[PD]Creating noise in Gem
IOhannes zmoelnig
zmoelnig at iem.at
Mon Jun 3 10:43:24 CEST 2002
? wrote:
>> [Rory Walsh]->[Re: [PD]Creating noise in Gem]->[02-05-31 17:22]
>>
>> |I can't seem to find the sig2pix~ object? Is this an
>> |extern? I'm using windows 95, thanks again for the
>> |help.
>>
>> got it from
>> ftp://iem.at/pd/Externals/GEMEE/
>There is a pix_sig2pix~ in the new gem release, I am not sure
>if it is the same one that was in the gemee ....
pix_sig2pix~ (from GEM>0.87) is not exactly the same thing as sig2pix~
from GEMEE:::
GEMEE's version uses a per-line interpretation of signals (this is:
signal vector n is used as pixel-line k; in the next dsp-cycle, the same
signal-vector n will be used as pixel-line k+1 (at least, that is, how i
understood it)
GEM's object is a per-image interpretation: (thus signal vector n will
be interpreted as y consecuting lines if pixel-data; in the next
dsp-cycle, the old signal-vector will not be re-used) this is the same
approach as in the SOLVE-library by Michael Droettboom.
in practical terms this means:
GEMEE's object will eventually have scrolling images (which might be no
problem when generating noise) because the signal-vectors are reused.
i thought this somehow unpractical, especially, when converting images
to signals, doing some manipulation (in dsp-domain) and converting the
signals back to images.
images created by GEM's object will always stay synched to some external
clock and therefore will not scroll. however you can obtain a scrolling
effect by delaying the signals.
the major drawback is, that you will need large blocksizes if you want
high resolution. eg: for a 128x128-image, you need a [block~ 16384]
somewhere, else only a part of the image might get drawn (for example:
only 64 samples, which would equal to half of the first-line)
of course you might want to set an overlap-factor (or upsampling ???) to
get more than 4 images per second.
mfg.cdsa.dr
IOhannes
PS: there seems to be a bug in the reverse pix_pix2sig~ object, which
keeps crashing my machine...(have had no time yet)
More information about the Pd-list
mailing list