CyberGene
RIP They actually say that sampled pianos are lame because they have only a limited number of layers, whereas they have at least 127
I know that and I’ve seen people citing it with a wrong preconception. Even if it was only one single sample layer (and even if there wasn’t any filtering applied), it would still be replayed with 127 different gain levels, depending on the MIDI velocities, so that’s already 127 different sample layers. Or 1024, or whatever gain multiplier you apply. And many modern samplers use interpolation (and filtering), so the discrete sample layers are just reference points but you will have as many different spectral and volume differences in between, as there are MIDI velocities. Same as Pianoteq. So that’s a moot point.
I’ve mentioned this across different forums and threads, even in this discussion, but listening to a recorded piano (as in a commercial recording by an artist) doesn’t sound like listening to a real piano at a concert.
If you need actual piano experience, then the software (modeled or sampled) is only one part of the goal. There’s the reproduction system, vibrations in the keys and so many other things.
I personally don’t need that. I use software to produce studio quality recordings. Sampled libraries give me exactly the same quality and feel.