Is Callier Effect Necessarily a 'bad' thing?

  • Thread starter Thread starter WD
  • Start date Start date
W

WD

Folks,

I am posing this question relative to scanners with collimated light
sources (e.g. Nikon Coolscans) and let's say black and white silver
based negative films.
I have read many threads on here saying how this is a bad combination
(silver negs and collimated LED source) since it enhances grain
due to the Callier effect. While grain is generally modeled as a sort
of noise, seems to me that the image IS the grain. After all what else
is the image composed of but the silver 'grains' on the negative.
Furthermore, it seems that the Callier effect is a sort of signal
gain if one considers the silver on the negative to be the signal.
In fact it is an optical signal gain before any CCD or further downstream
electronics have a chance to add in there own electrical noise.
Think of a low density (almost clear) part of a negative.
The Callier effect would present the strongest optical 'signal'
to the CCD, i.e. it would 'enhance' the optical contrast prior to
electronic noise corruption. In the case of a dense (dark) portion
of the negative, today's better scanners have a strong enough DMAX
to handle what may be an enhanced density due to the Callier effect.
If the Callier effect creates an image that is too contrasty, in
today's scanner workflow it is trivial to reduce contrast (vs. enlarger
workflow where paper selection and exposure might make this a trickier task).

I would like to hear others thoughts on my above assertion that the
Callier effect may in fact have benefits in todays scanner based
workflow. Unfortunately, I don't have access to enough equipment to
empirically demonstrate my assertion.

W
 
WD said:
Folks,

I am posing this question relative to scanners with collimated
light sources (e.g. Nikon Coolscans) and let's say black and
white silver based negative films.
I have read many threads on here saying how this is a bad
combination (silver negs and collimated LED source) since
it enhances grain due to the Callier effect.
Click to expand...

Small refinement, the Callier effect describes the apparent (!)
density increase caused by scatter in the emulsion. When the light
source is diffuse, the additional scatter by grainy emulsions doesn't
make a difference, because the light was already coming in at a
variety of angles and some will be refracted/scattered in the
dorection of the sensor. Directional/collimated light can only be
scattered away from the sensor, causing the apparent higher density.
The Callier quotient is larger for silverbased film densities than for
semi-transparent chromogenic black and white or color dye based films.

The graininess we can see are 3D clusters of grains or dye clouds, and
the degree of clustering is an attribute of the film, and developer
combination.
While grain is generally modeled as a sort of noise, seems to
me that the image IS the grain. After all what else is the image
composed of but the silver 'grains' on the negative.
Click to expand...

Correct in a sense, the grains are samples of luminance exceeding a
threshold at a particular random position.
Furthermore, it seems that the Callier effect is a sort of signal
gain if one considers the silver on the negative to be the signal.
Click to expand...

It's a bit more complicated, because the Callier effect is different
at different densities.
In fact it is an optical signal gain before any CCD or further
downstream electronics have a chance to add in there own
electrical noise.
Think of a low density (almost clear) part of a negative.
The Callier effect would present the strongest optical 'signal'
to the CCD, i.e. it would 'enhance' the optical contrast prior
to electronic noise corruption. In the case of a dense (dark)
portion of the negative, today's better scanners have a strong
enough DMAX to handle what may be an enhanced density
due to the Callier effect.
Click to expand...

Not exactly. The effect increases the apparent density as (diffuse)
density increases to approx. 1.2-1.5, after which the effect reduces.
The reason is that at low densities there are too few grains to cause
scatter, at medium densities there is optimal scatter, and at high
densities there is already little non scattered light that reaches the
sensor (most is absorbed anyway).
If the Callier effect creates an image that is too contrasty, in
today's scanner workflow it is trivial to reduce contrast (vs.
enlarger workflow where paper selection and exposure might
make this a trickier task).
Click to expand...

As indicated above, the Callier effect causes an increased contrast
gradient up to densities 1.2-1.5, at the expense of a reduced contrast
gradient at higher densities. So in fact it causes a more non-linear
density as the collimation increases. Formally, densities are called
'diffuse optical densities', exactly for that reason.

Bart
 
Bart,

So it sounds like your saying that the bottom line is that the Callier
effect causes a non-linear transformation between film density and scanned
results, with contrast enhancement in mid-tones only. Is this correct?
If so, seems the proper curve or LUT could be applied to 'undue' this.
In the real world, images are tailored to be 'pleasing to the eye', i.e.
curve or other transformations may be applied until the user
likes the result. So it seems the bottom line is that collimated vs.
non-collimated is really a non-issue for photographic images in a
scanned workflow. Is this your view?

W
 
WD said:
Bart,

So it sounds like your saying that the bottom line is that the
Callier effect causes a non-linear transformation between film
density and scanned results, with contrast enhancement in mid-
tones only. Is this correct?
Click to expand...

