VRGDG UI Image Batch Multi

VRGDG UI Image Batch Multi is a convenience node with a boring job: read a list of image file paths from a text box, load every image, and hand you back one IMAGE batch. That's it. No dynamic noodle inputs to wire, no per-image loaders, no file browsing dance - you paste paths and get a batch.

The "UI-friendly" bit is the whole story. Most batch-loading nodes in the ecosystem use dynamic inputs or a folder-glob pattern, which is great until it isn't - especially in big packs like this one where a workflow runner or a UI panel needs to feed a list of paths it already has. This node accepts the path list in a couple of formats (one path per line, a JSON array of paths, or an object with image_paths/images), which makes it trivially easy to feed from other nodes that emit path strings.

Why you'd reach for it: you've got a dataset folder, a batch of renders, or a set of Z-Image enhanced crops sitting on disk and you want them all in one tensor to feed a grid, a comparison, or a video-encode step. It's the kind of node that turns "load 40 images" into copy-paste.

The inputs

• image_paths - the list of paths, multiline. One per line, or a JSON list/object. This is the only input you'll actually think about.
• upscale_method - used when images in the batch have mismatched sizes and have to be brought to a common resolution. Choices are the standard ComfyUI set: nearest-exact, bilinear, area, bicubic, lanczos. bilinear is the default; switch to lanczos if you're mixing large images and want the downscale to be clean, or nearest-exact if pixel-art integrity matters more than smoothness.

The output

• images - the single IMAGE batch, ready for anything downstream.

Installing it

Same pack install as everything else in this collection: Manager → search vrgamedev, or:

cd ComfyUI/custom_nodes
git clone https://github.com/vrgamegirl19/comfyui-vrgamedevgirl.git

Restart and hard-refresh. No models, no extra dependencies.

Common issues

• Nothing loads. Double-check the path format - trailing quotes or braces get stripped per-line, but a genuinely malformed entry (wrong drive letter, missing extension) will fail. One bad path in the list usually fails the whole batch.
• Mixed-size images come out warped. The node normalizes sizes via upscale_method. If the result looks stretched, that's the resize, not corruption - keep inputs the same aspect ratio or accept the normalization.
• Huge batch, huge memory. A batch of 4K images is a lot of VRAM/RAM in one tensor. Load in smaller groups if you're pushing the limits.

Not a flashy node, but when you need forty images in one tensor without rebuilding your graph, it's exactly the right tool. Keep it in mind for dataset review grids and batch enhancement passes.