Load Image from Path

ComfyUI's built-in image loader wants you to upload through the browser widget. This node skips that entirely - give it a filesystem path as a plain string, and it loads the image straight off disk. Small difference, but it matters the moment you're not manually clicking around a UI.

Where this earns its keep is automation. If you're driving a workflow via the API, batch-processing a folder of images, or feeding output from an earlier script into a graph without a human sitting there to upload anything, the standard upload widget doesn't really fit - there's no browser session to drag a file into. A path-based loader does, because the path itself can be generated, scripted, or passed in as a run parameter, and the node just picks it up.

The one input. img_path - a required STRING, defaulting to empty. That's genuinely the whole configuration surface: no resize, no format conversion, no color-space option exposed on the node itself. Point it at a valid image file and it loads it. The single output is image, typed IMAGE, which is ComfyUI's standard image type and plugs into anything downstream that expects one - a vision-capable LLM node in this pack, a preview node, a standard ComfyUI sampler pipeline, whatever you're building.

One thing worth being deliberate about if you're running this on a hosted or API-driven executor rather than your own machine: img_path has to point somewhere the ComfyUI process itself can actually read, at the moment it runs. That's obvious when everything's local, but it's the first thing to check if this node errors out on a hosted run - a path that made sense on your laptop generally won't resolve inside a remote execution environment, and you need to know where your input files actually land in that environment before wiring a static path into this node.

Installing it. Load Image from Path comes with the full comfyui_LLM_party pack - no separate install step. Get it through ComfyUI Manager (search "comfyui_LLM_party") or manually with cd ComfyUI/custom_nodes && git clone https://github.com/heshengtao/comfyui_LLM_party, then pip install -r requirements.txt from inside the project folder using your ComfyUI Python environment, and restart. It needs no model file and no external service - it's reading a file off disk - so there's nothing to configure beyond getting the path right.

Troubleshooting. If it errors, it's almost always the path: wrong location, wrong permissions, or a format the underlying image library can't parse (corrupted file, or something that isn't actually an image despite the extension). If the node's missing from your list entirely rather than just erroring on a bad path, that's the pack's much larger dependency set failing somewhere else during import - this project bundles requirements for a dozen different optional integrations at once, and a single conflicting version can take the whole node set down. Check the ComfyUI startup console for the real error before assuming this specific node broke.