Re: Matching lenses (was: Re: HTML help)

[erker@xxxxxxxxxxxxxx (Greg Erker)]

>Since my two image reference marks were 6' apart on the projected
>image, a difference of about 1/3 of an inch represents a 0.5% mismatch.
>It was fairly easy to pick lens pairs that had MUCH better matching
>than this. I picked three pairs with better than 1/8" matching, which
>is (trying to allow for measurement error) 0.2% or better.
>
>Thoughts? Anything I've overlooked?  I seem to recall that 0.5% is
>a generally-agreed on number. Is that right?
>
>Joel.

  That's a great method for cameras
with interchangable lenses. For my
TLR's I have to shoot an awful lot of
rolls of film to do this (or attempt
to reroll one film a bunch of times
and load it into the next camera.)

  Using two points is fine as long
as the lenses all have the same
amount and type of distortion.
Suppose your first lens has zero
distortions and is exactly 50mm FL.
The second one might be 2% shorter
FL and have 2% barrel distortion
that makes the two points come into
alignment, fooling you into thinking
the lenses match.

  So, I think you can use this method
to match up Pentax 50mm f2.0's with
each other, but wouldn't try to match
f2.0's with f1.7's or f1.4's or other
mfgr's lenses (due to the different
number of elements in the Pentax lenses
and construction differences with
other brands).

  We discussed this before, but, for
my TLR matching I'm going to use a
test slide at the film plane of my
cameras and us it to project the image
on the wall by shining a light thru
the back. The slide will lifted off
the camera film rails by a thick piece
of glass so that the image is in focus
at a reasonable distance, rather than
infinity.

  I haven't made up my test slide yet.
Plan to do it on the laser printer with
an overhead transparency sheet. To avoid
the 2 point measuring problem I will
probably make a sqaure from 8 dots. Four
forming the corners and the other 4 on
the middle of the straight edges.

  If they all have the same distortion
then I'll just measure one diagonal (to
have the best accuracy). If not, I'll
have to match them by distortion shape
and amount as well.

Greg