emulsion side down and histograms

  • Thread starter Thread starter Linda
  • Start date Start date
I've played around with the "little triangle in the middle" and see
what changes it makes. I thought it had to do with middle grays. Not
sure where I got that.

Probably Photoshop. If you open Levels or Curves dialog box you'll see
three "color samplers" (they look like eyedroppers) in the bottom
right corner. The middle one officially sets the "midtones".

This is often called the "gray point" which is actually misleading.
What it does is sets the *neutral* point. But isn't that the same
thing? Not really...

The difference is that gray implies to most people that the color they
should click on in the image will turn into R=127,G=127,B=127 after
the correction. This is *not* true. You can click on *any* color in
the image which should be "neutral" and it will be adjusted
accordingly and that is often quite different from 127,127,127!

Here's a little test: Load any image into PS. Blow it up to 1600% so
you can see individual pixels. Change image mode to 8-bit so the Info
palette displays real values. Next, open Levels or Curves and click on
the gray color sampler. Make sure the Info palette is visible.

Now click on any pixel and watch the values in the Info palette. As
long as Levels/Curves is open the Info palette will show the
before/after values in the RGB fields. For example, I just clicked on
105,137,102 with the "neutral" eyedropper and it was changed to
111,111,123!
What is it for and does it have a name other than "the little triangle
in the middle"

It depends... ;o) No, really, it does.

Officially it's called the gray Input Levels. Which means it darkens
and lightens the midtones.

BTW, even though Levels can be used to set gamma some people don't
like it because to set true gamma you need to apply a specific
formula. Before I wrote my own scanner program, where I set the gamma
with 16-bit lookup tables calculated on-the-fly, I used these
Photoshop AMP files instead:

http://www.aim-dtp.net/aim/download/gamma_maps.zip

The zip file contains a bunch of AMP files for gamma up to 5.0 (if
memory serves) in increments of 0.1.

To use them, open the Curves dialog box in Photoshop, then click on
the Load button. Next, go to "Files of type" field and click on the
drop down list. It contains 3 file types and you should select the
bottom one: "AMP". Then go to the directory where you extracted the
above zip file and select "inverse gamma 2.2.amp" file.

I found that to be much more pleasing than the "harsh" Levels gamma.

Don.
 
In the center left of the interface is a row of buttons, the left hand
one with a spanner on it. Clicking this opens "Image Controls"
including both Exposure and Gamma. The settings displayed being those
of the last preview.These settings will be arrived at automatically
unless Auto Exposure is switched off.

That sounds like the right place, although it's a bit strange that
when Auto Exposure is off the gamma is not applied - assuming I
understood the above correctly.

Don.
 
Don said:
That sounds like the right place, although it's a bit strange that
when Auto Exposure is off the gamma is not applied - assuming I
understood the above correctly.

Don.

I didn't express it very clearly.
When Auto Exposure is off exposure and gamma are applied at the last
used setting but can be changed using "Image Controls" accessed by
clicking the button with the spanner on it.

mlgh
 
Don said:
Officially it's called the gray Input Levels. Which means it darkens
and lightens the midtones.

BTW, even though Levels can be used to set gamma some people don't
like it because to set true gamma you need to apply a specific
formula.

Actually Don, the middle level control in PS (and for that matter, in
most scanner software, such as NikonScan) does apply the gamma formula.
Not only that, but in PS it applies the PS toe limited gamma formula.
You can check this with the 16-bit grey ramp file I mentioned in another
thread and compare the results with the gamma formula precisely. The
error is simply the 15-bit limitation of PS.
Before I wrote my own scanner program, where I set the gamma
with 16-bit lookup tables calculated on-the-fly, I used these
Photoshop AMP files instead:

http://www.aim-dtp.net/aim/download/gamma_maps.zip
The problem with these is that they are only 8-bit accurate, with linear
interpolation between the 8-bit points. Gamma in the levels control is
much more accurate, even accounting for the toe limit.
I found that to be much more pleasing than the "harsh" Levels gamma.
Probably because the 8-bit limit restricts the maximum contrast that can
be applied to deep shadows even more than the toe limit. However there
are ways of achieving this without the slope discontinuities that the
amp curves introduce.
 
Don said:
Therefore, I strongly suspect ("global") gamma is set somewhere else
in the Epson software. There is usually an explicit reference to which
gamma should be used when scanning.

On my Minolta DSE 5400 sw, I cannot find a specific gamma control
anywhere. The closest thing is an option to scan linear or non-linear. A
linear scan with a scanner profile conversion in PS looks the same as a
non-linear scan without any conversion in PS (suspect it is done in
Minolta sw). Is this how Minolta controls gamma?
 
Kennedy said:
Actually Don, the middle level control in PS (and for that matter, in
most scanner software, such as NikonScan) does apply the gamma formula.

As I understand it, dragging on any point in Level or Curves in PS or
scanner sw means applying some kind of formula. Some clarification on
your comments please:

- Don: What is "true gamma", and why is a "specific formula" or a "gamma
formula" necessary?

