Question For Kennedy

[Dave]

Hi Kennedy,

You've mentioned several times that you increase the Analog Gain in
NikonScan to get more accurate scans. How much do you increase it?
Do you have standard setting or do you adjust it differently for each
image or group of images? I just added .5 and it helped a lot.

Cheers,
Dave

 

[Kennedy McEwen]

Hi Kennedy,

You've mentioned several times that you increase the Analog Gain in
NikonScan to get more accurate scans. How much do you increase it?
Do you have standard setting or do you adjust it differently for each
image or group of images? I just added .5 and it helped a lot.

I don't adjust the analogue gain unless the histogram shown in the
Curves window indicates that one or more of the colour channels is
clipped or well underexposed. When I do find this, which isn't very
often with slides, then I adjust it by whatever it takes to get an
unclipped histogram, but there are no fixed settings.

 

[Dave]

I don't adjust the analogue gain unless the histogram shown in the
Curves window indicates that one or more of the colour channels is
clipped or well underexposed. When I do find this, which isn't very
often with slides, then I adjust it by whatever it takes to get an
unclipped histogram, but there are no fixed settings.

Very good. Thank you. I can see how Analog Gain shifts the entire
histogram right or left depending on how you use it. That's exactly
what I need right now because I'm trying to scan some slides taken in
cold, wet, overcast, rainy Alaska and they're very dark. Analog Gain
lightens them up nicely.

BTW, many months ago you gave me some great advice on settings for
NikonScan. They really helped and I saved a copy of your message and
I pull it out every time I fire up the program. Thanks again for that
too.

 

[Dave]

Very good. Thank you. I can see how Analog Gain shifts the entire

histogram right or left depending on how you use it. That's exactly
what I need right now because I'm trying to scan some slides taken in
cold, wet, overcast, rainy Alaska and they're very dark. Analog Gain
lightens them up nicely.

There is an unfortunate trade off when using Analog Gain. If I set
the Analog Gain to zero, this batch of slides is dark and difficult to
see, and changing midpoint Levels or Brightness in Photoshop Elements
gives an unpleasant tint. If I set Analog Gain to about +1.0, the
slides are correctly exposed, but some detail is lost that I can't
recover. It's a dilemma. I think the root cause is that I took the
slides with an Olympus camera. I've owned several Olympus's and they
always shoot dark. Olympus engineers just set them that way. I
normally correct for it, but this time I didn't.

 

[Kennedy McEwen]

There is an unfortunate trade off when using Analog Gain. If I set
the Analog Gain to zero, this batch of slides is dark and difficult to
see, and changing midpoint Levels or Brightness in Photoshop Elements
gives an unpleasant tint. If I set Analog Gain to about +1.0, the
slides are correctly exposed, but some detail is lost that I can't
recover. It's a dilemma.

You could just be reaching the dynamic range limits of the scanner of
course - if you have highlights in those slides then they will saturate
if you bring up the analogue gain to correct for very dark shadows. You
could try two exposures and then merge the results in Photoshop - I
understand that Vuescan actually offers this facility in the software
directly, which is an advantage over Nikonscan, and effectively enhances
the dynamic range of the scanner significantly.

I think the root cause is that I took the
slides with an Olympus camera. I've owned several Olympus's and they
always shoot dark. Olympus engineers just set them that way. I
normally correct for it, but this time I didn't.

Coincidentally, most of my 35mm equipment is Olympus too - a couple of
OM-4Ti's, an OM-4, OM-2n and a couple of manual OM-1's with a stack of
matching optics. Never noticed the underexposure problem you allude to
on any of them, even a slide film I accidentally underexposed by a stop
all the way through on an unrepeatable trip a few years ago produced
some excellent images after asking the lab to push process it - so good
that I was tempted to rate Velvia at 100ASA from then on!

 

[Dave]

You could just be reaching the dynamic range limits of the scanner of
course - if you have highlights in those slides then they will saturate
if you bring up the analogue gain to correct for very dark shadows. You
could try two exposures and then merge the results in Photoshop - I
understand that Vuescan actually offers this facility in the software
directly, which is an advantage over Nikonscan, and effectively enhances
the dynamic range of the scanner significantly.

I haven't found a good solution yet. If I turn up the Analog Gain
enough to make the majority of the image look properly exposed, some
parts are very over-exposed and detail is lost. For example, the
branches of trees silhouetted against the sky in the background become
washed out and discontinuous. You would have to look at the image
with a critical eye to see that, but the perfectionist in me wants the
whole image exactly right. I also want world peace and a beach house
in Malibu, so call me a dreamer.

