These scanners are provided with two different film holders for negs:
- a manual metal and plastic strip which holds 6 negatives and is manually inserted into the slide holder. This is called the FH-2 on the LS2000.
- an auto-feeding strip holder which has a motor to advance a strip of up to 6 negs. This is called the SA-20 on the LS2000.
The SA-20 is very fast and convenient to use but is even worse than the FH-2 at holding the film flat and parallel. In my experience the SA-20 is totally unacceptable if you want focused scans.
Test Method Overview
Nikonscan has a couple of neat features which make this test possible - it allows you to focus at any point on the image, and it gives you a numerical readout of the focus position. This readout is in arbitrary units from 0 to 332 in steps of 4 (at least it is on the LS2000).
The method described here uses these capabilities to measure the distance of the negative or slide from the scanning head at a number of different positions on the film. From this we can create an accurate picture of the "shape" and location of the film, and see how much variation there is in the measured distance over the surface of the image. If the film was perfectly flat and the scanner was well aligned, the film to scanner distance would be constant across the image. As you will see if you do these tests yourself, this is not the case.
Even if this distance was not constant, so long as the variation was less than the DOF of the scanner lens system, things would be fine. It is when the variation goes beyond the DOF that you start losing focus at those locations on the image which are outside the DOF.
The graphs
(see below in the Results) show effectively where the
film is, compared to the plane of the scan. They show a cross section of the film (across
the width of the film strip) at three locations 'along'
the negative - see Fig 1. The vertical axis on the graphs shows the distance measured from the scanning plane to the film in the arbitrary units discussed above. The horizontal axis shows the location across the width of the film. The three different curves show the results at the leading edge of the film, the middle, and the trailing edge.
If you do these any of tests, I would be very interested to hear your results
. If I receive enough responses I will publish the results here.
My
quick conclusions
In the units
used by Nikon, the usable DOF field is around 24 units
- that is, +/- 12 from the focus
setting. The film
needs to remain flat to within this range to give sharp
edge-to-edge scans. (But if you insist on grain
sharpness, you will only get about 12 units (+/- 6) of
DOF).
Unfortunately,
the film when loaded into my scanner is not this flat.
In
the best situation using the FH-2
strip holder, the "film distance from the
scan plane" varies over the range 200 to 230
units, i.e. DOF required is 30 units assuming you set the focus
point exactly to the middle of the range.
Using
the SA20 motorised strip feeder, the film distance
typically varies from 180 to 270 units, i.e. DOF
required is 90 units.
So I need
30-90 units of DOF, but I only have 12-24 units of DOF
available. This confirms my practical observations and
does at least explain why the edges of most of my scans
are out of focus.
A large part
of the problem in my case appears to be that the scanning
mechanism does not move parallel to the longitudinal
plane of the film. As the scanner moves along the frame,
it gradually moves further and further away from the
film, instead of remaining parallel. I don't know whether
this is a fault, or typical of this scanner. Actually it
would be two faults, since the effect is evident when
using the FH-2 holder (so there is some adjustment
required of the scanning mechanism inside the scanner),
and more severe when using the SA20 (so there is also
some adjustment required with the SA20).
So:
- if you want any chance of focused scans, don't use the motorised SA-20 feeder, and don't use curled film/slides.
- measure your machine's characteristics and see if you have a problem. If your variation with typical film/slides is less than about ± 8 units then you don't need to worry too much except for the occasions when you have a badly curled film.
- if your machine gives a variation of more than ±8 units, then to make the best of it, don't let the machine focus at the default position. This is at the very center of the image and is likely to be one extreme of the "bow" in the film (although this is more complicated if you have a front-to-back alignment problem like my scanner). My suggestion - assuming landscape orientation - is to focus in the middle horizontally, but a bit less than a quarter of the way from the top or bottom.
To be more accurate about this for an important scan you would need to measure the focus position all over the image, then choose a suitable location which is at the middle of the range and focus on that point.
I took a
couple of typical negs and used NikonScan 3.0 to measure
the focus point at a number of locations on each neg -
the locations shown as x's on the illustration...
To measure
each point - control-click on the focus button, then
click on the location of the required point on the
preview image. Read the value reported by Nikonscan under
Scanner Extras / Manual Focus Adjustment / Position.
Record the value in XL
and graph it.
Three sets of
measurements were taken:
Flat neg in normal position, and upside down. With
the flattest negative I could find in the SA-20
strip film feeder, record focus values for front,
middle and back of the neg in normal position
(emulsion down with scanner standing upright).
Then repeat with neg strip inserted emulsion side
up. The purpose of the inversion was to check if
this allowed the film to sit more flat,
particularly when the film was curled as in the
next set of measurements.
Curled neg. Same
as above but using a neg strip that was curved
side to side, and forward to back. It curved
towards the emulsion side both longways and across.
Compare the 2 different film holders. Compare
the same averagely curled neg in the SA-20 strip
film feeder and then in the FH-2 holder.
Results
Set
1 - Front-to-back alignment and effect of emulsion-side
up or down using a flat film.
The
focal plane was clearly not parallel to the
negative in a longitudinal direction. The film
was "tilted" along the long axis
towards and away from the scanning plane. This
effect dwarfed the side to side variations. The
extent of the misalignment is - I think -
enormous and unaccepable.
There
is also a much smaller "tilt" side to
side.
The
film becomes apparently curved side-to-side at
some points, presumably due to the fact that the
front of the film strip is forced to curl
longways into the holder which distorts the
side-to-side orientation in places.
There
is not a lot of difference between emulsion down
or up, except it curls the neg more at the back.
Some
parts of the neg are a l-o-n-g way from the
automatically determined default focus points for
the two cases - shown as the blue X's, but at
least the auto-focus position is kind of in the
middle of the range of values.
Set
2 - Same as 1, but with a curled film (curled
longitudinally and side-to-side)
The
longitudinal curl has a big effect on the
side-to-side curl, depending on which way the neg
strip is inserted - i.e. with the longways curl
upwards or downwards. Again, I guess this is
because the film is forced into the take-up
receptacle at the front, which requires it to
curl down. The big difference in the resulting
"shape" of the negs depends on whether
this forced curling goes with or against the
natural longitudinal curl of the film.
The
longitudinal curl causes the lateral curl to
invert at the front of the negative, which is a
bit surprising.
Again,
some parts of the neg are a long way from the
auto-determined focus point, and if you are truly
trying to get as much as possible in focus, you
would be better focussing at some other point.
Unfortunately without doing quite a few test
focuses, you can't tell what is the right
setting.
My
previous method of of focussing a third of the
way between the edge of the neg and the centre is
not helping a lot, but it does improve things
marginally. The advantage of doing this is
grossly outweighed by the longitudinal
differences.
The
sideways tilt is still evident. It appears to be
a characteristic of the SA20, since it is not
present when using the FH2 holder.
Set
3 - comparison of SA20 adapter against the FH-2 strip
holder.
The
FH2 holder is considerably better than the SA20.
Much
of the longitudinal focal plane error is due to
the SA20 holder - i.e. it does not position the
film parallel to the scanning plane as well as
the FH2 does.
But
some of the longitudinal error seems to be due to
the scanner alignment itself, as the FH2 still
shows the same sort of error in the same
direction, but less of it.
How to measure usable DOF
The actual DOF of the lens on the LS2000 was measured by focusing on a particular position of
a neg and manually varying the focus of scans (by typing in various focus positions to the "Position" box) in both
directions until the grain started to blur.
The depth of
field of the scanner is approximately +/- 6 units, or +/-
12 units if some softening of the grain was acceptable.
