[argyllcms] Re: More questions about gamma
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sun, 01 Mar 2009 12:02:15 +1100
Leonard Evens wrote:
Suppose I have an digital file containing a series of gray scale images of
squares each with a know value in the in the range 0..255. I can in
principle measure the values of the images on my monitor using my
Eye-One-Pro and argyll. I can then plot the results as a graph.
If I understand what is going on properly---which may be far from the
truth---the resulting graph should theoretically be that of a power
function y = x^g, where g would be gamma.
Only if displayed on a theoretical CRT based monitor. All real
world displays will depart from a power curve, and non-CRT based
displays only display power function behaviour because they
have been made to emulate CRTs. Things like LCD have a natural
behaviour that is closer to linear light.
But, in point of fact,the graph
will only be approximately that of a power function, and the approximation
may not be all that close. (Even in the best case, there may be
various corrections added to the graph to allow for a realistic result.)
I assume that the process of calibration/profiling---ignoring color
for the moment---is aimed at producing such a transfer function.
Yes, that is the assumption. A power like curve is the right shape
to appear perceptually linear, since our eyes tend to see contrast
in proportion to the ratio between color brightnesses. If you
start out with a certain step size ratio near black, and then
apply that up the scale, the steps increase in absolute size.
ratio = 1.1
ie. black = 1.0
Steps = 1.0, 1.10, 1.21, 1.33, 1.46, 1.61, 1.77, 1.95, 2.14, etc.
This is a power curve.
Second question: Is there some way to use argyll commands and the
data files it produces to produce the graph of the transfer function
being aimed at and/or the one actually produced, without going through
the process of making the measurements manually?
Sorry, there is no automated way of doing this. You can perform this by
creating a test chart that has just grey scale values (ie. targen -d3 -e0 -m0
-f0 -g40),
measure this with the display in the state you want (ie. uncalibrated,
calibrated),
and then drag the resulting .ti3 files into Excell/OOCalc, and plot the RGB
values
against the XYZ Y values.
[ When importing the .ti3 into OOCalc, you need to choose file type "Text CSV"
down in the spreadsheet section, and then tick the "Separated by space" box
in the next dialog. ]
Graeme Gill.
Other related posts:
