[scribus] How to use color management?

[Hal V. Engel]

On Friday 29 August 2008 05:13:22 pm dwain wrote:
> On 8/29/08, Hal V. Engel <hvengel at astound.net> wrote:
> >  > hi phillip,
> >  > all of the color profiles must be the same on all of yoursoftware and
> >  > hardware,
> >
> > Nonsense your profiles MUST be hardware specific for color management to
> > work. Each piece of hardware needs to be characterized (have a hardware
> > specific profile) so that the CMS knows how to handle it's colors.  If
> > you use the same profile for everything then you have in affect turned
> > color management off.
>
> why then are there so many color profiles to chose from?  

Because there is lot of different hardware out there and each one has 
different charaterisics and needs a different profile.  In order for this to 
really work you have to have profiles that correctly characterize the hardware 
you are using.  If this was as simple as you claim then there would not be "so 
many color profiles to chose from" as there would be no need for them.  After 
all then we would only need sRGB for RGB things and swop (or something 
similar) for CMYK things.  I think you just made my point.  Vendors are 
supplying lots of profiles because they understand that every piece of 
hardware needs a profile that is specific to that hardware.  Unforunately 
these vednor supplied profiles are model specific and there is no way for 
these to account for things like manufacturing variations, age and other 
factors that affect individual devices.

> your
> hardware can read a specific profile.  

There is no hardware that can read an ICC profile nor is it common for 
hardware drivers to understand or use ICC profiles and in no cases does the 
hardware directly use ICC profiles.  This is why this is normally handled by 
the user level software like Scribus, GIMP or Photoshop.  Please stop 
spreading misinformation.
 
> if you set the color profile of
> your monitor to sRGB_IEC61966-2-1_noBPC.icc then your monitor responds
> to that profile.  if you set it to swop v2 then it reads that and
> responds accordingly.  in short the profile makes the hardware respond
> to the profile, otherwise, why use the profile.

Again more misinformation.  The profile describes or characterizes the 
hardwares color and tonal response.  It does NOT change the way the hardware 
works it just describes how it works.   This is THE critical concept about 
profiles that users need to understand.  Once they understand that it only 
describes the devices color and tonal characterisics then they are well on 
thier way to usnderstanding how all of this works.

It is a common misconception that profiles are for calibration (IE. that they 
change the device response).   Part of the reason that many users have this 
misconception this is that some monitor profiles also have calibration 
information in the form of Video Card Gamma Tag (VCGT) data that does affect 
the device response if the user runs a LUT loader.  But this is NOT part of 
the ICC specification and this information is not used by the CMS.

In addition,  your claim that "If you set it to swop v2 then it reads that and 
responds accordingly" is clearly false since swop is a CMYK profile and your 
monitor is an RGB device.  There is no way that you can get your monitor to 
behave as a CMYK device.  Again the profile does not change how the device 
responds nor can it alter the devices characterisics.

The reason you use profiles is to make it possible for the CMS to convert 
color data between devices so that the results come close to matching.  If you 
use the same profile for all of your devices nothing happens in the CMS.

>
> >  > except for your camera, but you can assign the profile if
> >  > you have photoshop.
> >  >
> >  > i too do not have any calibrating equipment, but i have found that if
> >  > software and hardware have the same color profiles you are in the ball
> >  > park.
> >
> > In the ball park of what?
>
> the ball park of managed color.

How would you know if you don't have your hardware properly calibrated and 
characterized?  Because things match your uncalibrated unprofiled monitor?

>
> >  > i have a viewsonic vp171b that i have an srgb profile with no
> >  > black point compensation set as the profile.  the rest of my hardware
> >  > carry the same profile.
> >
> > Again in this case you have disabled color management.
>
> how can i have disabled color management when the hardware has a
> profile and is responding?

No your hardware is not responding since that is not what profiles do (see 
above).  All you have done is force everything to the same color space which 
causes the CMS to do nothing.  After all when you do a color transform from 
sRGB to sRGB what happens?  (hint - nothing) The whole point is to have the 
CMS convert the color information (color space) from your input devices 
(camera, scanner...) to your monitor color space so that it is color matched 
or from the device color spaces to the printers color space when printing.   

If you are using generic color spaces (like sRGB) and your devices happen to 
have color and tonal responses that are close to what is described by those 
generic profiles then it will sort of work and may in some cases work 
surprisingly well.  But the fact that it sort of works is like saying my 
Ferrari sort of runs on regular fuel.  Yes it works but perhaps it would work 
a lot better on higher octane fuel or in the case of color management with 
profiles that were actually for the specific devices you are using.  

