scanner gamma

[Bob Whatsima]

Hello all

Please can someone direct me to a good FAQ or tuturial that will help me set
the gamma for my scanner. I'm using vuescan & a super coolscan, and I'm not
sure if I should be dealing with my gamma setting before or after I make my
scanner profile.

Also, how is it possible to take advantage of ICE when using a RAW workflow?

 

[Bart van der Wolf]

Hello all

Please can someone direct me to a good FAQ or tuturial that will help me set
the gamma for my scanner. I'm using vuescan & a super coolscan, and I'm not
sure if I should be dealing with my gamma setting before or after I make my
scanner profile.

VueScan bases its profile on the Raw (=gamma 1.0) data. You gamma
settings will not be of influence to the creation of a profile.
Afterwards, set the Color balancing to None and the Output profile to
Device RGB. Of course you can experiment with other Color balancing
settings, but those will take you away from the profiled results. In
your Photoeditor, e.g. Photoshop you need to assign the Profile you
created in VueScan for your scanner/dye-set combination.

Also, how is it possible to take advantage of ICE when using a RAW

workflow?

Setting the input Bits per pixel to 64-bits, makes sure that the IR
channel is saved together with the Raw RGB channels.Subsequent reading
of the Raw file offers the opportunity to either use or disregard the
IR channel with the Infrared clean filter setting.

Bart

 

[Kennedy McEwen]

Hello all

Please can someone direct me to a good FAQ or tuturial that will help me set
the gamma for my scanner.

Gamma isn't generally a function of the scanner. Most modern scanners
are based on CCDs which are extremely linear in response, so the gamma
for the scanner is 1.0000. Some drum scanners are still based on PM
tubes which do have a non-linear response and thus require individual
gamma setting.

Gamma is a function of the viewing system because, unlike the CCD, your
eyes respond logarithmically to light intensity. It is a bit of a
coincidence that CRT response is very close to the inverse response of
the eye. So encoding data to precompensate for the gamma of the CRT
produces a perceived linear response in the eye.

Probably one of the best FAQs available on this topic is Charles
Poynton's Gamma FAQ, archived at:
http://people.ee.ethz.ch/~buc/brechbuehler/mirror/color/GammaFAQ.html
which summarises major sections of his book "A Technical Introduction to
Digital Video". Whilst both the FAQ and the book focus on HDTV as a
medium, the principles apply equally well to scanners and computers.

 

[Bob Whatsima]

Thank you both. The problem to me is that although I've made a profile using
vuescan, my scans generally seem a bit on the dark side. Although setting
the whitepoint in my photoshop levels lifts the image somewhat, nearly all
of them seem to require further quarter and midtone lifting in photoshop.
They are correctly exposed trannies, which is what baffles me.

I have read that the vuescan icc profile maker isn't that sophisticated. (?)
Would I get a better result out of something like gretag Macbeth or Monaco -
specifically wrt the quarter tones of my scans not being so dark.

TIA

Gamma isn't generally a function of the scanner. Most modern scanners
are based on CCDs which are extremely linear in response, so the gamma
for the scanner is 1.0000. Some drum scanners are still based on PM
tubes which do have a non-linear response and thus require individual
gamma setting.

Gamma is a function of the viewing system because, unlike the CCD, your
eyes respond logarithmically to light intensity. It is a bit of a
coincidence that CRT response is very close to the inverse response of
the eye. So encoding data to precompensate for the gamma of the CRT
produces a perceived linear response in the eye.

Probably one of the best FAQs available on this topic is Charles
Poynton's Gamma FAQ, archived at:
http://people.ee.ethz.ch/~buc/brechbuehler/mirror/color/GammaFAQ.html
which summarises major sections of his book "A Technical Introduction to
Digital Video". Whilst both the FAQ and the book focus on HDTV as a
medium, the principles apply equally well to scanners and computers.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers (replace 'nospam' with 'kennedym' when

replying)

 

[Bart van der Wolf]

Thank you both. The problem to me is that although I've made a profile using
vuescan, my scans generally seem a bit on the dark side. Although setting
the whitepoint in my photoshop levels lifts the image somewhat, nearly all
of them seem to require further quarter and midtone lifting in photoshop.
They are correctly exposed trannies, which is what baffles me.

