[argyllcms] Re: question about monitor calibration

  • From: Gerhard Fuernkranz <nospam456@xxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sun, 27 Jan 2013 15:48:05 +0100


If the *same* set of colors (i.e. same measured XYZ) is presented to an 
observer in two different viewing environments with different viewing 
conditions, then the *appearance* of these colors to the observer will be 
*different*, although the colors are the same. For instance, if a captured 
bright outdoor scene is reproduced colorimetrically correct on a monitor in a 
dim room, then the scene will appear washed-out on the display, and it won't 
look good.

The aim of a "viewing condition correction" is eventually the reproduction of 
the colors in the destination viewing environment (e.g. display in dim room) in a way so 
that the *appearance* of the originally captured scene is preserved. This implies that 
the XYZ colors of the reproduction actually need to be *altered* (and not reproduced 
exactly 1:1) in order to achieve this goal.

There exist rather complex color appearance models like CIECAM02 for this 
purpose, which attempt to model various psycho-physical color appearance 
phenomena mathematically. If simplicity is demanded, then a simple gamma 
adjustment can be used as coarse first-order approximation for such an 
appearance model.

If the aim is (non-color-managed) displaying of sRGB images on a not too bright 
monitor in a dim environment, then adjustment for viewing conditions is not 
necessary, but the monitor's tone curve should be calibrated directly to the 
sRGB tone curve, because sRGB images are *output-referred*, and are therefore 
supposed to be already adjusted for sRGB viewing conditions.

On the other hand, the Rec. 709 tone curve (-> HDTV television) is a 
*scene-referred* encoding, thus a viewing condition correction is required for 
viewing such a video on a display in a dim environment, in order that the video 
does not appear washed-out on the monitor.

Best Regards,
Gerhard



Am 26.01.2013 12:12, schrieb Alberto Ferrante:

Dear Graeme,
I apologize for the delay in replying to your post. Your reply was very
clear and useful. Still, it raises some further questions from me
(below, inline with your text) :-)


An ambient of 11 Lux is quite dark. That's why the gamma is increased.
The ambient light correction is based on assumptions about the
source viewing conditions ambient light level and brightness,
and the correction will increase gamma if the source is brighter
than the display, and decrease the gamma for the reverse.
The source assumptions are base on the gamma selection. A power
curve, Lab curve or sRGB assume sRGB viewing conditions as source (250 Lux).
Rec709 and SMPTE240M assume TV studio conditions (1000 Lux).

My understanding was that for sRGB a "dimly lit" room was assumed. I
normally use a room with a single diffused light source on my left. I
have just tried to perform a read of the ambient light (device on the
monitor with the sensor turned to face me and the white "cap" on) with
all the lights on and the curtains open and I got a read of 27.7 Lux. My
Sekonic light meter measures 4.9EV that is approximately 80Lux in the
same conditions.
I am a bit puzzled about this...


Because profiling records the response of the calibrated device,
then (of course!) when you link this with a source colorspace, the
CMM strives to reproduce the source color space accurately,
removing as much of the destination devices "personality" as
it can, which includes any effects of calibration. So any change
to the device calibration _should_ have no effect if color management
is working correctly. To change the appearance of an image you need to
alter the colorspace it is defined to be in (ie. the source profile
that it is tagged with), or introduce a deliberate color adjustment
in the CMM process (say, by using an abstract profile in between the
two device profiles, or introducing viewing condition appearance adjustments).

OK, clear.


Calibration doesn't directly affect profiles/color managed response (see above).
To make viewing condition adjustments in color managed workflows
you need to account for it in the CMM linking/gamut mapping.
See ArgyllCMS collink -c and -d parameters:
<http://www.argyllcms.com/doc/collink.html#c>

I will try that, but... So far the color profile without correction for
viewing condition adjustment has proved to give extremely good results.
Printing by using professional services has given consistent results
with what I see on the monitor also from the luminosity stand point.

Thanks again for your help!

Best,
    Alberto

Other related posts:

  1. Home
  2. argyllcms
  3. Archive
  4. 01-2013