Beto-Music wrote:The K model it's very good, but it was also created using references, sound samples to be analized to allow it to be recreated by modelling. It's a amazing tool to recreate existing piano models, or a piano models using elements from more than one piano (like they did in Grand K).
I mean create a modelled piano from scratch, like a grand piano is created. Starting from soundboard, wood characteristics, strings, shape, hammers materials. If one thay that became possible, they could use it to help the creation of true piano model, since the virtual modelled piano would work as a simulation.
But such thing would demmand a enormous amount of computer power.
Key Fumbler wrote:Beto-Music wrote:I wonder when will possible to design a digital piano from scratch, using modelling technology, without use (analyze) sounds of other pianos. I mean design in the digital realm in a similar way real piano designers create their pianos.
What do you mean by that?
Maybe I've got the wrong end of the stick?
Surely all the parameters are there already for Modartt?
Surely they know how each and every parameter relates to the real world equivalent?
Through perhaps as they measure each and every different model that they come across their knowledge base grows wider. This experience meaning such a fully digitally imagined piano could be that much more natural.
We already have an example of that creative process in software don't we?
The K model is only based loosely on multiple different pianos, and doesn't therefore exist in the real world.
Maybe we could see a new super K for a K Pack?
Ignoring the unfortunate Kevin Spacey name tie in!
We don't really know how the physical modelling engine works. I'm sure if they tried to tell me I wouldn't understand half of it anyway.
We don't know what short cuts are skillfully made to allow an average modern CPU to reproduce a modelled piano in real-time that can fool the human ear.
I'm sure Modartt could by now (with the existing engine) produce an excellent sounding believable piano with no point of reference at all, other than memory and wish list, together with extensive knowledge of every single parameter they have at their hands.
Or you could get a quantum computer to reproduce every single aspect of the physical behaviour of the hammers alone (the parameters we fallible humans are aware of). Each and every strings individual and collective behaviour, calculating how each and every resonance in the piano changes behaviour of all the other structures in the piano. Maybe you'll end up with 100 billion times as many calculations for 1% better sound. And you're still find people that actually prefer the sound from the Pianoteq engine, because it's all subjective.