Integer Pair

This node hands you two integers from one box. You type two numbers, it outputs two numbers. The obvious use is a width/height pair - set them together in one place, wire each out to where it's needed - but it's generic: any time you want two related integers defined side by side rather than scattered across the graph, this does it.

Straight up: it's in the pack's _deprecated_ category, kept for backward compatibility with older Searge workflows. In current ComfyUI you'd use a general primitives pack for this. But it's a simple, honest little node and there's nothing wrong with it if it's already in your workflow.

How it works

There's no computation here - it's two constant integer fields with two outputs. The point is grouping. When two numbers naturally belong together (a resolution, a min/max pair, two seeds you want visible in one spot), keeping them in a single node is tidier than two separate widgets, and it's one fewer place to hunt when you tweak.

The inputs and outputs that matter

• value1 and value2 (both INT, default 0) - the two numbers you set. The max is enormous (effectively unbounded for any real use), so resolutions, seeds, step counts all fit.
• Outputs value 1 and value 2 (both INT) - each wires out independently to wherever it's needed.

That's the entire node. No model inputs, no operations, no side effects.

How to install it

Manager: search SeargeSDXL, install, restart. Manual, opencv first (required for the pack to load):

python -m pip install opencv-python
cd ComfyUI/custom_nodes
git clone https://github.com/SeargeDP/SeargeSDXL.git

Restart ComfyUI. No model files.

Common issues & troubleshooting

Type mismatch on connection. These outputs are INT. If a target socket wants a FLOAT (a CFG scale, a denoise value), it won't connect - you'd want a float source for those, not this. Integer outputs are for seeds, dimensions, step counts, and the like.

Off-ratio resolutions on SDXL. If you're using this to set width and height for SDXL, remember SDXL is picky about aspect ratios - it was trained on specific buckets (1024x1024, 1152x896, 1216x832, and their rotations). Feed it an arbitrary off-ratio pair and you'll get stretched anatomy. Stick to the trained ratios and upscale afterward.

Use a maintained primitives node instead. This is deprecated, and it's exactly the sort of tiny utility that a general node pack (rgthree, ComfyUI's own primitives) covers with one maintained node for every type. Only keep using this if it's already wired into a Searge workflow you don't want to disturb.