ICC corrected slide scans way too dark

[Markus Plail]

Hi there!

I profiled a Minolta Dimage Scan Multi Pro with the Provia/Astia/Sensia
target from Wolf Faust. Now when I look at the corrected (PS CS) scanned
image of the target it is way too dark compared to the slide on my light
table.
The profiles were done using LCMS profiler and GretagMacBeth
Profilemaker. There wasn't much of a difference between the two.

When I let Vuescan output a profile generated from the corresponding IT8
target description file, it looks like the color corrected image - also
too dark.

So is there supposed to be a gamma correction after applying the ICC
profile? If yes, by how much? 2.2 seems to be too much. I am using a
windows system.

Thanks for any hints
Markus

 

[nikita]

Markus,

gammacorrection or not, that depends on what you have to deal with
from the scan. If it's a RAWscan in linear 1.0 … yes, make a
gammacorrection. If just smaller offs the profiling will take care of
it, even though an optimizing for the greypatach GS11 to 100-110 RGB
can be a good thing. Compare the profiles with/without finetuning of
gamma.

As the Vuescan profiling (with Vuescans own use of Little CMS) also is
dark, I would suspect that you're using the wrong descriptionfile
loaded for the IT8 target allover. Or that it has been corrupted in
any way. It's a simple textfile, you know.

Go and download a new reference/descriptionfile and be careful to
chose the right one.

nikita

 

[nikita]

Addition;

The eventual gammacorrection would be *before" profiling the target
and *before* ASSIGNING the profile onto the real pictures later.

Any gamma adjustmet after the profile is assigned is in the creative
part of the editing.

nikita

 

[Wolf Faust]

As the Vuescan profiling (with Vuescans own use of Little CMS) also is
dark, I would suspect that you're using the wrong descriptionfile
loaded for the IT8 target allover. Or that it has been corrupted in
any way. It's a simple textfile, you know.

Well, I assume he did select the correct file and contribute following
guesses ;-) :

Often the too dark output is caused by incompatibilities of the
profiler and application used. For instance, unless Ed did change
things in VueScan, lcms and basicly most (all?) profiles not generated
by VueScan can not be directly used in VueScan. Try using the free
tifficc or jpegicc tools shipped with the lcms CMM (not the profiler)
to apply the icc profile to the scanned image file and see if this
brighten things up. These console programs are rather easy to use and
you can hardly make much wrong. They are usualy rather reliable when
used with lcms profiles and thus a good starting point. If they
work... you can switch to PS CS... one step at a time until you find
the fault.... if PS CS still fails, you most likely have to check your
PS settings.....

Another possible cause for the wrong brightness of the output might be
the used output (monitor?) profile. Maybe you use a generic profile
shipped with your graphic card driver or found on the internet? Here
difference in setup of driver/monitor can cause the problem.

 

[Markus Plail]

Well, I assume he did select the correct file

I surely did...

and contribute following guesses ;-) :

Often the too dark output is caused by incompatibilities of the
profiler and application used. For instance, unless Ed did change
things in VueScan, lcms and basicly most (all?) profiles not generated
by VueScan can not be directly used in VueScan.

Yes I know. I normally use the RAW scan files and do all adjustments in
Photoshop.

Try using the free tifficc or jpegicc tools shipped with the lcms CMM
(not the profiler) to apply the icc profile to the scanned image file
and see if this brighten things up.

Not really. And tifficc also had problems here with the colors on the
very left. Dark red, brownish colors were quite a bit off, whereas they
were correct when the profile was applied in PS. Apart from that the
brightness wasn't better.

Another possible cause for the wrong brightness of the output might be
the used output (monitor?) profile. Maybe you use a generic profile
shipped with your graphic card driver or found on the internet? Here
difference in setup of driver/monitor can cause the problem.

I used Adobe Gama to "profile" my monitor.