Coincidentally, most of my 35mm equipment is Olympus too - a couple of
OM-4Ti's, an OM-4, OM-2n and a couple of manual OM-1's with a stack of
matching optics. Never noticed the underexposure problem you allude to
on any of them, even a slide film I accidentally underexposed by a stop
all the way through on an unrepeatable trip a few years ago produced
some excellent images after asking the lab to push process it - so good
that I was tempted to rate Velvia at 100ASA from then on!

No one else has ever agreed with me about Olympus products either, so
maybe it's just me. I started with a Konica 35mm SLR over 30 years
ago and it shot nice and bright and clean. My next camera was an
Olympus OM-2 and I had to manually adjust the F-stop to make it shoot
bright enough. My little fishing camera is an Olympus Weatherproof
automatic and it shoots dark. My digital is an Olympus C3040 Zoom and
once again I've adjusted the F-stop by seven tenths of a stop to get
bright images. Personal preference perhaps.

 

[Kennedy McEwen]

I haven't found a good solution yet. If I turn up the Analog Gain
enough to make the majority of the image look properly exposed, some
parts are very over-exposed and detail is lost. For example, the
branches of trees silhouetted against the sky in the background become
washed out and discontinuous. You would have to look at the image
with a critical eye to see that, but the perfectionist in me wants the
whole image exactly right. I also want world peace and a beach house
in Malibu, so call me a dreamer.

You should certainly be able to get almost all of the possible density
range of slide film recorded within the dynamic range of the scanner,
but with some emulsions and high contrast subjects you could be running
out of room with adequate SNR. However, your comments make me wonder
whether it is not so much a brightness/contrast issue as gamma.
Presumably you have tried adjusting the gamma after getting analogue
gain in the right level to just avoid saturating the highlights?

No one else has ever agreed with me about Olympus products either, so
maybe it's just me. I started with a Konica 35mm SLR over 30 years
ago and it shot nice and bright and clean. My next camera was an
Olympus OM-2 and I had to manually adjust the F-stop to make it shoot
bright enough.

Eh?? The OM-2 is an aperture priority autoexposure camera, without any
exposure lock - in fact, not requiring a pre-exposure lock was its key
selling point with the exposure time being adjusted to cope with light
changes in real time as the film was exposed. So adjusting the f/#
would just be compensated by the camera altering the exposure time -
result: same EV.

My little fishing camera is an Olympus Weatherproof
automatic and it shoots dark. My digital is an Olympus C3040 Zoom and
once again I've adjusted the F-stop by seven tenths of a stop to get
bright images. Personal preference perhaps.

Again, the 3040 will adjust exposure time to compensate. Are you sure
you are actually making a difference? The correct adjustment, and only
one on an automatic camera, is over or under rating the film speed. Only
in a completely manual mode will adjusting the lens aperture influence
exposure. In your case, where the exposure produces darker than
required results, decrease the film speed slightly. Most Oly cameras
allow 1/3rd EV increments of film speed - and if you are concerned about
less than that then you are kidding yourself.

 

[Dave]

range of slide film recorded within the dynamic range of the scanner,
but with some emulsions and high contrast subjects you could be running
out of room with adequate SNR. However, your comments make me wonder
whether it is not so much a brightness/contrast issue as gamma.
Presumably you have tried adjusting the gamma after getting analogue
gain in the right level to just avoid saturating the highlights?

Gamma? I have gamma set as correctly as possible for my monitor using
Quick Gamma, and gamma set at 2.20 in Preferences (I have a PC) in
NikonScan and the little contrast chart is pretty close. Is there
more to it than that? My understanding is that gamma is just the
nonlinear relationship between the computer and the monitor.

Eh?? The OM-2 is an aperture priority autoexposure camera, without any
exposure lock - in fact, not requiring a pre-exposure lock was its key
selling point with the exposure time being adjusted to cope with light
changes in real time as the film was exposed. So adjusting the f/#
would just be compensated by the camera altering the exposure time -
result: same EV.

Again, the 3040 will adjust exposure time to compensate. Are you sure
you are actually making a difference? The correct adjustment, and only
one on an automatic camera, is over or under rating the film speed. Only
in a completely manual mode will adjusting the lens aperture influence
exposure. In your case, where the exposure produces darker than
required results, decrease the film speed slightly. Most Oly cameras
allow 1/3rd EV increments of film speed - and if you are concerned about
less than that then you are kidding yourself.

You're absolutely right. My mistake. I actually adjusted the
exposure on my C3040 by +.7 by going into the menu system and it makes
a noticeable difference. I don't have the OM-2 anymore, but I'm sure
I did something similar with it too because I usually shot it in
automatic mode and what you say about adjusting exposure time to
compensate is certainly correct. I don't know why I said F-stop.

