Re: Monitor calibration software/hardware
Re: Monitor calibration software/hardware
- Subject: Re: Monitor calibration software/hardware
- From: Andrew Rodney <email@hidden>
- Date: Thu, 20 Sep 2007 20:01:49 -0600
- Thread-topic: Monitor calibration software/hardware
On 9/20/07 7:39 PM, "Derrick Brown" wrote:
> If a user has a monitor(simple LCD or CRT, no ddc) and they are
> trying to achieve a certain calibration parameter set , lets say a
> brightness of 120cd/m2 at a gamma of 2.2 with a white point of
> 5000kelvin, and they are using a Spyder2 Pro, DTP94, EyeOne Display2
> or Eye One Pro as their measurement device.
Yes, that's what I expect to get (what I asked for). Is there a reason you
can't provide that?
> Now lets say those values (only for conversation sake) say that your
> lowest black measured has a dE of 2.0, and a white of dE 1.5 all
> based on your original target parameters for calibration.
I'd expect your error checking would tell me this or at least give me what I
wanted from square one. And I don't necessarily buy that you've provided
this level of accuracy (or lack thereof) unless I have another device of
known qualities to tell me I got this or not.
I don't see how measuring with a rubber ruler twice makes the results any
more useful. How many iterations of validation should I conduct to satisfy
myself that you did what I asked you to do initially?
> When the calibration is complete and a validation is executed
> (profile validation, not instrument validation) you are left with a
> set of values that reflect performance characteristics of the
> calibration, like dE values of a set of LAB values put thru the given
> profile then measured with that same instrument.
It tells me more about the CMM, the round trip/math you hope is correct
(which may not be) than the Absolute accuracy of what I asked for. You can
also use whatever set of LAB values you want to check and in a way, stack
the deck in your report favor. Some patches will have very low deltaE
values, others very high. Where in the color space you ask to measure a
color plays a big role in the results, (a role I as the user may have no way
to control).
> Now lets say those values (only for conversation sake) say that your
> lowest black measured has a dE of 2.0, and a white of dE 1.5 all
> based on your original target parameters for calibration.
> So now you have one data point.
My first question would be, why in the first place didn't your product give
me what I asked for? Why is it now different, why do I have to ask you again
to give me what I wanted? If you give me what I wanted and, as some have
suggested here, they view a color cast, now what? The math is correct, the
user is wrong? Maybe. But that's not a useful solution to this problem.
> Like maybe if you run this given
> display at 5700 k at a slightly lower luminance you may well have
> more accurate highlight and shadow detail?
Again, all I have to do is look at the highlight and shadow detail. I don't
even know if you measured into these areas or if I can ask you to measure
more samples there. Its a solution in search of a problem.
The argument initially was that validation provides a report of the accuracy
of the calibration and profile. I submit that without an independent and
trusted auditor, this is silly. Then the argument was, well this tells the
user they did something wrong. I submit that I can see this, I don't need
someone measuring a few dozen pre-selected color patches to tell me that the
puck fell off during the process or the screen saver came on and toasted the
process. I can use my eyes.
What would be far more useful than telling me I asked for 140 cd/m2 and got
138 would be to tell me what I SHOULD calibrate to, based on the light box
and ambient light conditions and the dynamic range of the print.
I don't need to jump through these hoops to build a printer profile. I
simply make a print! Why is conducting what is really a pretty simple
process (profiling a display; you don't really have to calibrate it) so damn
complicated?
With the exception of color geeks and power users (vocal but limited in
numbers), the market by and large isn't asking for more Bells and Whistles
but easier color management processes. But I guess if you must market an
expensive software product that isn't bundled with the original hardware,
you have to target those markets. If the buttons make them feel better, well
then the buttons are feel good buttons.
Andrew Rodney
Author "Color Management for Photographers"
http://www.digitaldog.net/
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden