Originally Posted by selig
Here's my confusion: oversampling addresses aliasing. When Predator came out I compared it's aliasing to Thor and Predator was much worse (Thor was quite impressive, btw). Predator has been updated, and my trial expired before I could compare them again, but still…
Problem for me is, Predator has 32x oversampling and Thor (supposedly) has none. So I'm confused why folks think that oversampling will automagically fix things that in some cases aren't even an issue in the first place? As always, I'm open to the fact I may be mis-representing the issue here or unaware of things that could have led me to different conclusions…
So my question is; in exchange for the CPU hit, what do you expect oversampling to do for you?
It's not that simple. I really hate to bring this topic up in public but the Predator issue might have been due to bad coding. In short, as far as I understood, not the entire chain of Predator was oversampled, only the oscillators (and then an option for the filter, on the VSTi that is). This might have contributed to that fact (i.e. what's the point of only oversampling the oscillators if the other stuff happen at normal audio rate?). Fwiw, I don't know what they did but that update to Predator did improve the waveforms/reduce aliasing overall.
Other developers have stated (not about Predator, but oversampling in general) that it is not just about having X amount of oversampling. Depending on the synth layout you might need to downsample as well. I.e. oversample oscillators, downsample before it goes into the filter oversample inside the filter, etc.
But, all that said and done. The thing that actually speaks FOR oversampling when it comes to Reason devices in particular is that some of them actually sound better when running at an higher audio rate (i.e. both Subtractor and Thor sound more detailed/alias less at i.e. 96 kHz compared to 44.1 kHz) Thor is less prone to aliasing artifacts in the FM Pair Osc when running at higher rate as well.