That's what tonescaling is for. You know that thanks to profiling
you'll start with reasonably accurate colors as caught on film, but
there is nobody that prevents you to adjust overall
brightness/contrast/colorbalance to your liking.
If e.g. you shot under an artificial (say 3200K) lighting on daylight
film, the scene will (thanks to profiling!) look very warm with yellow
whites. You most likely want to color balance that towards a more
neutral rendition.

I have read that the vuescan icc profile maker isn't that

sophisticated. (?)

I'd rather say it produces simple matrix profiles which are good
enough for most purposes. Only if you (or a customer) have to
pre-judge results before expensive production runs get approved, it
pays to invest in a spectrophotometer produced profiling system. The
simpler colorimeter based solutions can produce results somewhere
halfway the optimal aproach.

Would I get a better result out of something like gretag Macbeth or Monaco -
specifically wrt the quarter tones of my scans not being so dark.

Maybe, maybe not much. I suspect the issue is more with the (wrong)
expectation that profiling will automagically take care of all color
issues, it won't. It cannot read your mind and it doesn't interpret a
scene. But it's a common misconception, which you can now leave
behind... ;-)

Bart

 

[Kennedy McEwen]

Thank you both. The problem to me is that although I've made a profile using
vuescan, my scans generally seem a bit on the dark side. Although setting
the whitepoint in my photoshop levels lifts the image somewhat, nearly all
of them seem to require further quarter and midtone lifting in photoshop.
They are correctly exposed trannies, which is what baffles me.

I have read that the vuescan icc profile maker isn't that sophisticated. (?)
Would I get a better result out of something like gretag Macbeth or Monaco -
specifically wrt the quarter tones of my scans not being so dark.

Almost certainly not.

Profiling is not the panacea of colour ailments that it is generally
believed to be. In particular, if you don't know the underlying
principles and what you are profiling then you can end up with worse
results than you started with.

I would recommend switching all of the profiles and colour management
off to begin with. Adjust the black and white points (brightness and
contrast) of your monitor correctly either using Nokia Ntest or any of
the available facilities on the net, taking particular care over the
black level. You might need to adjust brightness and contrast several
times in sequence to get both black and white just right. Then make a
few scans at the default gamma settings (2.2 for a PC) and see if that
gets you into the right region. You can then decide if you really need
colour management and profiling to correct any minor deficiencies that
remain.

Unless you have a properly adjusted system to begin with, profiling is
just a waste of time and even if you get decent colour from it, you will
be throwing a lot of display dynamic range away.

 

[Leonard Evens]

Gamma isn't generally a function of the scanner. Most modern scanners
are based on CCDs which are extremely linear in response, so the gamma
for the scanner is 1.0000. Some drum scanners are still based on PM
tubes which do have a non-linear response and thus require individual
gamma setting.

Gamma is a function of the viewing system because, unlike the CCD, your
eyes respond logarithmically to light intensity. It is a bit of a
coincidence that CRT response is very close to the inverse response of
the eye. So encoding data to precompensate for the gamma of the CRT
produces a perceived linear response in the eye.

Kennedy,

I am probably misunderstanding something, but I've never understood the
significance of the coincidence. If the various gammas are chosen
properly, the light intensities on the monitor screen bear some
reasonable relationship to the original intensities in the scene. In
that case, your visual system will see the screen much as it sees the
scene, so it won't matter what transformation it applies to what the
retina records. I'm sure there is more to it than that, but could you
explain why the coincidence really matters?

