diffimg

compute a pixel-by-pixel difference image between two input images

TLDR

Compute the difference of two images and save it

$ diffimg [image1.png] [image2.png] [diff.png]

Use it for visual regression tests

$ diffimg [expected.png] [actual.png] [report/diff.png]

diffimg computes the per-pixel difference between two equally sized images and writes the result to a third file. Each output pixel encodes the delta between the corresponding pixels of the two inputs — identical areas appear black, changed areas light up. This makes it an easy drop-in for visual regression checks, before/after comparisons, and QA pipelines.Implementations vary — some are Python/PIL wrappers, some are Qt-based GUIs with a CLI front end, and some are Go/Rust rewrites. All agree on the basic three-argument signature and on support for PNG, JPEG, and GIF input.

PARAMETERS

image1

First input image.

image2

Second input image. Must be the same dimensions as image1.

output

Path where the difference image will be written. The extension selects the output format.

SUPPORTED FORMATS

PNG, JPEG, GIF, and (where ImageMagick is available under the hood) PostScript and TIFF.

CAVEATS

Both images must have the same dimensions; resize or crop first if they don't. Output reflects raw pixel deltas, not perceptual differences — a 1-pixel shift lights up every edge. For perceptual or structural comparisons, use compare (ImageMagick) with `-metric SSIM` or a dedicated tool like `pixelmatch`, `odiff`, or `perceptualdiff`.Different tools also named `diffimg` exist — check `diffimg --version` or the package description to know which you have.

HISTORY

Multiple tools named diffimg have circulated since the early 2000s. The most widely packaged is a Python CLI/GUI by Jonathan Zurflueh (thebulb). Other implementations include Qt-based desktop GUIs and Go/Rust rewrites.