简易加载图像 (Simple Load Image)

Short answer, if you're in a hurry: this is a stripped-down clone of ComfyUI's built-in Load Image node, bundled inside a much bigger pack. One dropdown in, an image and a mask out. If you already use the stock loader, you don't need this one - but if a workflow you downloaded calls for SimpleLoadImage specifically, here's what it actually does and why it's probably there.

SimpleLoadImage ships inside ComfyUI-Danbooru-Gallery, a 23-node grab-bag built by a solo dev - their own framing on release was "this is my first node... I had to make it myself with the assistance of AI" - originally as the engine room for a companion project called ShiQi_Workflow. Most of the pack is about pulling images and tags off Danbooru, cleaning up prompts, and managing big multi-character graphs. SimpleLoadImage is one of the handful of plumbing nodes that just move an image in or out of the graph, so the rest of the pack's nodes have something self-contained to plug into without reaching back into core ComfyUI for a loader.

How it works

It does exactly what the name says. Point it at a file already sitting in your ComfyUI input/ folder - same convention the stock loader uses - and it hands back the pixels as an IMAGE, plus a MASK. Nothing configurable beyond that: no upload widget, no channel picker, no resize option in the schema. The mask is almost certainly pulled from the image's alpha channel, the same way the core loader does it, so a plain JPG or a PNG with no transparency will give you a blank mask. That's normal behavior, not a bug.

Inputs and outputs

• image (input) - a dropdown of files already sitting in ComfyUI's input directory. The tooltip just says "select image file," and that's the whole feature.
• image (output, IMAGE) - the loaded picture. Wire it into whatever's next: a preview, a VAE encode, a ControlNet preprocessor.
• mask (output, MASK) - the alpha-channel mask, useful if you're feeding this into inpainting or anything else mask-aware.

Installing it

Two ways in, same as the rest of the pack:

• ComfyUI Manager - search "ComfyUI-Danbooru-Gallery," install, restart.
• Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/Aaalice233/ComfyUI-Danbooru-Gallery.git
cd ComfyUI-Danbooru-Gallery
pip install -r requirements.txt

then restart ComfyUI. The requirements file pulls in requests, aiohttp, aiosqlite, and psutil on top of the usual Pillow/torch/numpy - none of that is needed just to run this one node, but it's a single shared install for all 23 nodes in the pack, so don't skip it thinking you can get away with less because you only want the loader.

Common issues & troubleshooting

Node doesn't show up in the search. Confirm the folder is literally ComfyUI/custom_nodes/ComfyUI-Danbooru-Gallery (not nested one level deeper from a zip extraction), that the pip install actually finished, and that you restarted - then check the ComfyUI console on boot for an import traceback, which is the README's own first troubleshooting step. Custom nodes all share one Python environment with no isolation, so if the install throws a version conflict on aiohttp or similar, that's the well-known tradeoff of how ComfyUI's node ecosystem works generally, not something specific to this pack.

Dropdown is empty or missing your file. Same rule as the stock Load Image node: the file has to already be sitting in ComfyUI/input/ before this node's dropdown will list it. Drop it there - or run whatever earlier step in your graph writes it there - and refresh the browser tab.

Widget text looks like Chinese. The pack's front end defaults to Chinese, with multi-language support built in as one of its JS extensions. Other users report a gear/settings icon elsewhere in the pack that switches the display language to English.