Minimax H3 Refmod workflow: character-consistent video from photos plus a voice sample, no LoRA training
AlsterwasserHH · reddit · 2026-10-13
A detailed ComfyUI workflow for Minimax H3 Refmod: skip LoRA training and create a character identity from a folder of good images plus a 2-minute voice sample — refmod training takes seconds. The bundle includes a refmod-creation workflow (identityencode via Create H3 RefMod From Folder) and a main video workflow with refmod stacking, resolution/scheduler settings, audio VAE decode, and 24fps export. Color-coded nodes, CTRL+B to bypass unused inputs, with Dainamo's video tutorial and workflow download linked.
More from Multimodal
- Vixel launches: chat with an AI Director to turn drawings and songs into short films — lmoroney · 2026-10-13
- Vixel launches an AI filmmaking workspace built around an AI Director — lmoroney · 2026-10-13
- One-shot by Claude: the stunning AI short film 'Astra - Chromatique' — Extraaltodeus · 2026-10-13
- Procedural MetaHumans in three.js With an Easy Skinning Approach — jasonkneen · 2026-10-13
- Long Shot Studio 0.8.2 adds 64-shot chaining and auto ComfyUI crash recovery — R34vspec · 2026-10-13
- Surreal AI image series "Sometimes, we need fire" goes viral — gizakdag · 2026-10-13