InputShareNode

This one isn't an image node at all - it's a graph-organization utility, and it's built to solve exactly the problem the pack's own README opens with: workflows that get so wired-up and dense they become hard to read and error-prone to maintain. InputShareNode takes one value - any type at all, since its input accepts a wildcard - and re-exposes it as twenty separate outputs, so you can drop this one node somewhere central and pull a clean single wire from it to wherever you need that value next, instead of dragging one long wire across a crowded canvas to every consumer.

Why this exists when ComfyUI already lets one output feed many inputs

It's worth being upfront: vanilla ComfyUI already lets a single output socket connect to multiple downstream inputs - you don't strictly need a fan-out node to reuse a value in several places. What this node buys you is layout, not new capability. In a big graph - and this pack's whole premise is taming big graphs full of loader/controller/sampler stacks - routing one wire across the entire canvas to five different nodes gets visually tangled fast. Plant an InputShareNode near the source, and its twenty output ports give you twenty short, local wires instead of five long ones criss-crossing the graph. It's a tidiness tool, not a functionality one.

The inputs and outputs that matter

• source_node (type *, required) - the value to share. Because it's typed as a wildcard, it'll accept an image, a model, a conditioning, a string, a number - whatever you plug in.
• value_1 through value_20 - twenty outputs, all wildcard-typed, all carrying the exact same value you fed into source_node. Each one binds to whatever type its own downstream connection expects, independently of the others.

The node is marked as an output node in its schema, but functionally it's a pass-through - connect as many or as few of the twenty ports as you actually need; unused ones don't need to be wired.

How to install it

ComfyUI Manager: search ComfyUI-Apt_Preset, install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/cardenluo/ComfyUI-Apt_Preset
pip install -r requirements.txt

install.bat is Windows-only; on Linux (comfy.icu's executor included) run pip install -r requirements.txt directly. No model files involved - this is pure graph plumbing.

Common issues

Because every port is typed *, ComfyUI won't stop you from wiring, say, an IMAGE into source_node and then trying to plug value_1 into a slot that expects a MODEL - the wildcard typing only resolves once you actually connect an output, so a mismatched connection downstream will fail at that node, not at InputShareNode itself. If something several nodes away errors on a type it wasn't expecting, trace back through the InputShareNode to what was actually fed into source_node, since the error won't point here directly.

Twenty ports is a fixed number - if you genuinely need more than twenty consumers for one value (rare, but possible in a very large graph), chain a second InputShareNode off one of the first one's outputs rather than expecting more slots to appear.

Standard pack-wide note: this node ships inside a large suite, and a from-scratch ComfyUI-Apt_Preset install pulls in its full requirements.txt (onnxruntime, gguf, transparent-background, scenedetect, and more, for the pack's other 100+ nodes) - none of which this specific node needs, but all of which has to import cleanly for the pack to load at all. There's a documented case of a fresh install throwing IMPORT FAILED; if that happens, the real error is printed in the ComfyUI console above the generic failure line.