Video Face Swap Frame Picker
Scrub to a clear moment, capture the decoded frame at source resolution and check whether it contains one usable source face before spending credits.
Choose a source video
The video is decoded on this device and is not uploaded by this tool.
Browser-playable MP4, WebM or MOV, up to 250 MB
The video stays on this device. Only after you choose Check this frame is a reduced copy of that still sent to the FaceSwapAI detector; it returns face count and boxes, creates no generation task and uses no credits.
A source-frame workflow you can inspect
The picker separates local video decoding from the optional detector request, and reports only observable input properties.
- 1Choose a video and scrub the real decoded timeline without creating an account or generation task.
- 2Capture the current decoded frame as a full-resolution PNG; no screenshot controls are burned into the image.
- 3Optionally check one reduced still for face count, face-region sharpness, light, contrast and clipping before continuing.
Video frame picker questions
Does this upload my whole video?
No. The video is opened as a local browser object. If you explicitly check a captured frame, only a reduced copy of that one still is sent to the face detector.
Is the downloaded frame full resolution?
Yes. The PNG canvas uses the decoded video width and height reported by the browser. It does not upscale a smaller video.
Does a high readiness score guarantee a good face swap?
No. The score checks observable input conditions only. Identity resemblance, pose compatibility, occlusion, temporal stability and provider behavior remain separate questions.
Which moment should I choose?
Prefer one visible person, a large unobstructed face, low motion blur, even light and a pose close to how that identity should appear in the target video.