Performance Optimization Implementation Steps
For users with hardware limitations, processing efficiency can be significantly improved by the following methods:
- Lightweight model selection: Use u2netp by default (5x smaller than u2net), or set model="isnet-anime" (for anime material)
- Resolution degradation processing: First complete the keying in 1/4 resolution (Fusion page → Settings → Proxy Mode), and then switch back to Full for the final rendering.
- CUDA Acceleration Configuration: Make sure the rembg[gpu] version is installed and GPU processing is enabled in Da Vinci Preferences → Memory and GPUs
- Fragment Caching Policy: Right-click on the timeline clip → Enable Render Cache (Smart/Fusion), blue cache bar appears for smooth previewing
Alternative: For long videos, it is recommended to export a PNG sequence with alpha channel (format selected QuickTime+RGB+Alpha) before re-importing it for use.
This answer comes from the articleRembg-Fuse: Video Remove Background Plugin for Video Editing Software Da VinciThe
May not be reproduced without permission:AI productivity tools " How can I increase the speed of Rembg-Fuse on a low configuration computer?
