← sources

LONG MiniMax-H3 videos situation. Let's talk

source: reddit (reddit:StableDiffusion)published 2026-09-27 16:20ingested 2026-09-28 02:39

retained text

There are now several different approaches floating around (latent continuation, motion context, extenders, etc.), but the info is scattered across GitHub, Discord, and random comments.  
I want to collect everything in this single post with updated infos so we don't have to dig through a dozen different threads.

**Please reply with any of this:**

* Which method/workflow you're currently using (or tried)
* Pros and cons from your experience (quality, consistency, speed, VRAM, ease of use, audio continuity, etc.)
* Any tips, gotchas, or recommended settings
* Whether you're getting truly seamless results or still seeing degradation after a few extensions
* Is there a clear winner for quality vs. speed vs. simplicity?

Drop your experiences below so we can finally have one solid reference thread. Thanks!

I'll start with listing some methods/repos I've found and tested so far,

there should be way more methods and forks around at this point.

[https://github.com/NikoDemon80/ComfyUI-H3-Motion-Context](https://github.com/NikoDemon80/ComfyUI-H3-Motion-Context)

[https://github.com/seitanism/ComfyUI-H3-Motion-Context-MultiRef](https://github.com/seitanism/ComfyUI-H3-Motion-Context-MultiRef)

[https://github.com/pmhaidn/ComfyUI-Minimimax-H3-Extender](https://github.com/pmhaidn/ComfyUI-Minimimax-H3-Extender)

[https://github.com/ttulttul/ComfyUI-MinimimaxH3-Continuation](https://github.com/ttulttul/ComfyUI-MinimimaxH3-Continuation)

[https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum](https://github.com/ukr8b3g-cmyk/ComfyUI-H3-Continuum)

document fetched 2026-09-28 00:31 · canonical https://reddit.com/r/StableDiffusion/comments/1wrodxx/long_minimaxh3_videos_situation_lets_talk