Timeline Segment Output

A timeline is a storyboard, but ComfyUI generates one clip at a time. Timeline Segment Output is the node that bridges the two: give it the timeline and a segment index, and it hands you that one shot's prompt, length, image indexes, and media - ready to feed a video model. Loop over the segment count with the index incrementing and you've turned a whole timeline into a batch of individual generations.

The inputs

• timeline_info - from Timeline Editor. Non-negotiable.
• segment_index - which segment to pull (zero-based). The value your loop increments.
• prompt_format - default, or promptRelay for frame-range-annotated prompts that pair with a Prompt Relay encoder.
• optional images / audio - overrides for the segment's media, if you're driving the timeline from external inputs.

The outputs

• PROMPT - this segment's prompt string. Straight into a text encoder or conditioning node.
• TYPE - segment type as an INT (different segment kinds encode differently).
• NO_IMAGES - a boolean telling you whether this segment has any images. Genuinely useful for branching logic: an i2v segment expects an image, a pure t2v segment doesn't, and this flag lets your graph decide which path to take without inspecting the media yourself.
• IMAGE_INDEXES - which images in the timeline's image list belong to this segment, as a string.
• LENGTH - segment duration in frames.
• IMAGES and AUDIO - the segment's actual media.

That combination is why it beats a generic "get item from list" node: it's not just the data, it's the semantics - type, image presence, and indexes - all unpacked for you.

The loop pattern

The idiomatic use: Timeline Info Output for dimensions/fps, Timeline Segment Count for the bound, and this node inside a repeat loop with segment_index as the counter. Each pass pulls one shot, conditions a generation, and saves. When a shot comes out wrong, bump the loop to just that index and regenerate only that segment - which is exactly how the community uses these timeline workflows to fix one clip without redoing the whole video.

Install

It's part of ComfyUI-Easy-Media:

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

restart ComfyUI (ComfyUI Manager → "Easy-Media"). No extra models; FFmpeg system-wide is the pack-level requirement.

Where people get burned

segment_index out of range returns empty data - wire it to Segment Count so the bound and the index can't disagree. And the prompt_format choice matters more than it looks: pick promptRelay without a relay encoder downstream and you'll see the raw [start-end] frame annotations in your prompt string. Match the format to your conditioning stack. Otherwise, this is one of the most reliable nodes in the pack - feed it a valid index and it gives you a clean, ready-to-generate segment.