Many simple schemes to read rolls into other formats are not concerned
with the exact punch events, while sophisticated systems intend to
create exact replicas of each punch event and the increment spacing
as the roll is advanced.  While there are quite a few ways to do this
electronically, here is a method which relies on a simple optical
principle and is adapted to direct output or to graphic manipulation
of scans.

If the roll is spaced slightly away from a piece of photo sensitive
paper and illuminated by a ring shaped light, the hole will let light
through in a large, diffuse pattern, making quite a big overlapping
blob of the result.  If the diameters of the hole and ring light and
the distance of the light from the roll and film from the back of the
roll are set just right, the light will overlap many times at a point
directly under the center of the hole.  If you increase the contrast of
an exposed print so that the gray area is washed out but the center
hot-spots are left, it is logical that only the centers of the original
punches will remain, even where the original punches overlapped.

The opportunity then is much simpler to impose a grid or roll advance
rate to match that spacing.

If, instead of a photosensitive paper, a bank of light sensors was
used, and the sensors were set at a threshold so that only the
brightest overlap of light would trigger them, then simply advancing
the paper at a close rate to the original would give quite accurate
punch events.  The advance rate could be fine tuned by matching the
arrival of a bright level at multiple positions, in physical terms the
same as the electronic phase-locked loop.

It is possible, I suppose, to make an image of the holes, copy it in
a circular array at almost half the diameter of the holes, then
increase the contrast until only the center nugget remains, but I have
not demonstrated this.

Karl Petersen
Washington, Illinois