 

[Kennedy McEwen]

Kennedy,

I am probably misunderstanding something, but I've never understood the
significance of the coincidence. If the various gammas are chosen
properly, the light intensities on the monitor screen bear some
reasonable relationship to the original intensities in the scene. In
that case, your visual system will see the screen much as it sees the
scene, so it won't matter what transformation it applies to what the
retina records. I'm sure there is more to it than that, but could you
explain why the coincidence really matters?

It matters when you consider the precision that you need to specify the
data in. You are correct when you say that after the gamma of all of
the relevant components are properly accounted that you will see the
same relationship between different tones in the output as you see in
the screen - and that is true if you consider a continuous gamma
correction transform. That breaks down when you consider the effect of
noise and, in particular for digital systems, quantisation of the data
in the gamma corrected space.

 

[Bob Whatsima]

Maybe, maybe not much. I suspect the issue is more with the (wrong)
expectation that profiling will automagically take care of all color
issues, it won't. It cannot read your mind and it doesn't interpret a
scene. But it's a common misconception, which you can now leave
behind... ;-)

Ah crap, that's not the answer you were supposed to give!

The scanner profile I've made does seem to give me accurate colours. The
colour balance of my scans generally looks good. It's the luminosity of the
scans that I'm battling with. I have a Colorvision calibrated monitor, 6500k
& 2.2 gamma. I'm happy with the monitor profile. I'm scanning the images
into Adobe RGB.

If I lay out a bunch of slides on the lightbox, they all look correctly
exposed to me. None of them are overexposed and none of them are
underexposed. Yet the scanner uses a different exposure to scan each one -
which confuses me to start with, and then some of them require lots of
colour lifting, especially in the mid and quartertones as I mentioned
earlier. (Although I do concede that some scans come thru "almost correct"
ito brightness).

Why some images come thru correct and others come thru dark and requiring
lots of work is yet another of scanning's mysteries to me. When I take the
trannies of the dark scans back to the lightbox, they look like a perfectly
exposed photographs.

Although I can get the images all looking good with a bit of effort in
photoshop, I really come unstuck when I have to scan "similars" a day or
three later, and I can't remember how much I lifted the 1/4 and 1/2 tones on
my previous scan. So I end up with "similars" that don't look similar!

Should I perhaps be locking the exposure at a particular magic value instead
of letting it autoexpose? Will this help me get a more accurate scan ito
brightness? I want my scans to look as bright on screen as they look on the
lightbox, but that seems damn impossible to me!

 

[Fernando]

If I lay out a bunch of slides on the lightbox, they all look correctly
exposed to me. None of them are overexposed and none of them are
underexposed. Yet the scanner uses a different exposure to scan each one -
which confuses me to start with, and then some of them require lots of
colour lifting, especially in the mid and quartertones as I mentioned
earlier. (Although I do concede that some scans come thru "almost correct"
ito brightness).

Sound like your scanner has some troubles getting the correct
exposure. Maybe it measures it not on the whole frame, but only from
the central zone, for example.
Try locking the exposure as explained in the "Advanced Workflow
Suggestions" section of Vuescan's user manual (by measuring the
exposure of the clear film head).

Fernando

 

[RSD99]

"Bob Whatsima" posted:
"...
Should I perhaps be locking the exposure at a particular magic value instead
of letting it autoexpose?
...."

Probably.

"...
Will this help me get a more accurate scan ito
brightness?
...."

Probably.

"...
I want my scans to look as bright on screen as they look on the
lightbox, but that seems damn impossible to me!
...."

They can ... and maybe even better ... if the scanner and software is "set up properly."
In my case, when I first got my current scanner (Epson 4870) the "Auto-Everything"
literally drove me nuts *My* solution was learning how to use the software, and doing it
MANUALLY.

One thing that you might try is exposing some test charts, such as the Kodak Q-13 Gray
Scale or the Macbeth Colorchecker ... and working for a while with the images. Worst case:
You will at least have a better idea of what your combination if camera-film-scanner can
do.

 

[DavidTT]

"Yet the scanner uses a different exposure to scan each one..."

