Grainy skies with a 4000dpi scan: why?

[Marco Cinnirella]

Hi,

I was wondering why, when comparing scans made with an old 2820dpi
Minolta scanner to those from a Nikon Coolscan V (4000dpi), scans of
the same slide seem to have a grainy sky when scanned with the
Coolscan and a nice smooth sky when scanned with the Minolta at
2820dpi. The Minolta (a Scan Dual 2) is softer than the Nikon in terms
of the sharpness of the optics and the accuracy of the autofocus. In
both cases, I have not sharpened the sky portion of the scan. I have
tried to apply ICE and GEM in the Nikon scans and GEM does make the
sky smoother but still not as smooth as scans from the Minolta. When
printing 10X8 inkjet prints the difference in how smooth the sky tones
are is certainly noticeable to the naked eye.

I am thinking that this may be due to:-

1. The extra dpi of the Nikon scanner is resolving the film grain and
this is why the sky looks grainier (the film was Fuji Sensia 100 Slide
film by the way)
2. The 'grainy' effect in the sky is sensor noise on the Nikon -
unlikely though, it's not supposed to be a noisy scanner and I
certainly haven't noticed noise in shadow areas - unless I have a
defective example of the scanner...
3. The Nikon is resolving the so-called "pepper spots" that have been
documented to be present in Fuji slide emulsions, which show up as
black spots in light-toned areas of images under high levels of
magnification.
4. The 2820 dpi Minolta scanner's focus was so off that the sky looked
'cleaner' and less grainy simply because it was quite blurred.

I know that there are various ways I can deal with a grainy sky,
including GEM, use of Gaussian blur or median filters in PhotoShop,
and so on, but I was wondering which of the above reasons you guys
thought might explain my grainier skies with the 4000dpi Nikon scans?

Marco

 

[Bruce Graham]

Hi,

I was wondering why, when comparing scans made with an old 2820dpi
Minolta scanner to those from a Nikon Coolscan V (4000dpi), scans of
the same slide seem to have a grainy sky when scanned with the
Coolscan and a nice smooth sky when scanned with the Minolta at
2820dpi. The Minolta (a Scan Dual 2) is softer than the Nikon in terms
of the sharpness of the optics and the accuracy of the autofocus. In
both cases, I have not sharpened the sky portion of the scan. I have
tried to apply ICE and GEM in the Nikon scans and GEM does make the
sky smoother but still not as smooth as scans from the Minolta. When
printing 10X8 inkjet prints the difference in how smooth the sky tones
are is certainly noticeable to the naked eye.

I am thinking that this may be due to:-

1. The extra dpi of the Nikon scanner is resolving the film grain and
this is why the sky looks grainier (the film was Fuji Sensia 100 Slide
film by the way)
2. The 'grainy' effect in the sky is sensor noise on the Nikon -
unlikely though, it's not supposed to be a noisy scanner and I
certainly haven't noticed noise in shadow areas - unless I have a
defective example of the scanner...
3. The Nikon is resolving the so-called "pepper spots" that have been
documented to be present in Fuji slide emulsions, which show up as
black spots in light-toned areas of images under high levels of
magnification.
4. The 2820 dpi Minolta scanner's focus was so off that the sky looked
'cleaner' and less grainy simply because it was quite blurred.

I know that there are various ways I can deal with a grainy sky,
including GEM, use of Gaussian blur or median filters in PhotoShop,
and so on, but I was wondering which of the above reasons you guys
thought might explain my grainier skies with the 4000dpi Nikon scans?

Marco

Try a significant manual defocus on the Nikon. If that works, look up
"grain aliasing". If not, forget that theory!

Noise should not be an issue in sky with slide film. (it can be with
overexposed negs.)

Also, you say you tried ICE, but I'm not sure with what result. Not sure
but that may remove the "pepper spots" if that is what you have.

(I have a Canon FS4000 and I sometimes see lots of "dots" on the IR
channel with Fuji consumer negative stock - I'm not sure if that is noise
on the IR channel or "pepper spots" that people talk about).

