GR Checkered Board

A small, single-purpose generator: it draws a checkerboard pattern from scratch, no source image needed. Rows, columns, tile size, two colors, and a couple of border options - that's the whole node. It's the kind of utility that doesn't look exciting until you need it: a transparency-style backdrop for compositing tests, a calibration pattern for checking your crop or resize math actually landed where you think it did, or just placeholder/test art while you're building out a graph before real content is ready.

How it works

rows and columns set the grid, tile_size sets how big each square is in pixels, and color1/color2 (32 named colors each) alternate across the grid the way you'd expect from any checkerboard. border and border_color add a line between individual tiles; outer_border and outer_border_color add a separate frame around the whole board - these are independent controls, so you can have a thick outer frame with no inner tile borders, or the reverse.

The inputs and outputs that matter

• rows, columns, tile_size - the three that determine overall canvas size: total width/height comes out to roughly columns/rows times tile_size, so pick these together if you need a specific final resolution.
• color1 / color2 - the two alternating fill colors.
• border / border_color and outer_border / outer_border_color - optional, both default to 0 thickness (off).

Two outputs: IMAGE (the generated pattern) and MASK.

How to install it

Search GraftingRayman in ComfyUI Manager, or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/GraftingRayman/ComfyUI_GraftingRayman

Restart. As with every node in this pack, OpenAI's CLIP needs to be installed as a separate step or nothing imports - portable ComfyUI: .\python_embeded\python.exe -m pip install git+https://github.com/openai/CLIP.git; system Python: pip install git+https://github.com/openai/CLIP.git.

Common issues & troubleshooting

Pack fails to load. Check your console at startup for the CLIP import error and run the pip install above - this affects every node in the pack, this one included even though it's pure pattern generation with no vision model involved.

Output resolution is bigger or smaller than expected. Remember the final canvas size is a product of rows/columns and tile_size, not a size you set directly. If you need an exact final resolution, work backward from it to pick your grid and tile size.

Border isn't showing up. Both border and outer_border default to 0, meaning off. Set a nonzero value on the one you actually want visible - they're independent, so setting one doesn't enable the other.

Colors look wrong for what you needed. The color choices are named presets (32 options), not a free color picker - if you need an exact hex value that isn't in the list, this node's palette is the constraint, and you may need to recolor the output downstream instead.