proper way to convert from linear gamma negatives?

[false_dmitrii]

After neglecting it for far too long, I have everything I need to use
my scanner in low-dust, low-light, calibrated-monitor safety. So I'm
*finally* ready to put it to serious use.

I think I have the hang of Bart vdW's exposure gain approach to
removing the negative mask. Colors are starting to look very good.
However, I'm not sure how to best approach the contrast.

I've been trying out Picture Window Pro for a while, since it has all
the higher-end features I'd need (16-bit functions, large image, color
management) and costs much less than Photoshop. All I've been doing so
far is a simple negative inversion followed by a gamma adjustment.
However, I think I'm doing the gamma all wrong. The negatives are
scanned as 16-bit linear. I tried PWP's global gamma conversion
command, going from 1 to 2.2, but this leaves images looking a bit
dark, shadowy, and flat. Should I be using a different fixed value
(such as 1.8 or 2.5), or is it better to bring up the curves or levels
control instead and eyeball all contrast changes from the outset?

In other words, is there a "proper" gamma value that will pull the
linear negative image toward its "natural" contrast range, thus giving
a more "true to life" starting point for further adjustments--the
equivalent of scanning a slide at gamma 2.2 or thereabouts? Or do
negatives vary too widely for any one value to suffice?

Hope I communicated this properly. Thanks in advance.

false_dmitrii

 

[false_dmitrii]

Do I just have the gamma value backwards, the film being *negative*?
Is it closer to 1/2.2=.4545, plus a white point tweak? Looks much
better that way, though still somewhat dark.

false_dmitrii

 

[false_dmitrii]

Do I just have the gamma value backwards, the film being *negative*?
Is it closer to 1/2.2=.4545, plus a white point tweak? Looks much
better that way, though still somewhat dark.

Looks like it's more a case of PWP using the inverse of PS Elements 3's
gamma values for whatever reason. So the original question remains:
is there a "default" gamma value to use when converting from linear
gamma 1.0 after inverting a negative? In a Google search, I turned up
..60 as a possible starting point, and it seems to come pretty close to
a natural appearance. It's still necessary to bring the white point in
to get the full dynamic range, and there's still some color curve
tweaking to be done...does it sound like I'm on the right track?

false_dmitrii

(Way OT and a bit late, but it didn't cross my mind at first: I hope no
one from this rather international NG was directly affected by the
London attacks. My sympathies to anyone who was.)

 

[Kennedy McEwen]

Looks like it's more a case of PWP using the inverse of PS Elements 3's
gamma values for whatever reason. So the original question remains:
is there a "default" gamma value to use when converting from linear
gamma 1.0 after inverting a negative?

Not really. The issue is at least three-fold.

Firstly, the inversion process in Photoshop (and I assume PWP) assumes
that you are working in perceptual space to begin with, so that
inversion simply means the data is subtracted from the white level
rather than being mathematically inverted to 1/x. This means that you
need to apply a gamma compensation of something like 2.2 before you
invert the negative image in the first place. You will probably find
that this produces a much closer approximation to a proper negative scan
in the first place, and you might not need to go any further.

However the next issue is that the effect of gamma is most dominant in
the shadow region, so it is critical to define the black level in order
to get the correct gamma. Of course, since you are working with a
negative, the black level is actually the highlight prior to inversion
and consequently dependent on your exposure and how well you actually
determine and remove the orange mask and base fog level on your image.

Finally, there isn't really a fixed gamma for film as such - there is
another curvature in the opposite direction at the highlight end, so the
film response is really an "S" curve.

In a Google search, I turned up
.60 as a possible starting point, and it seems to come pretty close to
a natural appearance. It's still necessary to bring the white point in
to get the full dynamic range, and there's still some color curve
tweaking to be done...does it sound like I'm on the right track?

Could be...

I just tried a little test on some figures based on the Characteristic
Curve given for the green layer of Fuji Sensia Reala 100, in the Fuji
Film Datasheet. This shows a log-log plot of dye density against
exposure.

