ComfyUI Node
BSAI TaoMate Stream Chain (Long Video)
TaoMate one-click streaming long video: generates chunk by chunk from prompts, 3-step sampling per chunk, automatically continues each chunk from the previous chunk's last frame + tail audio, and outputs the concatenated full audio-video. cfg fixed at 1.0 (3-step distilled models do not use CFG).
BSAI TaoMate Stream Chain (Long Video)
- model
- clip
- vae
- audio_vae
- sigmas
- first_frame
- video
- audio
- last_frame
- tail_audio
- chunks
◄promptsScene 1: ... (one prompt per line; one line = one chunk)
Scene 2: ...
Scene 3: ...►
◄width1344►
◄height768►
◄length124►
◄seed0►
◄anchor_tail_audiotrue►
◄tail_audio_seconds1.5►
◄shift_video12.00►
◄shift_audio3.00►
◄verbosetrue►
CategoryBSAI/TaoMate
Inputs (16)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | — | |
| clip | CLIP | — | |
| vae | VAE | — | |
| audio_vae | VAE | — | |
| sigmas | SIGMAS | — | |
| prompts | STRING | Scene 1: ... (one prompt per line; one line = one chunk) Scene 2: ... Scene 3: ... | One chunk prompt per line; each chunk generates an audio-video segment and auto-continues from the previous chunk's last frame + tail audio for appearance/voice continuity |
| width | INT | 134432–8192 | — |
| height | INT | 76832–8192 | — |
| length | INT | 1245–362 | Frames per chunk (24fps, auto-snaps to 17k+5 grid; 124~5.2s, training range ~124-362) |
| seed | INT | 00–18446744073709550000 | — |
| anchor_tail_audio | BOOLEAN | true | Uses the previous chunk's tail audio as the next chunk's frame-0 audio anchor for voice/scene continuity |
| tail_audio_seconds | FLOAT | 1.50.2–6 | — |
| shift_video | FLOAT | 12.000.01–100 | — |
| shift_audio | FLOAT | 3.000.01–100 | — |
| verbose | BOOLEAN | true | — |
| first_frameopt | IMAGE | Starting anchor image for the first chunk (optional) |
Outputs (5)
| Name | Type | Description |
|---|---|---|
| video | IMAGE | — |
| audio | AUDIO | — |
| last_frame | IMAGE | — |
| tail_audio | AUDIO | — |
| chunks | INT | — |