[whatwg] Colour correction (was: Canvas ImageData comments)
hsivonen at iki.fi
Sun Jan 20 11:03:56 PST 2008
On Jan 20, 2008, at 19:58, Darin Adler wrote:
> On Jan 20, 2008, at 9:10 AM, Henri Sivonen wrote:
>> Most of the time, the solutions to the color space problem are
>> worse that the problem itself. The easiest fix for this whole mess
>> would be making Mac OS X default to 2.2 gamma (i.e. be compatible
>> with the overall legacy instead of the Mac legacy) and then
>> continue to treat Web color values as being in the system color
>> At least in order to avoid Breaking the Web, browsers need to treat
>> all untagged colors in a mutually consistent way within a browser
>> window regardless of the source of the color: image files, CSS,
>> plugins, video, legacy HTML attributes, etc. The usual way to do
>> this is to treat all untagged color values as being in the system
>> color space.
> Good explanation.
> The proposal from color experts here at Apple is to interpret
> untagged colors in the sRGB color space. This is what's done with
> most other untagged color in Mac OS X.
It sure looks like most untagged color is taken to be in the system
> But this rule not yet implemented in WebKit. Instead, when
> displaying on screen, today's WebKit treats untagged color as if it
> was in the system's primary display device's color space. This means
> that no color correction is applied to such colors.
Yes, and the sRGB approach would Break the Web as long as the Flash
plug-in doesn't participate.
> I'm not certain exactly what the "system" color space is
I mean the profile selected in System Preferences > Displays > Color >
> or whether the Mac OS X gamma difference from Windows is important
> when designing this.
The gamma difference is the foremost problem being "solved" but it
would be by far easier to solve by changing the Mac OS X default gamma
to match Windows. Computing color space transformation just because
Mac OS X is stuck with an unusual default gamma value is an overkill.
Sure, there are other color space differences, but most of the time
for most people, the non-gamma differences are less important.
> I think the sRGB design is a good one.
I disagree. Why would you want a brand new Cinema display emulate the
gamut of an office CRT from the previous millennium potentially by
clipping instead of stretching the colors to gamut of the device at
> When displaying color on devices with unusual color characteristics,
> it doesn't make sense to display color with no correction at all.
Yeah, but most Web browsing systems don't have too unusual
characteristics except for the Mac default gamma.
hsivonen at iki.fi
More information about the whatwg