Scan workflow for negatives using Minolta 5400

[Patrick P.]

Hello all,

I've used the 5400 for almost a year now with slides but not with
negatives. I'm using the Minolta software and have custom profile for
my slides. I've got a load of negatives I want to scan but before I
begin I thought I'd ask what the best method is to get the most out of
the negatives.

I've read that the 5400 isn't very good with negatives. Is it better
to scan as positive and invert in PS? If so what's the best way to do
this?

What settings should I use in the Minolta software? Auto Exposure,
color settings and all that.

Of course I'm assuming that I can't use any of the profiles I created
with my IT8 target slides.

Thanks,
Patrick

ps- color negs.

 

[Bart van der Wolf]

Hello all,

I've used the 5400 for almost a year now with slides but not
with negatives. I'm using the Minolta software and have
custom profile for my slides.

Just to be able and understand better, did you create it yourself?

I've got a load of negatives I want to scan but before I begin
I thought I'd ask what the best method is to get the most out
of the negatives.

"The best method" also depends on the intended use, but I'll assume
you want to extract all information the film has to offer. That will
allow you to store the originals in a safe place, and use the scans as
a digital negative.

I've read that the 5400 isn't very good with negatives. Is it
better to scan as positive and invert in PS? If so what's the
best way to do this?

In my experience the Minolta Scan Utility tends to clip the shadows
and/or the highlights, the scanner itself has no problems with
negatives.

What settings should I use in the Minolta software? Auto
Exposure, color settings and all that.

I suggest, others may have different preferences, the following
settings:
- Check, Auto expose for slides. That will get the exposure in the
direction you need, but you'll override this by adding a bias later.
- Uncheck, Auto focus at scan. The film base density makes it more
difficult to focus, and I want to make sure the focus is done where I
want it.
- Select, Color depth 16-bit linear. This will allow to adjust gamma
without too much problems later.
- Select, Prescan size large. Because this will allow more accurate
sampling colors in the following steps.
- Uncheck, Color matching on. This will avoid uncontrollable color
shifts.
So far for the Preference settings.

Now, position the film in the strip holder so you can see a bit of the
un-exposed inter-image space, and Preview (Scan as "Color positive").
Focus by placing the Autofocus point on a light film area with some
fine detail or a sharp edge. You may then want to crop the image to at
least exclude the dark mask, but keep the un-exposed bit of film
inside the crop area.

Now switch to the "Exposure control" Tab and adjust the Master, R, G,
and B exposure sliders in order to obtain a reading like 254,254,254
on the preview of the unexposed film part. I usually start with the
Master control set to +2, so the other controls need less adjustment.
Hit the "Apply settings to preview image" between changes.

Now reposition the film in the holder to exclude anything but the
image, and do another Preview, refocus, and scan. The result will be a
negative image with a neutral film base color.

In Photoshop, read the file, ignore any warnings about missing
profiles (don't assign one), and invert it. You can save this as your
digital negative, or you can first use Levels to adjust the highlights
slider control for the individual colors without clipping them
(pressing the Alt/Option key while adjusting will show the clipping
when it occurs), then save.

You can then try and Apply the default Posi Linear gamma profile for
the Minolta Scan Elite (don't use the other ones higher up the list),
followed by curves adjustments for Gamma, etc.

Of course I'm assuming that I can't use any of the profiles
I created with my IT8 target slides.

Probably not, but perhaps you can create a new profile based on your
Linear workflow after gamma adjustment. I'll need more details to give
better suggestions.

Bart

 

[silverfox]

Before applying this workflow, do you crop the scan area to exclude
misleading areas such as the unexposed film area or the corners of a
slide mount? After setting the preferences and control, can you crop the
scan area again (to include the slide mount corners for a full frame,
for example) before scanning? What changes would you make in this
workflow to scan slides? Thanks.

 

[Bart van der Wolf]

Before applying this workflow, do you crop the scan area to
exclude misleading areas such as the unexposed film area or
the corners of a slide mount?

