多视频合并

The short version: you feed this node a list of video file paths and it hands you back one merged video. That's it. In a world where video generation still happens in short bursts - a few seconds per pass, then you re-run for the next shot - a "merge these N files into one" node is the boring glue that makes the whole pipeline produce something you can actually upload.

It's called 多视频合并 ("multi-video merge") in the UI, and it lives in the ComfyUI_Lam pack, a Chinese kitchen-sink of utilities that also brings loops, face fusion and CapCut draft assembly. This node is one of the genuinely practical ones.

How it works

Under the hood it hands your list of paths to ffmpeg's concat machinery (the pack ships ffmpy as a dependency), then reports the merged file back as a path string. The order you put paths into pathList is the order they come out - there's no reordering, no smart cut detection, just concatenation. It merges what you give it; it does not re-encode to make mismatched clips play nice.

The inputs that matter

Only two, which is refreshing:

• pathList (LIST) - a list of video file paths. Feed it from a load-video node, a path node from this same pack, or any node that outputs a list of strings.
• filename_prefix (STRING, default ComfyUI) - the base name for the merged output file. You'll usually set this to something meaningful per job.

The single output, 视频地址 ("video path"), is a STRING. Wire it into a preview node or a save/copy step, or pass it to another tool node. Because it's a plain string you can also feed it into a text-display node to double check where the file landed.

Install

Same as the rest of the pack - it's one repo:

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

or search "ComfyUI_Lam" in ComfyUI Manager and let it do the clone. Then restart ComfyUI. Note the README is written for the Windows portable build and asks you to also extract insightface.rar/venv.rar into python_embeded\Lib\site-packages and run install.bat - for a merge-only workflow none of that is needed, and honestly the pinned old dependencies (tensorflow 2.12, numpy 1.23.4) will fight a modern ComfyUI install. If you're only here for this node, skip the model downloads entirely.

Common issues

• "Merged video has no audio" - expected if your source clips are silent, but also if they differ in codec. The node concatenates streams; mismatched audio/video codecs are the classic cause of a failed or glitchy merge.
• Fails or stutters on mixed-resolution clips - ffmpeg concat wants consistent resolution, fps and codec across inputs. Normalize your clips first (same encode settings) and the merge gets boring again.
• Video missing after run - check the path string in the output; the pack writes relative to its own working assumptions, so confirm the file actually exists where the string points before hunting deeper.

For anything fancier - cuts, transitions, captions burned in - this is the wrong tool. That's what the pack's CapCut/Jianying compose nodes are for. This one just makes one video out of many, and it does it without ceremony.