Bruce

 

[Raphael Bustin]

Hi,

I was wondering why, when comparing scans made with an old 2820dpi
Minolta scanner to those from a Nikon Coolscan V (4000dpi), scans of
the same slide seem to have a grainy sky when scanned with the
Coolscan and a nice smooth sky when scanned with the Minolta at
2820dpi. The Minolta (a Scan Dual 2) is softer than the Nikon in terms
of the sharpness of the optics and the accuracy of the autofocus. In
both cases, I have not sharpened the sky portion of the scan. I have
tried to apply ICE and GEM in the Nikon scans and GEM does make the
sky smoother but still not as smooth as scans from the Minolta. When
printing 10X8 inkjet prints the difference in how smooth the sky tones
are is certainly noticeable to the naked eye.

I am thinking that this may be due to:-

1. The extra dpi of the Nikon scanner is resolving the film grain and
this is why the sky looks grainier (the film was Fuji Sensia 100 Slide
film by the way)
2. The 'grainy' effect in the sky is sensor noise on the Nikon -
unlikely though, it's not supposed to be a noisy scanner and I
certainly haven't noticed noise in shadow areas - unless I have a
defective example of the scanner...
3. The Nikon is resolving the so-called "pepper spots" that have been
documented to be present in Fuji slide emulsions, which show up as
black spots in light-toned areas of images under high levels of
magnification.
4. The 2820 dpi Minolta scanner's focus was so off that the sky looked
'cleaner' and less grainy simply because it was quite blurred.

I know that there are various ways I can deal with a grainy sky,
including GEM, use of Gaussian blur or median filters in PhotoShop,
and so on, but I was wondering which of the above reasons you guys
thought might explain my grainier skies with the 4000dpi Nikon scans?

The Nikon scanners use LED illumination and
are fairly unique in that. Most other film scanners
use flourescent (cold cathode) bulbs. That is
partly, if not wholly the explanation. It's a bit like
diffuser vs. condenser enlargers; the Nikon
scanners are like the latter.

ICE won't help at all with the grain, it might help
with the pepper noise. I think GEM is designed
to alleviate grain, but I don't use it. Another
useful tool here is NeatImage -- very effective,
easy to use, and inexpensive.

I do see some grain in skies when scanning
color negatives, but less so from slides.


rafe b.
http://www.terrapinphoto.com

 

[Steve]

Hi Marco

I've made a similar (Minolta Scan Speed to Nikon 4000ED)upgrade in
scanners and can make the following observations:

1. The extra dpi of the Nikon scanner is resolving the film grain and
this is why the sky looks grainier (the film was Fuji Sensia 100 Slide
film by the way)

This may be true viewing the scans at 100% but at the same relative
size the Nikon scans are generally smoother, probably due to the fact
that the higher resolution resolves the grain better.

2. The 'grainy' effect in the sky is sensor noise on the Nikon -
unlikely though, it's not supposed to be a noisy scanner and I
certainly haven't noticed noise in shadow areas - unless I have a
defective example of the scanner...

I think this is very unlikely.

3. The Nikon is resolving the so-called "pepper spots" that have been
documented to be present in Fuji slide emulsions, which show up as
black spots in light-toned areas of images under high levels of
magnification.

I can verify that this is true, at 2820 dpi the "pepper spots" are not
resolved, but are VERY noticable at 4000 dpi. ICE should remove these
though.
"Pepper spots" are in my experience anyway only a problem with older
Fuji slides, they seem to have fixed the problem with their new
emulsions.

4. The 2820 dpi Minolta scanner's focus was so off that the sky looked
'cleaner' and less grainy simply because it was quite blurred.

This is quite possible, I know that my Minolta would sometimes give
very soft scans with some slides, I never did really find out why.

I know that there are various ways I can deal with a grainy sky,
including GEM, use of Gaussian blur or median filters in PhotoShop,
and so on, but I was wondering which of the above reasons you guys
thought might explain my grainier skies with the 4000dpi Nikon scans?

I'm suprised that you've got grainy skies from Sensia 100 film this is
usually only a problem with faster slide and negative films.

Is this a problem with only one film, or are all your films like this?
I wonder if you've experienced poor processing of the film?

Hope this helps

Steve.

 

[Wilfred van der Vegte]

I've made a similar (Minolta Scan Speed to Nikon 4000ED)upgrade in
scanners and can make the following observations:



I can verify that this is true, at 2820 dpi the "pepper spots" are not
resolved, but are VERY noticable at 4000 dpi. ICE should remove these
though.
"Pepper spots" are in my experience anyway only a problem with older
Fuji slides, they seem to have fixed the problem with their new
emulsions.

