ComfyUI Node
MultiTalk Prompt Schedule
Scene director for MultiTalk: assign prompts to frame ranges. Format: one line per scene <frames>: <prompt> <frames>: <prompt> ||| <NAG negative prompt> Lines without ||| use the global negative_prompt field. Per-scene negatives feed NAG (active at cfg=1). Connect negative_text_embeds to WanVideoApplyNAG → nag_text_embeds. Example: 120: Person 1 speaks to camera ||| looking sideways, turned away 120: Person 2 responds, person 1 nods 120: Both look at the camera and smile Each number is the duration in frames (120 = 5s at 24fps).
MultiTalk Prompt Schedule
- t5
- text_embeds
- negative_text_embeds
◄schedule120: Person 1 speaks to the camera, person 2 listens
120: Person 2 responds, person 1 nods
120: Both look at the camera and smile►
◄negative_prompt►
◄force_offloadtrue►
◄devicegpu►
CategoryWanVideoWrapper
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| t5 | WANTEXTENCODER | — | |
| schedule | STRING | 120: Person 1 speaks to the camera, person 2 listens 120: Person 2 responds, person 1 nods 120: Both look at the camera and smile | One line per scene. Format: frames: prompt text ||| optional NAG negative prompt |
| negative_prompt | STRING | Default negative prompt for NAG guidance. Used for all scenes without a per-scene negative (|||). Also used as the negative_text_embeds output for WanVideoApplyNAG. | |
| force_offloadopt | BOOLEAN | true | — |
| deviceopt | COMBO | gpu | 2 options: gpu, cpu |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| text_embeds | WANVIDEOTEXTEMBEDS | — |
| negative_text_embeds | WANVIDEOTEXTEMBEDS | — |