OBS Background Removal is an OBS Studio plugin that removes backgrounds from portrait video and images in real time, enabling virtual green-screen effects during recording or streaming.
The plugin uses a neural network to segment the portrait subject from its background, generating a mask that isolates the foreground. This mask can then be composited with other OBS sources to replace the background with images, colors, or transparency. The tool leverages ONNX Runtime for model inference, with support for GPU acceleration to improve performance. Beyond background removal, the plugin also includes low-light enhancement capabilities for improving video quality in dim conditions.
Adopters should know that the plugin is designed primarily for portrait segmentation—human subjects in framing typical of video calls or streaming. The tool integrates directly into OBS Studio's filter system, making it straightforward to layer with other effects and sources. Building from source is officially supported on Debian-based systems for both x86_64 and ARM64 architectures, though pre-built binaries are available for download. The README mentions a Lite version as an alternative if the main plugin encounters issues.
Development activity shows consistent engagement with users through GitHub Discussions and the OBS Plugins forum, with bug reports and feature requests tracked via issues. The project maintains an active sponsorship program offering both recurring and one-time support options, including private consultation for hands-on help. Documentation includes a dedicated usage guide and a series of tutorial videos covering core features, depth-of-field effects, low-light enhancement, and segmentation of non-human objects.
