Spreadsheet2Video Output Image

If Spreadsheet2Video Input Image is where a spreadsheet row enters your workflow, Spreadsheet2Video Output Image is where it leaves. In the niknah/Spreadsheet2Video-ComfyUI pack, this node marks the exact point where a row's result gets captured - and that captured frame is what the next row starts from. Without at least one of these, the loop has nothing to chain, and you're basically running disconnected rows.

It looks like an endpoint and behaves like one, but with a twist: it's never actually executed. When you run the workflow, the pack replaces this node internally with its Spreadsheet2VideoProcessImage machinery, which saves the frame to ComfyUI/temp as a lossless PNG and passes the last image forward. The Output Image node you see on the canvas is just the handle you wire up.

The inputs that matter

• images (required) - link the image output of your segment here. If your segment ends in a video node, link the video's output: the pack takes the last frame of that video and uses it as the next row's starting image. That's the core of the pack's "long video" trick - each row is one clip, and each clip starts where the last one ended.
• audio (optional) - link an audio output here if your row produces sound. The pack saves it per-row (as FLAC in the temp folder) and the main node stitches all rows' audio together at the end.

Where you put it

The README's rule: "Put the output image here. Link the image to the Spreadsheet2Video Input Image node if you have no image output. The next row will use this image or the last frame if it is a video."

A few practical notes. You can have as many Output Image nodes as you want - every segment gets one at its end. They must be independent of each other, same as the Input Image groups. And since version 1.0.2 the pack no longer requires an Output Image node for plain non-long-video runs - if you're just doing a batch of unrelated rows, you can skip it.

Where people get burned

• Resolution drift. The final stitch requires every row's output to be the same pixel size. Video pipelines love sneaking in a resize node, and the pack errors out with "All images must be the same size" when they disagree. The fix, per the author's own log message: put a resize node before this one so every row lands at the same dimensions.
• Forgetting it entirely in a long-video setup - if there's no output image, there's no last frame to chain, and the "endless" video collapses into independent clips.

It's not a node that does much on its own, but it's the seam where the pack's whole sequential-video idea comes together. Wire it right, keep resolutions consistent, and the rows chain end-to-end exactly the way you'd want.