Nikon LS-2000 film scanner vs HP 4C scanjet

[Glen A Stromquist]

I just picked up a hardly used LS-2000, I did a few slides first and was
very pleased with the results, just using the default settings on the
newest software from Nikon for this scanner. Then I tried a negative,
thinking that I should get similar results, but was sorely disappointed,
the output was extremely grainy, and overall a very poor rendition of
the original photo.

Just for comparison I scanned the actual photo in my scanjet 4c, and the
result was far better, even zooming the scan 200% on a 4x6 photo.

I was under the impression that the slide scanner would produce equal,
if not superior results to the flatbed, but this is plainly not so. I'm
wondering if I'm doing something wrong here in the setup, but from
reading the limited doc's I have I cant really see anything obvious.

Is this normal, for a film scan to be that much poorer than a photo
scan? I realize that the LS-2000 is not by any means one of Nikons best
scanners, but I expected better than this!!

 

[winddancing]

When playing with an elephant you must be careful to distinguish between the
tail and the snout. They are different animals. Slides work well, B/W negs
do not.

 

[Kennedy McEwen]

I just picked up a hardly used LS-2000, I did a few slides first and
was very pleased with the results, just using the default settings on
the newest software from Nikon for this scanner. Then I tried a
negative, thinking that I should get similar results, but was sorely
disappointed, the output was extremely grainy, and overall a very poor
rendition of the original photo.

Just for comparison I scanned the actual photo in my scanjet 4c, and
the result was far better, even zooming the scan 200% on a 4x6 photo.

I was under the impression that the slide scanner would produce equal,
if not superior results to the flatbed, but this is plainly not so. I'm
wondering if I'm doing something wrong here in the setup, but from
reading the limited doc's I have I cant really see anything obvious.

Is this normal, for a film scan to be that much poorer than a photo
scan? I realize that the LS-2000 is not by any means one of Nikons best
scanners, but I expected better than this!!

It may not be their flagship scanner any longer but its only a few years
since it held that position.

Irrespective of its current age however, I always found the LS-2000 to
be just as capable on negatives as positives. By their very nature
however, negatives do produce more grain than transparencies simply
because the image is compressed into a smaller density range when it
recorded on a negative. Blacks on a negative are nowhere near as dark
as the blacks on slide film and the whites on a negative are, of course,
limited by the orange mask whereas on a slide they are almost completely
transparent - so the overall density of the image on negative film is
lower. As part of the printing process, this compressed density range
is stretched out to produce deep black and clear white on the final
print. This contrast stretching stage also applies to the grain in the
negative as well, making it more visible. The other side of the same
coin is that negative film provides much wider exposure latitude than
slide film, because the density range is compressed on the emulsion.
That said, though, these differences are equally true of standard
chemical prints as they are of scanned images.

However, when you make a comparison of scanning a negative directly and
scanning a chemical photo printed from it there is a significant
difference - resolution. Most standard 6x4" chemical photos contain
little, if any, detail beyond 300ppi. This is especially true if the
photo has been produced by one of the modern digital developing and
printing establishments, which use 300ppi on the page as standard - and
even then the contrast achieved at 300ppi is well below 100%.

300ppi on a 6x4" print corresponds to approximately 1200ppi on 35mm
film, while the native resolution of the LS-2000 is 2700ppi - more than
4 pixels for every fully resolved pixel on the print! That means the
scanner has much more resolution than the print and it appears that, in
your case, film grain becomes significant in the gap between the two.
You don't mention the film type or speed, but I would be surprised if
you found this to be a significant problem scanning ISO100/21 media such
as Fuji Reala. Higher speed film with coarser grain is more likely to
show a significant difference.

In short, you are getting more grain from the LS-2000 scanner than from
a flatbed scan of the print simply because the print is unable to
resolve the grain in the first place, while the scanner is capable of
doing so. By contrast, you will be able to see much finer detail on the
LS-2000 scan than you could ever pick out of the flatbed scan from the
print - assuming that the negative contains such detail in the first
place. If you apply a small amount of image blur to the LS-2000 results
then you will be able to reduce the visibility of the grain (and of the
fine detail) to a point where both are comparable to the scanned photo.
You will note, however, that you are throwing information away to do
that - which is not what is generally required.