But as I wrote in my initial posting I think that the brightness is also
too low, when creating a tiff image from the IT8 description file when
comparing it to what I see on my light table (5000K).

regards
Markus

 

[Wolf Faust]

I used Adobe Gama to "profile" my monitor.

So, obviously the output and not the input is your problem.

Well, maybe the gamma setting used in Adobe Gamma is not working
correctly. The method used by Adobe Gamma to determine your monitor
gamma is not always working precissly. So, if your output is too
dark, increase/descrease your gamma setting in the Adobe tool until
the display shows known data correctly. From my experience with a
similar tool I did develop on the Amiga in the late 80's, the method
used by Adobe Gamma and similar tools work decently... but sometimes
require a bit stronger manual gamma readjustment by the user. After
all, the tool is just a sort of "better than nothing" approach for
users without measuring devices or self calibrating monitors and far
off from the dE 1 (max dE 5) most high end solutions offer.

Another issue might also be the graphics board software. I do not know
how Adobe Gamma works exactly with Windows, but I think there were
some problems with certain gfx boards and LUTs or conflicts with other
software... Maybe you should check if the Adobe Gamma tool is
compatible with your setup... Can't help on this issue.. better check
the web for those who know better about Adobe Gamma actuall
functioning...

 

[Markus Plail]

So, obviously the output and not the input is your problem.

I am not too sure.

I uploaded a corrected scan of your Provia/Astia/Sensia target to
http://www.gitteundmarkus.de/bilder/it8_2.jpg

It was scanned as RAW scan using vuescan, assigned the proper ICC
profile and the converted to sRGB (perceptive).

So you (and others of course can judge for themselves. IMHO it is too
dark, when comparing it to the target on my light table.

regards
Markus

PS: Why is it that the Provia target doesn't work for Astia 100F and
that there is one target for Astia and Velvia? Shouldn't the Astia be
closer to Provia than Velvia when it comes to saturation and contrast?

 

[Hecate]

I am not too sure.

I uploaded a corrected scan of your Provia/Astia/Sensia target to
http://www.gitteundmarkus.de/bilder/it8_2.jpg

It was scanned as RAW scan using vuescan, assigned the proper ICC
profile and the converted to sRGB (perceptive).

So you (and others of course can judge for themselves. IMHO it is too
dark, when comparing it to the target on my light table.

Gamma is only an approximate method. If you need real accuracy you
should use something like the Gretag Macbeth system.

 

[Markus Plail]

Gamma is only an approximate method. If you need real accuracy you
should use something like the Gretag Macbeth system.

Yes I know, but I do not aim for "perfect" color correctness. I am happy
that what I see on my monitor is approximetaly what I get back from
photo labs. So I take it that you find the scan correct?

regards
Markus

 

[nikita]

Markus,

I did compare your corrected IT8 in my Photoshop with my own IT8 slide
RAW scanned using Silverfast and profiled with Gretag ProfileMaker.
Assigned and converted to s-RGB as yours. Very similar. My darkend
goes much deeper with a lot more "punch". Patch 11 actually reads
sligthly lower than yours. Yours are sligtly lighter in other words.
But visually it's very close allover. Some of your red/"orange"
patches were oversaturated compared to mine. But that could be a
perceptual thing.

So, THAT IT8 you are showing us isn't to dark. You are probably using
a real high luminancevalue at that lighttable.....

To check a CALIBRATION vsv PROFILING go and download a gammatarget
for the value you've been calibrating your monitor to (2.2?) at the
link below. Drag it out on the desktop from the browser. Open in
Photoshop , use leave as is, no assigning. Then use the softproof :
View>proofsetup> MONITOR RGB.

Use Ctrl+ Y to activate/unactivate the proofview.

When MONITOR is activated as proof it will disconnect the
Monitorprofile from the Photoshop view. You're now looking at the
monitor CALIBRATION only as the RGB (workingspace set in Photoshop) is
sent directly to the monitor. Tha's when the file is untagged. When
assigning the scannerprofile it will use that as the RGB space sent to
the monitor. The normal conversion from Workingspace into the
monitorspace is taken out of the loop. It means that the ADJUSTMENT
the Adobe Gamma did to the monitors native behaviour is the only thing
you look at now– the adjusted view. Exacly like the non colormanaged
browser or another software in that field is using a
monitorCALIBRATION.

When UN-activating the proof and using Photoshop as normal without any
proofview, the conversion is going from the workingspace to what the
profile DESCRIBES. In otherwords the profiled monitor and the look of
it AFTER it was calibrated.

IF the look is darker of the gammatarget when the MONITOR RGB PROOF is
ACTIVATED – then the Adobe Gamma CALIBRATION hasn't taken it to 2.2.
BUT the profile is describing this and Photoshop knows how dark it is
and adjust the VIEW on the open file only (in Photoshop) for this. Old
Windows NT4 is braindead regarding CMS and and coulden't adjust the
screen (calibrate) but it was still possible to use it with Photoshop
thanks to the fact that the profile described it correctly in the
NATIVE state and compensated the view....it lightened up the open
image in Photoshop instead.


The Adobe Gamma is loading this CALIBRATION to the LUT using Adobe
gamma Loader in the startupfolder of a PC as you know. On the mac the
vcgt tag in the profile is loaded directly to the LUT
systemwise....switching a profile switch also the calibration.

If the loader doesn't work on the PC, that could be checked on a PC
[at least with w2k) by going to the monitorpanel. I'm not on a PC now
but if you push the advanced button or option something you'll get
into a tabbed dialog. When you tuch one of the tabs ( I don't remember
wich one) the lutload will realease and the calibration goes off. It's
now in it's native state. You can load it again by getting out of this
monitorpanel and then go to START button > Startupfolder or similar
and doubbleclick on the adobe gammaloader. That will load the
calibration again.

So,

using a gamma target you can check;

-Native value compared to calibrated.
-Calibrated with workingspace gamma 2.2 sent directly.
-The PROFILED view compared to the "sent directly view=calibration
only"

Bottomlines:

You will ALWAYS have a difference between NATIVE and THE OTHERS if
calibrating to 2.2.

You can live with a difference between CALIBRATED and PROFILED – as in
NT4 – while still having a correct view in Photoshop if the profiled
part of the profile describes it correctly.

Ideally the Calibrated and the Profiled will not show any diffrence
gammawise. Then the midtonelightness will be the same in
noncolormanaged and colormanaged aware applications. That's quite
nice.


If BOTH the calibrated AND the profiled is showing anything else than
the 2.2 when using the target – then that's where the problem is as
Wolf said, but if the profile DESCRIBES this non- lutloaded-view then
the view in Photoshop would be correct anyway and the
workingspace/documentspace will show it's gamma on the open picture.

Glad to see that you told Wolf that you did load the right
descriptiofile and it was nothing wrong with it

nikita

Scroll down to the "monitor tespatterns and grab one of them to load
in Photoshop and don't choke yourself with the rest of the stuff on
this page. Do the simple test first from my post:

http://www.normankoren.com/makingfineprints1A.html#gamadjust

 

[nikita]

To try to make my previous post a bit more clear .........:

Just looking at the gamma testimage in the webbrowser shows if the
CALIBRATION naild the target 2.2 or not.

Dragging it into Photoshop gives the opportunity to check the
Calibration and the Profiling using the *same* application and
minimizes the risk for diffs between the browser and Photoshop.

Going to the monitorpanel and release the calibration will put the
monitor in it's native gamma and can then be compared to the others.

These things together will show you whether the visual "darkness" you
feel has anything to do with a screwed monitorview – and where the
problem is in such case.



nikita

 

[Hecate]

Yes I know, but I do not aim for "perfect" color correctness. I am happy
that what I see on my monitor is approximetaly what I get back from
photo labs. So I take it that you find the scan correct?

It's within the parameters of a workflow using Gamma.