Extensions/Fixed Soundtrack
ComfyUI Extension

Fixed Soundtrack

Generate continuous multi-clip video over a song you already have: the soundtrack stays fixed while each clip's video is generated. MiniMax H3 Ref2VA template included.

By bno84·Created about 22 hours ago·Updated about 17 hours ago· 0
bno84/ComfyUI-FixedSoundtrack
Nodes—
On cloudLocal install
Stars0
Updatedabout 17 hours ago
Readme
<!-- Describe the pack as it is now: key features, usage and caveats. This is not a changelog. Replace outdated text instead of appending to it. -->

Fixed Soundtrack for ComfyUI

Generate continuous, multi-clip MiniMax H3 video over a song you already have. The song is encoded once and held fixed during sampling; the original audio is used as the final soundtrack.

Install

  • ComfyUI-Manager: search for Fixed Soundtrack (comfyui-fixedsoundtrack).
  • comfy-cli: comfy node install comfyui-fixedsoundtrack.

Requires ComfyUI 0.38.0+ and ComfyUI-Minimax-H3-Continuation 0.6.3+. Install the continuation pack from Git or choose nightly in ComfyUI-Manager (Registry version issue). Tested at e1768d5.

Template: H3 Ref2VA, three clips

Open FixedSoundtrack - H3 Ref2VA 3 clips under ComfyUI-FixedSoundtrack in the template browser. Load your song and reference images, then write each clip's prompt.

  • Clips continue the previous latent, with one final video decode.
  • Each clip accepts two reference images and an optional third. References can be shared across clips.
  • Clip N uses seed + N − 1 from the shared seed.
  • use_original_soundtrack defaults on. Turn it off to use the decoded audio latent instead of the trimmed source song.

H3 Audio Timeline Slice

Slices the encoded song on a continuous timeline and carries the previous clip's audio tail through the hidden overlap. It validates timing without encoding or sampling. Follow it with Set Latent Noise Mask (zero) and LTXV Concat AV Latent, as shown in the template.

Save Video (Fixed Soundtrack)

An alternative to core Save Video with additional codec choices:

| Container | Video | Audio | | --- | --- | --- | | mp4, mkv | h264, h265, av1 | flac, opus, aac | | webm | av1 | opus |

The template defaults to H.265 + Opus in MP4, CRF 18 and 128 kbit/s audio. Choose FLAC for lossless 24-bit audio. Set video quality with crf or video_bitrate_kbps; audio_bitrate_kbps controls AAC and Opus.

Development

With ComfyUI on PYTHONPATH and the continuation pack beside this pack in custom_nodes/ (or at H3_CONTINUATION_PATH):

PYTHONPATH=/path/to/ComfyUI python -m unittest discover -s tests -v

Edit workflows directly in ComfyUI and commit the exported JSON to example_workflows/.