Several years ago Greg Erker put together a group buy of Tominon
lenses for people on this list. As part of the deal he checked all
the lenses against each other and provided the best matches that could
be arranged from among the samples we bought. He did this by
projecting a slide through the lens and measuring the image on a wall.
As I recall even the worst possible pairing of the lenses we got was
better than 1%. When he was done picking the best pairings they all
matched by much better than 1%.

One way to measure the flange to film plane distance at infinity focus
is to use an autocollimator. You can do this with a view camera, a
first surface mirror, a pinhole, a bright light and a focusing loupe
(4X is good enough). Mount the lens on the view camera. Place the
first surface mirror over the front of the lens pointing back towards
the film plane. Place the pinhole on the ground glass and then
position the light behind it. With the loupe on the ground glass you
should be able to see the reflection of the light behind the pinhole.
When that image is in focus the camera/lens is focused for infinity.
By measuring from the front of the lensboard to the ground glass you
have the flange distance for infinity focus.

Mark the focus rail at this position. Focus the camera/lens (after
removing the mirror) so that you produce a 1:1 magnification.
(QuickDisc <http://www.salzgeber.at/disc/index.html> is a free target
for determining magnification and exposure compensation.) Mark the
focus rail at this second position. The distance between the two
marks is the exact focal length of the lens.

Richard Knoppow has written a lot about lens design and figuring out
lens specifications like focal length. If you use Google Groups
<http://groups.google.com> advanced search to search for his USENET
rec.photo.* newsgroups postings you can find more detailed
instructions and cites to various books.

--
Brian Reynolds | "It's just like flying a spaceship.
reynolds@panix.com | You push some buttons and see
http://www.panix.com/~reynolds/ | what happens." -- Zapp Brannigan
NAR# 54438 |

> Several years ago Greg Erker put together a group buy of Tominon
> lenses for people on this list. As part of the deal he checked all
> the lenses against each other and provided the best matches that could
> be arranged from among the samples we bought. He did this by
> projecting a slide through the lens and measuring the image on a wall.
> As I recall even the worst possible pairing of the lenses we got was
> better than 1%. When he was done picking the best pairings they all
> matched by much better than 1%.
>

You did not mention the details of how
this test was executed..... regardless, I think a home-brew execution of
such a test has too many variables to control AND measure - thereby
making the results very questionable. Without getting into a long
winded response (I think the list has endured enough), the variables
associated with this method are very similar to those with macro
photography (but in reverse) as it relates to image scale. As you
focus (film to lens distance), you change image scale (which is a
function of lens fl, which is unknown here), and when you alter the
lens-to-wall distance, you also change image scale. This is similar to
trying to solve an algebraic equation with two unknowns...... I would
be curious how this can be done, but I doubt the exercise is worth it,
as scanning the film is simple, cost effective and highly accurate
....assuming a decent scanner and as mentioned by a previous poster,
being consistent with film scanned. CS3 has a measuring function which
really increases the accuracy.

> Mark the focus rail at this position. Focus the camera/lens (after
> removing the mirror) so that you produce a 1:1 magnification.
> (QuickDisc <http://www.salzgeber.at/disc/index.html> is a free target
> for determining magnification and exposure compensation.) Mark the
> focus rail at this second position. The distance between the two
> marks is the exact focal length of the lens.
>

To set infinity focus, its probably easier to
focus on an infinity subject...but regardless, the premise of this test
is correct, and it will get you in the ballpark....however, nowhere near
as close as scanning film. The reason is....you have two variables
here... the change in position of the rear standard (from infinity to
1:1) can be measured quite accurately.... assuming you measure at the
rail, not at the standard, but, to measure 1:1 on a ground glass is very
difficult to the degree of accuracy required. For example, if your
measuring error is 2% on the gg, this error will translate to the same
2% error in the lens fl. If the image on the object is near 50mm on the
ground glass, 2% = 1mm or 1/25th of an inch. Considering a gg is a
very low resolution screen, with a lot of bleed-through at the object
edges. As a result of the gg shortcoming, I would NOT consider this
method very accurate, at least for the intended purpose here. If you
had a fiber optic plate at the gg, the test would probably be
acceptable...... but a fiber optic plate of this size would cost more
than a high end scanner. Remember, 2% in fl is very significant
here.... that is the entire purpose of the test....


Other than scanning film, determining actual fl (vs.
lens fl variance) to the level needed for accurate lens pairing, would
probably require an optical bench at a bear min. These home-brew
methods are designed to get you close, and in most all forms of
photography are very acceptable. But IMO, pairing lenses is one of the
details unique to stereo photography in which these home-brew methods do
not address.


The film scanner method simplifies this entire
process, as you have fixed all the variables.... same camera body, same
lens-to-subject distance, same lens-to-film distance, same apt., same
scanner, same measurement technique, AND you have the benefit of
magnifying the errors! Assuming the lenses tested are at least close
to their nominal fl, then even focus can be fixed. So the only
variable that remains is the object size on film, which fortunately can
be measured very accurately in CS3, as you are magnifying the object
size 20x, which will provide 20x the measuring accuracy (actually
more). And even the scanner method will only get you close to the
actual fl, but it will be VERY accurate in determining the variance in
lens fl's, which is all that matters here.