- Kennedy: Did you mean that a "gamma formula" is *only* applied for the
"middle level control", and a different formula is applied for points
other than the middle? Does the same statement holds true for Curves?
 
As I understand it, dragging on any point in Level or Curves in PS or
scanner sw means applying some kind of formula. Some clarification on
your comments please:
- Kennedy: Did you mean that a "gamma formula" is *only* applied for the
"middle level control", and a different formula is applied for points
other than the middle?

The left-most triangle controls where the black point is. The
right-most triangle controls the white point. The central triangle
controls the gamma that is applied to the data between those two points.
Only the central triangular control adjusts gamma, the other two
controls do not affect gamma at all. When the central triangle is
equidistant from the black and white point triangles then the gamma is
set to 1. As the central control moves towards the white point, the
gamma reduces and as it moves towards the black point the gamma
increases.
Does the same statement holds true for Curves?
No - the curves controls are much more flexible in that a transfer
function can be defined which is the cubic spline through the defined
points. There are no gamma adjustment controls in the curves dialog of
PS, although predefined gamma curves can be created and loaded as
required.
 
Kennedy McEwen said:
the curves controls are much more flexible in that a transfer function
can be defined which is the cubic spline through the defined points.

I should have mentioned that even the option to create a cubic spline
from the defined points can be deselected, making the "curve" even more
general still.
 
I didn't express it very clearly.
When Auto Exposure is off exposure and gamma are applied at the last
used setting but can be changed using "Image Controls" accessed by
clicking the button with the spanner on it.

Ah, OK! That's the way it should be. Once you set a gamma it gets
applied to all subsequent scans automatically.

Don.
 
Actually Don, the middle level control in PS (and for that matter, in
most scanner software, such as NikonScan) does apply the gamma formula.
Not only that, but in PS it applies the PS toe limited gamma formula.

Yes, that's what I meant although I did not know the terminology ("toe
limited").
The problem with these is that they are only 8-bit accurate, with linear
interpolation between the 8-bit points. Gamma in the levels control is
much more accurate, even accounting for the toe limit.

The 8-bit limitation is exactly why I wrote my own gamma (as well as
some other reasons like the PS' 15-bit limit) but that was the key.
Probably because the 8-bit limit restricts the maximum contrast that can
be applied to deep shadows even more than the toe limit. However there
are ways of achieving this without the slope discontinuities that the
amp curves introduce.

I don't know why but I found the AMP curves produced a smooth
transition as opposed to Levels gamma which appeared more "choppy".
(Note: Visual "test", no data.) This could indeed have been (and
probably was) the effect of "toe limitation".

Anyway, it's all academic for me now, but it's definitely interesting.

Don.
 
As I understand it, dragging on any point in Level or Curves in PS or
scanner sw means applying some kind of formula. Some clarification on
your comments please:

- Don: What is "true gamma",

Bad wording! ;o) But seriously...
and why is a "specific formula" or a "gamma
formula" necessary?

.... what I meant is I do not try to do other things (which Kennedy
identified i.e. the toe limitation) but apply a straightforward gamma
formula. Another thing is that I use all 16 bits.

Don.
 
On my Minolta DSE 5400 sw, I cannot find a specific gamma control
anywhere. The closest thing is an option to scan linear or non-linear. A
linear scan with a scanner profile conversion in PS looks the same as a
non-linear scan without any conversion in PS (suspect it is done in
Minolta sw). Is this how Minolta controls gamma?

I'm not familiar with Minolta but - and this is a wild guess - maybe
if you try a linear scan with no scanner profile you'll get a true
gamma 1.0 scan?

If Minolta documentation is in PDF format - and you haven't done this
already - try a search for "gamma". Maybe that will come up with
something?

But instead of my speculations, Minolta owners, now is the time to
jump in! ;o)

Don.
 
Kennedy said:
The left-most triangle controls where the black point is. The
right-most triangle controls the white point. The central triangle
controls the gamma that is applied to the data between those two points.
Only the central triangular control adjusts gamma, the other two
controls do not affect gamma at all. When the central triangle is
equidistant from the black and white point triangles then the gamma is
set to 1. As the central control moves towards the white point, the
gamma reduces and as it moves towards the black point the gamma
increases.

From this description, I will conclude that there is only one formula in
level, regardless of which points are moved.
No - the curves controls are much more flexible in that a transfer
function can be defined which is the cubic spline through the defined
points. There are no gamma adjustment controls in the curves dialog of
PS, although predefined gamma curves can be created and loaded as
required.

I'm aware that curves offer more flexibility over level, and always
thought that any edits done with level can be done with curves, but not
vice versa. In the case of gamma adjustment, isn't moving the level's
mid point the same as moving the curves' mid point (vertically up and
down)?
 
Don said:
I'm not familiar with Minolta but - and this is a wild guess - maybe
if you try a linear scan with no scanner profile you'll get a true
gamma 1.0 scan?

