Message: 5
Date: Wed, 10 Sep 2008 20:37:59 +0100
From: Martin Orpen <email@hidden>
Subject: Re: Ugra/Fogra Media Wedge v3.0
To: email@hidden
Message-ID: <email@hidden>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
On 10 Sep 2008, at 19:52, Mike Strickler wrote:
Profilemaker does a perfectly fine job and does not "fall apart in
the shadows," so my guess is that there is a more fundamental
problem there, possibly in your ink limiting in the RIP or the way
you're setting up PM.
OK then "Mr Certified Installer" you tell me the absolute optimum
profile settings for a 9880 that needs to match ISO Coated v2 and I'll
build and test yet another profile against the vanilla Argyll profile
built from the same data and send you the results.
I've tried multiple profiles output from Profilemaker including MaxK,
GCR3 etc and 0, 5 & 10 black starts. All do a poor job in matching the
new v3 dark patches in the media wedge. And all the results fall apart
after 3 iterations from the neutral.3cc.
Base linearization was the same for *all* the profiles tested so I
can't see where the "fundamental* problems are with the RIP ink
limiting.
If you're getting better results using Profilemaker then feel free to
post your sticker -- I'm sure you can find the time, how long does it
take to to press command-shift-4 and press the space bar?
--
Martin Orpen
Idea Digital Imaging Ltd
------------------------------
Message: 6
Date: Wed, 10 Sep 2008 22:36:23 +0200
From: Klaus Karcher <email@hidden>
Subject: Re: Ugra/Fogra Media Wedge v3.0 [was: Colorsync-users Digest,
Vol 5, Issue 299]
To: Mike Strickler <email@hidden>
Cc: colorsync-users <email@hidden>
Message-ID: <email@hidden>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Mike Strickler wrote:
If you absolutely MUST have a shootout (does anyone really have
time for
this?), use a full ECI2002 or IT8.7/4 chart. The results can be quite
different numerically--one "problem" color on a control wedge can
skew
the result misleadingly. [...]
Even better would be to use separate profiling- and test-sets. When
you
use the same data set for profiling and verification, you can't get
much
information about the true quality of the profile: the grid points of
the profiling target should be "hit" by any decent profiler -- but how
about the huge space between them? This especially applies to
iteratively optimized profiles ... and there are apparently better and
worse implementations -- see
<http://lists.apple.com/archives/colorsync-users/2005/Nov/
msg00230.html>
Profilemaker does a perfectly fine job and does not "fall apart
in the shadows," so my guess is that there is a more fundamental
problem
there, possibly in your ink limiting in the RIP or the way you're
setting up PM.
I can confirm that Martin has *no* fundamental problems with his ink
limits or profiling parameters. I have 3 profiles, all of them made
with his data set, all with 400 TIL / 100% K. Two of them shorten the
available gamut in the shadows, one does not:
PrintOpen:
<http://digitalproof.info/colorsync-users/
Epson880_GMGSemiMatte_MaxK_vs_ic2.png>
ProfileMaker:
<http://digitalproof.info/colorsync-users/E9880PM_vs_iC2.png>
Argyll:
<http://digitalproof.info/colorsync-users/E9880_vs_IC2.png>
Klaus
------------------------------
Message: 7
Date: Thu, 11 Sep 2008 00:07:00 +0100
From: Bob Marchant <email@hidden>
Subject: Re: Media Testing for maclife.de
To: ColorSync Mailing <email@hidden>
Cc: Uli Zappe <email@hidden>
Message-ID: <email@hidden>
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
On 10 Sep 2008, at 18:17, Uli Zappe wrote:
b) 1 studio flash in 45 degree position -> uneven lighting (RGB
230-242 for different white patches in my case), but only one color
temperature
Turns out that b) delivers the slightly better profiles.
Hi Uli .
From my own experience one light at a distance works best . We have
the luxury of a large studio with miles of blackout material . this
enables even lighting of the target ( and a less than 45 degree
angle ).
Regards,
Bob Marchant.