No, increased contrast for densities lower than 1.2-1.5 and contrast
reduction above that.
If so, seems the proper curve or LUT could be applied to
'undue' this.
Click to expand...

It would seem so, however the film itself also has a non-linear
density response to exposure. It compresses shadows and highlights
with a lower contrast. So if the goal is an accurate linear density
response it should be achieved through calibration, but for a pleasing
tonescaling I would not worry about it. Tonescaling is better done on
image content, because the eye/brain is easily fooled and it is not
very accurate in absolute luminance (see e.g.
http://web.mit.edu/persci/people/adelson/checkershadow_illusion.html).
In the real world, images are tailored to be 'pleasing to the
eye', i.e. curve or other transformations may be applied until
the user likes the result. So it seems the bottom line is that
collimated vs. non-collimated is really a non-issue for
photographic images in a scanned workflow. Is this your view?
Click to expand...

Fact is that the grain clusters do look more grainy in collimated
light, and scratches/dust are also more prominent. I really like the
result of a diffuse lightsource, partly because ICE doesn't function
on a silver based black and white film and partly because it reduces
graininess. Tonality can be adjusted quite easily once the image is
digtized. Since the Callier effect is marginal in dye densities there
is less of a tonality effect on those, but it still reduces the
graininess (at the cost of longer exposure times when scanning).

Bart
 
Bart,

Thanks for spending the time to share your insights and knowledge. Two
more related questions:

1. Where can I find more reference material on the Callier effect?
2. Any ideas how I might try to diffuse the light in my Nikon scanner
(e.g. laying a piece of translucent plastic of the frame in an FH-3
film strip holder)?

I think it would be interesting to experiment to get a feel for how
strong
this effect is in practice. I do scan a fair amount of BW (mostly
TMAX-100) and am generally quite please with the results but there is
always
room for improvement.

W
 
WD said:
Bart,

Thanks for spending the time to share your insights and knowledge.
Two more related questions:

1. Where can I find more reference material on the Callier effect?
Click to expand...

I'm not sure. Google just turns up bits of info related to enlargers
and contactprinting. An encyclopedia on photography might have
something.
2. Any ideas how I might try to diffuse the light in my Nikon
scanner (e.g. laying a piece of translucent plastic of the
frame in an FH-3 film strip holder)?
Click to expand...

You need a thin material that has perfect Lambertian diffusion, e.g.
opaline glass, closer to the lightsource. It will reduce the amount of
light reaching the sensor, so maybe the LEDs will prove to be too weak
for this. You could experiment with some materials from Lee Filters
Lighting products (http://www.leefiltersusa.com), but try to avoid any
structure in the diffusor.
I think it would be interesting to experiment to get a feel for
how strong this effect is in practice.
Click to expand...

You may find some historical info in the Minolta Multi Pro scanner
forum on Yahoo, it explains some of the considerations behind the
"ScanHancer" (http://www.allari-photo.com/scanhancer.html).

Bart
 
Bart,

Again, thanks. Interestingly enough, I pulled out the adapter from my
Nikon Scanner to take a look inside. I turned on the power and saw that
the light source in fact has some kind of cover on it. It started
flashing red, green, and blue. I was staring at the source at an
angle at least 45 deg. off perpendicular. Doesn't the fact that I could
see the LEDs flashing at this off angle imply that the light is not
perfectly collimated? Do we really know in fact how
collimated this light source is in the first place?

W
 
Bart,

Is this the same effect we wet darkroom workers get with a condensor (or
more radically point source) enlarger as compared to a diffuser enlarger? If
that is the case then it seems that at least in the "old days" the
enhancement of grain was considered a plus, as diffused-light-source
enlargers were mostly used for color printing; the softer look was not
considered desirable in B&W printing at all.

Toby
 
Toby said:
Bart,

Is this the same effect we wet darkroom workers get with a
condensor (or more radically point source) enlarger as
compared to a diffuser enlarger?
Click to expand...

Yes, the condenser concentrates the lightbulb's output on the film
area, thus increasing output which allows reduced exposure times.
There are variations with clear bulbs, frosted bulbs and on the other
side of the lighting methods is a softbox which provides diffuse
light.
If that is the case then it seems that at least in the "old days"
the enhancement of grain was considered a plus, as diffused-
light-source enlargers were mostly used for color printing;
the softer look was not considered desirable in B&W
printing at all.
Click to expand...

That is partly correct, although also a matter of taste, because paper
contrast is also a factor. The ultimate quality was/is considered to
be produced by contact printing, where the Callier effect is
practically non-existent.

However, in a situation where you sample the trans-illumination in
regular distances (discrete sampling), the "grain-cluster shadows"
cast by a point light source or a very collimated one, will increase
grain-aliasing (sharp edges have higher spatial frequencies than the
sampling density can resolve). So in a sampling system like a scanner,
the more diffuse lighting will reduce apparent graininess. From
thereon we are dealing with a digital image, and tonality is highly
adjustable.

