MiniMax H3 Ref2V Workflow: Face Swaps, Pose Transfer, Multi-Subject Mixes in One Chain

TBG______ · reddit · 2026-09-01

Building on an earlier lip-sync demo, the author built a general editing chain with MiniMax H3's Ref2V that treats a source clip as the "performance master" (motion, timing, camera) while identity/appearance is driven by reference images and audio.

Verified combinations:

Key practical details:

Related event: Community ComfyUI Workflows Unlock MiniMax H3 Video Editing Power(3 posts)→

Original post →

More from Multimodal

Multimodal channel →