There is another issue called "grain aliasing" which would be
significant if comparing the grain present on large chemical prints to
the scanned result, but it isn't relevant in the comparison you are
performing and I only mention it should you care to do some on-line
research on the topic. Nevertheless, the primary reason I upgraded to
the Nikon LS-4000 was to reduce grain aliasing by sampling at a finer
resolution.

In summary, the grainy effect that you are seeing is because the film
scanner is capable of pulling the detail out of your film where the
scanned print method simply is not. If you want to reduce the grain -
at the expense of a little detail - there are several standard
techniques and even several packages and Photoshop plug-ins which have
been developed specifically for this.

 

[Glen A Stromquist]

It may not be their flagship scanner any longer but its only a few years
since it held that position.

Irrespective of its current age however, I always found the LS-2000 to
be just as capable on negatives as positives. By their very nature
however, negatives do produce more grain than transparencies simply
because the image is compressed into a smaller density range when it
recorded on a negative. Blacks on a negative are nowhere near as dark
as the blacks on slide film and the whites on a negative are, of course,
limited by the orange mask whereas on a slide they are almost completely
transparent - so the overall density of the image on negative film is
lower. As part of the printing process, this compressed density range
is stretched out to produce deep black and clear white on the final
print. This contrast stretching stage also applies to the grain in the
negative as well, making it more visible. The other side of the same
coin is that negative film provides much wider exposure latitude than
slide film, because the density range is compressed on the emulsion.
That said, though, these differences are equally true of standard
chemical prints as they are of scanned images.

However, when you make a comparison of scanning a negative directly and
scanning a chemical photo printed from it there is a significant
difference - resolution. Most standard 6x4" chemical photos contain
little, if any, detail beyond 300ppi. This is especially true if the
photo has been produced by one of the modern digital developing and
printing establishments, which use 300ppi on the page as standard - and
even then the contrast achieved at 300ppi is well below 100%.

300ppi on a 6x4" print corresponds to approximately 1200ppi on 35mm
film, while the native resolution of the LS-2000 is 2700ppi - more than
4 pixels for every fully resolved pixel on the print! That means the
scanner has much more resolution than the print and it appears that, in
your case, film grain becomes significant in the gap between the two.
You don't mention the film type or speed, but I would be surprised if
you found this to be a significant problem scanning ISO100/21 media such
as Fuji Reala. Higher speed film with coarser grain is more likely to
show a significant difference.

In short, you are getting more grain from the LS-2000 scanner than from
a flatbed scan of the print simply because the print is unable to
resolve the grain in the first place, while the scanner is capable of
doing so. By contrast, you will be able to see much finer detail on the
LS-2000 scan than you could ever pick out of the flatbed scan from the
print - assuming that the negative contains such detail in the first
place. If you apply a small amount of image blur to the LS-2000 results
then you will be able to reduce the visibility of the grain (and of the
fine detail) to a point where both are comparable to the scanned photo.
You will note, however, that you are throwing information away to do
that - which is not what is generally required.

There is another issue called "grain aliasing" which would be
significant if comparing the grain present on large chemical prints to
the scanned result, but it isn't relevant in the comparison you are
performing and I only mention it should you care to do some on-line
research on the topic. Nevertheless, the primary reason I upgraded to
the Nikon LS-4000 was to reduce grain aliasing by sampling at a finer
resolution.

In summary, the grainy effect that you are seeing is because the film
scanner is capable of pulling the detail out of your film where the
scanned print method simply is not. If you want to reduce the grain -
at the expense of a little detail - there are several standard
techniques and even several packages and Photoshop plug-ins which have
been developed specifically for this.

Thanks, that makes sense, I did a bit of playing with paint shop pro and
shrunk the negative scan to the size of the print scan and they were
somewhat closer, but I'd still have to give the nod to the print scan
for clarity. However I am relatively at new editing graphics with these
programs, so perhaps with some more playing I'll see better results. The
other thing I forgot to mention was that there was an obvious streak on
the negative scan, perhaps from dust but the negative strip was freshly
developed and should have been clean.

