trotzig
/
diffux
/
104
/
1
99%
master: 99%
master: 99%
|
Ran
|
Files
19
|
Run time
–
|
Badge
Embed ▾
README BADGES
|
coverage: 98.322% (+0.04%) from 98.282%
push
travis-ci
lencioniUse translucent before-image + bright color for diff Previously we were generating a diff image that only contained grayscale pixels that were different between the two images. While this was effective at helping us spot that there were differences, we spent a lot of time looking at the three images to figure out where the differences actually were. To help remedy this, we believe that it makes sense to render a faded version of the "before" image (baseline) underneath the diff pixels. To make sure that the diff pixels aren't lost on top of this translucent image, we are rendering them in a bright pink color. We chose bright pink because it is very noticeable and also unlikely to dominate most snapshots. We set the opacity of the diff pixels to be at minimum double the opacity of the translucent image. This will help even the smallest differences actually be noticeable. At the same time, I changed the magic array-like access on the ChunkyPNG images to use `get_pixel` and `set_pixel` because these methods do the same thing as the magic array-like accessors but without checking of the bounds. Because we are in a tight loop and because we have already defined the bounds of the image, this should improve the performance of the diffs without sacrificing any robustness. Change-Id: I964ac3af1
293 of 298 relevant lines covered (98.32%)
17.27 hits per line
