From the page description, it seems they're using the same algorithm by Matthew Wickline and HCIRN as I use in Sim Daltonism, which is part of jsColorblindSimulator. The results are pretty similar too, except for that purplish tint in the deeply saturated blues.
There appears to be a problem with very saturated blues. This is with Sim Daltonism:
And this is with Coblis:
From the page description, it seems they're using the same algorithm by Matthew Wickline and HCIRN as I use in Sim Daltonism, which is part of jsColorblindSimulator. The results are pretty similar too, except for that purplish tint in the deeply saturated blues.