In this case it sort of works because many devices were designed to come close 
to sRGB because this is the assumed color space for all WIndows imaging 
devices and the web.  But most of these devices are at best sort of sRGB and 
many can be fairly far from sRGB due to factors like manufacturing variation 
and age.  In addition, in the case of input devices like scanners the sRGB 
like results are because of the device driver.  The native hardware response 
is normally far from sRGB so users on *nix systems who are not using the 
vendors driver software will likely not have results that are sRGB like.   

For newer devices as manufacturers are introducing new technology there is a 
general shift away from sRGB as a target device response.   This is the case 
because many of these newer devices have significantly more gamut and dynamic 
range than sRGB and constraining the device to sRGB will have a significant 
negative impact on the devices capabilities. This has been true for digital 
cameras and scanners for some time now and it has been happening at an 
accellerated rate for monitors the last couple of years.    We have also seen 
printers getting wider gamuts and wider dynamic ranges and high end photo 
printers are now capable to much wider gamuts and dynamic ranges than typical 
generic CMYK profiles such as swop.  This is an ongoing trend and I expect it 
will continue for some time.   Again why would anyone constrain any of the 
wider gaumt devices by using generic low gamut color spaces like sRGB or swop?

On the other hand if you have an RGB image from an uknown source that has no 
embedded profile the best GUESS for it's color space is sRGB.   But at best 
this is a guess.

>
> >  > i use the user settings with the rgb respectively set at 50% for a
> >  > middle gray.  no problems so far.  you could go the calibrator route,
> >  > but you have had some good suggestions so far.  as a monitor ages it
> >  > needs to be calibrated more often.
> >
> > No not more often with age.  As monitors age (at least for CRTs)
> > calibration becomes more important since the phosphors wear out and
> > change with use.  For CRTs calibration should happen at least once a
> > month and once a week is perhaps a little better.  For LCD monitors
> > things are more stable which means you can use longer intervals between
> > calibrations.
>
> not more often with age you say and then you say that as a crt
> monitors age phosphors wear out, which usually happens with age, you
> need to calibrate more often.  you are trying to have it both ways.

You are confusing total drift with drift rate.   Over time total drift will 
keep increasing but the drift rate remains fairly constant.   It is the drift 
rate that determines how often you need to calibrate and profile not the total 
drift.   If the drift rate did change with age then more often (or perhaps 
less often if the drift rate slowed down) would be the correct answer but the 
drift rate does not change with age in a significant way.  

>
> >  > this profile is recommended by the international color consortium for
> >  > on screen viewing.
> >
> > They do?  Where is this recommendation located?
>
> look at the line below for your answer.
>
> >  > here's the starting point to get the profile:
> >  > http://www.color.org/resourcemain2.xalter

I did look there and I didn't see any specific recommendation concerning what 
hardware to use this profile with and I would have found it surprising if the 
ICC had made such a recommendation.  

For many monitors, particularly CRTs,  sRGB may be better than not using any 
profile at all.  But don't confuse better with best or correct.   Many newer 
monitors have responses that are totally different from sRGB with some newer 
wide gammut LED backlighted LCD monitors having almost 50% more gamut than 
sRGB as well as significantly different primaries.  For these monitors using 
the sRGB profile would be completely wrong.  In addition, OLED monitors are 
starting to appear on the market (these are currently expensive and only a few 
models are available) and these have wider gamuts and wider dynamic ranges 
then any of the prior monitor technologies including wide gamut LCD monitors 
so this trend away from sRGB like responses will only increase over time.  

As a side note OLED monitors have a combination of the best characterisics of 
CRTs (very wide viewing angle, true additive color and very good blacks) and 
LCDs (slim profile, very bright white level) and some characteristics that are 
better than any prior technologies (wider gamut, wider dynamic range, much 
lower power consumption) and these are likely going to become the dominate 
monitor technology at some point in the not too distant future (IE. as 
production costs come down). 

> >  >
> >  > i use a commercial printer who prints digitally from pdf files.  he
> >  > told me that he can get the color right whether i send a rgb or cmyk
> >  > or mixed (gasp) color file.  i have given him all three and what i see
> >  > on my monitor is what comes off the press (printer).
> >
> > This is called luck.  Most of his devices are close to sRGB and in the
> > process of creating pdf files for the printer most of the conversions
> > done by the CMS are close to correct.  Again this is because he is lucky
> > rather than becaue his setup is correct.
>
> and you know this how?  i've been studying color management for going
> on 10 years.  i've tried all of the profiles to get what i see on
> screen on paper and i finally got it right.  yet, you say i have no
> color management because i have the same color profile in my software
> and hardware.  i do believe you are wrong.  check out
> http://www.color.org/  this is where i get my information.  where do
> you get yours?