------------------------------
Message: 8
Date: Thu, 11 Sep 2008 00:07:03 +0100
From: Bob Marchant <email@hidden>
Subject: Re: Media Testing for maclife.de
To: ColorSync Mailing <email@hidden>
Message-ID: <email@hidden>
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
On 10 Sep 2008, at 17:18, Fleisher, Ken wrote:
Developers of the commercial products go to great
lengths to achieve pleasing color, not accurate color.
Hi Ken.
Although there is always a certain amount of smoke and mirrors with
proprietary software , but some manufacturers do attempt to tip the
balance towards accuracy over pleasing colour . The Sinar system
that we use employs a system of camera profile + calibration , and
has proved to be a great system for advertising photography (BTW
there are no options for "looks "in the Sinar software). I believe
that Hasselblad and Lightphase in addition to looks have the option
of "accuracy". These systems ( CCDs as opposed to the mostly CMOS
chips used in the smaller DSLRs ) are used by photographers who do
prize accuracy over looks in the areas of colour critical advertising
photography , catalogue fashion photography , catalogue product
photography , beauty etc .
I appreciate that the Canon's , Nikons etc appeal more to a wider
market , and that for many , the look is more important than
accuracy , but I feel that it's too broad a brush to say that all
digital camera systems are geared towards prettiness at the expense
of everything else.
One additional aspect of camera calibration that has not been
discussed here
(Uli, perhaps you can tell us if you included this step for your
testing) is
flat-fielding. This is the process of correcting for spatial
non-uniformities in the lighting, lens falloff, etc. This is a
critical step
for camera calibration and can become very problematic if not done
(and done
correctly). The only stand-alone application that I know of for
this purpose
is Robin Myers' Equalight
Sinar have ( or should I say had as one has been removed from the
present software version ) systems in place . The first is a
diffuser placed over the lens at the working aperture that enables
the production of a white reference file which then takes into
account any difference in lens fall off due to camera movements /
vignetting.. The second is a two part calibration procedure where
the macbeth chart is captured followed by a white card capture
( this is the feature that has been removed , but hopefully will be
replaced on the next software release ).
In reality though , in order to get any substantial uneveness in the
first instance you would have to be shooting with either a wide angle
lens , loads of camera movement ( read tilt and shift lens on small
DSLR ) or some problem with the shutter.
Would it be possible to explain what you mean by spatial non-
uniformities in the lighting ?
Bob Marchant.
------------------------------
Message: 9
Date: Thu, 11 Sep 2008 00:07:05 +0100
From: Bob Marchant <email@hidden>
Subject: Re: Media Testing for maclife.de
To: ColorSync Mailing <email@hidden>
Message-ID: <email@hidden>
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
On 10 Sep 2008, at 20:25, Fleisher, Ken wrote:
Yes, we use Sinar here at the National Gallery of Art for capturing
artwork
and I agree that it does very well with the calibration. In that
use, I'd
say accuracy is the greater goal. Nevertheless, we are still not 100%
satisfied with the Sinar results because even when extracting the
raw data,
there is still some pre-processing that is baked into it. It's
pretty much
impossible to get just the raw, linear, cameral signals (after de-
mosaicing,
etc.) out of the file--
Yep...plus some of the problems ( more apparent in the past ) with
the obscure take on colur management and handling of camera profiles.
s, but Hasselblad also does a lot of preprocessing, such as over
sharpening, that is baked into the raw data. They do this so they
can claim
a greater MTF, but it's a false result. I don't know specifically
about
Lightphase in this regard.
Over sharpening and in the case of Lightphase overuse of noise
removal.
It's true though, that there is a "reproduction" mode in some of
these
systems, but the problem for us is that this repro mode usually
includes
their idea of pre-processing for the purpose of prepress. It's
still not a
scene-referred image capture, which is what we really need. That's
why I
claim that none of them are really aiming for accuracy.
Point taken.
Would it be possible to explain what you mean by spatial non-
uniformities in the lighting ?
Well, just that. Uneven lighting.
But that is a fundamental part of the target capture , and as such
prequisite for the use of the software in most cases.
. In the Sinar
software, this is taken care of when you do a "scene reference".
The camera
basically measures the non-uniformity by asking you to photograph a
white
card that covers your whole active image area, then makes
adjustments to
account for it so that the light is "even" throughout
Yes...I've got a couple of them <BG>
Regards,
Bob Marchant.
