📁 Eric Load Image Extended

ComfyUI's built-in Load Image is fine, until it isn't. It handles PNG and JPG, and it quietly loses the path of whatever you loaded - which matters more than you'd think once you want to read a file's metadata or save next to it. Eric Load Image Extended is the loader from the AAA Metadata System pack that fixes the second problem and adds a pile of formats to the first. If you've already installed that pack for its save-with-metadata nodes, this is the natural front end: it hands you the filename and full_path strings those nodes and the metadata query nodes want to be fed.

The headline feature is format support. Beyond the usual .jpg, .png, .bmp, .gif, the picker accepts .tiff, .webp, .psd, and .svg. PSDs load flattened (as a composite), and SVGs get rasterized into pixels - so it's "load SVG," not "load and keep vector," and that's worth knowing before you feed one into a sampler expecting a clean 1024×1024. Everything else behaves like the stock node: you pick from the dropdown, it scans ComfyUI/input.

What actually sets it apart is the outputs. It returns the standard image (IMAGE) and mask (MASK), same as always, but also filename and full_path as strings. That second one is the quiet MVP of this pack. The metadata side of AAA - the Query node, the Debugger - wants a filepath as input, and this node is the intended source for it. You can also route filename into anything that builds a save prefix, so your output files end up named after their input.

For a beginner the wiring is dead simple: image → your VAE decode / sampler chain as usual, and full_path → whatever metadata or file-node you want to interrogate. That's the whole job. It doesn't sample, doesn't condition, doesn't upscale.

One gotcha: the file picker lists files in the root of your input directory, not subfolders, and this node inherits the pack's install requirements (it's a normal ComfyUI node, so just install the pack and restart). Also remember this is a niche, one-author pack - there's no community army behind it, so treat it as a helpful utility rather than something every workflow you download will already have.

Install via ComfyUI Manager by searching AAA Metadata System, or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/EricRollei/AAA_Metadata_System
cd AAA_Metadata_System
pip install -r requirements.txt

Then restart ComfyUI and you'll find it under the Image category. If the dropdown is empty, drop your file into ComfyUI/input and refresh. It's a small node that fixes a small but real annoyance - the kind you stop noticing, which is exactly the point.