better scans with minolta 5400

[keith taylor]

just thought i would post my nugget of information.

when i first purchased my 5400 about 2 1/2 yrs ago i was ever swapping
between minolta software and vuescan, i was never really that impressed
with the colours from the minolta software and it also caused a few blue
screen crashes when i pushed my pc with other processes. eventually i
bought one of wolf faust colour profiles for velvia stuck it in vuescan
and got much better colours so i moved over to using vuescan, missing
out occasional releases which had serious bugs with my scanner, working
round the minor bugs. i was always very happy with vuescan, nothing is
perfect after all!

a few months ago i sent some medium format trans off to be scanned (i
can't afford, or justify a medium format scanner) the colours came back
quite bad, but the scan quality was good sharp, lots of detail. i tried
adjusting colours in ps with limited success. i then tried using the
tiffs as raw sans and bringing them into vuescan, hey presto it was very
easy to get really very good colours out a revelation indeed.

along these lines i thought i might give the minolta software 1.1.5
another go. although this is quite a tedious workflow i find that i'm
able to get cleaner, sharper and more colour accurate scans with
considerably more dynamic range than with using vuescan alone.

1: scan with minolta software manual focus, (this makes quite a
difference, especially with landscape shots, macro shots seem to always
come out sharper), ice on which in turn switches grain dissolver on.
don't adjust any colours at this stage.

2: save final scan then colour adjust it in vuescan, re-save.

3: bring into ps and adjust levels, final bit of touching up.

i know this may seem a bit lengthy, but i've never found the grain
dissolver to work particularly well with vuescan and it does clean
images up nicely. i thought ice & gd as implemented by the minolta
software caused scans to come out a little soft, however when you focus
manually they come out pin sharp.

hope this may be of help to a few folk out there still scanning slides,
now i've found a way of getting better scans i may hold off the
increasing urge to go digital till canon make a 12 megapixel camera for
under £1000. now all i need to do is go back and re-scan lots of slides
which i just rattled through vuescan which lack dynamic range and are
not quite a sharp as they could be at 100%

keith

 

[Robert Feinman]

just thought i would post my nugget of information.

when i first purchased my 5400 about 2 1/2 yrs ago i was ever swapping
between minolta software and vuescan, i was never really that impressed
with the colours from the minolta software and it also caused a few blue
screen crashes when i pushed my pc with other processes. eventually i
bought one of wolf faust colour profiles for velvia stuck it in vuescan
and got much better colours so i moved over to using vuescan, missing
out occasional releases which had serious bugs with my scanner, working
round the minor bugs. i was always very happy with vuescan, nothing is
perfect after all!

a few months ago i sent some medium format trans off to be scanned (i
can't afford, or justify a medium format scanner) the colours came back
quite bad, but the scan quality was good sharp, lots of detail. i tried
adjusting colours in ps with limited success. i then tried using the
tiffs as raw sans and bringing them into vuescan, hey presto it was very
easy to get really very good colours out a revelation indeed.

along these lines i thought i might give the minolta software 1.1.5
another go. although this is quite a tedious workflow i find that i'm
able to get cleaner, sharper and more colour accurate scans with
considerably more dynamic range than with using vuescan alone.

1: scan with minolta software manual focus, (this makes quite a
difference, especially with landscape shots, macro shots seem to always
come out sharper), ice on which in turn switches grain dissolver on.
don't adjust any colours at this stage.

2: save final scan then colour adjust it in vuescan, re-save.

3: bring into ps and adjust levels, final bit of touching up.

i know this may seem a bit lengthy, but i've never found the grain
dissolver to work particularly well with vuescan and it does clean
images up nicely. i thought ice & gd as implemented by the minolta
software caused scans to come out a little soft, however when you focus
manually they come out pin sharp.

hope this may be of help to a few folk out there still scanning slides,
now i've found a way of getting better scans i may hold off the
increasing urge to go digital till canon make a 12 megapixel camera for
under £1000. now all i need to do is go back and re-scan lots of slides
which i just rattled through vuescan which lack dynamic range and are
not quite a sharp as they could be at 100%

keith

I don't see why you need the vuescan step, what adjustments are making
at this point? Why not just go straight to
Photoshop? The problem with the Minolta software is that if you make
*any* adjustments it will clip the dynamic range and process as 8 bit
even if you save as 16 bit. If you are just scanning with no adjustments
then you won't have this problem.

