basicIn_INOUT

The honest description is in the source code: pass_through(self, image=None): return (image,). That's the entire node. An optional IMAGE goes in, the exact same IMAGE comes out, and nothing is computed, resized, or touched along the way. If you're used to utility nodes that at least flip a channel or add metadata, basicIn_INOUT will feel like a wire wearing a hat.

So why does it exist? Because the pack's whole philosophy is taming graphs that have grown dense enough to be unreadable. The author's README opens by complaining that workflows get "increasingly dense, making them difficult to understand and operate" - and one of the cleanup techniques is turning a spaghetti of wires into named entry points. This node is the pack's standard "input port" marker: it gives a commonly-used image input a fixed place in the graph, so a workflow reads as "basic input → controller → sampler" instead of thirty crossing wires. It also slots into the pack's IO_Port category, which is where these in/out anchor nodes live - think of it as a label that happens to carry data.

How it works

Under the hood it's a zero-copy identity: whatever tensor batch arrives (or None, if nothing is connected) is returned unchanged. That means it's safe to drop into any image path without changing results - it's purely organizational. There's no hidden resizing, no clamping, no dtype conversion, so you won't get surprised by an output that doesn't match your input.

Inputs and outputs

One optional input, image (IMAGE). One output, image (IMAGE). That's the full inventory. It's one member of the pack's basicIn_* family - there are siblings like basicIn_Seed, basicIn_float, basicIn_int, basicIn_string, basicIn_Sampler, and basicIn_Boolean serving the same "named entry point" role for their data types.

Installing it

It's part of ComfyUI-Apt_Preset, so:

cd ComfyUI/custom_nodes
git clone https://github.com/cardenluo/ComfyUI-Apt_Preset
# then install.bat or pip install -r requirements.txt, and restart ComfyUI

Or use ComfyUI Manager and search for "ComfyUI-Apt_Preset". Note this pack installs a long dependency list regardless of which of its nodes you actually use - it's a big all-in-one, not a lean single-purpose install.

The trap

The only real trap with this node is expecting it to do something. It won't convert types, it won't hold a value when the input is unplugged (it passes None through), and it won't save you from a missing connection upstream. If you want an image constant you can wire into a socket, this isn't it - grab a Load Image node or a solid-color generator instead. Use it for what it's for: a named, visible entry point in a graph that was getting too messy to follow.