Paul Bell [in 241230 MMDigest] raised an interesting point and
a problem that I've also made compromises on with producing rolls
by compressing the length.

Since the first run of rolls I made I've been using 0.066 inch punch
diameter. This is the same size used by QRS since the 1980s. Due to
this smaller punch diameter I had to shorten the piano roll masters
to make bridging normal thickness (otherwise it was way too thick).

I made the punch diameter smaller mainly to give a larger margin for
error should the rolls decide to mis-track for whatever reason (or if
the holes wandered relative to the edges of the roll which are trimmed
first).

So far repetition hasn't been an issue with the smaller punches in my
case. I had more problems playing QRS's yellow paper rolls which have
some punches so small they won't trigger _at all_.

I've compressed the length on the rolls only enough to leave very
thin bridging (as the bridging in the end is only there so that my
perforator won't try to cut slots longer than an inch. (I need to
figure out Spencer's segmenter to only insert bridging every 20 rows
period instead of having 20 rows empty slot at the start and end of
each note.)

Compressing the music using the bridging as reference is pretty
much bullet-proof as far as I can tell and the slower paper speed
compensates for repeating notes being closer together, giving adequate
time for things to reset.

So far the compression scheme has not had any adverse effects on
standard rolls. I've had mixed results from people with reproducing
rolls. Nick Brown (his channel on YouTube has many recordings of his
Knabe collection) had no issues playing two recuts I made for him of
Ampico rolls. However, another customer (who received rolls from the
same run) had issues with perforations triggering the rewind and
marginally close tracking that was almost off enough to cause playback
problems.

One thing that I have taken into consideration is the fact that the
first run of rolls was on old paper that was being cut oversized and
then dried to shrink down to normal dimensions (which in the end was
extremely close to being too narrow, 285 mm which is also QRS's roll
cutting width since the 80s).

I've since replaced the paper with new (and higher quality) material
which allowed me to cut at normal width again and the cuts I made in
that run have played exceptionally well for everyone that got them,
as far as I'm aware.

Something to note about compression for Ampico rolls in particular.
Mike Swanson sent me punch master MIDIs for several Ampico rolls and
they were all mastered at 29 punches per inch, I believe. However,
when I opened them in Perforation Editor to visually inspect them,
I found all the bridging was too thick. I set the step rate to 36
punches per inch. (Duo-Art's standard, I think? Julian Dyer might
have some more insight on this.)

That change completely nulled all the issues with bridging except for
one missing punch that I fixed myself. This did result in compression
of sorts but all other shortening I've done using the scale file in
M2G and have worked out ratchets for every major stepping standard.

I haven't attempted to cut any Duo-Art rolls as of yet. Spencer Chase
and Bill Luecht have made a Duo-Art scale for the Gcode converter so
I might give it a shot; however, I think those rolls will have to be
pretty much cut exactly as they were mastered, as from what I've heard
the Duo-Art is extremely finicky with snakebite timing, among other
things.

On a different note entirely -- this is for anyone who has my cuts
-- I did _not_ cut the leaders and trailers long enough to play well
on Duo-Arts unless you hold the play button for a few seconds to get
the roll wound around the spool a couple times. Frederick Hodges let
me know about this issue that he had run into with the rolls on his
Steinway Duo-Art.

I'm probably going to redo some of the masters and fix my configuration
so that the leaders are long enough to work on pianos that need a full
wrap around and also I will be adding the universal rewind holes at
the end of every roll (Ampico, Duo-Art, and Welte rewind triggers).

Piotr Barcz
Upstate New York