Only after setting the gain/channel exposure to produce a neutral
filmbase color to a maximum, non-clipping, exposure level. Before the
actual scan I re-adjust the (Color) Negative film in the holder, and
the exact cropping.
Slides don't need this kind of workflow, because they don't get
clipped by default like CN film does.

After setting the preferences and control, can you crop the
scan area again (to include the slide mount corners for a full
frame, for example) before scanning?

If you want to include the borders, I'd first uncheck the Auto
exposure for slides. Depending on how the software deals with edges,
it may try to increase total exposure, which could blow-out / clip the
most transparent image areas. On the other hand, if your image is
heavily overexposed, there's probably a benefit for a slightly higher
overall exposure level, because there are no image areas as
transparent as the film base to begin with.

What changes would you make in this workflow to scan slides?

I wouldn't use it for slides, other than tuning the overall exposure
level to just avoid clipping of the highlights you want to keep.
Unfortunately there is no clipping indicator option in the Minolta
software, so you'll have to use your own judgement.

Bart

 

[silverfox]

Only after setting the gain/channel exposure to produce a neutral
filmbase color to a maximum, non-clipping, exposure level. Before the
actual scan I re-adjust the (Color) Negative film in the holder, and
the exact cropping.
Slides don't need this kind of workflow, because they don't get
clipped by default like CN film does.


If you want to include the borders, I'd first uncheck the Auto
exposure for slides. Depending on how the software deals with edges,
it may try to increase total exposure, which could blow-out / clip the
most transparent image areas. On the other hand, if your image is
heavily overexposed, there's probably a benefit for a slightly higher
overall exposure level, because there are no image areas as
transparent as the film base to begin with.

I have no intention to include the slide mount borders in a scan. But
the paper slide mounts have round corners and fuzz along the borders.
These are unavoidable for a full frame scan, and my question is how to
deal with them. I typically would crop these out in the preview,
manually set the exposure and other adjustments, crop again in full
frame, keep the settings the same and scan.

I wouldn't use it for slides, other than tuning the overall exposure
level to just avoid clipping of the highlights you want to keep.
Unfortunately there is no clipping indicator option in the Minolta
software, so you'll have to use your own judgement.

Would you still use depth 16-bit linear for scanning slides?

 

[Bart van der Wolf]

Would you still use depth 16-bit linear for scanning slides?

Yes, linear should be 14-bit/channel or more, if you don't want
posterization problems from postprocessing after gamma adjustment. You
also need 16-bits to represent the dynamic range from slide films
(without DR compression) in the first place.

Bart

 

[WD]

Bart,

For color negatives, after you capture the negative (as a positive),
do you apply gamma/inversion curves by 'eye' or do you attempt to
use a true inverse film curve? (It seems it would be hard to
know how to 'align' such a curve with the captured data)

W

 

[Bart van der Wolf]

Bart,

For color negatives, after you capture the negative (as a positive),
do you apply gamma/inversion curves by 'eye' or do you attempt to
use a true inverse film curve? (It seems it would be hard to
know how to 'align' such a curve with the captured data)

First invert, then use Channel white points for overall color balance.
You then adjust gamma (or curves) to compress the brightness range so
it fits in the limited monitor/print dynamic range. The original scene
probably has such a high luminance range that an exact true inverse
film and display curve would not fit in the 16-bit/channel file (let
alone 8-b/ch output). Compression is inevitable, so use your
creativity to achieve a natural looking impression. Starting with a
gamma adjustment should get you in the right direction, but for some
images a logarithmic mapping works better.

Accurate colors are a whole different ball-game than tonemapping, but
they have more to do with the saturation of the primaries, and
perception.

Bart

 

[Patrick P.]

Hello,

Thanks Bart for sharing your workflow. Sorry I've been out of the
thread but I only just got around to trying out this workflow (busy at
work).

"The best method" also depends on the intended use, but I'll assume
you want to extract all information the film has to offer. That will
allow you to store the originals in a safe place, and use the scans as
a digital negative.

Yes, that would be ideal!

