Load Image (ussoewwin)

There's nothing exotic here, and that's the whole point of flagging it plainly: this is a Load Image node, functionally the same job as ComfyUI's built-in one, just packaged under this pack's own namespace instead of core. You'll see this pattern across a lot of node packs that bundle image or ControlNet processing - an author ships their own copy of a basic loader so their pack's node chain doesn't depend on core node internals staying exactly the same shape release to release, or so it shows up conveniently grouped with the rest of their nodes in the menu instead of scattered under core's own categories. If you weren't specifically looking for it, the stock Load Image node does the same job and there's no meaningful reason to prefer one over the other beyond menu organization.

The interface matches: one required input, image, which is a dropdown listing files ComfyUI can see in its input directory (the schema's sample shows the classic example.png placeholder - in practice this fills with whatever you've actually got uploaded or dropped in). Two outputs: IMAGE, the loaded picture as a standard image tensor, and MASK, pulled from the image's alpha channel if it has one. If your source file has no transparency, expect a blank/full mask rather than an error - that's how ComfyUI's image loading generally behaves, not something specific to this pack.

Where it plugs in: anywhere your workflow needs a starting image - feeding it to an img2img sampler, a ControlNet preprocessor, a captioning node like Florence-2 or WD14 Tagger ahead of this pack's Color Filter, or a Model Patch Loader chain that needs a reference photo. The MASK output is there for the cases where you need to isolate part of the image - inpainting, region-specific edits - and don't want to build a mask from scratch elsewhere.

Installing it: comes bundled with the whole pack, nothing separate to fetch - through ComfyUI Manager, search "ComfyUI-NunchakuFluxLoraStacker," or:

cd ComfyUI/custom_nodes
git clone https://github.com/ussoewwin/ComfyUI-NunchakuFluxLoraStacker.git

Restart ComfyUI and look for it under image/ussoewwin if you're browsing the node menu by category, or just search "Load Image" and pick the one from this pack if you specifically want it.

Being straight about troubleshooting here: there's genuinely not much to go wrong with a node this simple, and inventing failure modes for it wouldn't be honest. If the dropdown is empty, that means ComfyUI's input folder has nothing in it yet - drag a file onto the node or use ComfyUI's normal upload flow first. If you're unsure whether to reach for this node or the stock one, default to whichever one is already wired into the rest of your graph; mixing both in the same workflow works fine but adds nothing.