Re: Colormatch vs Adobe 98
Re: Colormatch vs Adobe 98
- Subject: Re: Colormatch vs Adobe 98
- From: Don Hutcheson <email@hidden>
- Date: Thu, 31 Jan 2002 12:19:40 -0500
David Scharf wrote,
>
I noticed that the DonRGB was said to be created especially for working
>
with D50 and 2.2 gamma working conditions. What are the ramifications of
>
#1. working on a Mac with a 1.8 gamma? #2. working at a different
>
color temperature such as D65 or 6500K? Or are the monitor settings
>
irrelevant?
In responding to this I found myself getting into some depth, so for those
of you in a hurry here's the executive summary ...
SHORT ANSWER:
In theory and mostly in practice your monitor gamma and white point are
irrelevant or at least do not have to match the RGB working space. That's
the whole point of device-independent ICC profiles.
(END OF SHORT ANSWER)
*************************************
Now, for those of you with time to kill, here's some extra detail, some food
for thought and a bit of a soap-box harangue about our current viewing
standards ...
REALLY LONG HUTCH-TYPE ANSWER:
MONITOR:
I usually recommend calibrating your Mac monitor to 1.8 because that's what
you're used to. Changing it to 2.2 just makes your other apps look darker
and accomplishes little if any improvement in soft proofing.
WORKING SPACE:
You absolutely SHOULD use a gamma 2.2 working space rather than 1.8, as 2.2
distributes tonal values more evenly throughout the visible lightness range
and preserves more shadow detail for later exhumation. In fact higher gammas
like 2.4, 2.6 or even 2.8 are theoretically closer to the ideal L* curve and
distribute the 256 levels even better from whites to blacks, but may
increase the risk of banding in lighter tones.
IMPLICATIONS OF GAMMA MISAMATCH:
In theory any mismatch between monitor gamma and working space gamma will
contribute to quantization errors, visible as 'banding' or 'contouring' on
your display. In practice some level of banding is inescapable anyway with
Apple's cheap use of 8-bit video LUTs and DACs (long overdue for upgrading
to 16 bits per channel), even if the monitor and working space gammas DO
match, and I find making them equal is of little practical value.
The good news is that the banding you see on screen due to v-LUT/DAC
quantization is ONLY on screen and not in the final output.
WHITE POINT:
Here's why you should avoid D65 like the plague:
1. Three-and four-band colorimeters are normally 'hardwired' to D50 and
CANNOT measure D65 correctly.
2. Any D65 color space will give you a blue cast if you convert from it
with Absolute intent, or a yellow-red cast if you convert to it with
Absolute intent.
3. All graphic arts viewing booths and viewing standards specify D50 for
color comparisons.
4. Virtually every printer, separator, agency, photolab, photographer or
publisher has access to some form of D50 viewing condition and uses that to
make color judgements. Changing to D65 puts you out of sync with your
clients, vendors and industry standards.
5. The ICC PCS (Profile Connection Space) defaults to D50.
6. All CIE spaces by default assume D50, in fact 'CIELab' should really be
written 'CIELab(D50)'. CIELab(D65) is computed quite differently and there
is no reliable translation between the two without going back to the
original multi-spectra data, which is thrown away in most profiles.
7. Converting D65 samples or profiles to D50 uses a processes (typically a
Bradford transform) which only works properly if all light sources used for
capture and viewing have a full D50 spectrum with 100 CRI (Color rendering
Index.) This is virtually impossible. The transform cannot solve metamerism
errors or work properly for low-CRI sources. For that we would need profiles
containing multi-spectral samples, not just the three L, a, b, variables.
WHAT'S BEHIND THE D50-D65 DEBATE:
The whole debate about monitor white points arises out of two main issues:
1. When people first started instrument-profiling their monitors, they
found that asking for D50 gave a screen that was too warm or reddish when
compared to a standard D50 viewing booth. This seems to be due either to a
failure in hardware measuring technology or a breakdown in the standard
observer model when viewing phosphor-based CRT displays. The error seems to
be much less with LCDs, which is one reason I'll never buy another CRT.
2. Even if you get a good visual match between screen and booth, D50 still
looks too warm for most people's taste. You notice this most acutely on CRTs
because they start out at about 9600K (very bluish), so reducing them to
5000K produces a massive color change which looks 'redish' for a few seconds
until our eyes adjust to the change. Unfortunately it also reduces
brightness because we're dimming the blue gun so much. LCDs are generally
about D50 - D65 out of the box so any change is less noticeable.
There is a widely-held misunderstanding that selecting D65 in your
calibration/profiling software 'solves' one or both of these problems.
Sometimes it will happen to come a little closer to matching your viewing
booth but usually it ends up being much cooler than D50, so you're no better
off when it comes to comparing the screen to a proof or original.
In practice the only effective way of matching the screen color to the booth
is to FIRST tweak the monitor guns by eye until a white screen matches the
white of the viewing booth, and THEN calibrate with software that lets you
select the current hardware-calibrated state of the monitor as a white
reference. This is why I continue to freely endorse what is by far the best
monitor profiling package in the world - ProfileCity's ICC Display. It's the
only software with which I can regularly and easily create a stunningly
close match between ANY display (CRT or LCD) and the viewing booth.
Incidentally, ICC Display may report the actual 'color temperature' to be
anything from about 4500K to about 6500K, but (a) I no longer care what it
is, so long as the screen matches the booth, and (b) you can't express a
tricolor source like a monitor in Degrees Kelvin anyway. The Kelvin scale
only applies to light sources like tungsten bulbs, that act as Plankian
'black body' radiators.
THE VALIDIDTY OF D50 AS 'STANDARD DAYLIGHT' (SOAP BOX HARANGUE):
Issue number 2 (above) throws into question the choice of 5000K as a
'standard daylight' white reference (back in 1963 on a rooftop in Rochester
NY). Most color scientists today would agree that 5000K is 'warmer' than
what the average human calls 'normal' daylight and would welcome a change in
the standard towards a 'cooler' temperature, perhaps 6500K. I too would like
to see us move in that direction but it's not one of my hot buttons.
Although I have a slight preference for D65, the point is that D50 is
perfectly acceptable and does the job. It's the only standard we have today,
it's been widely accepted for over thirty years, and it's available in some
form or another at virtually every graphic arts establishment!
In an industry bereft of useful standards, D50 is light at the end of the
punnel. By all means I support a future change in the standard towards
something cooler, but in the meantime for heaven's sake let's use what we've
got!
(END OF REALLY LONG HUTCH-TYPE ANSWER)
*************************************
I hope this is of some help.
Don
*************************************
Don Hutcheson
Hutcheson Consulting
(Color Management Solutions)
Phone: (908) 689 7403
Mobile: (908) 500 0341
email@hidden
*************************************