 

[Kennedy McEwen]

Gamma? I have gamma set as correctly as possible for my monitor using
Quick Gamma, and gamma set at 2.20 in Preferences (I have a PC) in
NikonScan and the little contrast chart is pretty close. Is there
more to it than that? My understanding is that gamma is just the
nonlinear relationship between the computer and the monitor.

Not *just* the relationship between the computer and the monitor.
Certainly that has a relationship which is approximately described by a
gamma curve, or more correctly, the relationship between video DAC data
and monitor output luminance does. That is merely the default gamma
that is applied to the image data for nominal viewing over the 8-bit
video range though.

However the film also has a response which can be approximately
described by a similar curve. Film response is only linear over a very
limited range of exposure values. If the film is under or over exposed
then its response, in terms of reproduced optical density for a given
exposure to light intensity, will be approximate a gamma curve and
require an inverse curve to correct for it, rather than simply analogue
gain or brightness and contrast adjustment. Most decent film scanners
these days have a linear response range which vastly exceeds the film
they scan, so exposure errors on the film will still be well within the
response range of the scanner but require additional gamma adjustment to
get the correct density distribution in the final image.

Even for correctly exposed film though there are cases where gamma is
preferable to brightness and contrast, such as pulling detail up out of
shadows without losing the shadow density etc. That is why the control
exists in Photoshop levels etc. in the first place.

It might be more than a coincidence that you disagree with others on the
nominal exposure your cameras produce and also find the scanned images
generally dark once they have been exposure compensated into the range
of the scanner.

 

[Dave]

Not *just* the relationship between the computer and the monitor.
Certainly that has a relationship which is approximately described by a
gamma curve, or more correctly, the relationship between video DAC data
and monitor output luminance does. That is merely the default gamma
that is applied to the image data for nominal viewing over the 8-bit
video range though.

However the film also has a response which can be approximately
described by a similar curve. Film response is only linear over a very
limited range of exposure values. If the film is under or over exposed
then its response, in terms of reproduced optical density for a given
exposure to light intensity, will be approximate a gamma curve and
require an inverse curve to correct for it, rather than simply analogue
gain or brightness and contrast adjustment. Most decent film scanners
these days have a linear response range which vastly exceeds the film
they scan, so exposure errors on the film will still be well within the
response range of the scanner but require additional gamma adjustment to
get the correct density distribution in the final image.

Even for correctly exposed film though there are cases where gamma is
preferable to brightness and contrast, such as pulling detail up out of
shadows without losing the shadow density etc. That is why the control
exists in Photoshop levels etc. in the first place.

It might be more than a coincidence that you disagree with others on the
nominal exposure your cameras produce and also find the scanned images
generally dark once they have been exposure compensated into the range
of the scanner.

Makes sense. Not much in this life is linear. I looked again at the
controls in NikonScan and didn't find an adjustment labeled "Gamma"
except in the Preferences section and I already set that for my
monitor, so I'm guessing that either Curves in NikonScan or Levels in
PhotoShop Elements is required to adjust the gamma-like Input/Output
relationship. To that end, I've compared using the Curves tool to
adjusting Analog Gain on the same scan and come up with about the same
results. Increased brightness means decreased detail.

It might be more than a coincidence. My girlfriend is a former
professional photographer and she thinks a lot of the images I adjust
are too bright.

I'm using bright and brightness here with their dictionary
definitions. There's probably a stricter technical definition.

I'm leaving town for a week. Thanks again for all your help.

 

[Kennedy McEwen]

I looked again at the
controls in NikonScan and didn't find an adjustment labeled "Gamma"
except in the Preferences section and I already set that for my
monitor, so I'm guessing that either Curves in NikonScan or Levels in
PhotoShop Elements is required to adjust the gamma-like Input/Output
relationship.

In both the Curves window of NikonScan and the Levels window of
Photoshop (and presumably PS-Elements too) the gamma is the central data
entry point, or the middle adjustment tag. This defaults to 1.00 and
can be adjusted up or down and in doing so you will observe that the
transfer function shown by the line becomes a curve - a gamma curve.
This can also be adjusted independently for each colour as well as the
overall luminance.

 

[Dave]

In both the Curves window of NikonScan and the Levels window of
Photoshop (and presumably PS-Elements too) the gamma is the central data
entry point, or the middle adjustment tag. This defaults to 1.00 and
can be adjusted up or down and in doing so you will observe that the
transfer function shown by the line becomes a curve - a gamma curve.
This can also be adjusted independently for each colour as well as the
overall luminance.

Makes sense. The midpoint (grey point) slider will move the
input/output relationship out of linearity and create a gamma-like
curve.

Thanks. Gotta run.