------------------------------
Message: 10
Date: Wed, 10 Sep 2008 19:09:42 -0400
From: Todd Shirley <email@hidden>
Subject: Re: Ugra/Fogra Media Wedge v3.0
To: colorsync-users List <email@hidden>
Message-ID: <email@hidden>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
On Sep 10, 2008, at 2:18 PM, Martin Orpen wrote:
My earlier problems with the darker patches looks to have been
caused by the paper profile.
Hi Martin
I only have experience using ORIS ColorTuner, so I am confused by your
term "paper profile". How do you use this profile in your RIP? In
ColorTuner, I linearize my printer, calibrate, then plug in the target
I am trying to color match to (Like ISOcoated_v2_eci) and run 3 or 4
iterations until I get numbers similar to yours. I don't have to
create a "paper profile" with a 3rd party software - the only ICC
profile I use is the target. Forgive my ignorance on this, I'm just
curious how the EFI rip works.
Also, what software are you using to generate your sticker? I have
ORIS certified proof, which creates a similar sticker but won't
separate out each solid color.
-Todd Shirley
------------------------------
Message: 11
Date: Wed, 10 Sep 2008 16:39:04 -0700 (PDT)
From: Eric Chan <email@hidden>
Subject: Re: Colorsync-users Digest, Vol 5, Issue 292
To: email@hidden
Message-ID: <email@hidden>
Content-Type: text/plain; charset=us-ascii
I have a small piece of software that computes
matrices on the fly. Each image file from my
Phase back has a different matrix associated
corresponding to a different illuminant.
How can I get ACR or Lightroom to accept
these matrices ? If they were profiles I could
just embed them in the file ...
This entry on the DNG Profile FAQ provides sample code:
http://labs.adobe.com/wiki/index.php/DNG_Profiles_FAQ#SampleCode
More specifically, here is how you could set your own matrix values:
#include "dng_camera_profile.h"
#include "dng_file_stream.h"
#include "dng_image_writer.h"
//
// Read a profile in from disk
//
dng_file_stream inStream ("MyProfile.dcp");
dng_camera_profile profile;
profile.ParseExtended (inStream);
//
// Stick your own matrix values in here ...
//
profile.SetColorMatrix1 (dng_matrix_3by3 (1.0, 0.0, 0.0,
0.0, 1.0, 0.0,
0.0, 0.0, 1.0));
profile.SetColorMatrix2 (dng_matrix_3by3 (1.0, 0.0, 0.0,
0.0, 1.0, 0.0,
0.0, 0.0, 1.0));
//
// Write the modified camera profile out to disk
//
dng_file_stream outStream ("MyModifiedProfile.dcp", true);
tiff_dng_extended_color_profile writer (profile);
writer.Put (outStream);
outStream.Flush ();
------------------------------
Message: 12
Date: Wed, 10 Sep 2008 16:44:14 -0700 (PDT)
From: Eric Chan <email@hidden>
Subject: Re: Media Testing for maclife.de
To: email@hidden
Message-ID: <email@hidden>
Content-Type: text/plain; charset=us-ascii
Measurement data are not from the CC24, though, but from the
Digital ColorChecker SG, which provides many more color patches
and therefore, more information
That is fine, but be careful. There is not as much "information" as
you might expect, because there are many dependencies among the
patches of the SG or DC charts, unlike the CC24, for which the
spectra were chosen very carefully.
(I could print my own gazillion-patch target with my Epson 3800 but
at most a few of the printed patches would be useful.)
Eric
------------------------------
Message: 13
Date: Wed, 10 Sep 2008 17:11:02 -0700
From: Chris Cox <email@hidden>
Subject: Re: Media Testing for maclife.de
To: ColorSync Mailing <email@hidden>
Message-ID: <C4EDB026.7632EĢemail@hidden>
Content-Type: text/plain; charset="US-ASCII"
How exactly do YOU go about getting controlled lighting?
Most people use a lightbooth of some sort or another.
BTW - "lightbooth" covers a wide range of products and quality.
Chris
On 9/10/08 12:58 AM, "edmund ronald" <email@hidden> wrote:
As Chris Cox so nicely pointed out, it's not the model, it's how
it is
implemented.
