Blur Faces in Video
Apply an elliptical blur to faces detected frame by frame by an on-device AI model, then review the result for missed or partial detections.
Frame-by-frame detection attempt
A WebAssembly face detector evaluates processed frames and can miss or lose a face during movement, occlusion, blur, or pose changes.
Elliptical, padded masks
Masks extend beyond detected boxes and use a soft elliptical clip, but coverage can still be incomplete; review hairlines, chins, and partially hidden faces.
Scoped local processing
Supported video contents are processed on your device and are not uploaded for processing; site analytics and the assistant are separate.
How to blur faces in a video
- Open the tool and add your video Select a compatible file for Blur Faces in Video.
- Review and adjust Choose a blur strength, process the clip, then review every section for missed, drifting, or partial face masks before export.
- Export and download Review the exported result from Blur Faces in Video, then download it.
Frequently asked questions
Does it track faces between frames?
The detector evaluates frames and may reuse recent boxes across short gaps. Tracking can drift or miss a face during occlusion, fast motion, or large pose changes, so review the complete output.
Why is there a 30-second cap?
Per-frame neural inference is heavy. Split longer footage with Trim Video first, then blur each part.
Can I choose the blur amount?
Yes — light, medium, or strong.
Will it blur faces that turn away or are partly hidden?
Detection is more likely to work on visible front or three-quarter faces. Turned-away, small, blurred, or occluded faces can be missed or only partly covered; use an appropriate blur strength and review every section.
Is this enough for GDPR or anonymization?
No compliance or anonymization guarantee is provided. The tool can assist with face masking, but you must review the full result, correct missed or partial masks, and confirm that it meets your legal and publishing requirements.