〽️ Image Ops Constant

ImageOps Constant is the least glamorous node in the ComfyUI-Majoor-ImageOps pack, and probably the one you'll use more than you expect. It generates solid-color or checkerboard RGBA plates - flat images with a color, a size, an alpha, and even a frame count. No input, no model, just a source of pixels. That sounds boring until you're staring at a graph that needs a background plate, a placeholder, a test pattern, or a canvas to composite something onto.

What it makes

mode switches between constant (a solid fill of color) and checkerboard (color and color_b alternating, with tile_size plus offset_x/offset_y to shift the pattern). Output resolution comes from width and height (1–8192), with an aspect_ratio preset helper that derives them from common formats. frame_count (1–4096) generates a whole batch of identical frames - which is the batch-first flex: one node produces a multi-frame background plate for an entire video, ready to comp over.

alpha sets the output opacity (0–1), and here's the neat bit: that same value is what the node returns as its mask output. So a single alpha setting drives both the transparency of the image and the matte it hands you. Wire mask into a merge and the plate composites as a semi-transparent layer immediately.

Outputs

image, mask, and width, height, frame_count as plain INTs. Those INTs are the sleeper feature - they give you the dimensions without counting pixels yourself, which is genuinely handy when you're syncing sizes across the graph.

Where you'll actually use it

• Background plates for compositing - a solid or checkerboard canvas to drop a subject onto.
• Test patterns - check alignment, crop behavior, or whether a node is doing something to flat color. A checkerboard instantly shows warping, tiling, and resampling artifacts that a photo hides.
• Placeholders while you build a workflow - wire in Constant, get the graph flowing, swap the real footage in later.
• Canvas sizing - generate a plate at a specific resolution and let the INT outputs drive downstream sizes.

Install

ComfyUI Manager (search "ComfyUI-Majoor-ImageOps"), or:

cd ComfyUI/custom_nodes
git clone https://github.com/MajoorWaldi/ComfyUI-Majoor-ImageOps

Restart and hard-refresh the browser (Ctrl+F5 / Cmd+Shift+R). No models, no extra dependencies - it's torch tensor generation.

The honest take

There are fancier plate generators with gradients and procedural detail (this pack's own ImageOps Noise and Ramp cover that ground). Constant is deliberately simple, and that's the point: when you need a flat color or a test pattern, you want zero thinking. The one behavior to remember is that alpha drives both the image and the returned mask - if your plate looks too transparent, it's the alpha slider, not a bug. And don't expect anti-aliasing magic on the checkerboard edges at tiny tile sizes; that's a resampling thing downstream, not this node's job.