Automatic Face Obscure for OBS · OBS video filter

Automatic face blurring for OBS.

An OBS video filter that detects and tracks faces and hides them with pixelation or blur.

Automatic trackingPixelate or blurDirectML acceleration
Automatic Face Obscure for OBS pixelating detected faces in OBS Studio

Track through movement

Stable face IDs and an adjustable tracking hold keep the effect attached as people move or briefly turn away.

Pixelate or blur

Use block pixelation or adjustable multi-pass blur, with padding to cover more of the area around each face.

Select who is hidden

Hide every detected face, only selected face IDs, or everyone except the IDs you choose.

CPU or GPU

Choose the tracking processor.

Run the bundled YuNet detector on the CPU, or use SCRFD‑500M through DirectML on a compatible NVIDIA, AMD or Intel GPU. If GPU inference fails, the filter falls back to CPU detection.

Supported sources

Video Capture Device and Media Source. Scene, Browser Source, Game Capture and Display Capture are not supported in this version.

Windows x64OBS StudioVideo filter

Options

Crowd mode raises detection resolution and lowers the confidence threshold to find smaller faces, at a higher processing cost. An extra mirrored detection pass helps with faces seen from the side. Optional diagnostic boxes show tracking IDs while you configure the filter.

Version 0.6.0

Download

Download the compiled Windows x64 plugin package. Extract the included folder into C:\ProgramData\obs-studio\plugins\obs-auto-face-obscure, restart OBS, then add Automatic Face Obscure for OBS under Effect Filters.

Windows x64OBS StudioMIT plugin code
Download for Windows ↓Compiled package · Includes runtime dependencies

Important: Face detection can miss people because of occlusion, lighting, pose or distance. Monitor the programme output when privacy is critical. The bundled SCRFD‑500M weights are licensed by their owner for non-commercial research use; remove or replace them before commercial use or distribution. The package includes the applicable model and runtime notices.

Support OmniaTV

Help independent tools grow.

Contributions fund development, maintenance, documentation and infrastructure.