I tried another negative and its corresponding print tonight, results
were better this time, both scanners using defaults and ICE turned off
on the LS-2000. I still think the print scan gets the nod in this case
as well, but these are not retouched in any way from PSP, just resized.
Please have a look if you have time at
http://community.webshots.com/album/115038246MIsmHz

I quickly took a look at PSP and could not see any readily apparent ways
to reduce the grain, so any pointers in this are well appreciated as well

TIA

 

[Kennedy McEwen]

Thanks, that makes sense, I did a bit of playing with paint shop pro and
shrunk the negative scan to the size of the print scan and they were
somewhat closer, but I'd still have to give the nod to the print scan
for clarity. However I am relatively at new editing graphics with these
programs, so perhaps with some more playing I'll see better results. The
other thing I forgot to mention was that there was an obvious streak on
the negative scan, perhaps from dust but the negative strip was freshly
developed and should have been clean.

I tried another negative and its corresponding print tonight, results
were better this time, both scanners using defaults and ICE turned off
on the LS-2000. I still think the print scan gets the nod in this case
as well, but these are not retouched in any way from PSP, just resized.
Please have a look if you have time at
http://community.webshots.com/album/115038246MIsmHz

I quickly took a look at PSP and could not see any readily apparent
ways to reduce the grain, so any pointers in this are well appreciated
as well

I had a look at your comparison images and, even though these are
compressed and resized images, it is quite clear that the scan from the
negative has more detail in it. For example, zoom in on the badge on
the player's jersey - it is much better resolved on the negative scan
than on the print scan, as expected. So I would still expect grain to
be more visible in the scanned negative than in the print. However,
there are a couple of other differences which may be biasing your view
of what is the better image:

* colour balance - the print scan looks a little pink to me, whilst the
film scan looks a little blue. I am sure the buildings in the
background should be white. On the unlit side they may be reflecting
the grass and have a hint of green, but on the lit side they should be
white. PSP auto-colorbalance might help here, but I suspect that the
dominance of green grass may result in some loss of green saturation in
both images, so you might be left with no option but to adjust them
manually.

* levels - the print scan has very clearly clipped shadows. Whilst this
results in a very deep, pleasing, black in the final image, it also
means that detail, such as the studs on the sole of the player's kicking
boot or the folds of the black shorts have been lost. In comparison,
the negative scan loses far less shadow detail but does have a higher
than desirable black level, giving the overall impression of a 'grubby'
image. You can reduce the black level and adjust the gamma in the
negative scan to get a deep black while retaining the shadow detail, but
you cannot recover the clipped detail from the print.

I would also suggest using ICE on this negative because there are quite
serious scratches and dust on it. This may well have happened during
processing and post print handling so don't necessarily think you have
done it yourself. If it is a continual problem, consider changing your
processing establishment. ICE will completely clear it though - as well
as the couple of dust specs present - far better than any post scan
filtering will. ICE detects the presence of dust and scratches and only
processes those parts of the image. Unlike some other scanners, the
defect detection required by ICE is captured in exactly the same pass as
the image data, so there is no chance of misalignment and the correction
placement is exact.

Good though it is, Paintshop Pro doesn't really have a built in tool for
grain removal. You can use a *small* amount of gaussian blur or smart
blur experimenting with the levels. There are other packages
specifically for grain and noise removal, such as Kodak's GEM plug-in
which will integrate with PSP or NeatImage which is a stand alone suite.

GEM is at http://www.asf.com/products/plugins/gem/pluginGEM.asp
NeatImage is at http://www.neatimage.com/

Both are available with free demo versions so you can try them out first
and decide which suits you best.

Alternatively, you can use an alternative scanning package, such as
Vuescan, which has grain reduction built in as well as a host of other
features such as automatic colour balance etc. As downloaded, this is a
demo version that places a watermark on your scans and this feature is
removed if you decide to buy it. A major benefit of Vuescan is that it
will work with a lot of scanner types, including flatbeds, so it might
be the only scanner interface package you ever need - meaning that you
can learn it thoroughly and forget about all the other manufacturer
specific interfaces.

