Add perceptualTolerance parameter to Diffing+bitmap to compare perceptual differences using Delta E 1994

[JoelWhitney]

Differences from swift-snapshot-testing

Similar to what was implemented in swift-snapshot-testing, with some slight differences

To match the established pattern of tolerance, this PR adds an optional perceptualTolerance parameter that allows you to specify the percentage each pixel can be different from source pixel and still considered a match. This parameter complements the existing tolerance parameter that determines the percentage of pixels that must be considered matching in order to consider the whole image matching.

The default value of 0.0 means pixels must match perfectly (and performs existing equality check). Any other value greater than 0.0 and less than or equal to 1.0, will be used when comparing each pixel where 0.0 (0%) means no difference and 1.0 (100%) means completely opposite. A suggested value of 0.01-0.02 is similar to the precision of the human eye.

Perceptual tolerance value	Description
0.0	Must be exactly equal
≤ 0.01	Allows differences not perceptible by human eyes
≤ 0.02	Allows differences possibly perceptible through close observation
≤ 0.1	Allows differences perceptible at a glance
≤ 0.5	Allows differences when more similar than opposite
≤ 1.0	Allows any differences

Unlike swift-snapshot-testing, which uses a CILabDeltaE CoreImage Filter, we must do a byte-by-byte comparison, calculating Delta E values as on-the-fly.

Formula resource: http://www.brucelindbloom.com/index.html?Eqn_XYZ_to_Lab.html

Description of pointfreeco/swift-snapshot-testing#628...

Problem

The existing image matching precision strategy is not good at differentiating between a significant difference in a relatively small portion of a snapshot, and imperceivable differences in a large portion of the snapshot. For example, these snapshots below show that a 99.5% precision value fails an imperceivable background color change while allowing noticeable changes (text and color) to pass:

Reference	Fails	Passes
	Imperceivable background color difference	Significant text and color changes

Solution

This PR adds a new optional perceptualPrecision parameter to image snapshotting which determines how perceptually similar a pixel must be to consider it matching. This parameter complements the existing precision parameter that determines the percentage of pixels that must be considered matching in order to consider the whole image matching.

This approach is similar to pointfreeco/swift-snapshot-testing#571 and pointfreeco/swift-snapshot-testing#580 but uses perceptual distance rather than Euclidean distance of sRGB values. This is significant because the sRGB color space is not perceptually uniform. Pairs of colors with the same Euclidean distance can have large perceptual differences. The left and right colors of each row have the same Euclidean distance:

[JoelWhitney]

Hey @balazstothofficial, I know there hasn't been much activity here recently, is this PR better directed towards my fork?

[balazstothofficial]

@KlausNie @Numoy Maybe one of you can take this over?

[KlausNie]

@JoelWhitney sorry for the delay
and thank you, @balazstothofficial, for tagging me, I'd have missed it!

@JoelWhitney, I've set @eldrig0 on it, and he's getting into the topic with the intention of getting it merged!

[JoelWhitney]

That's great, thank you @KlausNie @eldrig0!

I was able to test the changes in a test framework where I had a reproducible case. I meant to circle back and add unit tests but never got to it. I'll see if I can add some tests while the changes are being reviewed

[JoelWhitney]

Hey @eldrig0 @KlausNie, is there going to be a new release published following this merge?

[eldrig0]

Yup, will do it tomorrow