Transferring that data manually into Excel allowed me to play around
with it and apply different corrections to see which produced the most
linear fit between exposure and final image density. Applying a simple
gamma curve with a value of around 0.7 (which may be 1.4, depending on
whether PWP uses gamma directly or gamma compensation) produces the most
straight line. A value of 0.6 pulls the shadows up too much with a
convex up curve, whilst a value of 0.8 keeps the shadows too low with a
convex down curve. So based on that, I would go with 0.7 (or 1.4) for
the gamma of Fuji Superia Reala 100.

Other films may be different of course, and indeed the gamma looks
slightly different on the other colours on Fuji Superia Reala, but I
haven't checked those yet. Nor have I tried going through the rigmarole
of positive scanning negatives, as I get good results directly on my
scanner.

(Way OT and a bit late, but it didn't cross my mind at first: I hope no
one from this rather international NG was directly affected by the
London attacks. My sympathies to anyone who was.)

On the fairly regular occasions when I work at our head office or have
meetings with MOD, I would be travelling on the Circle Line from
Liverpool Street just around the time of the incident there. Fortunately
I was at a meeting miles away at our detector manufacturing plant in
Southampton on the day. One of my colleagues at the same meeting lives
around Russell Square, apparently just over one of the other incidents,
so he decided to stay an extra night in Southampton rather than struggle
to get home.

By a strange co-incidence I was also at the same venue on 9/11 and
remember that the first I had heard of that was on leaving the facility
mid-afternoon to drive home. As I turned the car ignition the radio
came on and the first sound that came from it were "amounts to a
declaration of war against the USA". At first I thought I was listening
to an "Orson Welles War of the Worlds" situation on BBC Radio 4, and it
was some minutes before I realised that it was a real news report!

Like many Brits though, having lived through 25 years of Irish terrorist
activities (and lost several school-friends to that) there was no
surprise on hearing the news on Thursday.

On this occasion, I heard about it by phone from a relative working in
London. Expecting me to be in the area, he called me to make sure I was
OK and let me know he was - and suggest meeting for a beer to celebrate
Wednesday's great news that the 2012 Olympics will be a short walk and a
15 minute train ride from my front door. What a difference a day makes,
but that is the way most Brits have become accustomed to reacting to
these outrages - sympathy for the victims while we thank our luck it
wasn't us and get on with life. Anything else means *they* win.

 

[Don]

Like many Brits though, having lived through 25 years of Irish terrorist
activities (and lost several school-friends to that) there was no
surprise on hearing the news on Thursday.

On this occasion, I heard about it by phone from a relative working in
London. Expecting me to be in the area, he called me to make sure I was
OK and let me know he was - and suggest meeting for a beer to celebrate
Wednesday's great news that the 2012 Olympics will be a short walk and a
15 minute train ride from my front door. What a difference a day makes,
but that is the way most Brits have become accustomed to reacting to
these outrages - sympathy for the victims while we thank our luck it
wasn't us and get on with life. Anything else means *they* win.

Glad to hear you're OK, Kennedy!

On my various travels I have lived in London, on and off, for about
3-4 years and that included some of the Irish bombings. While they
would (at least in theory) call the authorities in advance - and these
current atrocities came unannounced - the odds of getting hurt in one
are really very miniscule. Indeed, the day after the bombings was
probably the safest day to go on the tube!

To keep things in perspective, more people die from car accidents or
various self-inflicted diseases (smoking!) etc. so worrying about a
terrorist incident - be it on the ground or in the air - is really not
rational.

Yes, we all get butterflies but if we look at the facts it really
doesn't make any sense to worry. I'm sure the number of traffic and
other "regular" fatalities was probably reduced as London came to a
standstill, so it all "balances out" in a morbid way.

Going off on a philosophical tangent - and taking a Buddhist approach
- we all have to go one day and, in the larger scheme of things,
worrying about it also doesn't make sense. Religious people should not
fear because they are "going to a better place" (or so they think...)
while devout atheist know there's nothing to fear.

Still, eons of evolution have instilled in us the urge for
self-preservation. As Woody Allen would say: I don't mind dying, I
just don't want to be there when it happens! ;o)

Condolences to the affected, of course, but the Brits' "stiff upper
lip" is certainly the motto to be followed here! So, I wholeheartedly
second Kennedy's "just get on with life".

And now we return you to our regular programming... ;o)

Don.

