ComfyUI Node
Causal Forcing Generate (T2V / I2V)
832x480, 81 frames at 16 fps. Runs the official Causal Forcing pipeline in the runtime and returns real VIDEO outputs and a JSON report.
Causal Forcing Generate (T2V / I2V)
- runtime
- image
- video
- report
◄prompt►
◄seed0►
◄modelcfpp_framewise_2step►
◄videos1►
CategoryCausalForcing
Inputs (6)
| Name | Type | Default | Description |
|---|---|---|---|
| runtime | CAUSALFORCING_RUNTIME | — | |
| prompt | STRING | Used as written (no prompt rewriting). | |
| seed | INT | 00–2147483647 | — |
| model | COMBO | cfpp_framewise_2step | cfpp_framewise_2step: Causal Forcing++ frame-wise 2-step (official checkpoint causal-forcing++/framewise-2step.pt, EMA): 2 denoising steps per frame, 4 for the first frame. | cf_framewise_4step: Causal Forcing frame-wise 4-step (official checkpoint framewise/causal_forcing.pt, EMA): the 4-step baseline. | cfpp_framewise_1step: Causal Forcing++ frame-wise 1-step (official checkpoint causal-forcing++/framewise-1step.pt, EMA): 1 step per frame, 4 for the first frame. | cf_chunkwise_4step: Causal Forcing chunk-wise 4-step (official checkpoint chunkwise/causal_forcing.pt, non-EMA weights as in the official command): text-to-video only. |
| videos | INT | 11–4 | Videos in this job (seeds seed, seed+1, ...), one runtime process. |
| imageopt | IMAGE | Optional first frame for image-to-video (frame-wise models only). Resized to 832x480 by the runtime, as in the official script. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| video | VIDEO | — |
| report | STRING | — |