DIY AI music video: audio-locked lip sync workflow rendered on an RTX 3070 Ti
illumnat · reddit · 2026-10-24
A creator shares a full lip-sync music video workflow: extract a clean voice-only stem as an "audio lock" guide so instruments don't break the mouth movement, render with Singularity v2 at 1.6MP (1728x960, 20 steps) on a RunPod GPU, then upscale in DaVinci Resolve with Super Scale + 4x NVIDIA RTX Video and finish with Halation, mild motion blur, and 35mm grain. Done on a humble RTX 3070 Ti.
More from Multimodal
- AI Meme: 'The Eagle' Turns Khabib Into a Half-Man Half-Eagle UFC Fighter — egeberkina · 2026-10-25
- Creator makes 30-second wuxia anime fight with Seedance 2.5, shares full prompt — azed_ai · 2026-10-25
- Fan releases AI-made manga Arcade Hunter's #1 using NijiJourney and ChatGPT — azed_ai · 2026-10-25
- Soundbox: free local ElevenLabs/Suno alternative for game audio, with 21 audio tools exposed to agents via one MCP line — WeakTrifle3475 · 2026-10-25
- Disabled creator's AI music video 'Scorch': AI as an accessibility tool, not a ghostwriter — Mpire2025 · 2026-10-25
- Hybrid Claude + Seedance pipeline renders EDM video with 60%+ archive footage to cut costs — AIandDesign · 2026-10-25