 

[false_dmitrii]

(e-mail address removed) writes

Firstly, the inversion process in Photoshop (and I assume PWP) assumes
that you are working in perceptual space to begin with, so that
inversion simply means the data is subtracted from the white level
rather than being mathematically inverted to 1/x. This means that you
need to apply a gamma compensation of something like 2.2 before you
invert the negative image in the first place. You will probably find
that this produces a much closer approximation to a proper negative scan
in the first place, and you might not need to go any further.

Okay, sounds good. When performing perceptual inversion, does it
matter if there's a big gap between the negative image's black point
and the start of the image data? Would it make more sense to correct
this, either globally or per-channel, before rather than after
inversion?

However the next issue is that the effect of gamma is most dominant in
the shadow region, so it is critical to define the black level in order
to get the correct gamma. Of course, since you are working with a
negative, the black level is actually the highlight prior to inversion
and consequently dependent on your exposure and how well you actually
determine and remove the orange mask and base fog level on your image.

Right, this was my starting point. I *think* I have this part working,
though I'll post more elsewhere.

Finally, there isn't really a fixed gamma for film as such - there is
another curvature in the opposite direction at the highlight end, so the
film response is really an "S" curve.

Aaargh, this is what I was afraid of. I was hoping the HW gain
adjustments + linear-to-monitor gamma compensation would be enough.
Once we get into the physical film behavior, my nonexistent photography
background becomes a huge obstacle. I know how to use the various
software color and contrast tools to get results I like, but I can't
realistically see myself acquiring enough knowledge to make *accurate*
judgments about the original response of the film. Not anytime soon,
at least.

My piecemeal "learn as needed" approach worked up to now, but it's at
its limit. Is there a good photography reference text you know of that
would give me enough of an understanding of gamma-related film issues
that I could apply it when scanning negatives? I know I can always
spend some money to be able to throw the negative into Silverfast's
Negafix or Vuescan, but I like knowing *what* manual adjustments to
make and *why* to make them. Right now, it feels a bit like I tried
to learn how to work a manual transmission only to find out that I
first needed to learn how to put one together.

Not that I regret having the problem presented.

Could be...

I just tried a little test on some figures based on the Characteristic
Curve given for the green layer of Fuji Sensia Reala 100, in the Fuji
Film Datasheet. This shows a log-log plot of dye density against
exposure.

Transferring that data manually into Excel allowed me to play around
with it and apply different corrections to see which produced the most
linear fit between exposure and final image density. Applying a simple
gamma curve with a value of around 0.7 (which may be 1.4, depending on
whether PWP uses gamma directly or gamma compensation) produces the most
straight line. A value of 0.6 pulls the shadows up too much with a
convex up curve, whilst a value of 0.8 keeps the shadows too low with a
convex down curve. So based on that, I would go with 0.7 (or 1.4) for
the gamma of Fuji Superia Reala 100.

I can grasp what you're saying, but I don't think I could put it into
practice on my own. I still have only a general sense of how film
gamma operates and none of the mathematical background; applying and
evaluating a true-to-film gamma curve from scratch sounds beyond me at
the moment.

Other films may be different of course, and indeed the gamma looks
slightly different on the other colours on Fuji Superia Reala, but I
haven't checked those yet. Nor have I tried going through the rigmarole
of positive scanning negatives, as I get good results directly on my
scanner.

Negative handling is definitely a low point of Dimage Scan. Do you do
many manual negative adjustments in Nikon Scan, or does the software do
most of the work automatically?

On the fairly regular occasions when I work at our head office or have
meetings with MOD, I would be travelling on the Circle Line from
Liverpool Street just around the time of the incident there. Fortunately
I was at a meeting miles away at our detector manufacturing plant in
Southampton on the day. One of my colleagues at the same meeting lives
around Russell Square, apparently just over one of the other incidents,
so he decided to stay an extra night in Southampton rather than struggle
to get home.

Ugh. After the WTC attacks I found out that a friend and his wife were
one flight removed from one of the NY-to-LA planes. It's always
unnerving to hear about such close calls, let alone live them out.
Glad you brought your luck with you that day.

On this occasion, I heard about it by phone from a relative working in
London. Expecting me to be in the area, he called me to make sure I was
OK and let me know he was - and suggest meeting for a beer to celebrate
Wednesday's great news that the 2012 Olympics will be a short walk and a
15 minute train ride from my front door. What a difference a day makes,
but that is the way most Brits have become accustomed to reacting to
these outrages - sympathy for the victims while we thank our luck it
wasn't us and get on with life. Anything else means *they* win.