Bart
 
WD said:
Bart,

Again, thanks. Interestingly enough, I pulled out the adapter
from my Nikon Scanner to take a look inside. I turned on the
power and saw that the light source in fact has some kind of
cover on it. It started flashing red, green, and blue. I was
staring at the source at an angle at least 45 deg. off
perpendicular. Doesn't the fact that I could see the LEDs
flashing at this off angle imply that the light is not perfectly
collimated?
Click to expand...

The emission you can see, will never reach the film and as such
doesn't contribute to the image. What matters is how the lightsource
looks from the perspective of the film. Small lightsources will cast
sharp (film grain-cluster) shadows (also depends a bit on the type of
developer used). Large light sources will cast diffuse shadows, and
these reduce the grain-aliasing in a discrete sampling situation like
scanning.
Do we really know in fact how collimated this light source
is in the first place?
Click to expand...

Not in absolute terms, but those that have looked inside the scanner
report a cylindrical lens/bar that acts as a condensor, just like in a
projector/enlarger. Empirically, it can be shown that the scanner
doesn't produce a linear optical density when scanning a (calibrated)
silver based stepwedge. The input is of known luminance, the
transmission deviates from the expected values as a function of
density (assuming a linear scanner ADC response).
The Minolta Scan Elite 5400 allows to demonstrate that difference (ADC
is then a constant) by shifting a built-in diffusor in front of the
lightsource. The same stepwedge produces a more linear response as a
result.

You can find the empirical results for a particular stepwedge and that
particular scanner (both mine) in the following thread:
<http://groups.google.com/groups?selm=3f47efad$0$49112$e4fe514c%
40news.xs4all.nl&output=gplain>
The diffusor used in the DSE-5400 scanner is not a perfect diffusor
but a compromise, because it would otherwise create *very* long
exposure times. However, the result is already very noticable as it is
implemented, graininess is significantly reduced, and measured
resolution is practically unchanged.

Bart
 
2. Any ideas how I might try to diffuse the light in my Nikon scanner
(e.g. laying a piece of translucent plastic of the frame in an FH-3
film strip holder)?
Click to expand...

One suggestion I read around these parts is to unfocus slightly. This
does appear to reduce grain somewhat.

For what is worth, when I performed my very first scan (then LS-30,
now LS-50) I was *very* disappointed at the amount of grain. However,
after examining the film more carefully, I realized that the grain was
there all along! We just didn't see it or pay attention to it.

The problem is that in the analog world we have put up with and
accepted "fuzziness" (be it when a tiny slide is projected to a wall
size screen or, when we print to "fuzzy" photographic paper). The
thing is, one usually had not examined the images at the level of
magnification we do after we scan. The way I look at it now is that
the scanners just caught up with this and are simply "too good".

So, I - for one - have actually learned to "love" the grain. It
assures me that that the image is sharp. The fact that it's visible at
100% (or more) magnification is (for me) beside the point when I
consider the context (i.e. how we normally view or print them).

Don.
 
My personal feeling is that the grain *should* be a major player in 35 mm
photography. I used to see so many articles in the photo rags recommending
ultra-fine grain films so that your 35 mm pix will look like medium or even
large format, but anybody who has shot medium and large format knows that it
just ain't so.

I've never been able to reproduce the smooth tonality of the larger formats
using 35 mm, grain or no grain. In fact my own experience has been that
ultra-fine grain 35 mm photography just looks like a poor imitation of its
larger format siblings.

The grain is what says "35 mm" to me. In fact much contemporary 35mm
Japanese (B&W) photography emphasizes both grain and contrast as a
deliberate style choice. I've been very happy with the results of my Nikon
4000 scans, and often use USM to further accentuate the grain on my pushed
high-speed 35mm B&W negs with excellent results.

Toby
 
Folks,

These are all valid points. I have recently done a fair amount of shooting
in TMAX-100 35mm. Scanned on a Coolscan 5000 I have been impressed with
the razor sharpness. The grain varies depending on exposure. I have also
shot Plus-X, not as sharp and grainier with its own look. It depends
on the subject and the shot, exposure etc. Both 'looks' are valid.
From a scanning point of view I am always curious about how the scanner
will affect the look, how does it compare to a traditional print etc.
The answer as usual is that everything affects the look. I am quite happy
with the Nikon but wonder about gee if the lightshource was diffuse would
it look different etc. I have a drum scan of a particular 35mm color negative
and compared it to the Nikon and no big surprise but the drum scan 'looks'
better. In this case the overall image had more 'pop' or clarity. Tweaking
in Photoshop I could get the Nikon scan close but not quite. This is
no surprise given the difference in technology (and cost!!).

W
 
Back
Top