 

[keith taylor]

I don't see why you need the vuescan step, what adjustments are making
at this point? Why not just go straight to
Photoshop? The problem with the Minolta software is that if you make
*any* adjustments it will clip the dynamic range and process as 8 bit
even if you save as 16 bit. If you are just scanning with no adjustments
then you won't have this problem.

robert, you don't really need the vuescan step, but it makes life for me
easier than a lot a colour fiddling in ps. vuescan's poor dynamic range
is direct from the scanner not when you import a tiff. i'm only dealing
with 8bit files i really don't see the point with 16bit.

i could go direct to photoshop, but i find the colour management very
easy to use in vuescan, adjusting the red, green and blue colour
exposure settings.

keith

 

[Mendel Leisk]

Kieth, I'm doing the same, with same scanner. Output 16 bit linear in
MSU, use it for scan-from-disk with Vuescan. Or something just work
them in PS.

With some color neg scans I'm doing through this MSU/Vuescan workflow,
I get much improved highlight definition, compared to converting the
16bit lin in PS. This is using "generic color neg" profile.

When scanning color negs in MSU, I start by scanning leader without
auto-exposure, and get the rgb levels all just under 255.

 

[JJSrock]

I don't see why you need the vuescan step, what adjustments are making
at this point? Why not just go straight to
Photoshop? The problem with the Minolta software is that if you make
*any* adjustments it will clip the dynamic range and process as 8 bit
even if you save as 16 bit. If you are just scanning with no adjustments
then you won't have this problem.

I see no evidence that making adjustments with the Minolta 5400 sw will
"clip the dynamic range and process as 8 bit even if you save as 16
bit". Specifically, adjust the Exposure Control tab and you can see
whether any channel is clipped in the histograms. How did you know the
sw is processing in 8 bit mode?

 

[Wilfred]

Robert Feinman wrote:




I see no evidence that making adjustments with the Minolta 5400 sw will
"clip the dynamic range and process as 8 bit even if you save as 16
bit". Specifically, adjust the Exposure Control tab and you can see
whether any channel is clipped in the histograms. How did you know the
sw is processing in 8 bit mode?

If you check the histograms in an image editor you will see 'combing'
(for an example of combing not created with the Minolta DSU, look here:
http://www.sphoto.com/techinfo/histograms/histograms3.htm, at the bottom
of the page). This is a typical sign of 8-bit processing.
The Minolta software also tends to clip the highlights, especially with
negatives; recently I found that it even does this with slides scanned
in 16-bit linear mode.

 

[Hecate]

If you check the histograms in an image editor you will see 'combing'
(for an example of combing not created with the Minolta DSU, look here:
http://www.sphoto.com/techinfo/histograms/histograms3.htm, at the bottom
of the page). This is a typical sign of 8-bit processing.
The Minolta software also tends to clip the highlights, especially with
negatives; recently I found that it even does this with slides scanned
in 16-bit linear mode.

So combing is a typical sign of 8 bit processing. That would be why my
16 bit images get combing when using adjustment layers in Photoshop
then....

Yeah, right.

--

Hecate - The Real One
(e-mail address removed)
Fashion: Buying things you don't need, with money
you don't have, to impress people you don't like...

 

[Wilfred]

So combing is a typical sign of 8 bit processing. That would be why my
16 bit images get combing when using adjustment layers in Photoshop
then....

Well, my 16-bit images from the 5400 don't get combing when using
adjustment layers in PS, which is something I do frequently. But then
again, I'm using VueScan and you're using the DSU.
Apparently, the DSU slightly corrupts the 16-bit data even when no
adjustments are made in the scanning process (assuming that's what
you're doing).

 

[JJSrock]

Well, my 16-bit images from the 5400 don't get combing when using
adjustment layers in PS, which is something I do frequently. But then
again, I'm using VueScan and you're using the DSU.
Apparently, the DSU slightly corrupts the 16-bit data even when no
adjustments are made in the scanning process (assuming that's what
you're doing).