On the one hand, 50-some dead out of many millions is a miniscule
amount, and as with the WTC, a small fraction of what the bombers must
have hoped for.

On the other hand, it's still 50 too many, and the harm done extends
beyond deaths caused.

But on the third hand, if China and the Soviet Republics could endure
the loss of many 10s of millions, the current targets of terrorism can
get through far worse than what's happened so far. It's not as though
our respective countries would just fold up and disappear if the
largest cities were struck down.

And AQ and other known groups can't ever "win" anyway. Half of America
could disappear, yet OBL & friends *still* could never live out in the
open without someone coming after them. No one wins these conflicts,
everyone loses....

That's all the politics I'll bring to this NG. Pure math and physics
are much more uplifting.

false_dmitrii

 

[Kennedy McEwen]

Okay, sounds good. When performing perceptual inversion, does it
matter if there's a big gap between the negative image's black point
and the start of the image data? Would it make more sense to correct
this, either globally or per-channel, before rather than after
inversion?

Well, it will make some difference to the compression or expanding of
the highlights in the final image, but it would have to be a significant
offset to make a noticeable difference because at the highlight end the
gamma approximation to perceptual space is pretty linear.

Right, this was my starting point. I *think* I have this part working,
though I'll post more elsewhere.


Aaargh, this is what I was afraid of. I was hoping the HW gain
adjustments + linear-to-monitor gamma compensation would be enough.
Once we get into the physical film behavior, my nonexistent photography
background becomes a huge obstacle. I know how to use the various
software color and contrast tools to get results I like, but I can't
realistically see myself acquiring enough knowledge to make *accurate*
judgments about the original response of the film. Not anytime soon,
at least.

Real world effects, I'm afraid - nothing is ever particularly simple
once you delve into the details. That is why the software option exists
- to hide you from all of this complexity. ;-)

My piecemeal "learn as needed" approach worked up to now, but it's at
its limit. Is there a good photography reference text you know of that
would give me enough of an understanding of gamma-related film issues
that I could apply it when scanning negatives? I know I can always
spend some money to be able to throw the negative into Silverfast's
Negafix or Vuescan, but I like knowing *what* manual adjustments to
make and *why* to make them. Right now, it feels a bit like I tried
to learn how to work a manual transmission only to find out that I
first needed to learn how to put one together.

Not that I regret having the problem presented.


I can grasp what you're saying, but I don't think I could put it into
practice on my own. I still have only a general sense of how film
gamma operates and none of the mathematical background; applying and
evaluating a true-to-film gamma curve from scratch sounds beyond me at
the moment.

Well, you don't really need to be able to do the analysis - all that
does is prove whether you are doing the right thing or not. The
empirical approach is to apply gamma of about 2.2, invert, then apply a
gamma compensation of around 0.7. Just for clarity, in case PWP applies
the inverse function, when you apply the gamma of 2.2 the negative image
will lighten, when you apply the 0.7 gamma the positive image will
darken, particularly in the shadows. In effect the film is compressing
the shadows and the 0.7 gamma stretches them back out again.

Negative handling is definitely a low point of Dimage Scan. Do you do
many manual negative adjustments in Nikon Scan, or does the software do
most of the work automatically?

I scan as a negative on the Nikon and then set the black point to the
film base level (with a small margin of safety, about 2 levels maximum).
Having done that I save the setting as a reference that I can pull up
any time I want to scan film of that type again.

<OT comments snipped. No need to feed their publicity.>

 

[Don]

Aaargh, this is what I was afraid of.

It's like reading my own posts! ;o) I, for one, *really* sympathize
with your frustration having gone through it all myself (only with
slides). And still do, actually...

My piecemeal "learn as needed" approach worked up to now, but it's at
its limit. Is there a good photography reference text you know of that
would give me enough of an understanding of gamma-related film issues
that I could apply it when scanning negatives?

I don't have specific pointers - other than to collect Kennedy's posts
;o) but searching the Net has been very valuable, albeit very time
consuming, mostly because often I didn't even know what to search for!

