Image Accumulator Start

Image Accumulator Start is the pack's manual image-collection node: it gathers generated images across multiple runs until you've hit a target count, passes them on, and emits a status signal that its partner node - Image Accumulator End - uses to know the batch is complete. It's the same "accumulate then release" idea the pack's XY Grid Accumulator is built on, but without the grid: you tell it how many images you want, and it gates the rest of the workflow until you've queued that many.

A small honesty note before the details: the current version of kenjiqq/qq-nodes-comfyui has since folded this pattern into the specialized XY Grid Accumulator, and the standalone Start/End pair no longer appears in the pack's main source. If your installed copy doesn't show them, you're on a newer build - the grid accumulator is the replacement. But plenty of workflows and older installs still use the pair, so it's worth knowing what they do.

What it does

• images (required) - your IMAGE tensor, straight off the VAE decode.
• count (required) - how many images to collect before the batch is "complete."
• reset (optional, INT, default 1) - the reset counter, in the same spirit as the Reset button on XY Grid Helper: bump it to start a fresh accumulation session.
• images (output) - the incoming images, passed through while collecting.
• status (output) - an IMAGE_ACCUMULATOR_STATUS signal you wire to Image Accumulator End so it knows whether the batch is done.

How to use it

The pairing is the whole point: Start declares a session with a target count and hands off a status token; End consumes your images plus that status and releases the combined batch only when the count is met. Between Start and End, downstream stays gated - the accumulator pattern holds the workflow until the queue has produced enough.

Where you'd reach for it: collecting N frames or N variations that you want combined into one tensor before a single downstream op - say, stacking multiple generations before feeding a comparison or a batch post-processor. If your goal is a labeled XY grid, you want the grid accumulator instead; this is the unlabeled general-purpose version.

Troubleshooting

• Nothing comes out at the end. You queued fewer runs than count. Each prompt produces one image; you need exactly count of them for the release.
• New batch mixing with old. Forgot to bump reset. Start a fresh session by changing the reset value.
• Status mismatch. Start and End must be paired - an End fed a status from a different Start session will misjudge completion.

Installing it

Part of qq-nodes-comfyui:

cd ComfyUI/custom_nodes
git clone https://github.com/kenjiqq/qq-nodes-comfyui

Restart, or install qq-nodes-comfyui via ComfyUI Manager. It lives under QQNodes/Image. No pip requirements - the pack is plain Python on ComfyUI's own libraries.