After 400+ generations: a working MiniMax H3 prompt for character replacement video editing
Darqsat · reddit · 2026-08-20
The author spent 6 hours and 400+ generations reverse-engineering MiniMax H3's reference mode for SCAIL-style character replacement, sharing a full working prompt and key findings:
- retentionanalysis is the main driver: trigger words like fullypreserved and attributetransfer determine quality
- The [video editing] keyword in summary enables frame-by-frame editing; [audio reuse] works but audio gets rewritten
- Action descriptions in detaileddescription barely matter; stronger visual anchoring of subjects (hair, clothing, position) stabilizes identity
- Edits fail most when the original character is barely recognizable
Tested with prunedint8 quantization + 4-step turbo LoRA, mostly at 5s duration.
More from Multimodal
- MLP-Splatting (ECCV 2026) code released: MLPs as light-field primitives instead of Gaussians — AjdDavison · 2026-10-16
- Digital human Fae codes her own facial expressions frame-by-frame with Claude Opus 5.5 and FACS — repligate · 2026-10-16
- Meta's Omnilingual ASR, covering 1,600 languages, named a TIME Best Invention of 2026 — AIatMeta · 2026-10-16
- ATAS: Zhejiang-led team builds world-state memory framework for consistent agentic video generation — CSProfKGD · 2026-10-16
- Flux 3 Klein's Default Templates Are Broken Without Its Prompt Upscaler — StrictMove5839 · 2026-10-16
- Kroma Touted as the Most Well-Rounded Image Model Yet on Civitai — Different_Fix_2217 · 2026-10-16