Strange results with display profile created with Eye One Display

I've just received my Eye One Display, and have calibrated my display for
D65, gamma 2.2.

If I open a greyscale wedge in Photoshop 7.01, and assign it the sRGB
profile, it looks fine.

If I now assign this profile a D50 profile (ColorMatch), again, it looks
fine.

If I now do a soft proof, selecting my display profile as the proof profile,
with an absolute intent,
it does *not* look ok - the colour barely shifts at all. It *should* turn
slightly yellow. (this is what
happens when I use another D65 display profile, such as sRGB)

Anyone else notice this kind of behaviour?

I'm using Eye One Match V2.03A.

I've double checked that there's no other display lookup table loader
running, or residing in my Startup folder.

Greg.
p.s Will contact Eye One support as well.

 

This *might* be a bug in the Eye One Match software. According to an ICC
profile inspector utility,
the profiles always seem to have a white point which appears to be close to
D50, regardless of the
white point I selected in Eye One Match.

As further evidence, if I assign the greyscale wedge sRGB, and do an
absolute proof to my D65 display,
it turns blue. All this is starting to add up - the D50 wedge didn't change
because the white point
matches the display white point - i.e - both a relative conversion and an
absolute conversion will appear
to be identical.

I am using the Advanced mode of Eye One Match.

This isn't a good first impression of Eye One Display, assuming that I'm not
doing something wrong.

Greg.

 

More evidence:
I've now tried the beta Little CMS profiler, which actually supports the Eye
One Display puck.
This profiler is extremely basic at the moment - it merely profiles the
display in it's current settings.
I.e, it does not force the user to adjust brightness, contrast, and try to
hit a whitepoint target - it
measures how the display is currently set, and just creates a profile. It
reports a whitepoint which is
very close to the whitepoint I entered in to Eye One Match. This means that
Eye One Match *is*
allowing me to *calibrate* my monitor correctly - the only problem is in the
creation of the ICC file
afterwards. ;^)

The problem is present in both the version of software I received (V2.0.1),
and the current version
available for download (V2.0.3A).

I have *not* yet tried the "easy" mode of Eye One Match.

Greg.

 

This is what seems to be the case, after speaking with others, and having
done some more testing:
1. The Eye One Display profiles comply with the V4 ICC specification, which
mandates that the whitepoint be exactly D50 for
displays.
2. The profiles are entirely usable for normal display
3. It seems highly likely that the profiles are working fine for Photoshop's
soft proofing as well. (the appearance matches that of
a traditional monitor profile, overall.)
4. At the moment, the profiles do *not* seem to be usable in Photoshop for
absolute intent conversions, if the monitor is
calibrated/profiled for a whitepoint other than D50. As an example, if
one converts from ColorMatch to the non-D50 monitor
profile, using absolute intent, incorrect pixel values will result.

Point 3 (soft proofing) was the one which I was most concerned about, and
I'm really glad that this does seem to work ok.
I'm not too happy about 4, but it's not nearly as important to me, and I
suspect, most of us.

Greg.

 

If you've found a practical use for absolute colormetric rendering, my hat's
off to you; with no white point reference, it's been useless to me.

I hope you're not going through all of this just to get a decent
screen-printer match. I've built dozens of profiles using Eye-One,
calibrate my monitor with it, and have zero problems...so I guess I'm just
really puzzled by your efforts. What are you trying to achieve that has
practical application?

 

The http://www.color.org/ICC1-V41_ForPublicReview.pdf does not mandate
that at all.

You possibly have confused with the whitepoint of the Profile
Connection Space (PCS) that is completely another issue, it has to be
D50 and this requirement has been there from the start.

Color-space conversions are made through the PCS and the requirement
of the D50 PCS whitepoint does not make any requirements for the
device whitepoint (mediaWhitePointTag in case the monitor) that is
just what it measurably is.

Timo Autiokari

 

I quote, from that document:
6.4.25 mediaWhitePointTag

Tag Type: XYZType

Tag Signature: 'wtpt' (77747074h)

This tag, which is used for generating ICC-absolute colorimetric intent,
specifies the XYZ tristimulus values

of the media white point. If the media is measured under an illumination
source which has a chromaticity

other than D50, the measured values must be adjusted to D50 using the
chromaticAdaptationTag matrix

before recording in the tag. For reflecting and transmitting media, the tag
values are specified relative to

the perfect diffuser (which is normalized to a Y value of 1,0) for
illuminant D50. For displays, the values

specified must be those of D50 (i.e. 0,9642, 1,0 0,8249) normalized such
that Y = 1,0.

See Annex A for a more complete description of the use of the media white
point.

Greg.

 

Flycaster,
As you probably now understand, I was mainly concerned about soft proofing.
I thought that a soft proof involved an absolute intent conversion as the
final step,
from the output profile (printer profile, being proofed), to the monitor
profile.
As far as I can tell, Photoshop doesn't do this. (I don't yet understand
fully what it
does do, but I think I understand the *result* of what it does, now)

Other than that, though, I use absolute intent for testing purposes. For
example,
when I printed the two test patches for that monitor gamut experiment that
you and
I did together, I use absolute intent, because I wanted to print these
colours *exactly*, without
a whitepoint shift. (this wasn't an absolute intent conversion to the
monitor, of
course - it was from Lab space to printer space). I have also needed
absolute
intent for other tests, which have been from the document space to the
monitor
space. I needed it just recently in fact, and it served me very well. I
could not
use my Eye One Display monitor profile for this - I had to use a traditional
monitor profile

If the Eye One monitor profiles really are entirely to spec, then it seems
that
either a) Photoshop has a problem with them, or b) the V4 ICC spec has
removed
functionality. In the latter case, I'm not sure why it was felt necessary to
do this.

