ComfyUI Node
BD Face Texture Bake
Bake source photos into the canonical UV layout using nvdiffrast. By DEFAULT bakes every detected view and outputs (uv_textures, confidences) batches that plug straight into BD_UVConfidenceBlend. uv_layout_mask is view-independent: it marks every UV-atlas texel that maps to a mesh triangle (the full target region). Wire it into BD_UVConfidenceBlend.target_mask so the inpaint_mask covers gaps no view managed to fill. Set view_index >= 0 to bake just one view (debug/inspection); the output is still a 1-length batch so wiring is unchanged. Undetected views (rear / failed detection) are skipped — they carry no usable landmarks, so they contribute nothing to a face-UV bake.
BD Face Texture Bake
- images
- face_fit
- uv_textures
- confidences
- uv_layout_mask
- status
◄view_index-1►
◄texture_size2048►
◄min_confidence0.00►
Category🧠BrainDead/FaceWrap
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| images | IMAGE | Source photo batch — same order/count as the LANDMARKS_BATCH the FACE_FIT was built from. | |
| face_fit | BD_FACE_FIT | FACE_FIT from BD_FaceFit. Provides canonical mesh + per-view 2D vertex positions. | |
| view_index | INT | -1-1–63 | -1 = bake ALL detected views (default). >= 0 = bake only that view index. |
| texture_size | INT | 2048256–8192 | Output UV texture resolution. |
| min_confidenceopt | FLOAT | 0.000–1 | Texels with face-cosine below this are zeroed in the mask. |
Outputs (4)
| Name | Type | Description |
|---|---|---|
| uv_textures | IMAGE | — |
| confidences | MASK | — |
| uv_layout_mask | MASK | — |
| status | STRING | — |