I missed your original post and do not know how you scan. How do you
know the scanner is using different exposures? Do you think the scanner
is using different exposures in hw or in sw? If you take the raw scan
approach, you tell the scanner's hw to scan at a fixed setting every
time and tell the scanner's sw not to manipulate the raw scan. The raw
scans will not look pretty or close to the slides and will need a lot of
work in PS, but at least you know that the scanner's sw has not been
manipulating the raw scans.

 

[Bob Whatsima]

Ta to everyone for responses. I'm trying the advanced workflow suggestion
that Fernando has mentioned below. I presume clear-head means the
transparent part of a roll before frame 1 and after frame 36? The image I
have chosen has some perfect detail one side, then some orange-yellow-red
burnout area which fades into a segment where the chemicals have had no
affect on the celluloid.

So I mount aforementioned clear-head image, insert it into the scanner, do a
preview, then pull a scan for prosperity. I then lock exposure to see what
exposure value was generated by VS for my clear-head image. It comes out at
4-comma-something.

Assuming all the images that i have to scan are shot on the same stock film,
albeit different days and different lighting conditions, can I use this
4-comma-something "lock exposure" value for all my future scans? Very few of
these originals look under- or over-exposed, regardless of the lighting
conditions they were shot under. Would this be a safer method than the
default which I think changes the CCD exposure on an image-by-image
evaluation?

I have noticed that my scans that look the most incorrect are the ones with
heavy contrast - eg a small white boat floating on a large area of deep blue
water. Perhaps by nailing down an exposure value that is correct for
so-and-so stock film, I will get more consistency out of my scans?

Any advice / criticisms / suggestions gladly accepted.

 

[nikita]

Bob,

if you're scanning slides and want the scanner to behave exacly the
same way each time it runs;

Lock the exposure, scan the IT8 slide, apply any readymade gammacurve
that sets the greypatch 11 in the range of RGB 100-110, run the
freestanding profiler like ProfileMaker, Monaco or the free little CMS
to generate out a scannerprofile. Any freestanding profiler needs a
completely locked state of the scanner!

Scan the real slide at the same locked exposure, apply the same gamma
curve, ASSIGN the profile in Photoshop, Convert to workingspace Adobe
RGB.

Make the Creative editing from this always the same scanner behaviour.
Forget the internal Vuescan calibration flow.

Some people are using the NON setting in Vuescan and generates out a
"cropped" tiff instead of the pure RAW. This will output the file in a
gammacorrected state thus no gammacurve has to be used manually. But
that is depending on what Ed is doing with the gamma function of
Vuescan in the future. Earlier it was different things connected to
that one. If that was in the NON or the other settings I don't
remember.

Anyway, disconnect as much as possible and profile externally using
thirdparty profilers. Some tests has shown that there's little
difference between the expensive profilers and the free little CMS.
Bruce Fraser has said things about tests between profilers like that
isn't going to show the real world differences. But that doesn't
matter. As long as it delivers what your eyes sees on the light table,
it's working regardless of pricelevel. I'm running them all and prefer
Profilemaker 5 which is quite good at scannerprofiles now. I can see
differences between PM 4 and 5.

nikita



http://www.littlecms.com/profilers.htm

At this site chose "Gammaterminology and calculations", scroll down
the page till you find downloads for gammacurves.

http://www.aim-dtp.net/aim/calibration/index.htm

 

[Bob Whatsima]

Lock the exposure, scan the IT8 slide, apply any readymade gammacurve

that sets the greypatch 11 in the range of RGB 100-110, run the
freestanding profiler like ProfileMaker, Monaco or the free little CMS
to generate out a scannerprofile. Any freestanding profiler needs a
completely locked state of the scanner!

You mean scan as a 48bit RAW file? And apply the gammacurve in photoshop?
Because I open the RAW file in photoshop, it's going to ask me to leave
colour as is / assign profile / convert to working space, etc.

A RAW file is RGB, right? So even if I choose "leave as is / don't
colour manage this document", it is still doing some colour management
in the background? Whether I like it or not.

