Wan Multi-Image Loader

This one isn't a Wan conditioning node at all - it's plumbing. It's a small utility that lets you upload a batch of images once, arrange them visually, and pull any one of them out by index. The rest of this pack is built around feeding several reference images into a single generation, and before this node existed, doing that meant creating a folder, renaming files in the order you wanted them, and pointing separate Load Image nodes at each one. This is the pack author's fix for that busywork.

Why it exists

Read the pack's changelog and you'll see this described as a "high-performance image selection node," and the framing tells you what problem it was actually solving. The first version stored image data as base64 directly in the workflow - which meant huge workflow JSON files and, worse, browsers hitting QuotaExceededError from LocalStorage limits once you had more than a handful of images loaded. The rewrite switched to ComfyUI's standard /upload/image and /view server endpoints instead, so images live as actual files on disk and the node just references them. Smaller workflow files, faster loading, and - the part that actually matters day to day - a working manual reorder so you can drag images into the sequence you want without touching a filesystem.

Inputs and outputs that matter

There isn't much to configure, deliberately:

• index (INT, required, default 0, 0-999) - which image in your loaded batch to output. This is the only thing you're meant to change per-use; wire it to a different value (or increment it in a loop) to pull a different image out.
• images_data (STRING, optional) - the internal payload that tracks which files you've uploaded and their order. In practice you don't hand-edit this; it's populated by the node's own upload/reorder UI when you add images through the widget, not something you type into by hand.

Output: image - a single IMAGE, whichever one index points at. That plugs straight into a start_image, middle_image, ref_images, or frame_N_image slot on any of the other Wan22FMLF nodes.

Installing it

ComfyUI Manager: search "ComfyUI-Wan22FMLF". Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/wallen0322/ComfyUI-Wan22FMLF

Restart ComfyUI. No model downloads or heavy dependencies for this specific node - it's frontend/backend plumbing on top of ComfyUI's existing upload machinery, not something that touches the GPU or needs its own checkpoint.

Common issues

Old workflows still store images as base64. If you're opening a workflow JSON from before this node's rewrite, expect it to be large and possibly slow to load, and expect it to error if you paste it into a fresh install without the referenced images. The fix is simply to re-upload through the current node's widget rather than trying to preserve the old embedded data.

Index out of range does nothing useful. If index points past however many images you've actually loaded, you won't get a helpful error - you'll get either the last image again or an empty output, depending on how your downstream node handles a missing IMAGE. Double-check the count in the widget before assuming a wiring problem elsewhere.

Reordering doesn't feel like it "stuck." Because the ordering lives in images_data and gets baked into the node's saved state, make sure you're dragging to reorder inside the node's own widget UI and not trying to change order by re-uploading images in a different sequence - re-uploads append, they don't replace.

This is a convenience node, not a required one. If you only ever need one or two reference images, a plain Load Image node is simpler and there's no real reason to add this pack's dependency for that alone. It earns its keep once you're juggling four, five, six reference images across the multi-frame nodes and don't want a folder full of manually renamed files.