Overall, and in my experience, it's better to grasp the nettle early
and be done with it. Like you, I also did not have the mathematical
background or the knowledge of physics involved but after avoiding it
for a long time, in the end I just had to wrap my brain around at
least some of it.

You're probably familiar with these but here goes anyway (a random
sample off the top of my head and in no particular order):

http://www.aim-dtp.net/ - sometimes controversial site but lots of
stuff including:
http://www.aim-dtp.net/aim/download/gamma_maps.zip AMP gamma files for
Photoshop

www.computer-darkroom.com - a generic site

www.brucelindbloom.com - look for "RGB16Million.tif.zip", handy for
tests

http://www.reindeergraphics.com/free.shtml - free 12-bit histogram
Photoshop plug-in also handy for tests

....and I'm sure many more which escape me at this moment...

Right now, it feels a bit like I tried
to learn how to work a manual transmission only to find out that I
first needed to learn how to put one together.

Which - after a lot of frustration - is exactly what I ended up doing
i.e. I wrote my own software. Not that I recommend it (because it does
take a lot of time) but it was definitely very educational even though
there are still gaping holes in my knowledge

But, again, I really sympathize with what you're trying to do because
I also like to know what I'm doing rather than just blindly click on
pre-thunk buttons.

Don.

 

[Don]

I just tried a little test on some figures based on the Characteristic
Curve given for the green layer of Fuji Sensia Reala 100, in the Fuji
Film Datasheet. This shows a log-log plot of dye density against
exposure.

On a slight tangent, lately, I've been trying to figure what does
Nikon actually do with their Kodachrome option (in order to "boost"
its effects) and this raised a number of questions...

To this end, I ran a "compare" between a Kodachrome and a Positive
scan. This was done by tabulating RGB values for pixel pairs in both
images and averaging them out to produce a 16-bit look-up table (LUT).
Applying this LUT to the Positive scan then produces an image
virtually identical to the Kodachrome scan.

Next, I converted this LUT to an 8-bit AMP file so it can be "plotted"
in Photoshop for further testing and playing around.

Here's an example of such curves:

http://members.aol.com/tempdon100164833/nikon/P2K.amp

http://members.aol.com/tempdon100164833/nikon/P2K-R.jpg
http://members.aol.com/tempdon100164833/nikon/P2K-G.jpg
http://members.aol.com/tempdon100164833/nikon/P2K-B.jpg

N.B.
1. All this in gamma 2.2 and, originally, in 16-bit.
2. Some bins are missing even after conversion to 8-bit. Such bins
have the value of 0 in the AMP files. Usually, this is visible in the
dark areas.

A couple of interesting points jumped at me. In spite of the curves
appearing fairly linear there are noticeable deviations at extremes.
At first, I attributed this to noise/clipping respectively, but after
normalizing into a straight curve and applying that, the result was
vastly inferior. Therefore, it appears these are not deviations but
quite intentional.

On closer inspection, the curves also have a slight, well, curvature
in that they bulge upwards in the middle. Again, this is not a
side-effect of conversion etc. but appears quite intentional.

Initially, the idea was to replace these curves with the corresponding
analog gain (AG) boost but due to these "anomalies" (ends, bulging)
that approach would require additional processing.

Anyway, do these curves remind you of anything? Other than the obvious
i.e., a compare of Kodachrome to Positive... ;o)

What are the reasons for the deviations at either end? Again, assuming
they have a deeper significance. Or the slight "gammaesque" bulging?

Finally, any suggestions about creating a generic Kodachrome curve out
of this data/method, other than the obvious averaging of a number of
samples? My preference, as always, is for an empirical,
self-correcting approach to avoid the infinite theoretical
permutations and ambushes... ;o)

As I say, the idea is to amplify the effect because some slides need
"more Kodachrome". Curiously, slides where nominal exposure
(*absolute* AG = 0) is optimal seem perfect with the default Nikon
Kodachrome correction. However, for slides where optimal exposure is
higher (e.g. *absolute* AG = 2.0) a seemingly linear "Kodachrome
boost" is required to achieve the same color balance as at AG=0. (All
those conclusions, however, are still very preliminary.) I attribute
this to the characteristic film curve e.g.
http://www.kodak.com/global/en/professional/support/techPubs/e55/f002_0486ac.gif
but do not know this for sure.

