I don’t think it’s timing…you can null any sample based synth or drum machine fine which suggests render timing is fine in general.
I guess it’s something about the analogue modeled synths specifically (prologue and the couple of korgs I tested were the same)…but confusingly you can make multiple renders that will all null with each other which seems to rule out purposely introduced randomness.
Yes, a major problem comes from the free running oscillator of any synth.
Scenario 1:
Set the Retrologue to fixed oscillator phase in the instrument track
duplicated this instrument track
polarity inverted the duplicated track. The result:
it nulled instantly as it should be!
Scenario 2:
I RIP this fix phased osc MIDI pattern.
The audio is then phase inverted.
Now the original instrument track and the polarity inverted RIP audio track runs parallel, the result:
I get sounds occasionally, so RIP does not seem to be sample accurate!
Oh, so you ARE a techie testing audio.
For what it’s worth, MIDI and analog synths have always phased for me.
Not that it actually makes any difference in the music.