NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/comp/graphics/12124

Internet Message Format | 1992-11-19 | 5.3 KB

Path: sparky!uunet!mcsun!uknet!mucs!lilleyc
From: [e-mail] (Chris Lilley)
Newsgroups: comp.graphics
Subject: Re: Color Sample to RGB Conversion
Keywords: color conversion
Message-ID: <[e-mail]>
Date: 19 Nov 92 13:52:19 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]>
Sender: [e-mail]
Organization: Dept Computer Science, University of Manchester, U.K.
Lines: 124

In article <[e-mail]> [e-mail]
(Michael Sierchio) writes:

>In article <[e-mail]> [e-mail] (Chris Lilley)
writes:

>>In article <[e-mail]> [e-mail] (W. Joe
>>Hewitt) writes:

>>>     I am looking for a way to generate a color on a workstation
>>>that closely  matches a given sample from a color plate.

>>The simple answer is, mix it by eye.

>This is less than satisfying, and less than satisfactory.

As indeed I stated in my article. It will however give accurate
results, simply.

>>>     I have tried scanning the image (24 bit scanner) with little success.
>>>The scanning software only gives percentages of the R,G, & B components.
>>>Using a little math I converted each percentage to a value in the range
>>>0..255 (24 bit color system) for each RGB component.

>>>The problem is that the new color doesn't look very much like the
>>>original sample.

>>No, it won't. Red, green, and blue are very loose specifications for
>>colours. They differ between different monitors, and between different
>>scanners.

>This is false.

Sorry to trouble you, but my statement is absolutely correct and it is
you that is wrong. If you want figures, just let me know. Or you could
look up an introductory book on colour science.

>The RGB of NTSC video are quite standard.

Yes, they are. So?  Did Joe Hewitt say he had an NTSC video monitor perched
on his workstation? I find that unlikely. They are pretty rare things
- even TV studios don't use them nowadays, they use SMPTE monitors
instead.

And if you know of a scanner whose filters are closely matched to the
NTSC phosphor chromaticities and gamma, please post the details  ;-)

>A properly
>adjusted RGB monitor will give good results.  The key is the get the
>input from an RGB Camera, whose RGB signals are designed to match an
>RGB monitor.

This sad misconception is one reason why people have so many problems
with colour matching.

>>As you have realised, RGB is not the way to go. You basically have two
>>options.

>RGB is how the colors are represented on input and display devices.

This statement is correct, but the inferences you draw from it are not.
If you had bothered to read my post properly you would have realised
that each devices RGB is different. I suggest you go read it again.

>There exist color spaces that attempt to preserve the perceptual range --
>this isn't relevant to the discussion.

The original poster stated that he was having a problem with colours
not matching perceptually. I gave a brief introduction to the
standardised, tried and tested methods of achieving a match.
Therefore, those colour spaces are precisely relevant to the
discussion. It is your sadly naive view that all RGB devices are
exactly equivalent and all precisely matched to the NTSC standard that
is irrelevant. Go read up a little.

>>your scanner RGB --> CIE XYZ --> your monitor RGB

>Nope.  RGB -> XYZ -> RGB -> screen

>What's the point?

Let me spell it out for you.

The scanner red value will not be the same number as the monitor red
value. The scanner green value will not be the same number as the
monitor green value. The scanner blue value will not be the same
number as the monitor blue value.

However - and *this* is the point - the perceived colour on
screen will now match the colour of the plate, which is what the
original poster was asking about.

>There will never be any more information than in
>the original RGB image.

The number of bits of information has not changed, but the bit
patterns have. The extra information, if you want to look at it that
way, has come from the data on the chromaticities of the monitor
phosphors and the scanner filters.

Your general point about scanning in a colour image as RGB and
displaying it on an RGB monitor is an acceptable way to get colour
into the computer for low quality applications where the colour
matching is irrelevant. Putting pictures of military jets and half
dressed women on a workstation root window, for example ;-)

I do not blame you for not understanding this - it is a common
misconception. But understand that your suggested approach is
inadequate when the colour match is required to be of moderate or
high quality.

>>>     I am looking for a way to generate a color on a workstation
>>>that closely  matches a given sample from a color plate.

I think that counts as a requirement for high quality.

--
Chris Lilley
---------------------------------------------------------------------------
Technical Author, ITTI Computer Graphics and Visualisation Training Project
Computer Graphics Unit, Manchester Computing Centre, Manchester, UK
Internet:   [e-mail]        Janet:   [e-mail]
Voice:      [phone]          Fax:     [phone]
---------------------------------------------------------------------------