Vuescan 8.0- Nikon LS4000- analog gain and raw file questions

[Marcello Penso]

Dear All,

I am just starting to scan for archiving purposes my 20,000 slides or so
and I saw that Vuescan 8.0 has 'analog gain' settings for Nikon
scanners. I've tinkered with those settings a bit and they can produce
dramatically different results in raw file scans, as opposed to the
exposure clipping setting or the long exposure pass setting, which seem
to have little effect on raw scans (at least looking at them through
Irfanview.)

I've noticed that 'auto levels' or 'white balance' color correction (in
a second pass to produce jpgs for contact sheets) has consistently
produced files which are a bit on the red side. I'm not sure whether
this is scanner related, monitor related, or Vuescan related, but I do
notice reddish cast in the raw scans. My next step will be to try
viewing the raw scan tiffs I've scanned already on another monitor to
see if this reddish cast is monitor related, since my 21" Viewsonic G810
is 4.5 years old.

I have a few questions:

Does long exposure pass produce noticeable differences in quality in
well lit exterior shots, or should I reserve that for underexposed
slides, or slides with strong silhouettes or dark (winter sunset
lighting) slides?

How does the exposure clip affect the exposure timing? I don't quite
understand the description in the Users Guide. Does this setting enhance
image quaility (color rendition) as well, or only affect overall
brightness of the raw scans?

Should the analog gain settings be used for known scanner deficiencies
in CCD exposure?

Is getting an IT8 slide recommendable for checking monitor/scanner color
calibration? I went to Wolf's site, but those slides seem film specific,
and I have slides with a variety of different films, a good portion
Agfachromes.

Any suggestions would be welcome. Thanks.

Marcello

 

[nikita]

Marcello,

20.000 slides - that is work.

If you've said negs I would consider Vuescan perhaps. However, this
software is based on so many variables that will influence the
scanningresult. Those variables shift importance from one version to
the other. The result is to unpredictable. At least if your goal is to
have any consistent flow from slide 1 to 20.000...
There is a great risk that you find yourself REscanning bunches of
slides with new parameters set when gone through half of that mass of
slides and at that point find a bad sideffect or a better setting.
That is killing even the best scanneroperator.....

I've been using Vuescan for years. Today, I've found more stable flows
with Silverfast. An application that I did earlier reject much because
the pricelevel and the toolarsenal which I did not need. Now I really
appreciate what it delivers at the BASIC level – Quality. For that
amount of slides I would invest that money it costs to get an Ai6 for
the Nikon 4000. I'm using that combo myself. The built in IT8 does the
job to nail the CM and it comes with an IT8 slide in the package. You
have the option to use analog gain with a very usable histogram to max
out each compåonentchannel if you wish. The RAW outputs are extremely
clean. You can use ICE on the RAWS.

Even when scanning negs on the 4000 the RAW files that it spits out
when set to negative mode is a dream to work on manually. Instead of
using negafix, just invert
the RAW in Phtoshop (set gamma 2.2 box active in the preferences of
Silverfast to get the RAW in gamma 2.2 and you don't have any gamma
work to do on the RAW) . Then ballance each componentchannel manually
or use the Autocolor a la Bruce Fraser method with snap midgrey. If
this is done in 16 bit bpc you have an extrenely well working semiauto
scansoftware in Photoshop. No clipping of endpoints whatsoever. The
negative RAW highbitoutput (not with neg direct) is a "prepaired" RAW
to be used with silverfast HDR and negafix. the result with a manually
made editing directly works better if done with feeling. It's NOT hard
work.

If you're running your own third party profiling application ( Gretag
Profilemaker for example) on RAWs of slides, it will take you as far
as it's ever possible. It's in the ballpark without editing. Even the
built in calibration will do much better than what Vuescans built in
IT8 will do. It's worth the extra money which these days isn't that
large anyway. Especially when Silverfast comes with the IT8 slide
target. But remember to keep the work in Silverfast as simple as
possible without any extreme editing. Just find a decent basic flow in
high bit and do the rest on copys of the RAWs.

Try the demoversion before starting up that huge projekt!

Nikita

PS. The IT8 calibration isn't filmdependent i real life. It's waay
much more important that the referencefile is telling the right
story.....and no...I'm not connected in any way to Lasersoft. Instead
I've been giving them a hard time on the support level to max out my
quality.

 

[Philip Homburg]

I am just starting to scan for archiving purposes my 20,000 slides or so
and I saw that Vuescan 8.0 has 'analog gain' settings for Nikon
scanners.

If you set the analog gain to high, the sensor in the scanner will saturate
and clip the highlights.

Should the analog gain settings be used for known scanner deficiencies
in CCD exposure?

Analog gain is basically linear. If both white and mid grey can be corrected
with a single change in the analog gain for one channel, then that is a good
solution. If white and mid grey require different settings, you have to
move on to profiling or manual corrections with levels or curves.

 

[SJS]

I am just starting to scan for archiving purposes my 20,000 slides or so
and I saw that Vuescan 8.0 has 'analog gain' settings for Nikon
scanners. I've tinkered with those settings a bit and they can produce
dramatically different results in raw file scans

This sounds wonderful. I wish Vuescan had this for the FS4000 (assuming
the scanner has such an adjustment).

I've noticed that 'auto levels' or 'white balance' color correction (in
a second pass to produce jpgs for contact sheets) has consistently
produced files which are a bit on the red side. I'm not sure whether
this is scanner related, monitor related, or Vuescan related, but I do
notice reddish cast in the raw scans.

I get a tint with Vuescan if I set the input to 'slide' rather than
'image'. I assume Vuescan is compensating for the Kodachrome blueness.
Now I just use the raw scan file and brighten it in Photoshop as
required.

Does long exposure pass produce noticeable differences in quality in
well lit exterior shots, or should I reserve that for underexposed
slides, or slides with strong silhouettes or dark (winter sunset
lighting) slides?

I haven't had any success with this on my FS4000. I understand it is
meant to improve the detail in dark areas but may cause artifacts in the
transistions between light and dark.

How does the exposure clip affect the exposure timing? I don't quite
understand the description in the Users Guide. Does this setting enhance
image quaility (color rendition) as well, or only affect overall
brightness of the raw scans?

I think this tells Vuescan to brighten the image until the specified
proportion of pixels clip to white. It doesn't seem to affect the raw
scans.

Regards,

Steven

 

[Marcello Penso]

If you set the analog gain to high, the sensor in the scanner will saturate
and clip the highlights.


Analog gain is basically linear. If both white and mid grey can be corrected
with a single change in the analog gain for one channel, then that is a good
solution. If white and mid grey require different settings, you have to
move on to profiling or manual corrections with levels or curves.

I'm sorry, but I don't understand this last paragraph. In Vuescan there
is analog gain for red blue and green channels. Ed Hamrick in the manual
says that analog gain settings really relate to increase CCD exposure
time, which I understand. But I don't understand what white and grey has
to do with that.

Marcello

 

[Marcello Penso]

Marcello,

20.000 slides - that is work.

If you've said negs I would consider Vuescan perhaps. However, this
software is based on so many variables that will influence the
scanningresult. Those variables shift importance from one version to
the other. The result is to unpredictable. At least if your goal is to
have any consistent flow from slide 1 to 20.000...
There is a great risk that you find yourself REscanning bunches of
slides with new parameters set when gone through half of that mass of
slides and at that point find a bad sideffect or a better setting.
That is killing even the best scanneroperator.....

I've been using Vuescan for years. Today, I've found more stable flows
with Silverfast. An application that I did earlier reject much because
the pricelevel and the toolarsenal which I did not need. Now I really
appreciate what it delivers at the BASIC level ? Quality. For that
amount of slides I would invest that money it costs to get an Ai6 for
the Nikon 4000. I'm using that combo myself. The built in IT8 does the
job to nail the CM and it comes with an IT8 slide in the package. You
have the option to use analog gain with a very usable histogram to max
out each compåonentchannel if you wish. The RAW outputs are extremely
clean. You can use ICE on the RAWS.

Even when scanning negs on the 4000 the RAW files that it spits out
when set to negative mode is a dream to work on manually. Instead of
using negafix, just invert
the RAW in Phtoshop (set gamma 2.2 box active in the preferences of
Silverfast to get the RAW in gamma 2.2 and you don't have any gamma
work to do on the RAW) . Then ballance each componentchannel manually
or use the Autocolor a la Bruce Fraser method with snap midgrey. If
this is done in 16 bit bpc you have an extrenely well working semiauto
scansoftware in Photoshop. No clipping of endpoints whatsoever. The
negative RAW highbitoutput (not with neg direct) is a "prepaired" RAW
to be used with silverfast HDR and negafix. the result with a manually
made editing directly works better if done with feeling. It's NOT hard
work.

If you're running your own third party profiling application ( Gretag
Profilemaker for example) on RAWs of slides, it will take you as far
as it's ever possible. It's in the ballpark without editing. Even the
built in calibration will do much better than what Vuescans built in
IT8 will do. It's worth the extra money which these days isn't that
large anyway. Especially when Silverfast comes with the IT8 slide
target. But remember to keep the work in Silverfast as simple as
possible without any extreme editing. Just find a decent basic flow in
high bit and do the rest on copys of the RAWs.

Try the demoversion before starting up that huge projekt!

Nikita

PS. The IT8 calibration isn't filmdependent i real life. It's waay
much more important that the referencefile is telling the right
story.....and no...I'm not connected in any way to Lasersoft. Instead
I've been giving them a hard time on the support level to max out my
quality.

