FastH3 Omni open-sourced: reference-conditioned video with synced audio, 5.6-11.9x lower latency
mnmunknown · reddit · 2026-10-21
hao-ai-lab (FastVideo) open-sourced reference-conditioned FastH3 Omni, built on MiniMax-H3, with training code included. Give it a prompt plus images, video clips, and a voice, and it returns a finished video with synchronized sound.
- Up to 9 image references, 3 video clips, and a voice per request; tag each (<Picture 1>, <Video 1>, <Audio 1>) to control whose face, motion, and voice go where
- PDD + DMD distillation cuts 50 sampling steps to 8 forward passes
- 5.6-11.9x lower end-to-end latency than base MiniMax-H3 at the same 1344x768 output
- Runs on consumer GPUs; training code and reference preprocessing are open
- Built with Nuva Lab and Reactor, with support from NVIDIA
More from Multimodal
- Megalith 008: stunning AI-generated colossal architecture series — azed_ai · 2026-10-21
- Higgsfield offers $25K prizes for open-source ComfyUI rebuilds of its apps — MuziqueComfyUI · 2026-10-21
- Stunning AI-generated elderly portrait blurs the line with real photography — LudovicCreator · 2026-10-21
- Writer with no drawing skills uses AI to publish third episode of original manga — OriginalMangaStudio · 2026-10-21
- HeyGen's Professional Voice Clone needs just 30 min of audio to sound exactly like you — rohanpaul_ai · 2026-10-21
- Flux 3 Max ranks #1 on physics with 1.5M+ human judgments — isidentical · 2026-10-21