CyberGene
QuasiUnaFantasia there has been a speculation that Pianoteq is not about modeling the piano as an object that produces sound but rather a clever sound compression which works with samples as an input and will then analyze the partials, their amplitude curves with the time, etc., and then when the sound is generated, sine waves will be summed and their amplitudes will be controlled by following the analyzed curves. That's based on the only patent that Modartt have published many years ago. They have recently stated that it's old and they are doing better modeling, however they have never uncovered what that means.
I think many people get confused over what modeling actually is. They imagine that the computer holds a virtual piano model in memory and the mechanical system is modeled with its geometry, materials, moving parts and ultimately we have a virtual hammer striking virtual strings that produce audio. That's called finite analysis and it's been stated that it's so difficult to do it in real time that (if I remember correctly) to produce 1 second of audio from such a model would require 24 hours of intense computation by a very powerful computer. With that in mind, I think the term "modeling" is used pretty vaguely nowadays and in the case of Pianoteq most certainly means (IMO) that they model input audio samples rather than the actual piano object as a mechanical system. Or, in other words, it's more a synthesizer with parameters that allow it to behave like a piano better than a conventional sampler would do. But it's definitely not a physical model the way most people imagine it.
