ComfyUI Node
H3MD Segment Continue
A ComfyUI node in MiniMax H3/H3MD with 11 inputs and 4 outputs.
H3MD Segment Continue
- positive
- latent
- model
- previous_latent
- locked_audio
- audio_vae
- positive
- latent
- model
- actual_overlap
◄overlap_frames0►
◄continue_audiotrue►
◄drift_strength0.60►
◄drift_lock_tokens2►
◄release_vramtrue►
CategoryMiniMax H3/H3MD
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| positive | CONDITIONING | — | |
| latent | LATENT | Target AV latent from H3MD Segment Encode | |
| model | MODEL | — | |
| overlap_frames | INT | 00–3600 | — |
| continue_audio | BOOLEAN | true | — |
| drift_strength | FLOAT | 0.600–1 | Drift: peak re-noise weight for the overlap prefix token farthest from the seam (the prefix is restored exactly at the end). 0 = prefix kept unchanged during sampling too (the former static_lock). |
| drift_lock_tokens | INT | 20–64 | Prefix tokens nearest the seam that stay locked (1 token = 3.4 frames). |
| release_vram | BOOLEAN | true | Between segments: unload idle models (text encoder, VAE, upscaler) and soft_empty_cache. The base UNet is kept loaded. |
| previous_latentopt | LATENT | FINAL sampled latent of the previous segment. Only 'samples' is read; all other keys are dropped. | |
| locked_audioopt | AUDIO | — | |
| audio_vaeopt | VAE | — |
Outputs (4)
| Name | Type | Description |
|---|---|---|
| positive | CONDITIONING | — |
| latent | LATENT | — |
| model | MODEL | — |
| actual_overlap | INT | — |