Color blindness emulation

[orangemug]

An initial attempt at adding 'color blindness emulation' to the editor. To use it add a query string param color-blindness-emulation with one of the following values. Note the links point to the branch build of the editor for testing.

• protanopia
• protanomaly
• deuteranopia
• deuteranomaly
• tritanopia
• tritanomaly
• achromatopsia
• achromatomaly

This is the first stage, if it's correct we'll merge and work out a way to create a UI

Note: Most of the credit goes to https://github.com/hail2u/color-blindness-emulation

[orangemug]

First up, here is a collection of tools to test against and reference implementations

• https://gist.github.com/Lokno/df7c3bfdc9ad32558bb7
• Lacks description of the matrix components mlaily/NegativeScreen#9
• https://gist.github.com/dannyid/c45be2b9d695ac0d9120
• https://github.com/MaPePeR/jsColorblindSimulator

To test I installed Color Oracle however taking a screenshot with Color Oracle enabled seemed to mess with the colors. I've manually tested with Color Oracle comparing with the osm-liberty style. This wasn't really a good enough test because it was only a visual (side by side) check. But it seemed correct.

Secondly I tested with the http://mapeper.github.io/jsColorblindSimulator/ uploading a screenshot of the editor, the results were as follows

• achromatomaly - very slight color variation
• achromatopsia - very slight color variation
• deuteranomaly - very slight color variation
• deuteranopia - same
• protanomaly - same
• protanopia - same
• tritanomaly - same
• tritanopia - very slight color variation

There was very slight color variations in some modes which could be to do with the generation of the screenshots, or some kind of compression. Either way the variation was barely noticeable in the modes highlighted above.

I've also checked the matrixes against other implementations and they are the same.

I believe this to be good to merge.