Image & Mask

Functionally, this is ComfyUI's core LoadImage node wearing RiceRound branding: pick a file from your local input folder and get back both the image and its alpha channel as a mask. What makes it worth a separate node rather than just using the built-in one is where it sits in RiceRound's system - this is the Input-category node you use for a local image parameter while you're building and testing a workflow you intend to publish through RiceRound.

Why not just use the core Load Image node

You can, while you're just experimenting. But ComfyUI_RiceRound's whole premise is that RiceRound Publish scans your graph for its own Input-category nodes to figure out what should become an editable field on the page it generates. A core LoadImage node isn't part of that vocabulary, so if your published workflow needs to let someone supply their own starting image, it needs to come from a RiceRound node RiceRound's own tooling recognizes - this one, or one of its URL/bridge siblings depending on where the image is actually coming from.

The inputs and outputs that matter

• image (required, a file-picker dropdown pulling from ComfyUI's input directory - the same widget core LoadImage uses, defaulting to whatever sample file, like example.png, is present) - this is what you're setting locally to test with.
• Two outputs: image (IMAGE) and mask (MASK), pulled from the alpha channel the same way core LoadImage does it. Wire image into your generation pipeline and mask anywhere you need the alpha, or ignore it if your source file doesn't have one.

When to reach for a different image node instead

This pack ships several image-related Input nodes and they're not interchangeable. Use this one (RiceRoundImageNode) when the image is a local file you're picking by hand. If the image needs to come in from a URL at runtime instead - the more realistic shape for what an end user of your published page would provide - use RiceRoundDownloadImageNode or RiceRoundDownloadImageAndMaskNode. If the image is already flowing through your graph from somewhere else and you just need to mark that point as an externally-provided input, that's RiceRoundImageBridgeNode.

Installing it

ComfyUI Manager: search "RiceRound," install, restart. By hand:

cd ComfyUI/custom_nodes
git clone https://github.com/RiceRound/ComfyUI_RiceRound

then restart. No dependencies or model downloads are documented for this pack in the README - it's a workflow-packaging layer, not a generation one, so there's nothing large to wait on during install.

Common issues & troubleshooting

The file picker is empty or missing the file you want. Same behavior as core LoadImage: it lists whatever's in ComfyUI's input/ folder, so a file has to actually be placed there (or uploaded through the UI) before it shows up in the dropdown here.

You're not sure if this belongs in a workflow you're about to publish. It does - it's Input-category, and per the README that's exactly the family of nodes meant to end up in the graph you feed into RiceRound Publish. The nodes to leave out are the separate Output-category ones, used for local debugging only.

Published page users need to upload their own image, and a static local file picker clearly can't do that. That's expected - this node is for your local testing default, not for how an end user actually supplies input once the workflow is live. Confirm how RiceRound's platform actually surfaces image upload on the generated page (the README's screenshots suggest it does, via the wider publish flow) rather than assuming this specific node is the upload mechanism itself.