AIHub Action New Video Segment

AIHub Action New Video Segment is AIHub Action New Video with a sense of where it belongs. The plain node just ships a generated clip back to the client; this one also says "place it before segment 4," "replace segment 2," or "slot it in after segment 1." That distinction is everything in a real video editor, where generation is usually one step of a multi-clip timeline rather than a one-shot render. Generate the missing shot, hand it back, and it lands exactly where the cut needs it.

Pack context, briefly: it's from otavanopisto's ComfyUI-aihub-workflow-exposer, the bridge that turns ComfyUI into the generation engine inside another app. Expose nodes pull the current segment in from the client; action nodes push results back over the pack's own websocket (port 8111). This one is outbound, with extra placement instructions attached.

How it works

Identical encoding to New Video - re-encode the VIDEO through ComfyUI's video stack (format/codec/crf), stream the bytes to the client - but the websocket event is NEW_VIDEO_SEGMENT and the envelope additionally carries a reference_segment_id and a reference_segment_action. The client interprets those as instructions for where the new clip goes in its segment list, then handles the file storage per the action field.

The inputs that matter

• video (required) - the path of the video file to export.
• reference_segment_id - which segment the new clip is positioned against. The client owns these ids; feed them in from an expose node or a condition.
• reference_segment_action - NEW_AFTER, NEW_BEFORE, or REPLACE. Replace is the one you'll reach for when regenerating a specific shot: it swaps the referenced segment's content instead of inserting another clip.
• format / codec - auto (pass through) or explicit mp4 / h264. Set them when your editor is picky about what it'll play.
• crf - H.264 quality factor, 0–51, default 23. Lower is better quality. Only applies to codecs that support it.
• action - file-level APPEND/REPLACE.
• name / file_name / autoplay - same as the plain video node.

No outputs - pure side effect.

Installing it

ComfyUI Manager (search "ComfyUI-aihub-workflow-exposer"), or:

cd ComfyUI/custom_nodes
git clone https://github.com/otavanopisto/ComfyUI-aihub-workflow-exposer
# restart ComfyUI

No downloads - the pack rides on ComfyUI's existing video tooling.

Where people get burned

The same two traps as its siblings. First, reference_segment_id must be a real id from the client - an unknown reference usually means the segment gets placed wherever the client's fallback logic puts it, which is rarely where you wanted. Second, an empty video input returns early without a peep, and no connected aihub client means the clip goes into the void; check the connection before you start suspecting the node.