Hi,
I demoed the M1 to Peter Kirn and some of his friends last Thursday. He
came up with a few interesting ideas:
(1) "Differentiate" the camera signal. Instead of drawing the direct
picture from the camera, draw the difference between the last two
frames, e.g. defined as:
diff(x,y).c = |f1(x,y).c - f2(x,y).c| for each color channel c=r,g,b
The TMU provides most of the infrastructure to perform this difference
operation quickly. We can modify the arithmetic circuit that performs
alpha blending so it computes this difference instead. The software will
then have to swing a bit the video buffers.
This can help make some patches more motion reactive (and static objects
like background are canceled out).
(2) Provide pre-built maps for common MIDI controllers. Maps are already
saved in performance files, but it could be convenient to have "partial"
performance files that only alter the MIDI map. Maybe Werner is on this
already.
(3) Make the internal texture buffer not rectangular but have a
user-defined shape. I'm not exactly sure how to do this, maybe with an
alpha mask and a PNG/JPG picture.
S.
_______________________________________________
http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org
IRC: #milkymist@Freenode