Thomas Knoll, who is a stellar programmer, publicly described how he
created the numbers for the original DNG matrices by measuring camera
under different illuminants *in a lightbooth*
ROTFL
I hope the color scientists on this list haven't exploded from
excessive laughter at this point. Please, if you dropped your teacup,
don't sue me.
Edmund
------------------------------
Message: 14
Date: Wed, 10 Sep 2008 17:18:27 -0700
From: Chris Cox <email@hidden>
Subject: Re: Media Testing for maclife.de
To: <email@hidden>
Message-ID: <C4EDB1E3.76330Ģemail@hidden>
Content-Type: text/plain; charset="US-ASCII"
Wait, who said that Adobe used lightbooths with fluorescent lamps?
Isn't that making a rather large assumption?
(especially since the higher quality X-Rite/Gretag-MacBeth booths use
filtered halogen lamps for daylight)
Chris
On 9/10/08 8:26 AM, "email@hidden"
<email@hidden> wrote:
Message: 8
Date: Wed, 10 Sep 2008 16:10:49 +0200
From: "edmund ronald" <email@hidden>
Subject: Re: Media Testing for maclife.de
To: "Bob Frost" <email@hidden>
Cc: email@hidden
Message-ID:
<email@hidden>
Content-Type: text/plain; charset=ISO-8859-1
On Wed, Sep 10, 2008 at 11:57 AM, Bob Frost
<email@hidden> wrote:
Edmund,
For those of us who are not color scientists, but photographers,
when you've
stopped ROTFLing would you explain the joke in a little more
detail please.
Thanks,
Here is a pointer to the spectrum of a fluorescent lamp
http://www.pl.euhou.net/docupload/files/Excersises/WorldAroundUs/
Spectroscope/
Spectra/Fluorescent_lighting_spectrum_labelled.gif
Taken from
http://www.euhou.net/index.php?
option=com_content&task=view&id=179&Itemid=13
As you can see, the spectrum of a fluorescent lamp is very spiky. The
color temperature may be 5000K or similar, but the way it is achieved
is really ugly. A camera caracterisation made with such an illuminant
is as much a caracterisation of the light as of the camera. Of
course,
the resulting "profile" will work very nicely in fluorescent-lit
offices, which is where programmers tend to sit. I think that before
accusing Ulli of being "unscientific", Adobe should take a hard look
at their own past methods and practices when establishing their
profiles. Adobe may have had a good model -the DNG 2-matrix
approach-,
but they combined it with bad experimental practice to match it with
reality.
Edmund
------------------------------
Message: 15
Date: Wed, 10 Sep 2008 17:26:40 -0700
From: Chris Cox <email@hidden>
Subject: Re: Lightbooth Profile Building
To: <email@hidden>
Message-ID: <C4EDB3D0.76332Ģemail@hidden>
Content-Type: text/plain; charset="US-ASCII"
No, Edmund unfortunately provided guesswork and inuendo.
Edmund made an incorrect assumption, and then tried to build on that
assumption.
Chris
On 9/10/08 11:20 AM, "email@hidden"
<email@hidden> wrote:
Message: 4
Date: Wed, 10 Sep 2008 09:32:08 -0700 (PDT)
From: Michael Papet <email@hidden>
Subject: Lightbooth Profile Building
To: email@hidden
Message-ID: <email@hidden>
Content-Type: text/plain; charset=us-ascii
Edmund,
Are you really sure you have enough detailed knowledge of Adobe's
"experimental practice" to make such a statement?
Mark,
Edmund provided a set of facts that illustrated his point. To
reiterate, the
process by which the profiles were built were problematic.
Replying "Are you
sure?" brings no logic and no new information to the group.
Following with a
qualifier "most likely" and then make up some magical environment to
invalidate is statement is very poor quality discussion. No one
learns
anything! Comments like this continue to degrade the quality of
discussion.
If you have an issue with his facts, please present them
professionally.
Michael
------------------------------
_______________________________________________
Colorsync-users mailing list
email@hidden
http://lists.apple.com/mailman/listinfo/colorsync-users
End of Colorsync-users Digest, Vol 5, Issue 300
***********************************************