ComfyUI Node
SceneCut Detect (Advanced)
Raw OmniShotCut output: every proposed shot and transition, nothing merged. Follow with SceneCut Refine. Its output is cached, so tuning later stages does not rerun detection.
SceneCut Detect (Advanced)
- video
- detector
- scenes
- summary
- images
◄overlap20►
◄confidence0.00►
◄batch_windows8►
◄decoderauto►
◄output_imagesoff►
CategorySceneCut/advanced
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| video | VIDEO | — | |
| overlap | INT | 200–90 | Frames shared by neighbouring 100-frame windows. Each window is trusted only for its middle; more overlap gives boundaries near window edges more context. |
| confidence | FLOAT | 0.000–0.95 | Drop boundaries whose localisation confidence is below this (0 keeps everything the model proposes). The dropped shot merges into the next one. |
| batch_windows | INT | 81–64 | Windows per forward pass. Speed only; results are identical. |
| decoder | COMBO | auto | auto uses the ffmpeg binary when present, otherwise PyAV. Both give identical frames. |
| output_images | COMBO | off | Also send frames to the images output, from the selected scenes (all scenes when none are selected). all frames: every frame, in order. first / middle frame per scene: one image per scene. off: the images output carries a single black frame. Frames are full resolution float32 (about 25 MB per 1080p frame), so prefer one per scene or select scenes first. |
| detectoropt | SCENECUT_DETECTOR | From SceneCut Detector Loader. Default: OmniShotCut on the GPU. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| scenes | SCENECUT_SCENES | — |
| summary | STRING | — |
| images | IMAGE | Frames chosen by output_images (a single black frame when off). |