ComfyUI long-video controller for MiniMax H3 Ref2VA singing and speaking workflows
HEALTH ? 62 / Stable
// readme
ComfyUI H3 Long Video
A MiniMax H3 long-audio segmented generation plugin with two independent control nodes: MiniMax H3 Β· MV Generator and MiniMax H3 Β· Talking Avatar. Both support local / RunningHub, segment review, generation, single-segment redo, preview, and merging.
The repository and install package include two complete example workflows: MV, Talking Avatar. Both let you choose local / RunningHub in the node. For the RH single-segment compute workflow, you must import it separately into your own RH workspace and fill in new IDs. The Talking Avatar and MV model conditioning chains differ, so each uses its own canvas. This plugin does not wrap all generation nodes into a single black box for you.
Installation
- Put the plugin in
ComfyUI/custom_nodes/ComfyUI-H3-LongVideo, avoiding duplicate installs or an extra nested directory. - Use ComfyUIβs Python to install
requirements.txt, and prepareffmpegandffprobe. - Install theβ¦