Image Combiner

WtlImageCombiner is what you want when you have several images of different sizes and you need to see them all at once and hand them onward without forcing them into a single crop. Where the pack's Image Batch node squashes everything to one resolution, this one keeps every image exactly as it is, previews the lot in RAM, and outputs them as an IMAGE list.

Connect image_1, image_2, and keep going - new slots appear as you connect, same as the batch node. Each input is a normal IMAGE tensor, and each one can be a different resolution; nothing gets resized or cropped. The output port, named "images," is a list of image tensors (OUTPUT_IS_LIST under the hood), so downstream nodes that iterate over a list of images - rather than expecting one uniform batch - will take it happily. That's the whole distinction in one sentence: batch node = forced to one size, combiner = a list where every image keeps its own size.

The other half of the job is the preview. It renders all the images together in the node's preview panel, RAM-only, no disk I/O. So it doubles as a "let me check all my outputs at once" inspector while also being the thing that forwards them. If you've got a workflow producing several crops, tiles, or stage outputs that need eyeballing before they feed a loop, this is a nice two-birds node.

Two things to know. First, it's an output node (OUTPUT_NODE = True), so it can sit at the end of a graph as the terminal preview - but it also emits the list, so it's not limited to being a dead end. Second, the preview-only philosophy means nothing is written to disk; the rendered preview lives in the browser session. If you need files on disk, add a save node downstream.

It ships in the WtlNodes pack (Scorpiosis0/ComfyUI-WtlNodes). Install via ComfyUI Manager by searching "WtlNodes", or:

cd ComfyUI/custom_nodes
git clone https://github.com/Scorpiosis0/ComfyUI-WtlNodes.git

then restart ComfyUI. Deps are numpy, scipy, and pillow - no models, no keys, nothing to download.

The pair to remember: Image Batch when you need one uniform tensor and accept the crop; Image Combiner when sizes must survive and a list is the right shape. Pick by what the next node actually eats.