From: Eric Rudd Subject: Re: colour blindness Date: 29 Feb 2000 00:00:00 GMT Message-ID: <38BC1C1A.30BABFD7@cyberoptics.com> Content-Transfer-Encoding: 7bit References: <38BAC91D.67FB9086@magma.ca> Content-Type: text/plain; charset=us-ascii X-Trace: client 951852060 38.196.93.9 (Tue, 29 Feb 2000 14:21:00 EST) Organization: CyberOptics MIME-Version: 1.0 Reply-To: rudd@cyberoptics.com NNTP-Posting-Date: Tue, 29 Feb 2000 14:21:00 EST Newsgroups: comp.graphics.algorithms Cory Brown wrote: > Are their any resources on the internet concerning converting RGB > channels so that they simulate various forms of colour blindness? Any > magic one-liners out there? There are many different kinds of color vision defects, but some of the most common forms are representable as reduced forms of normal color vision. One starts out with three color-matching functions representative of normal color vision, and boils them down to two, by linear combination. The color matching functions of both normal and color-defective observers are presented in great detail in the book Wyszecki, Guenter and W S Stiles, _Color Science: Concepts and Methods, Quantitative Data and Formulae_, 2nd edition (Wiley, 1982). This is an excellent text that is worth tracking down. There should be enough information there to construct the matrices to go from RGB to some other two-dimensional space. RGB is OK, since the normal trichromatic color matching functions are implicit in them. One then has to assign colors to the two axes that result, and it is a philosophical question as to what the colors actually "look like" to a color-defective observer. As for one-liners, one could do something like the following r' = 0.5*(r + g) g' = 0.5*(r + g) b' = b and get something that is a reasonable approximation to the two most common forms of dichromatism, protanopia and deuteranopia. This model is reliable at least in that it correctly simulates the confusion between red and green for such observers. -Eric Rudd rudd@cyberoptics.com