From what I understand "leave as is" would use your working space as a
reference (mine = Adobe RBG).

Is colour management not then being applied twice? Using the Adobe RGB
look up table (or whatever it's called) when it *first* opens my
document?

 

[Bob Whatsima]

Have got a profile going which I'll test drive at work tomorrow.

 

[nikita]

Yes, scan a RAW file if you want to use a separate gamma hit in
Photoshop, or use the NON setting and output a so called "cropped"
tiff (Ed Hamricks name on a processed file – not a RAW). With NON you
can specify a gamma value in the settings. A pure RAW comes in linear
1.0 gamma usually if not using a 24 bit output. Therefore the separate
gamma curve is needed in Photoshop. The NON is just gammacorrected but
not processed in any other aspect. You can profile a svcanner in any
state as long as it is LOCKED in a fixed state. It's a good idea to
minimize the involved variables and factors as far as possible.
Therfore a pure RAW from a locked exposure and a gammahit in
Photoshop, plus assigning the scannerprofile plus converting to
workingspace is the most controlled route. Put it all into an Action.
It just takes one button to push and it's there.

Leave as is doesn't change the number/colors in the file itself. It
just assumes that the file is resting in Adobe RGB (workingspace) and
makes a conversion under the hood to the Monitorprofile. Your view is
based on the assumtion that the file is resting in AdobeRGB or the
workingspace set in the PS prefs. As soon as you ASSIGN the
scannerprofile it all goes from that into the monitorspace. AND any
further conversions is going FROM the scannerprofile TO the
destination. The destination to CONVERT to before doing any editing
would be the Workingspace.

Actually it doesn't matter if you assign anything else when opening
the file – as soon as you assign something else it just change
profile. It doesn't change anything in the file. CONVERT changes the
numbers in the file. But an ASSIGNED profile will always be the source
for the CONVERSION. Thus the convertion will end up difrently
depending on what the file is tagged/assigned regarding profile. If
it's the true file – the scannerprofile – everything will be ok. If
not the truefile it will screw up colors. But you can always ASSIGN a
profile that looks good to your eyes and then convert. As long as that
is your choice and you're aware of that it's ok.

The most simple flow when you have done all the profiling is to use
the "missing profile" dialog upon the opening and chose the new
scannerprofile and check the checkbox for "and then convert to
workingspace". Then it is in the workingspace when you see it. But
using "leave as is" when opening will give you more choices when using
the Image>mode>Assign profile command instead after the file has get
opened. You can here chose/try different renderings like perceptual or
relative and see what looks best. When using the "missing profile"
dialog it uses the rendering that you set in the Colormanagment prefs
of photoshop. This is defaulting to Relative Colormetric. If that's
what you want, use it.

No colormanagment is done in the background when opening and leave as
is *more than* the assumtion for the source for the conversion to the
monitor. The only thing that you need to do is ASSIGN the
scannerprofile and then convert to the workingspace. As simple as
that. Then do the creative editing.

nikita

 

[Mike Engles]

Gamma isn't generally a function of the scanner. Most modern scanners
are based on CCDs which are extremely linear in response, so the gamma
for the scanner is 1.0000. Some drum scanners are still based on PM
tubes which do have a non-linear response and thus require individual
gamma setting.

Gamma is a function of the viewing system because, unlike the CCD, your
eyes respond logarithmically to light intensity. It is a bit of a
coincidence that CRT response is very close to the inverse response of
the eye. So encoding data to precompensate for the gamma of the CRT
produces a perceived linear response in the eye.

Probably one of the best FAQs available on this topic is Charles
Poynton's Gamma FAQ, archived at:
http://people.ee.ethz.ch/~buc/brechbuehler/mirror/color/GammaFAQ.html
which summarises major sections of his book "A Technical Introduction to
Digital Video". Whilst both the FAQ and the book focus on HDTV as a
medium, the principles apply equally well to scanners and computers.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)

Hello

I have a problem with the idea of inverse gamma to aid in preception.