I am the maintainer of an open source ICC profiler named LProf and I have a 
good understanding of how color management works and what are considered 
accepted practices by the color management community.  I actually write code 
used to produce ICC profiles and have studied the ICC specification documents 
in detail.   I am not claiming to be the worlds top expert in this area and 
there may be others on this list with more expertise in color management than 
I have but most members of this list have less expertise in this area than I 
do.  My appologies to the list I am not trying to put on airs here.  There are 
lots of folks out there, some of them members of this list, who I know have, 
in some cases way, more expertise in this area than I do.

I almost didn't respond to the original post by Dwain because I didn't want to 
get involved in a pissing contest.  On the OpenICC list we see the kinds of 
claims made by Dwain occationally.  On the OpenICC list, where almost everyone 
has a high level of expertise in color management, these kinds of posts are 
ignored because everyone knows the claims being made are BS (sorry for being 
blunt) and they also know that no one else there will take it seriously so 
there is no need to respond.  But on a list like this where the vast majority 
are not CM experts and many are totally new to the subject area some might 
actually believe the missinformation.  So the reason I am responding is to 
prevent the spread of misinformation.  Please everyone make sure that your 
sources of information about color management are authoritative.  There is 
simply too much misinformation out there and in an area that takes some effort 
to understand, like color management, it is very tempting to use simplistic 
approaches such as that offered by Dwain.

>
> when i try a new commercial printer i always ask what color profile
> for cmyk they use and if i don't have it then i ask for a copy.  also,
> i used another commercial printer for a magnetic sign. 

This is the correct thing to do and this is the only part of Dwain's color 
management work flow that is actually correct (or at least it would be correct 
if the profiles being used for the rest of the work flow were correct).  

The cost and effort needed to actually do the rest correctly is fairly 
insignificant now days so I am having a hard time understanding why Dwain is 
insisting on on using a workflow that is so far from accepted practice.  Of 
course if it works for him fine but everyone else should be aware that what he 
is advising is not accepted practice with in the color management community.  

In the past it was very expensive (IE. $300 or more for monitor measurement 
hardware, $1500 or more for a reflective measurement device and you also 
needed to buy expensive software in addition to the hardware) and because of 
this it was only those who were working professionaly in some field that 
involved color critical work that could afford the hardware and software to do 
this the right way.  Under those conditions I could understand why many users 
tried to figure out less costly ways to "come close" to the desired result 
(been there done that).  But today for less than $100 you can get all of the 
hardware (a measurement device like a Huey and an IT8 target) and software you 
need to calibrate and profile all of your input devices and your monitors.  

In addition, for users that need to have profiles for printers to do local 
printing you can find profiling services that will create custom printer 
profiles for very reasonable rates ($20 to $30).  For many ink jet printers 
you only need one profile for each paper/driver setup for the life of the 
printer.  For Windows and Mac users paper/printer/ink specific profiles are 
available from most printer, paper and ink vendors now days for free.  These 
profiles are not as good as a true custom profile but are generally very good 
and if your printer happens to be a very close match to the one used to create 
the profile the results can approach a custom profile.  For *nix users in most 
cases your only options are to have custom profiles created by a profiling 
service or get the hardware to do this yourself (this is still somewhat 
expensive however starting at around $450 for new hardware).  I do know of one 
ink/paper vendor (MediaStreet) that will provide profiles to anyone using 
their inks or papers and if you are the first user with a particular 
printer/paper/ink/driver combination, which is likely the case for most *nix 
users, they will ask you to print the profiling target and mail it to them and 
you will end up with a free custom printer profile (please do not abuse this 
service).  

Anyone who is serious about doing color management should start out by 
calibrating and profiling their monitor using a measurement device.  This is 
your basic first step and as I have said this is now farily inexpensive and 
not hard to do (there is a learning curve however so the first time will take 
some effort).   In addition if you have more than one computer or work in a 
shop with more than one user/computer doing color critical work the hardware 
and software can be shared which makes the cost very low on a per 
user/computer basis.  

The next step is to profile any input devices you use.  Again this is not 
expensive and depending on your input device could cost as little as $20 for a 
profiling target (Wolf Faust sells reflective 5" x 7" IT8 targets for $15 plus 
shipping that will do the job for most scanners).  If you are using a 
commerical printing service that understands color management then that is all 
you need to have a correctly characterized system.   At that point it is 
simply a matter of correctly configuring all of this into a working system.

Hal
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.scribus.net/pipermail/scribus/attachments/20080830/d847ffd6/attachment.htm>