通用打印输出

LamCommonPrint (通用打印输出, "generic print output") is the debug node: you wire any value into it, it prints that value to the ComfyUI console, and it passes the value through on its output so the rest of your graph keeps running on the same data. If you've ever stared at a workflow wondering what a node actually produced, this is the honest answer - it shows you, without breaking the wire.

The "通用" (generic) matters here: the input is typed *, so it accepts images, strings, ints, lists, masks - anything. That's its whole trick. Most debug nodes are typed to one data kind; this one swallows whatever you throw at it and reports it back.

How it works

The mechanism is deliberately dumb: on execution it prints the value to the console (that's the terminal window running ComfyUI, not the UI), then returns the same value unchanged. Nothing is transformed, normalized or truncated-by-design - you get the raw representation as Python prints it. For tensors that's usually a shape line like torch.Size([4, 512, 512, 3]); for strings and numbers it's the value itself.

The inputs and outputs

• obj (*) - the value to inspect. Wire anything here.
• Output obj (*) - the exact same value, passed through untouched. Chain it: wire the output to the next node and you get logging plus continuation for free.

One flag to know about: this node is marked as an output node. ComfyUI treats reaching it as "the workflow finished," which is great when it's genuinely your last node - but it means dropping it mid-graph can make ComfyUI report completion early. If you need the logging in the middle of a pipeline, use the pack's LamCommonPrintNoOutput variant instead, which does the identical print-and-pass but doesn't claim to finish the job.

Install

Pack install, nothing extra:

cd ComfyUI/custom_nodes
git clone https://github.com/yanlang0123/ComfyUI_Lam

and restart. No models, no patch.

Common issues

• Nothing in the UI - prints go to the console/terminal, not to a UI panel. Check the window you launched ComfyUI from.
• Workflow ends "early" - you placed the output-node version mid-graph. Swap in LamCommonPrintNoOutput.
• Tensor prints are just shapes - expected; that's Python's default repr for tensors. If you need pixel-level inspection, print the image through a preview node instead.

It's two seconds to wire and it will save you an hour of guessing. Every kitchen-sink pack needs one, and this one does the job without ceremony.