Native audio-video sync with Minimax H3: A ComfyUI workflow breakdown
CommentSignal9029 · reddit · 2026-08-31
The author created the short film 'IN TRANSIT' using Minimax H3, exploring a 'Brutalist Frequency' aesthetic. The workflow was split into two passes in ComfyUI:
- Generation (Ref2Vid): Used an Ollama node (Gemma4:26b) to format prompts for H3, integrating optimizations like Spectrum Apply and Comfy Kitchen attention to generate base clips at 864x480.
- Latent Upscale: Utilized custom MMH3Tools nodes to upscale directly in latent space to 1344x768 using Turbo LoRA, preserving the raw concrete texture without the 'plastic' look of standard upscalers.
The key achievement was native audio-video synchronization, where the model calculated physical reactions to reference sounds (e.g., pneumatic suction) perfectly in sync, requiring zero post-production sound design.
Related event: Making Audio-Video Synced Short Film with MiniMax H3(3 posts)→
More from Multimodal
- Pollo AI launches Recast video restyling, 60% off for launch — SimplyAnnisa · 2026-10-16
- Tripo hosts first world-building hackathon in SF with World Labs and Runway talks — mhdfaran · 2026-10-16
- Kling 4.0 boosts text-to-video with character consistency and lip sync — CurieuxExplorer · 2026-10-16
- 5.8M-param Speech Event Detector adds laughs, breaths and背景 sounds to Whisper transcripts — Ayush Kumar Bar · 2026-10-16
- Huawei's SUPRA-Tex turns one image into relightable 3D materials with WYSIWYG fidelity — Chen Yang · 2026-10-16
- Giant Michael Jackson AI fashion images show the power of extreme proportions — OVolosin82152 · 2026-10-16