implementation of this kind of functionality that's less "scratchy"-rendered from relatively weak attempts at satisfying prompt and Uniform Context Looping in AnimateDiff.

Related thread:
https://www.reddit.com/r/StableDiffusion/comments/1i1q907/hunyuan_videos_naturally_loop_with_length_of_201/

SVC

Begin by reviewing the rendering behavior, the Time slider, and the proposed "Closed Loop" option described in the issue, then read the linked Reddit thread for the 201-frame context. Define the supported loop lengths and rendering behavior before identifying the relevant implementation and tests; done means a reliable loop can be rendered without scratchy results.

Written by the indexing model from the issue text.