What you seem to be missing is that the PS histograms are only
*displaying* 256 values (i.e. 8-bits) whether the image is 8- or 16
bits. Tools like Levels and Curves also only *display* or let you
manipulate 256 values, while they actually *work* on a 16-bit image.

If Robert Feinman is based his claim on viewing histograms, he is dead
wrong.

 

[Wilfred]

What you seem to be missing is that the PS histograms are only
*displaying* 256 values (i.e. 8-bits) whether the image is 8- or 16
bits. Tools like Levels and Curves also only *display* or let you
manipulate 256 values, while they actually *work* on a 16-bit image.

If Robert Feinman is based his claim on viewing histograms, he is dead
wrong.

If 8-bits data is seriously manipulated, combing will be visible in an
8-bit histogram (and of course also in a high-bit histogram, such as the
one provided by the wide histogram plugin,
http://www.reindeergraphics.com/free.shtml#widehisto)

If 16-bits data is seriously manipulated, combing *will* be visible in a
16-bit histogram, but typically *not* in an 8-bit histogram (unless the
manipulation is severe).
This is not so problematic, because an 8-bit histogram is supposed to
cover the number of colors that the human eye can see. If the 8-bits
histogram is smooth you can expect that there's no posterization.
The surplus that 16 bits provide gives us room for manipulation.

BTW, the reason why Hecate noticed combing when using adjustment layers
on 16-bit images may also have been that she used a plugin for high-bit
histograms.

 

[Hecate]

Well, my 16-bit images from the 5400 don't get combing when using
adjustment layers in PS, which is something I do frequently. But then
again, I'm using VueScan and you're using the DSU.
Apparently, the DSU slightly corrupts the 16-bit data even when no
adjustments are made in the scanning process (assuming that's what
you're doing).

Actually, I realised you'd assume that and I was talking about a 16
bit RAW from a digital camera. So, the answer is no. It depends on
what you do the image as to whether you get combing or not. And yes,
I've seen combing on some 16 bit images. So your assumption is
incorrect.

--

Hecate - The Real One
(e-mail address removed)
Fashion: Buying things you don't need, with money
you don't have, to impress people you don't like...

 

[Wilfred]

Actually, I realised you'd assume that and I was talking about a 16
bit RAW from a digital camera. So, the answer is no. It depends on
what you do the image as to whether you get combing or not. And yes,
I've seen combing on some 16 bit images. So your assumption is
incorrect.

OK, so forget what I said beginning with 'Apparently'. That was based on
my false assumption.
Of course you can get combing with 16 bit data. Whether you'll get it or
not depends on the quality of the data and the amount of 'stretching'
you perform on it.
If I get combing after making tonal adjustments within the Minolta
software, and do *not* get combing when making tonal adjustments
afterwards, then something must be wrong with the way the Minolta
software performs its calculations. To me, the most likely assumption is
that it does its calculation with less than 16 bits, but it is possible
that its algorithms are flawed for other reasons. My conclusion is
"don't do curves and levels in the Minolta software".

 

[Hecate]

OK, so forget what I said beginning with 'Apparently'. That was based on
my false assumption.
Of course you can get combing with 16 bit data. Whether you'll get it or
not depends on the quality of the data and the amount of 'stretching'
you perform on it.

Yes. Which is why you *cannot* make the statement that "combing means
that the Minolta software is only using 8 bits". Glad we sorted that
out

If I get combing after making tonal adjustments within the Minolta
software, and do *not* get combing when making tonal adjustments
afterwards, then something must be wrong with the way the Minolta
software performs its calculations. To me, the most likely assumption is
that it does its calculation with less than 16 bits, but it is possible
that its algorithms are flawed for other reasons. My conclusion is
"don't do curves and levels in the Minolta software".

Whilst I wouldn't like to comment on this as I don't have any evidence
one way or the other, I completely agree with your conclusion and
would go on to apply that to most scanning software. I find it much
easier and more precise to do those corrections in Photoshop.

--

Hecate - The Real One
(e-mail address removed)
Fashion: Buying things you don't need, with money
you don't have, to impress people you don't like...