[PD] pix2sig and back

Jack jack at rybn.org
Wed Oct 25 00:33:22 CEST 2017


Hello,

Le 24/10/2017 à 11:24, Peter P. a écrit :
> * cyrille henry <ch at chnry.net> [2017-10-24 13:32]:
>>
>>
>> Le 24/10/2017 à 13:18, Peter P. a écrit :
>>> Hi list,
>>>
>>> I am trying to convert an audio signal to pixel data using pix_sig2pix~
>>> and then back to audio data using pix_pix2sig~ without further
>>> manipulations. Please see attached patch.
>>>
>>> I do scale the audio signal from -1 to 1 to lie in a range compliant
>>> to pixel data, which I assume to be 0 to 1.
>>>
>>> Not having connected an input to the subpatch, this signal will hence be
>>> 0.5 and will display in the red color channel.
>>>
>>> Converting that red channel back to audio signals and doing the opposite
>>> scaling operation, I am getting a 51dB constant signal at the output
>>> even without having attached a signal to the input of pix_sig2pix~.
>>> Checking with a snapshot~ object indicates a constant signal of 0.003921
>>>
>>> I assume this to be due to limited resolution numbers in the pixel data
>>> and wonder what their resolution is?
>>
>> 8 bits / channels
>>
>> so resolution is 1/256 = 0.0039...
>>
>> You can code MSB in Red, and LSB in green is you want 16 bits...
>>
>>>
>>> Then I texture the pixel data onto a rectangle. This rectangle
>>> does flicker and flip diagonally from time to time and I wonder if it is
>>> an error on my machine only and where it could come from.
>>
>> I don't see any flicker in your patch. (linux / intel GPU)
> 
> Thank you Cyrille for this helpful information!
> 
> Jack, I am on Debian GNU/Linux with an onboard Intel graphics controller
> (i915 variant).

OK, i guessed you were on nvidia because if you are not sync to vblank,
you often get a line diagonaly separating two pictures. It is called
"screen tearing".
I didn't test your patch yet...
++

Jack


> 
> P
> 
> _______________________________________________
> Pd-list at lists.iem.at mailing list
> UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
> 




More information about the Pd-list mailing list