ComfyUI Node
Face Fix - Prepare Full Face Video (Shot Aware)
Prepares every frame as a shot-aware 512px face video for LTX; no legacy Z-Image anchor controls.
Face Fix - Prepare Full Face Video (Shot Aware)
- video_frames
- face_video_512
- face_fix_context
◄detection_confidence0.40►
◄crop_padding0.08►
◄minimum_face_pixels12►
◄rotation_assistOff (fastest)►
◄short_gap_tracking2►
◄cut_sensitivity0.15►
◄crop_smoothing0.85►
CategoryVRGameDevGirl/Face Fix
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| video_frames | IMAGE | Complete input video frame sequence. | |
| detection_confidence | FLOAT | 0.400.05–0.95 | Minimum face detector confidence. |
| crop_padding | FLOAT | 0.080–0.5 | Extra space around each detected face crop. |
| minimum_face_pixels | INT | 124–2048 | Minimum detected face size before it is rejected. |
| rotation_assist | COMBO | Off (fastest) | Optional rotated detection for tilted faces. |
| short_gap_tracking | INT | 20–12 | Frames to bridge when detection briefly misses a face. |
| cut_sensitivity | FLOAT | 0.150.05–0.95 | Hard-cut threshold; lower values detect subtler cuts. |
| crop_smoothing | FLOAT | 0.850–0.95 | Stabilizes crop position and size within each shot. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| face_video_512 | IMAGE | — |
| face_fix_context | VRGDG_FACE_FIX_CONTEXT | — |