ComfyUI Node
Pose Video Driver (Pure Logic)
Validate a pose-frame payload (JSON text of keypoints) and a reference-image payload (JSON text with width/height), and produce a deterministic latent plan (JSON string) describing the video latent a downstream GPU node would build. Pure-logic: no torch, no VAE, no file I/O.
Pose Video Driver (Pure Logic)
- latent_plan_json
- reason
- valid
◄pose_frames{"frame_count":1,"keypoints":[]}►
◄reference_image{"width":512,"height":512}►
◄latent_channels4►
◄latent_scale8.0►
CategoryParametric/Video
Inputs (4)
| Name | Type | Default | Description |
|---|---|---|---|
| pose_frames | STRING | {"frame_count":1,"keypoints":[]} | Pose frames. Default {"frame_count":1,"keypoints":[]}. |
| reference_image | STRING | {"width":512,"height":512} | Reference image. Default {"width":512,"height":512}. |
| latent_channelsopt | INT | 41–64 | Latent channels. From 1 to 64. Default 4. |
| latent_scaleopt | FLOAT | 8.01–64 | Latent scale. From 1 to 64. Default 8.0. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| latent_plan_json | STRING | — |
| reason | STRING | — |
| valid | BOOLEAN | — |