Faishme Repeat BBOX

Faishme Repeat BBOX takes a single bounding box (the BBOX type, the same one emitted by YOLO-style detection nodes) and duplicates it a set number of times. Where the pack's other repeat nodes manipulate tensors, this one just multiplies a Python list: bbox * repeat yields a list of repeat identical boxes. It's a tiny, specific utility - and the whole point is in that type distinction.

How it works

The BBOX type in ComfyUI is not a tensor; it's a small Python object that detection nodes hand around to describe a region - typically [x, y, width, height] or similar coordinates. Python's * operator on a list replicates its contents, so bbox * repeat produces a list of repeat copies of the exact same box. The output stays typed as BBOX (a list of them), and repeat runs from 1 to 128.

Why you'd use it

Detection workflows have a shape problem: a detector emits a varying number of boxes, and many downstream nodes want their inputs as a batch or list. If you've detected exactly one region - say, a garment in a catalog photo - and the next node in line expects a list of boxes (or one box per item in a batch), you need that single box to become N boxes. That's this node's entire reason for existing: replicate the one detection so the rest of the pipeline can process it like a batch. Pair it with the pack's list-oriented stack/split nodes and you can massage single detections into whatever shape your graph wants.

Inputs and outputs

• bbox - the single BBOX detection to duplicate.
• repeat - how many copies, 1 to 128 (default 1).

Output is a bbox - now a list of repeat identical boxes.

Installing it

The standard pack install - ComfyUI Manager, search "ComfyUI_faishme", or:

cd ComfyUI/custom_nodes
git clone https://github.com/AkashKarnatak/ComfyUI_faishme

Restart ComfyUI. It appears under FaishmeNodes.

Where people get burned

The honest caveat: BBOX is a convention, not a standard. Whether this node's * even makes sense depends on which detection pack produced your boxes and what that pack's list semantics are - if your boxes come from a pack that represents them differently, the "repeat" may not compose the way you expect with that pack's downstream nodes. Also remember it makes N identical copies, not N different boxes; if you wanted to fan a single detection out into per-batch variations, this isn't the node. It's narrow, it's unglamorous, and for the exact problem it solves, it's the right five lines of code.