[ptx] Display gamma... and colour management

[Hal V. Engel]

On Tuesday 03 January 2006 12:24 pm, Jean-Luc Coulon (f5ibh) wrote:
> Le 03.01.2006 20:53:54, Hal V. Engel a écrit :
> > On Tuesday 03 January 2006 10:29 am, Andrew Mihal wrote:
> > > On Tue, 3 Jan 2006, douglas wilkins wrote:
> > > > Andrew has only implemented the "pass-through" of ICC profiles
> >
> > the profiles is critical.  But monitors are harder to profile and
> > printers
> > are far more difficult to profile.
>
> As there are no probes supported with linux for the moment, there is no
> real way to calibrate a monitor.

Yes this is a major issue.  I have been in contact with two different vendors 
to try to work out the details for getting the necessary technical 
information to write the code to add support for affordable currently 
available measurement devices to LPROF.  One of the vendors, who shall remain 
un-named, were total a$$es about the subject and were totally against giving 
that information to me or making high level binary libraries available for 
Linux users for anything other than x86 and at this point are no longer 
making that available.   

I am still talking with the second vendor but the results are uncertain at 
this point.  They do have linux binary libraries for x86 but I have not been 
able to get my hands on these and they are very uncertain about how to make 
the libraries available to Linux/Unix users.  But at least they are willing 
to talk to me about the subject.  If the talks drag on much longer I will 
start talking to the third vendor.  But the sad part is that there are only 
three possible vendors that I know of.

My impression so far is that the entire industry is clueless about how to deal 
with the Linux/Unix market segment.   They want this to be just like the 
Windows market where they provide a binary library for one architecture.  
When it is pointed out to them that for this market segment binary libraries 
need to be available from them for more than a dozen architectures (LPROF has 
been build on 15 architectures that I know of)  you can almost see their eyes 
glaze over when talking to them on the phone.   In addition,  they seem to 
have a hard time accepting that many of the applications in this segment are 
supported by "one man shops" so that makes things even more difficult.

In any case there is at least a small glimmer of hope that the problem will be 
solved sometime soon. 

>
> Two important things are to be taken in account :
> 1 - Does the colour management has to be done in a desktop-wide way or
> in an application-wide way?
> 2 - If you use a CRT, gamma makes (some) sense. If uour are using a
> LCD/TFT, there is no real meaning for "gamma" and it would probably
> tricky to "calibrate" a display without an appropriate tool
>
> If the display is not calibrated all what can be done is switching from
> a theoretical colour space to an other theoretical.
>
> The scarse[1] project was very interesting but this project is in a
> dead state and the monitor problem has no been solved in an other way
> than giving Trinitro (or what you want) theoretical icc profiles for a
> given colour temperature.
>

LPROF will create monitor profiles in two ways.  If you can somehow manage to 
get some actual measurements for your monitor you can feed it a measurement 
file and it will produce an accurate monitor profile.  It will also produce a 
rough profile using the monitor primaries and estimated color temperature 
plus a set of gamma numbers that are gotten by using the Norman Koren gamma 
charts which are close enough to get the gamma correct to about +- 0.1.  Not 
perfect but better than no profile.

In addition Argyll CMS has support for some older (not in production at this 
time) or very expensive (> $3000 US) instruments that the vendors had 
published interface specs for and can be used with those instruments to 
profile both monitors and printers.

> Regards
>
> Jean-Luc
>
> [1]http://www.scarse.org/