I'll look into Silverfast. Thanks for the tip.

Marcello

 

[Marcello Penso]

This sounds wonderful. I wish Vuescan had this for the FS4000 (assuming
the scanner has such an adjustment).


I get a tint with Vuescan if I set the input to 'slide' rather than
'image'. I assume Vuescan is compensating for the Kodachrome blueness.
Now I just use the raw scan file and brighten it in Photoshop as
required.


I haven't had any success with this on my FS4000. I understand it is
meant to improve the detail in dark areas but may cause artifacts in the
transistions between light and dark.


I think this tells Vuescan to brighten the image until the specified
proportion of pixels clip to white. It doesn't seem to affect the raw
scans.

But in the help file, the last paragraph for the 'RGB/Infrared
Exposure' line says:

Note that these exposure values don't directly control the brightness of
the final image. This is controlled by options in the Color tab. These
exposure values control the brightness of the raw scan file only.
^^^^^^^^^^^^^^^^^^^^^^^^

If clipping affects the calculation of how to brighten, wouldn't that
impact the raw scan file? And if brightness is affect, isn't color
rendition also?

Marcello

 

[degrub]

White and grey points are assumed to have equal R,G,B values. Otherwise
there is a color cast in the image which may or may not be desired. It
can require some care to boost the RGB gains individually and still keep
the white or gray where it should be for the image.

Frank

 

[nikita]

The analog gain is the exposure of the RAW file. It adjusts each
componentchannel FROM the basic exposure to peak them separately into
equal maximum of what the transparent filmbase can handle without
clipping those tonelevels near it. That will max out as much as it can
from the deep dense dark areas of the slide/neg = getting the maximum
dynamic range of the scanner. That's the reason for using it and not
stay with the basic exposure only. Staying with the basic exposure
only would leave unused space for one or two channels when the
strongest peaks based on the basic exposure. IF ballancing this up
afterwards in Photoshop, you get the ballance but trashing dynamic
range in the scanner. How much this means in real world seen by human
eyes and in any output is a good question....and it depends on each
situation – not just theoretical bullshit.

It's simply like comparing the RGB composite channel in Levels, with
going into each componentchannel and adjusting the ENDpoints (in the
white end) separately in all the channels. The difference is that the
analog gain does it directly with exposure of the CCD. The levels or
curves does it from what it gets from the basic exposure....

You use the histogram and adjust the analog gain till each channel
reach the top and not more than so. It effects the RAW files
colorballance allover as it cleans up the transparent filmbase to
neutral. It is adjusting the exposure TIME of the Nikons but could
also mean adjusting the intensity of three lightbulbs red - green -
blue. How the rest of the scale down to black looks depends on the
scanner. That's where the IT8 calibration/profiling comes in. It
cleans up the greyscale, sets the hues and saturation of all colors
allover the place.

Optimize the scanner with analog gain, lock it at that state, then
adjust to a gamma that sets an IT8 GS 10/11 greypatch in the range of
100-110 RGB. That's the HARDWARE calibration which is the basement for
the SOFTWAREcalibration using the internal calibrationsystem or a
thirdparty profilingapplication. These two acts give you a scanner
that uses full dynamic range and is colorcalibrated for slidework. The
ICC profile describes how the scanner looks when hardwarecalibrated.
This information is used to move the colors into a deviceindependent
workingspace>>monitorspace that is not connected to the scanner
anymore. The workingspace is greyballanced, which means that it is
giving a neutral greyballance response allover when your editing the
picture with Photoshop tools. Doing so in the scanner space would not
do so. The effect of the moves would not be equal everywhere in the
colorspace. That is one reason to why we always should convert to a
workingspace. The device independence is another.

Use the internal calibration system to be able to use the on-the-fly
colorsystem and beeing able to use the scannersoftwares own tools ON
TOP of that.

Use a third party profilig solution with a completely locked scanner
with no further exposure or tools variations at all. Just hit the file
with an ASSIGN the ICC profile in Photoshop and then CONVERT into a
deviceindependent workingspace like Adobe RGB. From that edit to taste
and creative goals.

nikita

 

[Philip Homburg]

I'm sorry, but I don't understand this last paragraph. In Vuescan there
is analog gain for red blue and green channels. Ed Hamrick in the manual
says that analog gain settings really relate to increase CCD exposure
time, which I understand. But I don't understand what white and grey has
to do with that.

You can use the analog gain for color correction if one channel is way off.
If what is supposed to be white on a slide comes out with a red color cast,
it might be a good idea to reduce the analog gain for the red channel.

However, if correcting the red in the whites creates a cyan color cast in
the greys then there is some non-linear effect that cannot be corrected
with the analog gain.

Futhermore, an analog gain correction of less than one stop gives you
less than one bit of extra information. When scanning in 16-bit mode,
one extra bit is probably not worth the trouble.