[Feature request] Ignorable areas

[alexwhitman]

Would it be possible to implement the ability to ignore areas of an image when diff'ing? For example, if I have an image of a web page which I want to compare against a baseline image I might want to ignore certain areas, such as dynamic text or images.

This could be done by specifying rectangle areas to ignore with top left x and y coordinates and bottom right x and y coordinates.

[cesutherland]

The easiest thing to do would be to white-out those areas before doing the diff. Might be beyond the scope of this tool, but I'll think on it.

[jreichenberg]

+1 on this feature request...For automated testing it would be really excellent to be able to define a region to diff. Maybe a reasonable MVP would be something as simple as an option to say "region": ["0,0", "200,100"] to specify the top left and bottom right corners of the diffable region.

Great project!

[baartz]

Agreed. Exclusion areas would be outstanding.

[hayesmaker]

👍 +1

[subhog]

I've added a pull request with this feature: #58.

The equal method accepts regions option. If passed, an initially empty mask is generated that describes which part of images to compare. regions is an array of rectangles, each rectangle passed in the form [x0, y0, x1, y1, include], where include decides on whether the rectangle should be added to the mask (and compared), or subtracted from the mask (and ignored). Example:

equal(a, b, 0, {regions: [
    [20,  200, 220, 500, true],
    [60,  400, 80,  420, false],
    [160, 400, 180, 420, false],
]})

will check the "face", but not the "eyes".

I didn't want to change the existing API in the PR, but I also suggest that the tolerance parameter should go inside the options object.