[argyllcms] Re: What does dispcal actually do?

  • From: Nikolay Pokhilchenko <nikolay_po@xxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Wed, 28 Nov 2012 23:13:00 +0400

Wed 28 Nov 2012 Richard Hughes wrote:
>Splitting up the screen "calibration" into two phases, calibration and
characterisation is easy enough to explain. Characterisation can be
explained as getting the accurate device response, but how do I
explain calibration?This is not exact as You've wrote (IMHO). Characterisation 
is the process of obtaining of device behaviour. Knowing the device behaviour 
allows find the right way to display color correctly.
Getting the accurate device response (by mine) is the calibration process.

>As far as I can understand it, dispcal does some kind of iterative
algorithm to set the vcgt tables, with an optional target white point.
Could someone please explain in fairly technical language (assume the
user knows about Lab and XYZ) what dispcal is actually doing?
>Performing the calibration, dispcal searches the best correction 2D curves for 
>certain display, one curve per channel. The curves allows to obtain desired 
>(choosen) display response - certain gamma or standard brightness curve (sRGB, 
>L-star, etc.). After calibration, while applying calibration curves to the 
>display data, the brightness response with R=G=B become desired and neutral as 
>possible. There is two aims of calibration: correct brigtness response and 
>neutrality along the neutral scale R=G=B. Calibration have no deal with 
>saturation or color tone but helps to characterize the display by simple 
>matrix at the profiling (characterization) stage if the display acts as 
>additive device.

Other related posts: