MultiTrack Editor

This is the heart of ComfyUI-Easy-Media, and it's the reason most people install the pack. MultiTrack Editor is a visual timeline that lives inside a ComfyUI node - you get video clips, audio clips, subtitle clips, and "task" segments you can drag around, trim, and stack, and the node hands the whole thing to the rest of your graph as one structured blob. It's a non-linear editor bolted onto a node graph, which sounds like it shouldn't work, and mostly it really does.

The pitch is decoupling. The editor only handles media editing and loading - it's not bound to any one model. That's the difference from a fixed t2v/i2v workflow: the same timeline data can feed MiniMax H3 (t2v, i2v, r2v), Wan, LTX, or Bernini depending on which generation nodes you hang off the outputs. Reddit's video crowd picked up on it fast, mostly via LTX 2.x timeline workflows that chain long multi-segment videos out of the editor.

The four tracks

• Task Track - the core. Each task segment defines a generation unit: a task type (t2v, i2v, r2v, v2v), an image, a duration, and user/system prompts. Workflows are designed to loop over task segments automatically.
• Video Track - import clips, multi-segment stitching, and (via OmniShotCut) intelligent shot segmentation. Drop a clip in and matching task segments are auto-added with the right duration.
• Audio Track - multi-segment audio stitching for dubbing and dialogue.
• Subtitle Track - subtitles recognized from audio/video, burned in downstream.

The inputs and outputs that matter

You'll set three things up front: resolution (or pick "Custom" for your own width/height), format (picks a model's default resolution and frame rate - "Wan" is the default), and track_data, which is the state your timeline was saved in. The optional inputs are where dynamic injection happens:

• prompt_override - a formatted string that replaces every segment's prompt. Handy when you want to drive the timeline from an agent or an app. The format uses @图片1 @音频1 ... [0-120] | ... [120-240] style placeholders with frame or second ranges, segments split by |. Yes, it's Chinese-native syntax - @image1/@img1/@图1/@图片1 all work.
• image, audio, video - media lists that slot into segments by index (use the pack's easy makeImageList / makeVideoList / makeAudioList to build them).

Outputs: TRACKS_INFO (the structured timeline blob every other MultiTrack node consumes), plus IMAGES, AUDIO, and VIDEO as lists. One thing to know: when a timeline has no slot-referenced media, the editor defers loading and those three outputs come back None - the task/audio output nodes materialize media only when they need it, which keeps long-form workflows from dragging gigabytes of frames around.

Installing

It ships with the whole pack - ComfyUI Manager, search "Easy-Media", or:

cd ComfyUI/custom_nodes
git clone https://github.com/yolain/ComfyUI-Easy-Media.git

Restart, then look in the Templates panel for example workflows under ComfyUI-Easy-Media. FFmpeg installed system-wide is the one requirement the README insists on. The editor UI itself is a bundled frontend (the pack ships compiled web assets) - if the node appears but the visual editor doesn't render, your install is stale, so pull the latest version.

Where people get burned

The editor UI is where you'll spend your time, and it's genuinely more complex than a normal node - expect a learning curve the first session. The prompt_override format is powerful but picky; get a placeholder or a frame range wrong and a segment silently uses its default prompt instead of failing loudly. And remember the tradeoff the README is upfront about: dynamic injection via prompt_override can't control audio duration and ranges the way direct visual editing can. If you need precise control, edit in the timeline, don't drive it from a string.