ComfyUI Node
JR MiniMax H3 Empty Audio Latent for Tail (Experimental)
Experimental empty audio matched to an H3 tail video or single frame. Uses the native 24 fps / 40 audio ticks/s timeline (7 video tokens -> 22 frames -> 37 audio ticks; legacy single frame -> 2 ticks), matching the input dtype/device. Connect with the same video latent to AV Latent Builder, then a normal second-pass sampler. Not encoded silence, not an original-video audio crop, and not an audio lock.
JR MiniMax H3 Empty Audio Latent for Tail (Experimental)
- video_latent
- audio_latent
- status
CategoryJR MiniMax H3/Latent
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| video_latent | LATENT | Connect the same video as AV Builder: tail_context_latent [1,24,5k+2,H,W] or legacy T=1. Not an AV pair or masked latent. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| audio_latent | LATENT | — |
| status | STRING | — |