Very enjoyable. Image manipulation/detection/recognition is a field that actually starts out pretty simple (I was surprised, I assumed it would be fiendishly math-heavy), but scales well into the dissertation-level.
The article is a pretty cool highlight of some low-hanging fruit. I'd like to see something using a n-neighbor-aware technique instead of pixel-by-pixel to detect shifting/scaling/rotation in a direction in addition to color change, but I'm sure those are much longer articles.
You're right, I wanted to keep this one short and simple. I have been thinking about writing an article about calculating image similarity, instead of just finding differences between two versions. Stay tuned! :)
The article is a pretty cool highlight of some low-hanging fruit. I'd like to see something using a n-neighbor-aware technique instead of pixel-by-pixel to detect shifting/scaling/rotation in a direction in addition to color change, but I'm sure those are much longer articles.