Don.

 

[Kennedy McEwen]

http://members.aol.com/tempdon100164833/nikon/P2K-R.jpg
http://members.aol.com/tempdon100164833/nikon/P2K-G.jpg
http://members.aol.com/tempdon100164833/nikon/P2K-B.jpg

Anyway, do these curves remind you of anything?

Maybe no surprise, but they remind me of the correction for a
characteristic curve for Kodachrome. ;-)

http://www.kodak.com/global/en/professional/support/techPubs/e55/f002_04
86ac.gif

Look, in particular, at the higher density of the red layer, and compare
that to how the shadows are lifted more for red in your curves above.

 

[Mike Engles]

After neglecting it for far too long, I have everything I need to use
my scanner in low-dust, low-light, calibrated-monitor safety. So I'm
*finally* ready to put it to serious use.

I think I have the hang of Bart vdW's exposure gain approach to
removing the negative mask. Colors are starting to look very good.
However, I'm not sure how to best approach the contrast.

I've been trying out Picture Window Pro for a while, since it has all
the higher-end features I'd need (16-bit functions, large image, color
management) and costs much less than Photoshop. All I've been doing so
far is a simple negative inversion followed by a gamma adjustment.
However, I think I'm doing the gamma all wrong. The negatives are
scanned as 16-bit linear. I tried PWP's global gamma conversion
command, going from 1 to 2.2, but this leaves images looking a bit
dark, shadowy, and flat. Should I be using a different fixed value
(such as 1.8 or 2.5), or is it better to bring up the curves or levels
control instead and eyeball all contrast changes from the outset?

In other words, is there a "proper" gamma value that will pull the
linear negative image toward its "natural" contrast range, thus giving
a more "true to life" starting point for further adjustments--the
equivalent of scanning a slide at gamma 2.2 or thereabouts? Or do
negatives vary too widely for any one value to suffice?

Hope I communicated this properly. Thanks in advance.

false_dmitrii

Hello

If you have a Nikon scanner, you can do the inversion in NikonScan.
Just pull up and down the high and low points on the diagonal line. Do
not use the ARROW points.
Try ROC and you will have to use some HUE shift as reds tend to be deep
magenta. Use the midtone picker to set the mid grey.
I have screen shot.

http://www.btinternet.com/~mike.engles/mike/NscanINV.jpg

Mike Engles

 

[Don]

Maybe no surprise, but they remind me of the correction for a
characteristic curve for Kodachrome. ;-)

So, it's like I suspected... ;o)

Look, in particular, at the higher density of the red layer, and compare
that to how the shadows are lifted more for red in your curves above.

Yes, I noticed that!

BTW, what do you think of the procedure in general (i.e. comparing a
few scans and creating a generic curve out of that - also, see below)?

How would you "boost" such a curve? Just lift the middle of each
channel or change the ratio in some way?

Another more complicated way is to abandon Nikon's Kodachrome option
completely, scan as positive, and then go through the torture of
manual adjustments, followed by an averaging of several such changes
to come up with a characteristic curve for my particular Kodachromes.

I like the self-correcting nature of that, but absolutely loath having
to, first, agonize about the subjective decisions needed to make the
image look like the slide... :-(

Don.

 

[Kennedy McEwen]

BTW, what do you think of the procedure in general (i.e. comparing a
few scans and creating a generic curve out of that - also, see below)?

If you used a Q60 or IT-8 slide, I would call it a film profile. ;-)

How would you "boost" such a curve? Just lift the middle of each
channel or change the ratio in some way?

If you just lift the middle then you will only change the colour balance
in the middle of the density range. If you want to change the balance
across the density range then you have to change the overall ratio - and
you could do that with the analogue gain though.

 

[Andreas]

Don, you wrote your own scanning software. If it's better than V******,
can't you share it with others?

I'm more or less happy with V****** for scanning negatives. I'm less
happy with it when it comes down to slides and I'm willing to embrace
any other piece of software that does them right. (This eliminates the
Minolta-software that scans at a lower resolution and results in
clipping.) What's the status of your own scanning program?

A.