I stress though that I did probably exaggerate this issue. I'm a lot more
comfortable
than I was initially.

Greg..

 

Note that this subject is also being discussed at dpreview.com.
This message seems key:
http://forums.dpreview.com/forums/read.asp?forum=1003&message=7347365
but I still don't understand it. Perhaps others here will benefit more
from that post.
In that post there is a further link, to a forum on imagingrevue.com. I
don't have an
account on that site, so I can't access it.

Greg.

 

the quote is OK but you need to understand it a little more.

In case of the CRT, the media whitepoint is the CRT screen itself when
driven with 255,255,255.

What the quote instructs is that the actual measured hardware
whitepoint of the CRT has to specified (as D50) XYZ value in that
profile tag. It does NOT require that the actual hardware whitepoint
of the CRT should be set to D50. Or you could set the media whitepoint
tag to D50 and set the chromatic adaptation tag in such way that it
converts the media whitepoint tag that holds the D50 to the actual
hardware whitepoint, the result is the same and does NOT require that
the actual hardware whitepoint of the CRT should be set to D50.

What you say is the very same as to require that the users of what
ever printed matter should "set" the actual paper surface of the
prints to D50 chromaticity. Pretty hard thing to do.

Timo Autiokari http://www.aim-dtp.net

 

Again, understood/agreed, and I never said this.

The media whitepoint tag *must* be set to D50. It's not correct to say that
"you could".
It *must* be.

But you are putting words into my mouth.

All I'm saying is this:
1. Eye One Match created profiles do, always, have a media whitepoint tag of
D50 (very very close, anyway). This is regardless of the hardware whitepoint
I calibrate to, and regardless of the whitepoint I enter into the profiling
software. The software definitely caters for different hardware
whitepoints. The fact that Eye One Match always writes a media whitepoint
tag of D50 is to the V4 specification.

2. The profiles are usable for most things, including Photoshop soft
proofing.

3. The profiles are *not* usable for absolute colorimetric conversions from
the document space to the monitor space, *if* the hardware whitepoint is not
D50.

4. I simply don't know whether there is something wrong with the profiles,
or something wrong with Photoshop, or whether the V4 ICC specification
itself has a problem. All I know is that I do not know how to use these
profiles for absolute colorimetric conversions, in Photoshop 7.01 (or
Photoshop CS), on Windows XP. I think something is broken, somewhere.

Greg.

 

Just by the way, the profiles do not have a chromatic adaption tag.
('chad'). Could that be the problem?

Greg.

 

Unfortunately, vastly different soft proofing results are obtained in
Picture Window Pro, when using
Eye One Display profiles, vs traditional monitor profiles. I'm reasonably
certain Picture Window Pro
does in fact do an absolute colorimetric conversion from the proofing
profile to the monitor profile,
and that would explain the difference.

Greg.

 

You said:

Quote: "1. The Eye One Display profiles comply with the V4 ICC
specification, which mandates that the whitepoint be exactly D50 for
displays."

I understood the above like: ...mandates that the whitepoint of
display devices should be (set) exactly to D50.

Particularly the word exactly seem to refer to the hardware
whitepoint, in case of numbers in the profile we rarely say that some
value has to be entered exactly as D50, instead we just say the value
has to be D50..

So, we agree that ICC spec does not mandate to adjust the monitor to
any particular hardware whitepoint.

Timo Autiokari

 

The chromatic adaptation tag is rather new, older profilers do not
include that tag.

Color managed SW should perform correctly with older and new ICC
profiles. Especially Photoshop should since Adobe is a member of the
International Color Con Sortium who publish the ICC spec.

Timo Autiokari

 

Definitely.

Greg.

 

Soft proofing *to* the Eye One Display profiles may not be yielding proper
gamut warning results. (I do this to see
whether the colours in my image are out of gamut of my monitor, sometimes)

I did the following:
- Profile/calibrate monitor to D65 using Eye One Display.
- Profile the monitor in the same calibration state with the Little CMS
beta 3 monitor profiler, using the Eye One Display puck.
- Open Photoshop 7.01, and create a blue patch (0,0,255) and assign
ColorMatch RGB.

- View | Proof Setup | Custom, proofing profile the i1 profile, intent
relative, blackpoint compensation off
- View | Gamut warning (enable). Observe that image is out of gamut.
- Image | Adjust | Hue Saturation, adjust saturation until gamut warning
disappears. Adjust value required: -41%

Repeat these three steps, using the Little CMS profile: Hue adjustment
required: -11%

Repat these steps using sRGB: Hue adjustment required: -15%

I have trouble believing that the gamut warning for the i1 profile
represents reality.

Greg.

 

(yes, I did reset the image to 0,0,255) inbetween each proof - always the
same starting point before
doing the saturation reduction)

 

Understood. However, with respect to your first paragraph,

"As you probably now understand, I was mainly concerned about soft proofing.
I thought that a soft proof involved an absolute intent conversion as the
final step,
from the output profile (printer profile, being proofed), to the monitor
profile. As far as I can tell, Photoshop doesn't do this. [snip]"

It does, if you toggle the AbCol rendering intent. The soft-proof dialogue
will allow you to select whatever rendering intent you want, as will the
final Print Space selection.

 

It converts from the document space to the *proofing* profile using the
selected intent.
As far as I can see, it does *not* convert from the proofing profile back to
the monitor's space using
absolute intent, regardless of whether "simulate paper white" is enabled or
not. I thought it did.
I believe the Microsoft CMM does, and my experiments in Picture Window Pro
3.5 support
this.

Greg.