I also made similar upgrade (Minolta Scan Speed to Minolta 5400) but in
my experience the pepper spots on older Fujichromes were also visible at
2820 dpi. I haven't tried these slides with my DSE 5400, yet.
OTOH I never scanned the newer emulsions at 2820 dpi but I can also
clearly see tiny dark spots in the sky when scanning Velvia 100F at 5400
dpi. Most of these are, indeed, removed by ICE, it seems.

BTW, I didn't know these pepper spots were widely discussed. I couldn't
find any refences to them by searching Google. If anyone has a link to a
discussion about these spots I would be grateful.

 

[Markus Plail]

BTW, I didn't know these pepper spots were widely discussed. I
couldn't find any refences to them by searching Google. If anyone has
a link to a discussion about these spots I would be grateful.

Search fpr pepper grain instead of pepper spots and you will find plenty
of links. For example:
http://www.luminous-landscape.com/reviews/film/fuji-pepper.shtml

regards
Markus

 

[Bart van der Wolf]

SNIP

The Nikon scanners use LED illumination and
are fairly unique in that. Most other film scanners
use flourescent (cold cathode) bulbs. That is
partly, if not wholly the explanation. It's a bit like
diffuser vs. condenser enlargers; the Nikon
scanners are like the latter.

That's correct. Also the higher magnification will allow to visualize High
ISO film graininess (clustered dye-clouds) more clearly.

Bart

 

[Marco Cinnirella]

Thanks for your comments. I don't think processing was a problem - I
always use Fuji's own processing lab in the UK for my Fuji processing,
and they are excellent. I guess I just expected scans from a more
expensive scanner to be 'smoother' in areas of continuous tone and
didn't bank on the fact that the better focus and higher dpi of the
Nikon scanner may just show the grain and 'pepper spots' more...

I wonder if anyone has compared Minolta's 'grain dissolver' to Nikon's
GEM?
Is it me or does GEM occasioanally produce a slightly 'painterly'
effect in some parts of an image? Still, it does reduce the grain in
blue skies.

 

[Markus Plail]

I wonder if anyone has compared Minolta's 'grain dissolver' to Nikon's
GEM? Is it me or does GEM occasioanally produce a slightly
'painterly' effect in some parts of an image? Still, it does reduce
the grain in blue skies.

Minolta also has GEM. The grain dissolver is just a little glas plate
between the light source and the film to diffuse the light and thus
reduce the emphasizing of film grain.

The painterly effect might come frome overcorrecting the image. It's the
same with every noise reduction utility. If you overdo it you will get
this effect in areas of high detail, for example fur or trees on a far
away mountain.

regards
Markus

 

[Simon Waldman]

Minolta also has GEM. The grain dissolver is just a little glas plate
between the light source and the film to diffuse the light and thus
reduce the emphasizing of film grain.

It can't be *just* that, as it does increase the scan time considerably.
This would make sense if it were doing something in software at the same
time. Perhaps GEM, but the manuals say that the 5400 does not have GEM.

 

[Kennedy McEwen]

It can't be *just* that, as it does increase the scan time
considerably. This would make sense if it were doing something in
software at the same time. Perhaps GEM, but the manuals say that the
5400 does not have GEM.

It *is* just that - although "just" somewhat understates the value of
the device on image quality! The reason the scan time increases is
because of the finite transmission of the grain dissolver and the
scattered light, meaning that less light actually reaches the CCD and
longer exposures are required.

Checkout http://scanhancer.iddo.nl/ for details on the original grain
dissolver.

 

[Kennedy McEwen]

I am thinking that this may be due to:-

1. The extra dpi of the Nikon scanner is resolving the film grain and
this is why the sky looks grainier (the film was Fuji Sensia 100 Slide
film by the way)
Yes.

2. The 'grainy' effect in the sky is sensor noise on the Nikon -
unlikely though, it's not supposed to be a noisy scanner and I
certainly haven't noticed noise in shadow areas - unless I have a
defective example of the scanner...
No.

3. The Nikon is resolving the so-called "pepper spots" that have been
documented to be present in Fuji slide emulsions, which show up as
black spots in light-toned areas of images under high levels of
magnification.

