Make Image Batch

You've got three images coming from three different nodes and you want them handled as one batch - for a batched sampler pass, a contact-sheet preview, an IPAdapter that takes multiple references, whatever. Make Image Batch stitches them together. You start with one image input and the node grows more slots as you connect them, so you can pile on as many sources as you like. And unlike a raw tensor stack, it reconciles differing image sizes for you, which is the feature that makes it the friendlier of Impact Pack's batch-building tools.

How it works

You connect images one at a time; each connection reveals the next empty input slot, so the node expands to fit however many you feed it. On execution it combines all of them into a single batched IMAGE. Per the pack's own docs, the inputs "can be scaled up as needed" - meaning if your images aren't all the same dimensions, it handles the reconciliation rather than throwing a mismatch error. That's the practical difference from Image List to Image Batch, which expects its items to already agree in size.

The inputs and outputs that matter

• image1 (IMAGE) - your first image. It's technically optional in the schema, but connecting something here is what starts the batch and unlocks the next slot.
• Additional image slots - appear as you wire in more sources. There's no fixed cap; the node scales to the number of connections.
• Output: a single batched IMAGE.

No numeric settings. It's a plumbing node - the behavior is entirely in what you connect.

How to install it

ComfyUI Manager: search ComfyUI Impact Pack, Install, restart.

Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/ltdrdata/ComfyUI-Impact-Pack
cd ComfyUI-Impact-Pack
pip install -r requirements.txt

Run the pip in ComfyUI's Python environment and restart. No models needed.

Common issues & troubleshooting

Make Image Batch vs Image List to Image Batch - which one? Use Make Image Batch when you have several separate image wires coming from different nodes and possibly at different sizes; it grows slots and reconciles dimensions. Use Image List to Image Batch when you already hold a single list output and just need it restacked as a batch. Both end in a batch; they differ in what they start from and how forgiving they are about size.

Only one image in the output. You probably only connected one slot. The extra inputs only appear after you wire the previous one, so keep connecting until you've added all your sources.

Order matters. The batch order follows the input slot order. If a downstream node treats batch index 0 specially, mind which image you connected first.

You wanted a grid image, not a batch. A batch is still N separate frames stacked in one tensor, not a single tiled picture. If you're after an actual contact sheet, you need a preview or an image-grid node downstream - this node just groups them.