XY Inputs: PosCond //EasyUse

The simplest way to describe this node: instead of typing four different prompts one at a time and eyeballing the difference across separate runs, you wire up four already-encoded conditionings and let a comparison grid show you all of them against the same seed, same settings, at once.

What it's actually for

This isn't a prompt-text sweep - that's a different tool. This node takes CONDITIONING objects, meaning whatever fed it already went through a CLIP Text Encode (or equivalent) upstream. That distinction matters: you can compare not just different prompt wording, but genuinely different conditioning pipelines - one encoded with a different CLIP skip, one built from a completely different technique like prompt weighting or a different text encoder path - and see the output difference directly, without the prompt text itself being the only variable.

The inputs and outputs that matter

Four optional slots - positive_1 through positive_4 - each taking a CONDITIONING. None are required, which is a little unusual for this node family, but makes sense given the purpose: you fill in as many as you're actually comparing, from two up to the four-slot ceiling. Leave slots empty if you're only comparing two or three variants.

The output is a single X_Y, same shape as every other node in the XY Inputs family, and it plugs into Easy-Use's XY Plot node to actually render the grid. This node itself does no encoding and no rendering - it's purely the axis definition, packaging your pre-built conditionings into something the plot node can iterate over.

When to reach for this instead of a text sweep

If what you're actually testing is prompt wording - "cat" vs "kitten" vs "feline" - a text-based sweep further upstream is usually more convenient, since you're not manually wiring separate CLIP Text Encode nodes for each variant. Reach for this node specifically when the variable isn't the prompt text itself but something about how it got encoded, or when you're comparing conditionings that came from genuinely different sources (a wildcard-populated prompt against a hand-written one, for instance) rather than a simple word swap.

Installing it

ComfyUI Manager: search ComfyUI Easy Use, install, restart. By hand:

cd ComfyUI/custom_nodes
git clone https://github.com/yolain/ComfyUI-Easy-Use

Windows: double-click install.bat. Linux/Mac: run pip install -r requirements.txt yourself. Restart ComfyUI.

Where people get stuck

The most common trip-up is wiring CONDITIONING from a source that isn't actually comparable across the four slots - mixing a positive conditioning meant for one checkpoint's CLIP with one encoded against a different checkpoint entirely tends to produce a grid that's confusing rather than informative, since you're not isolating the variable you meant to test. Also worth knowing: if you only need two or three points on the axis, just leave the unused slots disconnected rather than wiring in a duplicate conditioning - an empty slot is handled cleanly, a redundant one just wastes a render. And as with every node in this family, remember it produces nothing visible on its own; if you're not seeing a grid, check that the XY Plot node downstream is actually present and connected.