[SciPy-user] Maximally distinguishable colour map

[David Warde-Farley]

On 2-Mar-08, at 4:44 AM, Robert Kern wrote:

> On Sat, Mar 1, 2008 at 10:12 PM, David Warde-Farley <dwf at cs.toronto.edu 
> > wrote:
> The qualitative color palettes on the ColorBrewer are fairly
> rigorously designed. Just avoid the pastel palletes. Note that human
> vision can only reliably distinguish about 7 hues in the same scene.
> Their 'Dark' palette is probably the most suitable.

Thanks Robert. I suspected there was an upper limit like that;  
consequently I'm currently doing every 43rd when there are less than 8  
and an even spacing with less difference between them (our system  
prefers sparse solutions and so a user would have to force a non- 
sparse situation like this, in which case they can't reasonably expect  
to make sense of the colours anyway).

It seems as though none of the qualitative palettes are certifiably  
colourblindness-friendly, but I guess this comes with the territory.

David