Vuescan is at http://www.hamrick.com/vsm.html

There is an ongoing discussion at the moment in another thread in this
group concerning the relative merits of these three options for grain
reduction, so you might also want to read that before trying them out.
See the threat titled "Vuescan + LS40: grain reduction compared to Nikon
Scan GEM".

Finally, another technique for grain, and grain aliasing, reduction is
to offset the focus very slightly when you scan the image. The LS-2000
has a manual focus setting to allow you to do this. The real advantage
of this technique is that it reduces the grain before it even reaches
the CCD. Obviously you will lose some image sharpness as well, so don't
go overboard with it, though some sharpness can be recovered by
filtering, but if you are just trying to get as much from a scan as is
present on a 6x4" image this might be the cheapest and quickest solution
for you.

Good luck!

 

[Kennedy McEwen]

One other point which may still be relevant to PSP (it still wasn't
implemented at V7.01, which was the last version I examined in detail):

before downsizing an image which has fine structure such as grain or
noise, implement a gaussian blur with a radius at least half the scale
factor you intend to downsize by. So, if you have a 3000x2000 pixel
image an intend to produce am 800x533 image for web display, so the
scale is around 4x reduction, use a gaussian filter with a radius of at
least 2 pixels. This will filter out the majority of the information
that would otherwise alias, including the noise and grain.

This is particularly important if you use PSP's pixel resize or bicubic
options for downsizing - they seemed to have figured out how to do it
with bilinear and then introduced a "smart" resize option that just
selects bilinear for downsizing because it has the filter embedded in it
and bicubic for upsizing because they screwed up bilinear upsizing! It
was a real cludge, but at least they fixed the bilinear bug in V8! :-(

After the resize, you may want to implement a sharpening filter to
restore the sharpness of the image using the final information it
contains.

Photoshop implemented such a pre-filter by default on their downsizing
operations several versions back, and I am sure (although I have not
tried it) that this is also implemented in PS Elements.

 

[Glen A Stromquist]

One other point which may still be relevant to PSP (it still wasn't
implemented at V7.01, which was the last version I examined in detail):

before downsizing an image which has fine structure such as grain or
noise, implement a gaussian blur with a radius at least half the scale
factor you intend to downsize by. So, if you have a 3000x2000 pixel
image an intend to produce am 800x533 image for web display, so the
scale is around 4x reduction, use a gaussian filter with a radius of at
least 2 pixels. This will filter out the majority of the information
that would otherwise alias, including the noise and grain.

This is particularly important if you use PSP's pixel resize or bicubic
options for downsizing - they seemed to have figured out how to do it
with bilinear and then introduced a "smart" resize option that just
selects bilinear for downsizing because it has the filter embedded in it
and bicubic for upsizing because they screwed up bilinear upsizing! It
was a real cludge, but at least they fixed the bilinear bug in V8! :-(

After the resize, you may want to implement a sharpening filter to
restore the sharpness of the image using the final information it contains.

Photoshop implemented such a pre-filter by default on their downsizing
operations several versions back, and I am sure (although I have not
tried it) that this is also implemented in PS Elements.

Thanks for the great info! I'll check out vuescan as well as some
plugins, I may even try photoshop and give them a whirl as well. One
thing I did notice with this particular negative that the colors were
much more vivid and closer to the real print than the flatbed scan, and
it could be zoomed in on much better than the print scan without
pixelizing, but the graininess was still there. However I noticed that
much of the grain disappeared when I re-sized the image to 75% twice, so
it looks like this will be a learning experience for me.

As far as the negative scratches go, I thought this was the scanner
since every neg I have scanned shows this up. I blew the SA-20 out with
an air duster and dusted the negative as well, but still get the
scratch. This must have happened in the post-handling of the negative,
as it has never been removed from its sleeve before. Unfortunatly where
I live there are not a lot of options other than the local mega-mart 1
hour, but in the plus side my daughter is the one now doing the work so
I'll stress to her about being carefull with the negatives. One day I
hope to get that darkroom I've always wanted but by then digital SLR's
may be affordable

I was thinking I'd catalog all my neg's with this scanner, but given the
sheer number of them I have and the speed of the scanner I may have to
rethink this!

thanks again for your help!