Maybe - depends on the film.

4. The 2820 dpi Minolta scanner's focus was so off that the sky looked
'cleaner' and less grainy simply because it was quite blurred.

Maybe.

Also, note that the light source in the Nikon is partially collimated.
making it particularly prone to accentuating grain and defects in the
image. Diffuse light sources, used in some other scanners (not
Minolta's though) tends to reduce the effect.

 

[Kennedy McEwen]

I wonder if anyone has compared Minolta's 'grain dissolver' to Nikon's
GEM?

Completely different approaches to the problem of grain. GEM is a post
scan software algorithm which, as with all grain reduction software,
removes image detail along with the grain.

The grain dissolver is a light scattering system which effectively
softens the light source in the scanner, reducing grain. Since this is
done before the scan, it effectively prevents grain aliasing (which is
what makes the scanned grain look worse than it actually is) before it
can occur. Once the grain has been scanned and aliased, no software
algorithm can remove it without also removing significant detail.

You can, by *very* slight defocus, create a similar (but not the same or
as good as) effect to the grain dissolver on your Nikon scanner. Once
again, you just defocus sufficiently to reduce the amount of grain that
is aliased.

Is it me or does GEM occasioanally produce a slightly 'painterly'
effect in some parts of an image? Still, it does reduce the grain in
blue skies.

Consider the painterly effect to be telling you that too much detail is
being lost. ;-)

 

[Simon Waldman]

Kennedy McEwen wrote:

It *is* just that - although "just" somewhat understates the value of
the device on image quality! The reason the scan time increases is
because of the finite transmission of the grain dissolver and the
scattered light, meaning that less light actually reaches the CCD and
longer exposures are required.

ah! Fair enough.

TBH I've seen little effect from it, but I'm mostly using ISO400 B&W,
so I guess the grain is too big for it to make much difference.

Checkout http://scanhancer.iddo.nl/ for details on the original grain
dissolver.

I'll do that.

 

[Bart van der Wolf]

It can't be *just* that, as it does increase the scan time considerably.

SNIP

It is. Most desktop scanners have a (semi-) collimated lightsource, which
concentrates the light emitted by the lightsource and that helps to improve
the amount of light reaching the film. Because of the directional nature of
collimated light, it also becomes sensitive to scatter/diffraction by e.g.
film grain or dye clouds. The light that was originally focused/condensed in
the direction of the sensor, is partially deflected/diffracted away from it
which makes the scattering area seem to have additional density which
accentuates low- to medium-density contrast but also graininess and requires
additional exposure.

It the light reaching the film is already diffused, the film particles
cannot diffuse it much more so, the graininess is not accentuated. Contrast
is also modified, because the amount of scatter is density dependent. This
works positive for high density areas where contrast is increased (e.g. sky
in negatives or shadows in slide) but without an increase in graininess.

Bart

 

[Norman Black]

Minolta also has GEM. The grain dissolver is just a little glas
plate

It *is* just that - although "just" somewhat understates the value of
the device on image quality! The reason the scan time increases is
because of the finite transmission of the grain dissolver and the
scattered light, meaning that less light actually reaches the CCD and
longer exposures are required.

Acording to ASF, the guys to developed ICE, GEM and ROC. GEM is a
software algorithm. ICE is the only thing that needs hardware support.
That support being an infra-red channel. All grain reduction algorithms
will induce some softness.

Norman

 

[Kennedy McEwen]

Acording to ASF, the guys to developed ICE, GEM and ROC. GEM is a
software algorithm. ICE is the only thing that needs hardware support.
That support being an infra-red channel. All grain reduction algorithms
will induce some softness.

Keep up at the back please, the discussion is about the Minolta Grain
Dissolver, not ICE GEN ROC or DEE. ;-) The Grain Dissolver is a
hardware solution, GEM et al. are software with ICE requiring hardware
to detect the defects in the first place.

 

[Hecate]

Keep up at the back please, the discussion is about the Minolta Grain
Dissolver, not ICE GEN ROC or DEE. ;-) The Grain Dissolver is a
hardware solution, GEM et al. are software with ICE requiring hardware
to detect the defects in the first place.

I'm trying to follow this thread because I'm probably going to
purchase a Minolta 5400, but I'm drowning in TLA's