If we view a slide by holding it up to the light, or on a light table
or, by projection on to a screen, as far as I can see no adjustments
have been made, but the image is percieved as linear and correct.
It is a pretty good description of the original scene.

A scanner is in essence a system of projecting light through the slide
and then reading the transmitted light values with a sensor, which you
say has a linear response. Now to view the the resulting image on a CRT
we need to apply a inverse gamma of the order of 2.2 to 2.5, otherwise
the image is too dark.It is no longer, unless inverse gamma is applied,
a good description of the original scene.

The first three examples are the same as the fourth, except that in the
fourth a CRT is involved.

Mike Engles

 

[Wayne Fulton]

I have a problem with the idea of inverse gamma to aid in preception.

That is also my understanding Mike, the CRT output response should be
corrected to be perceived as linear intensity, and not because gamma matches
any eye perception response curve. A CRT is simply not a linear device (a
field effect at the CRT grid), and so this inverse gamma formula is invented
to first oppositely distort the input data so the CRT response will appear
linear, meaning linear output intensity at the tube phosphor, simply to
match the linear pre-encoded input data intensity from slide or scene.

So encoding data to precompensate for the gamma of the CRT
produces a perceived linear response in the eye.

Not wishing to speak for Kennedy, but I read it supposing he added the final
"in the eye" for that same reason... The CRT response should be linear so the
eye can perceive the image reproduction as linear. The eye/brain and the CRT
do have their non-linear ways of working, but from a CRT result point of view,
it doesnt really matter how the eye does that, nor how the CRT does it, so
long as the CRT response is corrected to show linear intensity, meaning it
looks like the slide or scene that created the data.

The human eye is said to have a 1/3 power law response, and inverse gamma of
1/2.2 is coincidentally similar, but this doesnt change WHY the CRT is
corrected to be linear (not to match the eye response). It would not matter if
the two formulas might have been entirely different somehow, linear intensity
is still the correct CRT response for the eye to view (to match slide or
scene). However this similarity surely must work out as convenient to use the
same gamma shape curve (Curve tool for example) to edit an image arbitrarily
brighter or darker in a satisfactory way to look natural to our eye. That
seems great, but in my opinion, that is the extent of it.

Gamma also has the strong side benefit of compressing the digital data so that
8 bits can be sufficient to show it somewhat like the eye/brain can perceive
it. That compression discards the right data that we wouldnt notice anyway
(discards more bright tones of which we have too many to be useful to the
eye/brain), and retains more important sparse dark data steps that we might
notice (the idea that 100 intensity steps differing by 1% are seen). Which is
fantastic, but it is just clutter in explanations, as it also is NOT WHY gamma
is done. Gamma is instead done to correct the CRT response to be linear, and
has been done from the earliest days of analog television shown on CRT.

 

[Kennedy McEwen]

Hello

I have a problem with the idea of inverse gamma to aid in preception.


If we view a slide by holding it up to the light, or on a light table
or, by projection on to a screen, as far as I can see no adjustments
have been made, but the image is percieved as linear and correct.
It is a pretty good description of the original scene.

A scanner is in essence a system of projecting light through the slide
and then reading the transmitted light values with a sensor, which you
say has a linear response. Now to view the the resulting image on a CRT
we need to apply a inverse gamma of the order of 2.2 to 2.5, otherwise
the image is too dark.It is no longer, unless inverse gamma is applied,
a good description of the original scene.

The first three examples are the same as the fourth, except that in the
fourth a CRT is involved.

I am not sure that I see what your difficulty is here. The output of
the scanner is linear and, if there was a linear way to display that,
then there would be no need for gamma, other than to correct for the
perception of quantisation noise throughout the range - which could be
achieved with sufficient bit depth in any case. The gamma compensation
is required because the CRT response is non-linear, not because the
image is too dark without it. If this was just a case of compensating
for image brightness, a linear offset to the levels or gain would be
sufficient. Most of this is explained in a lot more detail the FAQ that
I referenced earlier in the thread.