 

[Glen A Stromquist]

One other point which may still be relevant to PSP (it still wasn't
implemented at V7.01, which was the last version I examined in detail):

before downsizing an image which has fine structure such as grain or
noise, implement a gaussian blur with a radius at least half the scale
factor you intend to downsize by. So, if you have a 3000x2000 pixel
image an intend to produce am 800x533 image for web display, so the
scale is around 4x reduction, use a gaussian filter with a radius of at
least 2 pixels. This will filter out the majority of the information
that would otherwise alias, including the noise and grain.

This is particularly important if you use PSP's pixel resize or bicubic
options for downsizing - they seemed to have figured out how to do it
with bilinear and then introduced a "smart" resize option that just
selects bilinear for downsizing because it has the filter embedded in it
and bicubic for upsizing because they screwed up bilinear upsizing! It
was a real cludge, but at least they fixed the bilinear bug in V8! :-(

After the resize, you may want to implement a sharpening filter to
restore the sharpness of the image using the final information it contains.

Photoshop implemented such a pre-filter by default on their downsizing
operations several versions back, and I am sure (although I have not
tried it) that this is also implemented in PS Elements.

Thanks for the great info! I'll check out vuescan as well as some
plugins, I may even try photoshop and give them a whirl as well. One
thing I did notice with this particular negative that the colors were
much more vivid and closer to the real print than the flatbed scan, and
it could be zoomed in on much better than the print scan without
pixelizing, but the graininess was still there. However I noticed that
much of the grain disappeared when I re-sized the image to 75% twice, so
it looks like this will be a learning experience for me.

As far as the negative scratches go, I thought this was the scanner
since every neg I have scanned shows this up. I blew the SA-20 out with
an air duster and dusted the negative as well, but still get the
scratch. This must have happened in the post-handling of the negative,
as it has never been removed from its sleeve before. Unfortunatly where
I live there are not a lot of options other than the local mega-mart 1
hour, but in the plus side my daughter is the one now doing the work so
I'll stress to her about being carefull with the negatives. One day I
hope to get that darkroom I've always wanted but by then digital SLR's
may be affordable

I was thinking I'd catalog all my neg's with this scanner, but given the
sheer number of them I have and the speed of the scanner I may have to
rethink this!

thanks again for your help!

 

[Kennedy McEwen]

Thanks for the great info! I'll check out vuescan as well as some
plugins, I may even try photoshop and give them a whirl as well. One
thing I did notice with this particular negative that the colors were
much more vivid and closer to the real print than the flatbed scan, and
it could be zoomed in on much better than the print scan without
pixelizing, but the graininess was still there. However I noticed that
much of the grain disappeared when I re-sized the image to 75% twice,
so it looks like this will be a learning experience for me.

If you look in the bottom right hand side of the PSP window you will see
the actual size of the image in pixels - it sounds like you have far
fewer pixels on your print scan than you have in your negative scan.

As far as the negative scratches go, I thought this was the scanner
since every neg I have scanned shows this up.

Are you saying that this *same* scratch is in exactly the same place in
every negative you scan? If so, does it still appear if you use the
film strip holder (FH-2 or 3, I can't remember which one goes with the
LS-2000)? If the scratch only appears on the SA-20 then it may be
damaged and actually scratching your negatives as it feeds them. I
would have to say that I think this is most unlikely - but given the
potential consequences of scratching every negative you feed into it, it
is worth checking very carefully as soon as possible.

I was thinking I'd catalog all my neg's with this scanner, but given
the sheer number of them I have and the speed of the scanner I may have
to rethink this!

For cataloguing purposes you might find a flatbed scanner which takes
two film strips and scans, say, 12 frames at a time, a much better
option. Performance will be much poorer than the Coolscan, but for
making contact sheets they are a lot quicker. In fact, if you already
have a dedicated film scanner, its almost the only justifiable reason I
can see for the resolution offered by today's flatbeds. Of course, if
you are planning a digital archive then rigorously scanning each
individual frame at the maximum resolution of a dedicated film scanner
such as the one you are using is really the only way to go - and it does
take a *very* long time. ;-)