Simple Image Loader

Simple Image Loader is the most honest node in this pack: a STRING path goes in, an IMAGE tensor comes out. You type (or wire) a file path into image_path, it opens the file with PIL, converts to RGB, scales to 0–1 float, and hands you a batch-ready IMAGE you can feed straight into a KSampler, a mask node, or anything else that wants pixels.

Compared to ComfyUI's built-in LoadImage, this loses the convenient dropdown of files in your input folder. What it gains is the ability to load any absolute path - a file outside the ComfyUI input directory, a path generated by another node in this pack, a render you just dumped into output and want to re-import without moving it. When your workflow is path-driven - which is the design philosophy of lazniak's whole suite - a loader that takes a bare string is the right shape. Wire ComfyUIBasePath or FilePathProcessor into it and you've got a loader that resolves files without you ever touching a browser.

What you set

• image_path (STRING) - an absolute path to an image. The default placeholder is C:/path/to/your/image.png, which is the author being honest about what this expects: a real path, not a filename.

Output: IMAGE. That's it.

What to know before you use it

• Absolute paths only. Relative paths won't resolve the way you hope.
• It converts to RGB. Any alpha channel gets flattened. If you're loading a PNG with transparency to do masking work, this loader will silently drop the alpha - use core LoadImage for that job instead.
• It errors loudly. Missing file, bad path, unreadable format → it raises, it doesn't return a black image. That's good behavior; a silent black frame is how you waste an hour.
• No metadata handling. It reads pixels and nothing else - no EXIF, no workflow chunks. The KB's image-I/O essay is worth a skim if you expected the embedded workflow to survive; here it's just pixels in, pixels out.

Install

Standard videoclipgenerator install:

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

Restart, find it under "image". Pure PIL and torch, both already present, so no extra dependencies and no models to download. If you mostly work with images that live in ComfyUI's input folder, you'll probably keep using the core loader - this one earns its place in path-driven and multi-node workflows, where being a simple string-in/pixels-out component is exactly the point.