Wan Minimal Canvas Test

Let's be direct: this node outputs a blank white image. That's the entire node. WanMinimalCanvasTest takes a width and a height, produces a solid white rectangle, and hands it back to you as an IMAGE. If you installed ComfyUI-WanSeamlessFlow expecting all of its nodes to be useful, this is the one you can safely delete from your graph.

It exists because the pack's author was debugging canvas rendering while building WanEmbeddingPrevizCanvas, and left the test harness in. The class docstring says it outright: "Extremely minimal Canvas test node - just outputs a blank white canvas... For debugging TypeError." The author hit a type error while getting images out of the pack, built the smallest possible node to isolate it, and shipped it along with the fix. It lives in the WanSeamlessFlow/Debug category, which is ComfyUI's way of saying "developer scaffolding, not a tool."

How it works (such as it is)

It creates a white PIL image at your requested size and converts it to a tensor. That's the whole pipeline - there's a dedicated minimal tensor conversion function in the source that exists specifically to sidestep whatever the original error was. Notably, it accepts a text_embeds input, but the code ignores it entirely. The input is there so the author could test that a node accepting WANVIDEOTEXTEMBEDS could still emit an image without the dict breaking something - it's a canary, not a dependency.

The inputs

• width and height (defaults 800x400, range 64–2048) - the size of the blank canvas.
• text_embeds (required by the schema, ignored by the code) - wire anything in; nothing happens.

Output is a single canvas IMAGE. Route it to a preview and you get... a white rectangle.

Install

It comes free with the rest of the pack. ComfyUI Manager, search wanvideo - seamless flow (or "WanSeamlessFlow"), or:

cd ComfyUI/custom_nodes
git clone https://github.com/fblissjr/ComfyUI-WanSeamlessFlow

Restart ComfyUI. No extra dependencies, no models. If you're not using the pack's real nodes, you probably don't need the pack at all - but if you've cloned it, this node is already there.

Is it worth anything?

One genuine use: it's a smoke test. If you suspect the pack failed to load or its image-output path is broken, drop this node in - if it renders a white image, the pack imported cleanly and the tensor pipeline works, which means your problem is upstream, not in the install. Beyond that, treat it as a curiosity. It's not a canvas widget, it doesn't draw anything, and it has nothing to do with Wan video generation. It's the pack author's debugging scar tissue, shipped and documented by accident.

Honest take: this is exactly the kind of node you ignore in a workflow dump and wonder about later. Now you know. Delete it, keep it for the smoke test, and spend your time on WanSmartBlend and WanEmbeddingPrevizCanvas - the nodes that actually do something.