Olaf Jan Schmidt wrote:
> Hi!
> 
> The work on new guidelines for fonts and colour  use in KDE4  isalmost 
> finished. I have received some more helpful comments from Kenneth that I 
> still need to merge into the document:
> http://amen-online.de/~olafschmidt/colors/
> 
> While discussing this with him and Danny Allen on IRC, I realized that the 
> suggested default colours in the current draft of the document do not have 
> enough contrast for accessibility. I have therefore written a PHP script to 
> objectively evaluate the colour contrast of the Oxygen palette, and then 
> defined a new suggestion for the default colours in KDE4:
> http://accessibility.kde.org/oxygen.php
> 
> Of course I am interested in hearing feedback from artists on this.
> To make testing of the suggested settings possible, I am attaching two files:
> 
> 1. Most colours can be tested by copying KDE4.kcscr to  
> ~/.kde/share/apps/kdisplay/color-schemes and then selecting the "KDE4" colour 
> scheme in kcontrol.
> 
> 2. For seeing how the new colour roles could be used for syntax highlighting 
> in KDE4, colorscheme-kate needs to be attached (not with kwrite or kate, but 
> with some other editor like kedit) to the file 
> ~/.kde/share/config/katesyntaxhighlightingrc
> Then you need to select "KDE4" in the Kate colour settings, to switch on 
> syntax highlighting and to open an adequate file.

I applaud you for realizing that color is a usability issue.

I have been complaining about a certain icon theme in regard to 
usability issues for some time.  Perhaps CrystalSVG is OK for people 
with 20/20 vision.  But, I am slightly visually impaired.  I am legally 
blind without corrective lenses and with them my vision isn't 100%.

You are correct that color is important.  However, this is not the last 
word in usability considerations.  Information theory can also be 
applied to usability.  The idea of which color contrasts well with which 
is part of this.  And for color blindness, these colors would need to be 
weighted first.

Specifically, it is the spatial (2 dimension) frequencies that are 
relevant (e.g compare a checker board with a smooth gradient).  For 
monochrome images simply taking the 2 dimensional FFT and computing the 
volume under the curve would give a start.  Using a weighting function 
might improve the metric.  With a color image the issues are more 
complex.  Obviously, you are going to first need to decompose the color 
image into three images and take the 2 dimensional FFT of each of the 
three.  Not only would a weighting function be used for computing the 
weighted volume of each of the three but a a weighted sum should be used 
for combining the three weighted volumes to arrive at a final figure. 
The final consideration with color is which color space would be most 
appropriate -- would a color space other than RGB give better results? 
At first glance it appears that the color and brightness metrics for 
NTSC color TV might be a better choice.

-- 
JRT
______________________________________________________________________________
kde-artists@kde.org |  https://mail.kde.org/mailman/listinfo/kde-artists