ComfyUI Node
CS Video Segment (SAM3.1)
Define semantic, mask, bounding-box, and positive/negative point prompts on any video frame, then propagate them in both directions.
CS Video Segment (SAM3.1)
- model
- images
- video_input
- proxy_video
- MASK
- anchor_mask
- video_info
◄anchor_frame0►
◄propagation_directionboth►
◄max_objects16►
◄prompt_data{"version":2,"objects":[]}►
Category😺dzNodes/CineStyle/Video
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | Official ComfyUI SAM3/SAM3.1 model. | |
| anchor_frame | INT | 00–10000000 | Frame where the selector prompt is defined. |
| propagation_direction | COMBO | both | Propagate toward both sides of the anchor or only one side. |
| max_objects | INT | 161–64 | Maximum SAM3.1 multiplex object slots. |
| imagesopt | IMAGE | Video frames as an IMAGE batch. Connect CS Load Video for Selector input. | |
| video_inputopt | VIDEO | Optional VIDEO input from CS Load Video. | |
| proxy_videoopt | VIDEO | Optional VIDEO used only by Selector preview. | |
| prompt_dataopt | STRING | {"version":2,"objects":[]} | Selector multi-object semantic, mask, bbox, and point prompts. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| MASK | MASK | — |
| anchor_mask | MASK | — |
| video_info | DICT | — |