Before converting to the scanner's profile, a linear slide scan will
look very dark in PS. After converting to the profile, the scan will
look close to the slide.
If Minolta documentation is in PDF format - and you haven't done this
already - try a search for "gamma". Maybe that will come up with
something?

Gamma is indeed mentioned in the software manual. But there is no
"explicit reference to which gamma should be used when scanning".
Instead there is a description on how to adjust gamma in level (much
like what has been covered in this thread). Since I want a raw scan, I
have been avoiding any adjustments in the scanner sw. Interesting
enough, I now notice the manual says, "the gamma slider allows the
brightness of the image to be adjusted without loosing image
information." How is this possible since adjusting gamma (either by
level or by profile conversion) essentially means redistributing the
scan data?
But instead of my speculations, Minolta owners, now is the time to
jump in! ;o)

Yeah!
 
Before converting to the scanner's profile, a linear slide scan will
look very dark in PS. After converting to the profile, the scan will
look close to the slide.

No, I meant the Minolta scanner software. If it applies a profile then
this may mask the effects of gamma somewhat. I mean, gamma 1.0 should
still be noticeable but the application of a profile may make this
less obvious.

Another unknown (for me, because I'm not familiar with Minolta) is
whether Minolta scanner software actually applies the scanner profile
(i.e. changes values for all pixels in the image) or merely tags the
image with a profile (but leaves the image pixels untouched).

If it's the former then the original data is "lost" but if it's the
latter then, indeed, by rejecting the profile when the image is loaded
into Photoshop the original data - and its gamma state - will be
preserved.
Gamma is indeed mentioned in the software manual. But there is no
"explicit reference to which gamma should be used when scanning".
Instead there is a description on how to adjust gamma in level (much
like what has been covered in this thread). Since I want a raw scan, I
have been avoiding any adjustments in the scanner sw. Interesting
enough, I now notice the manual says, "the gamma slider allows the
brightness of the image to be adjusted without loosing image
information." How is this possible since adjusting gamma (either by
level or by profile conversion) essentially means redistributing the
scan data?

Exactly!

Maybe they mean no data at the extremes is lost - as in clipping when
setting the black and white point? In that sense, gamma will "keep"
"all" data but spread it out on one side and compress it on the other.
Of course, strictly speaking this will still lose data, as you say.

The trouble is these things are usually written by marketroids and
they are very "flexible" with their use of language... :-(

C'mon Minolta owners! ;o)

Don.
 
From this description, I will conclude that there is only one formula in
level, regardless of which points are moved.
Yes


I'm aware that curves offer more flexibility over level, and always
thought that any edits done with level can be done with curves, but not
vice versa. In the case of gamma adjustment, isn't moving the level's
mid point the same as moving the curves' mid point (vertically up and
down)?

No. You can prove this to yourself by moving the central point up or
down sufficiently. The result is that the black or white levels clip,
depending on the direction it was moved. The gamma function *never*
clips - although it may result in rounding of the result to a saturated
level, there is never a discontinuity in the slope of the curve.
 
What is the best method for cleaning the glass on a FB scanner?

I've tried everything from Windex to alcohol to lens cleaners to...
anything else I could think of (even using lens paper instead of
cloth). In the end, any "wet" cleaners just made matters worse. It may
look good to start with but after a few minutes a film or fog seems to
develop.

In my experience, the best is to use a good, dry (!) microfiber cloth.
It leaves no residue and removes all debris and film (that's the
evaporation from electronic components in the scanner - see below).

Some people suggest using a damp cloth. I personally don't do that,
but if you do make sure you use distilled water, not tap water!

The catch is finding a good microfiber cloth. They are not all made
equal and most cloths advertised as "microfiber" are anything but. One
sign of a good microfiber cloth is that when used on glass it almost
has a rubbery feel to it. It almost seems to "grip" the glass. In any
case get at least two so while one is the wash you can use the other.

However, cleaning the top of the glass is only half the story. Namely,
electronic components tend to evaporate an oily substance which then
condensates on the underside of the glass. It's therefore advisable to
open the scanner from time to time and clean the underside of the
glass as well. Sometimes the ribbon cable tends to rub against the
glass leaving streaks as well.

Removing the glass can be tricky and varies from scanner to scanner.
Some have screws hidden underneath rubber pads, some only have two
screws on one side and hinge on the other, etc.

Finally, I find that clearing the glass is best done in a darkened
room with the scanner lamp on and viewing the glass at a shallow angle
(squat down until your eyes are almost parallel to the glass). This
reveals all smudges which are invisible when you look at the glass
from above and in bright daylight.

There is also one neat trick to find out if the cleaning was done well
and that's to scan "nothing". Lift the scanner lid and, in a darkened
room (!), scan the whole surface. This produces a "black" image.
Brighten it up radically in an image editor and shriek in horror! ;o)

Don.
 
Back
Top