Load (Multiple) Images (Batch)

This is the batch-flavored twin of the vsLinx Load (Multiple) Images (List) node. Same multi-select button, same upload-to-input behavior, but the output is different in a way that changes how you use it: here the images come out as a single batch tensor, and the filenames come out as one comma-separated string instead of a list. Batch output is the fast path when everything downstream is happy to process images all at once - but it comes with a rule you need to respect.

The rule: batches are uniform

A batch is a single tensor, which means one shape. If your selected images have different heights or widths, the node resizes them to fit a common size based on the first image - even if that means cropping, padding or stretching to get there. The README is blunt about it: using a batch always implies using uniform dimensions. That's the whole reason the List sibling exists. Pick Batch when your images are already the same size (or you don't care about exact framing), pick List when you need each image to survive intact.

Inputs and outputs

• selected_paths - set via the "Select Images" button; you don't type here.
• fail_if_empty (default on) - error out if the selection resolves to nothing (e.g. files were deleted or moved). Off if you want the workflow to continue with an empty batch.
• filename_handling - "full filename" gives image (1) from image (1).png; "deduped filename" strips ComfyUI's auto-appended (1) counters.

Outputs: images (IMAGE batch) and filenames (STRING, comma-separated). That single filenames string is handy for stuffing into metadata or an Image Saver without iterating.

Also available: a max_images property to cap how many you can select (0 = unlimited).

When to reach for it

• Batch ControlNet conditioning on a set of same-sized references.
• Preview-grid workflows where you want all inputs in one tensor.
• Style-fix batch pipelines: load 8 reference tiles, push them through one encode pass, get a batch of latents out.

Install

Ships in the vsLinx pack. ComfyUI Manager → search "ComfyUI vsLinx Nodes", or:

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

Restart, find it under vsLinx/image as "Load (Multiple) Images (Batch)". No extra dependencies.

The honest take

It's the same good idea as the List node with a sharper trade-off. The batch tensor is faster to process and trivially feeds nodes that want a batch - but the silent resizing can ruin your input framing, and it will surprise you the first time a 512px and a 768px image come out looking like a 512px image with a squished companion. If your inputs are camera-straight screenshots at mixed sizes, prefer the List version or pre-normalize. If you control the input sizes, this is the one to grab - one click, one tensor, done.