[Openicc] [argyllcms] Re: Helping with colord

[Chris Murphy]

On Mar 7, 2011, at 2:02 PM, Richard Hughes wrote:

> On 7 March 2011 20:51, Chris Murphy <lists at colorremedies.com> wrote:
>> I wish a grant were possible to determine the accuracy of EDID primary information. With display technologies all over the map, a colorimeter is ill equipped to do a good job all of the time due to mismatching primaries to the color matrix in the colorimeter (or its software). I am presently unconvinced that most people need a colorimeter in this category. It really takes a spectro to be sure. I have seen colorimeters, including the Huey, do a worse job than default EDID based profiles, with common displays.
> 
> GCM already auto-generates rough profiles according to the EDID
> primaries. I'm still not sure how I should be generating the VCGT,
> should I just be assuming the table should be linear and then adjust
> the temperature to something like D50? Are there any helper functions
> in lcms to change the color temperature of a gamma table by linearly
> adjusting the blue and red channels? This seems a bit like a hack to
> me. Ideas welcome.

LCD's have a sigmoidal tone reproduction curve. To turn this into a gamma function defined tone reproduction curve requires vcgt to be an inverse function to the natural behavior of the display. Most stand alone LCDs on the market implement this in hardware. So linear vcgt presents the display as Windows would be default.

If white point is going to be changed, this is best done in hardware, not in the video card LUT. Most calibration packages should suggest the user select a close color temperature preset in the display itself, before fine tuning it with adjustment to the upper end of the vcgt curves. Aggressive change of these curves, especially 6-bit LCD laptop displays, will result in noticeable posterization of images and video.

Without measurements*, or EDID that describes the tone reproduction characteristics of each channel, you're kinda guessing. 


Chris Murphy


*measurements could be with eyeballs using a visual based calibration tool to determine a displays actual TRC per channel, so that you can compute an inverse function for vcgt so that you get the desired TRC (typically defined by the sRGB curve, or a 2.2 gamma function is acceptable as well).