CreatorOS Studio: open-weights local ComfyUI video suite hits 100+ features, removes people from clips
Key_Cap_8111 · reddit · 2026-10-30
Tired of paying per render, a developer open-sourced CreatorOS Studio: a local one-stop video studio on ComfyUI using open weights only (Apache-2.0), now with 100+ features (38 tools, 65 styles). New this week:
- Remove people or cars from clips, with the scene rebuilt behind them (LTX-2.5 Clean-Plate IC-LoRA)
- Type a line and get voice + lip-sync in one LTX-2.5 render
- Generated transitions between shots (Wan 2.2 first/last frame)
- New angles and 24 looks on any photo: profile, plush toy, comic, pixel art, golden hour...
- Instant green screen with SAM3
- Automatic clip QC: frozen video, jump cuts, blur, face likeness, with auto retake on serious defects
- CLI and MCP server; every render downloads as its ComfyUI workflow
Photos run on 12GB cards (FLUX.2 Klein 4B peaked at 10.7GB); video doesn't fit yet (Wan 2.2 Q4 OOM). Benchmarks in docs/LAB.md.
Related event: Open-source CreatorOS Studio update brings 100+ AI filmmaking tools(2 posts)→
More from Multimodal
- Zero-budget battle royale short film made entirely with MetaAI — 3_Moons · 2026-10-31
- AI-generated 20s perfume pitch spot delivers ink-and-smoke luxury aesthetics — EdenMakers-ai · 2026-10-31
- Open-source image-to-3dlab fixes garbled text in image-to-3D, runs fully locally — Scobleizer · 2026-10-31
- AI movies weekly: voice actors ratify AI protections with 94.95% yes vote, cold box office — lmoroney · 2026-10-31
- ZenTrainer: open-source LoRA training inside ComfyUI with live previews, NVIDIA and AMD support — Nynxz · 2026-10-31
- Four companies post-train their own video models after MiniMax H3 goes open source — videorouter · 2026-10-31