Eyeline Labs open-sources ID-V2V video model
Eyeline Labs has open-sourced ID-V2V, an identity-preserving video restylization framework designed to propagate visual and lighting edits from a single keyframe across an entire video sequence. By decoupling identity preservation from edit-driven synthesis using control signals like depth maps and facial normals, ID-V2V overcomes identity drift and temporal flickering common in video diffusion pipelines.
ID-V2V brings high-end VFX control to open-source video generation, addressing the main bottleneck holding back AI in post-production: consistency.
• Solves character identity degradation across edited frames by treating identity preservation as a specialized relighting problem.
• Keyframe-driven editing aligns with traditional VFX compositing and lighting pipelines instead of relying purely on text prompt manipulation.
• Open-sourcing the model and code enables widespread adoption and integration into tools like ComfyUI and Blender for indie creators and studios alike.
DISCOVERED
1h ago
2026-08-02
PUBLISHED
1h ago
2026-08-02
RELEVANCE
AUTHOR
AI Search