I've been unable to get a good result with your method. Perhaps I'm
doing something wrong or just not good enough with PS. Here are a few
questions-

If I set the master control to +2 all the histogram on the other
channels fall off the edge. I assume I don't want this. I fiddle until
I can get all 3 to 254, without any bit of the histogram cut off.
Right?

When you say invert the image in Photoshop all I'm doing is flipping
the curves. Is there a different way to do this?

In levels if I adjust the individual colors R G and B, I end up with
color cast. All I do is look at the histogram and move the black point
and white point sliders to either edge of the graph. I sometimes use
the alt key to make sure I don't clip. Is this right? Normally when
I'm scanning slides I'd never go into the individual channels unless I
was fixing a color cast.

The posi linear profile now really makes things bad.

I've been using Photoshop for a while now but today I really feel like
a beginner.

Just to be able and understand better, did you create it yourself?
Probably not, but perhaps you can create a new profile based on your
Linear workflow after gamma adjustment. I'll need more details to give
better suggestions.

Yes I made them myself with an IT8 target using Little CMS (I also
have MonacoEZ).

First invert, then use Channel white points for overall color balance.
You then adjust gamma (or curves) to compress the brightness range so
it fits in the limited monitor/print dynamic range. ---
Starting with a
gamma adjustment should get you in the right direction, but for some
images a logarithmic mapping works better.

Using the White and black points do correct the color cast and
adjusting the overall gamma does help the image but it still isn't
right. Not sure how to do a logarithmic mapping. Do you have any links
for information on this?

Accurate colors are a whole different ball-game than tonemapping, but
they have more to do with the saturation of the primaries, and
perception.

Bart

I was hoping to avoid the guessing of accurate colors. Does this
workflow allow for it or do we have to eye ball it?

Thanks,
Patrick

 

[Bart van der Wolf]

SNIP

If I set the master control to +2 all the histogram on the
other channels fall off the edge. I assume I don't want this.
I fiddle until I can get all 3 to 254, without any bit of the
histogram cut off. Right?

The goal is to get the film base color, the most transparant
(inter-image) film area, close to 255, 255, 255. This will, if the
image is in linear gamma space, effectively remove the (yellow/orange)
film base color. So measure the color of that area in the preview, as
you pull back or increase the RGB sliders after increasing the Master
gain to +2.

When you say invert the image in Photoshop all I'm doing
is flipping the curves. Is there a different way to do this?

You can do this with Image|Adjustments|Invert or with
Image|Adjustments|Levels. In the latter case you have more control
over the exact clipping points per channel. In Levels you can reverse
the "Output Levels" from 0-255 to 255-0.

In levels if I adjust the individual colors R G and B, I end
up with color cast.

Go back to the Master and R/G/B sliders in the Minolta scan utility,
and sample the film base color. If that part is not done right, the
rest of the procedure will fail.

All I do is look at the histogram and move the black point
and white point sliders to either edge of the graph. I
sometimes use the alt key to make sure I don't clip. Is this
right? Normally when I'm scanning slides I'd never go into
the individual channels unless I was fixing a color cast.

Correct, and you will use either the Levels or Curves control to
adjust, after inversion, what has become the White point.

The posi linear profile now really makes things bad.

Redo the procedure. The Gamma will need adjustment, but that is part
of the tonescaling. You may also forget the posi-linear profile, but
then you'll need to adjust per channel saturation. However, let's not
jump to solutions for what may be not a problem.

SNIP

Yes I made them myself with an IT8 target using Little
CMS (I also have MonacoEZ).

They were probably built with a gamma adjusted target, so the linear
workflow will not work with them.

SNIP

Using the White and black points do correct the color
cast and adjusting the overall gamma does help the image
but it still isn't right. Not sure how to do a logarithmic
mapping. Do you have any links for information on this?

Let's first get the basics right... There are various ways of getting
logarithmic mapping, but we'll get to that later.

SNIP

I was hoping to avoid the guessing of accurate colors. Does
this workflow allow for it or do we have to eye ball it?

Maybe, but because we want to adjust/prevent an already
compromised/clipped histogram, there are trade-offs. Again, let's get
the basics right first.

Bart