/wd/ - Web Design

our graphics team finally ditched that massive 18gb repo of golden images for something way more efficient. instead of doing a pixel-by-pixel diff, we now just compare MD5 hashes stored in a tiny 19kb json file. it is insanely much lighter on our embedded hardware since we arent processing huge files every time.
>the old way was such a bottleneck
it basically turned our entire test suite into just a bunch of string comparisons . does anyone else still deal with heavy image assets in their automation, or is everyone moving toward this kind of hashing approach?

article: https://dzone.com/articles/visual-regression-frame-buffer-hashing