剪映合成视频

JyComposeVideoOut (剪映合成视频) is the "this is the end of the line" version of the pack's Jianying/CapCut video-composition node. Feed it a group of media, set a canvas and frame rate, and it assembles the clips into one rendered video - then, because it's marked as an output node, ComfyUI knows the workflow is done and reports it as the deliverable.

"剪映" is Jianying, ByteDance's Chinese video editor and the domestic sibling of CapCut, so this whole corner of ComfyUI_Lam is aimed at a workflow style that's common in Chinese shares: generate clips in ComfyUI, then auto-assemble them into a finished video with audio, effects, captions and an extra track, all inside the graph.

Why there are two of these nodes

JyComposeVideo (the non-output version) and this one share the same inputs and outputs. The only real difference is the output-node flag. ComfyUI only finishes a graph when an output node is reached, and it decides which output nodes count by that flag. So:

• Use JyComposeVideoOut when the composed video is your final result - the thing you save and share. ComfyUI will treat it as the workflow's completion point.
• Use the non-Out JyComposeVideo when composition is an intermediate step and something else comes after.

Getting this wrong is a real (if subtle) gotcha: if you end a workflow on the non-output variant, the graph may never report "finished" cleanly, and if you use the Out variant mid-graph, ComfyUI can consider the job complete early while the rest of your pipeline still needs to run.

The inputs

• medias (MEIDA_GROUP) - required; the clips to compose, from this pack's media-group producer nodes.
• draft_name (STRING, default ComposeVideo) - the video's name ("视频名称" per the author's tooltip). Set it per job.
• width / height (INT, default 1920×1080) - canvas.
• fps (INT, default 30) - frame rate.
• audios / effects / captions / track0 - optional layers: soundtrack, transitions, subtitles, extra track.

Outputs: 视频路径 (STRING - the rendered file's location) and 视频时长 (FLOAT - its length in seconds).

Install

Pack install, no extra models:

cd ComfyUI/custom_nodes
git clone https://github.com/yanlang0123/ComfyUI_Lam

and restart. Same caveat as its twin: this is the Jianying/CapCut ecosystem, so expect Chinese-first workflows and the pack's Chinese UI labels.

Common issues

• Workflow finishes "early" - you used the Out variant mid-graph, or ended on the non-Out variant. Match the variant to the position: Out at the end, non-Out in the middle.
• "medias" unfillable - needs the pack's own media-group type, not raw files.
• Output path looks wrong - read the 视频路径 string; the render location is whatever that says, not necessarily the default output folder.

Pick the right variant, wire the media group, and this is the node that turns a pile of generated clips into a finished video in one queue run.