Ok, I'll do it: additive synthesis will give you "perfect" results as long as (1) you don't truncate the harmonic series too early (ie. you synthesise enough to reach Nyquist frequency) and (2) your sine-wave calculating is numerically accurate. The main restriction is that you need to be able to calculate (either analytically or by FFT of a reference waveform, or something) the amplitudes and phases of the harmonics individually, which can be (highly!) inconvenient when it comes to "creative modulation."chk071 wrote:OP seems to have a history of "great ideas". viewtopic.php?f=33&t=479007&p=6699751#p6699751 I was actually hoping for a dev to chime in, and put an end to this esoteric nonsense. Oh well, maybe in the OP's newly created thread, which is actually the same bla bla bla without any factual base.
Practical implementations usually don't produce "perfect" results because it's actually better to fade out the high harmonics gradually before Nyquist frequency just to avoid clicks with pitch modulation and to put some upper limit on the number of harmonics generated, so you don't run out CPU if the user decides he wants a 0.0001Hz waveform. Such a practical implementations are essentially "better than perfect."
Either way, if you think you can generate better waveforms than what you get from additive synthesis (when done correctly, duh), you need to stop eating mushrooms.
PS. If you think you can get similar quality for less CPU or less inconvenience for certain types of creative effects, then sure.. that's useful.. but please don't try to pretend you can beat the "perfect" in terms of quality, that's just non-sense.