Re: 3.7 DDC: Colors

On Wednesday, August 2, 2006, 7:56:42 AM, Susan wrote:

SL> Hello,

SL> A comment for the MWBP guidelines 3.7 Default Delivery Context [1]. Under
SL> Colors it says:

SL>     "Web safe.

SL>      (A Web safe color is one that has Red/Green/Blue components
SL>      chosen only from the values 0, 51, 102, 153, 204, and 255.)"

SL> The definition looks good  

I wouldn't call it "good", and the concept of 'web safe colors' is based on a number of misunderstandings. 

I'm frankly alarmed to see some piece of mythology like this turning up in a W3C spec. Next we are going to be claiming that using some magic set of colors always prevents dithering, and such like nonsense.

 
SL> but usage of the phrase "Web safe" doesn't always
SL> match the definition. Maybe the spec could say "216 colors" instead of Web
SL> safe (the 216 are the 6x6x6 cube that is part of some popular 8-bit video
SL> palettes). Chris Lilley, copied on this mail, may or may not have a moment to
SL> correct me there.

Thanks, Susan.

I do have a moment - both to correct the definition and to argue that the term is a misnomer and is neither safe nor useful for the Web. 

Its easier to describe it in hex, rather than arbitrary decimal numbers.

A 6x6x6 simulation of truecolor is produced by using three digit color values (eg #3CF )  where each digit is either 0,3,6,9,C or F.

There is no guarantee whatsoever that restricting colors in an image to these values will prevent dithering, so the 'safe' part of 'web safe' does not apply.

Furthermore, restricting an image to these colors often means that,on a limited palette system,the colors need to get mapped to the closest color in the palette so the original color is subject to two distortions one to move it to 'web safe' and one to move it to the actual device palette. This results in objectionable color shifts and banding which could have been avoided. 

'Web safe' colors are an anachronism based on faulty logic. They had some utility in the days of 8-bit Mac and PC graphics, giving some chance that the colors would not dither. However, they totally failed on other systems (eg XWindows) where they would *always* dither. Their applicability to mobile is highly questionable.


SL> [1] http://www.w3.org/TR/2006/CR-mobile-bp-20060627/#ddc

SL> Thank you,




-- 
 Chris Lilley                    mailto:chris@w3.org
 Interaction Domain Leader
 Co-Chair, W3C SVG Working Group
 W3C Graphics Activity Lead
 Co-Chair, W3C Hypertext CG

Received on Wednesday, 2 August 2006 11:09:03 UTC