Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!gatech!swrinde!pipex!bbc!alanr
From: alanr@rd.bbc.co.uk (Alan Roberts)
Subject: Re: cie to RGB conversion
Message-ID: <D81p8u.E9u@bbc.co.uk>
Sender: usenet@bbc.co.uk
Nntp-Posting-Host: sung0.rd.bbc.co.uk
Organization: British Broadcasting Corporation, UK
X-Newsreader: TIN [version 1.2 PL0]
References: <jspears-030595173331@f617-1d40-d5.hac.com>
Date: Thu, 4 May 1995 08:36:30 GMT
Lines: 43

Jim Spears (jspears@msmail2.hac.com) wrote:
: We are trying to convert color values specified in CIE xyL to RGB and keep
: running into problems understanding the process.  Would appreciate any tips
: or pointers as somebody must have stumbled across this before.  One serious
: question is the necessity for calibrating our CRT, do we really have to do
: this or is there a 'close enough' value for gamma for Sony Trinitron
: monitors.

If by CIE xyL you mean Yxy, then all you need is in FAQ documents
by Charles Poynton, Adrian Ford and myself, available via:

	http://www.wmin.ac.uk/~ajoec1
	ftp to wmin.ac.uk
	ftp to elaine.crcg.edu/pub/doc/colour
	http://www.inforamp.net/~poynton

If L means L* (as in CIELab CIELuv or CMC) then you have a non-linear
conversion to do first, but details of that are in the FAQs as well.

As to the gamma of CRTs, it all depends on the degree of accuracy you
want. The widely held belief that CRT gamma is 2.2 or 2.8 is wrong, the
theoretical value is 2.5 for cathode-voltage drive, but in practice it
is always a little lower. I have measured a fair number of display from
various manufacturers and always get values of between 2.25 and 2.4,
the most commonly found valkue is around 2.35. It's most unlikely that
your display differs from this. BUT, it depends on how precisely you set
the black level. The best way is to get or make a test signal consisting
of a black background, with areas at +2% and -2% (voltages), and to adjust
the black level such that the +2% area is distinguishable but the -2%
area is not. If you do this in a sensibly darkened room (incident light
on the screen lower than 0.5 cd/m-2, then you should get theoretical
performance from your monitor. If the light level is higher than that,
you can allow for it in the calculations, but the results get less
reliable.

Hope that helps.

--
************* Alan Roberts **************
* BBC Research & Development Department *
* My views, not necessarily Auntie's    *
*    but they might be, you never know. *
*****************************************