(add a t after the d in the email adress)

 

[Don]

Hi Andreas,

It's not really for public consumption and I don't think it would be
useful for most people because my main goal was to get a raw scan
without "interference" which most other scanner software seems to
insist on and is often difficult to turn off. However, many people
actually want their scanner software to "do everything".

So, it's just something I wrote for myself. Although, I did add a few
(for me) handy features like absolute exposure (instead of relative
based "auto"), a focus matrix (to get the most out of curved film),
individual channel gamma (for playing around with Kodachromes),
sub-pixel alignment for high dynamic range, 16-bit histogram, etc.

My dislike for Vuescan is well known here because of its many problems
and it's not really comparable to what I wrote because the goals are
quite different.

What problems exactly are you having with slides?

I don't know Minolta software but it should be able to scan at optical
resolution (anything above optical resolution is not real and can be
done later - and much better! - in an image editing software). As for
clipping, that's usually a problem of auto-exposure. You need to
adjust the % of clipping or, better still, turn auto-exposure off
completely and set the exposure manually.

Don.

 

[Don]

If you used a Q60 or IT-8 slide, I would call it a film profile. ;-)

It wouldn't be the first time I "invented" something that has already
invented... ;o)

However, I found the above "image comparison" (whereby I extract a
curve needed to convert from one into the other) very revealing in
many other instances and it's one of my main tools now!

If you just lift the middle then you will only change the colour balance
in the middle of the density range. If you want to change the balance
across the density range then you have to change the overall ratio - and
you could do that with the analogue gain though.

That's actually why I asked because after all the many (and elaborate)
wild goose chases I always keep getting back to analog gain which is
where I came in 3 (!) years ago... :-(

I found recently that boosting analog gain of a positive scan and then
setting the gray point (hence the "lifting the middle" reference
above) produces much better images than setting the gray point of a
Nikon's "Kodachrome mode" scan!

Upon closer inspection the Nikon's Kodachrome mode graypoint adjusted
scan actually has a moderate red tint! The analog gain corrected and
graypoint adjusted scan, by comparison, is much purer and much truer
to the original.

The catch, however, is jamming the "square peg" of linear analog gain
into the "round hole" of a non-linear characteristic film curve. :-/

Don.

 

[Andreas]

Surprise: the Minolta-software for my scanner (a Scan Dual II, USB) is
not able to deliver scans in the native resolution of the scanner, 2820
dpi. Yes, for some strange reason you can set the "input"-dpi to 2820
dpi but the output (that's what you get) is 2400 dpi at most.

One of the problems I have with V****** is clipping (when scanning
slides; negatives are fine in this respect) but I don't want to
elaborate on this here in the newsgroup when I did not report it to the
author.

 

[Mendel Leisk]

Andreas, I was able to up the output dpi to 5400 (from the software's
limit of 4000) on my Scan Elite 5400), by directly editting a settings
file. I forget the name/extension, but it's findable. It's in the
program directory, down in a sub-directory. and is one of the most
recent files. I think the word "pref" may be in it, or something like
that.

Save a .old version, just to be safe.

 

[Kennedy McEwen]

Surprise: the Minolta-software for my scanner (a Scan Dual II, USB) is
not able to deliver scans in the native resolution of the scanner, 2820
dpi. Yes, for some strange reason you can set the "input"-dpi to 2820
dpi but the output (that's what you get) is 2400 dpi at most.

I vaguely recall someone raising this issue with the SD-II in the dim
distant past, but I couldn't find a reference to it on Google.

IIRC, the input-dpi *is* what you actually get: that is what the scanner
captures the data at. The output resolution is merely a single number
that the scanner driver writes to the resulting file and is, in almost
all cases, completely irrelevant. There isn't much call for 1:1 digital
reproduction of 35mm frames. ;-)

The "output-dpi" is merely a scaling figure, it has nothing whatsoever
to do with the resolution the scanner operates at. If you examine the
image size in your output file you will find that at 2400dpi it is
actually larger than the original frame.

There is a similar discussion about such useless output resolution
figures going on in the thread called "Image does not report the same
resolution at which it is scanned" at the moment. It is superfluous
information in all but a few special cases.

 

[Mendel Leisk]

here's the exact path/file on my pc:

C:\Program Files\DiMAGEScan\DS_Elite5400\Job\Custom\Full resolution no
crop.135