Distortion by copying machines

Paul Butler (pgb@kau1.kodak.com)
Mon, 31 Jul 1995 09:04:17 +1000

Jess Stuart writes

----- Begin Included Message (part thereof) -----

I understand that copying machines distort large images. How do
you take the distortions into account? Are they significant enough to worry
about, or am I just acting like chicken little??

----- End Included Message -----

The following has been my experience:

1. A 1:1 reproduction on a typical photocopier seems to introduce negligible
distortion (also see below about problems with original drawing) but I
would expect a series of enlargments to compound any distortion that might
be present in the enlargement process (using a photocopier). To my knowledge
all lenses introduce some distortion particularly at the corners of a
rectangular image. But of course, this can be minimised with quality lense
systems used by a knowledgible photographer.

2. Many years ago (circa 1964), I had a professional photographic enlargment
shop enlarge the 1:48 scale drawing of the Stinson SR6 blown up to 1:12.
At the time I was very pleased but when I came to build the model I found
many size and shape discrepancies in the prints. Some were due to the
enlargment process, (each drawing had to be done in two parts due to the
limitations of the equipment) and some where probably due to discrepancies
in the originals. I built the model without too much trouble but it didn't
fly because it was too heavy.

3. In recent years I have investigated the quality of published originals
by carefully scanning them and flipping the resulting image. By overlaying
each image on the computer screen (each being a different colour) and
comparing them, differences are readily apparent. Due allowance must be made
for errors introduced by the scanning process itself. For instance, some
of the lower end scanners may not scan a perfect circle properly; one axis
being slightly bigger than the other but I have found the error is usually
less than 1% but it is still there. The other problem is that it requires
a lot of care to get the image square to the orthagonal axes of the scanner.
Generally I am happy if I am within 3 pixels over a wing span or fuselage
length on a 600dpi scan.

What I have found:

a) Wing times of different shape, usually slight but sometimes quite
noticable.
b) Wing and tail plan forms a little different between top surface and
bottom surface views.
c) Slightly different rib spacing.
d) Bent fuselage in plan view

It is very likely that few, if any of the above are present in the
draftsman's original drawing. The printing process may introduce errors and
paper is not a stable medium. It will shrink, expand and squirm with
change in moisture content. So when you are looking at very small
differences (like for a 1:72) model instability in the paper could easily
account for a few 1/100 's of a millimetre.

Of course you must be aware of the possiblity of such "errors" actually
being
present in the fullsize aeroplane (I think the Ansoldo S.V.a is an example)
but all the forgoing is mean't to point out the dangers of enlarging small
drawings to large scales without going to the trouble of confirming the
accuracy of the result.

Paul Butler

pgb@kau1.kodak.com