[scribus] Colour Profiles

[Hal V. Engel]

On Sunday 06 September 2009 07:03:03 am Peter Nermander wrote:
> > So are we saying that for every digital camera that ships, the
> > manufacturer should provide a profile which we should select with
> > Scribus ?  If so, what file extension are we looking for ?  I'm not
> > aware of having one, but happy to look.  What if we use pictures from
> > multiple cameras ?  Would we not then need to apply a profile per picture
> > ?
>
> The problem is that the profile for a camera would depend on the
> ambient light when you take the photo.

This is true to a certain extent but it is both simpler and more complex.  
Digital cameras can have white points set for different types of light sources 
for in camera processing (this also applies to raw processing but it happens 
later in the process) and this corrects the images to about a 5000K white 
point.  This is sort of like using different types of slide film under different 
lighting conditions for those old enough to have worked with slide film.  For 
light sources with a smooth light spectrum you can use a single profile since 
the images coming out of the camera or the RAW processing have all been 
corrected to the same white point and will have nearly the same gamma 
characteristics.  But for light sources with spikes or drop outs in the 
spectra you will likely need to create specialized profiles since there will be 
an interaction between the spectral characteristics of the light source and 
the sensor.  In general sun light, sky light, strobe, flash bulbs and all 
incandescent lights have smooth spectra.  So this covers the vast majority of 
light sources you are likely to use.  On the other hand most florescent lights 
have large spikes and dropouts in the spectra and will likely require 
specialized profiles. 

>
> But you don't need a profile for the camera, since the camera
> manufacturer put the profile into the camera. 

But what profile is that?  How do you know what profile to tag the images with 
so that downstream processing (like printing or viewing on a monitor) knows 
what the characteristics of the device that produced the image are?  Remember 
that any time you need to view or print something that the color management 
engine needs two profiles to do this correctly.  A source and a destination 
profile.  So I disagree - you do need a camera profile.  I think your assertion 
is that it is possible to use some "standard" or generic profile. 

> The pictures from the
> camera will be in a "standard" colorspace like sRGB or for better
> cameras maybe AdobeRGB.

Perhaps but perhaps this is mostly wishful thinking.

>
> Of course the professionals only shoot RAW, and then apply corrections
> to get what they want, but they will still save the resulting image
> (usually a TIFF) using AdobeRGB. 

They do?  I know that this is not how I do my RAW processing.  I create custom 
profiles for my cameras for my RAW processing work flow.  I then use these 
custom profiles as part of that work flow to convert from the cameras native 
color space to my standard working color space.  This is both easy to do and 
very powerful.  The only thing that is needed is a profiling target of some 
sort and a little effort.

> You could say that when you use RAW
> you create the profile for each picture while you apply the
> corrections.

You do?  I know that this is not how I handle it.  When using a good custom 
profile I seldom make any adjustments to the photo during RAW processing other 
than to make exposure adjustments.

>
> > OK -  but again, the paper and ink issues alone make this implausible
> > for those of us that aren't professional printers.
>
> Not really. The profile is valid for each combinaton of paper and ink.

and driver and driver settings.  For this to really work you need to have a 
very controlled, consistent and repeatable work flow.  The main thing is to 
select what your work flow(s) will look like (IE. paper, ink, driver settings 
...) and then always do it exactly the same way.  Of course even if you are 
not using color management you should still have a consistent work flow. 

> As long as you buy the same ink and use the same paper the colors will
> not differ much. 

Again you need to have tight controls on your printing work flow.  Everything 
must be the same for every print job.  Consistency is the key.   This is 
something that the professional print shops do.  They have institutionalized 
highly consistent work flows so that they get consistent results.  There is no 
reason that end users can't do the same thing in their home or office "print 
shop".  They just need to think through what they are doing and then remove 
variations in their processes. 

> And there are companies that create a profile from a
> printed testsheet you send them. You download a file from their
> homepage, print it, send the print by mail to them and get a profile
> back. Not sure of the cost though, but a lot less than buying your own
> equipment to do the calibration.

This is basically correct.  These services used to be expensive (like $100 to 
$200 per profile) but in the last 5 or so years there has been a lot of new 
service providers entering the market.  In addition the cost of the equipment 
has come down significantly and these factors have driven down the cost.  Now 
days it is not hard to find services that will do RGB printer profiles for $25 
and CMYK profiles for $35.  This makes getting good printer profiles well with 
in reach for almost anyone.  It is also a good idea for Windows and OS/X users 
to check with their paper/ink vendor to see if free high quality profiles are 
available.  In many cases they are and if your paper/ink vendor does not 
supply these then perhaps you should consider switching media vendors.

And a clarification profiling is NOT a calibration process it is a 
characterization process.   That is the profile describes the devices color and 
tonal characteristics.  Calibration is an adjustment to do the the device or 
it's driver that changes it's characteristics.

>
> If you want top quality you should make a new profile each time you
> change ink and each time you open a new pack of paper. But you
> probably don't need to make a new profile that often if you stick to
> the same ink and paper.

Basically I agree but to a certain extent this depends on what equipment you 
are using and how picky you are.  For example Epson ink jet printers have a 
high level of consistency and changing inks is less of an issue than the 
differences between having new (fuller) ink cartridges vs. ones with lower 
levels of ink and ink levels are a very minor issue.  But HP ink jet printers 
since the print head is built into the cart will have more variation from cart 
to cart since you are also getting new print heads every time you change inks.  
But in most cases with ink jet printers as long as you use the same ink vendor 
and the same paper (vendor and type) and the same driver settings you should 
be fine with one profile for the life of the device.  

>
> > What I'd like is a means of colour management that may well sacrifice
> > some accuracy, but is quick to use, and can be used with any printer,
> > but gives acceptable results at a cost, and for an amount of effort,
> > that is appropriate to the size and requirements of small jobs that
> > don't need perfect colour, but would benefit from colour that isn't
> > totally uncorrected.
>
> You have an image, where the printout does not match what you see on
> screen. May I ask why you don't go the easy route and adjust your
> screen so it matches the printer output? Then you can adjust the image
> on you screen until it looks as you want it, and it will print
> correctly?
>
> Becuase your result might be because your monitor is not correctly
> adjusted. If it is a matter of dark/light pictures, an incorrectly set
> gamma is the most likely cause. Creating a simple profile with gamma
> adjustments is quite easy. Look at for example lprof.

LProf will allow you to create a profile that has the gamma correctly 
characterized and also has some aids for setting the black level.  But if your 
display has inflections in the gamma curves (uncommon for CTRs but fairly 
common on LCDs) you will still have issues (IE. LProf assumes that your gamma 
curves are smooth and this may not always be true). 

Back when this was first put into LProf (around 1999) the equipment needed to 
do measurement based calibration and profiling of monitors was fairly expensive 
at around $400 for the least costly devices not including software.  Now days 
you can get measurement devices like the X-Rite Huey for around $50 plus 
shipping including basic calibration/profiling software.  The Huey is a better 
device than the $400 device was a decade ago since it will hold it's 
calibration better.   The older device would require re-calibration about 
every 2 years.  In other words the hardware is both cheaper and better now 
days than it has ever been.  There is no reason that anyone who is concerned 
about color should not have one.

Hal