Traveling Between Models: From Steinway City to Petrofland (e.g.)

Hi everyone,

I think I understand what physical modeling is, the simulation of the behavior of strings, hammers, soundboard, etc. But I’m not a physicist, and I don’t really know how far that “tuning” of the model goes until it becomes difficult, if not impossible, to distinguish it from its physical “analog” counterpart (and this could apply not only to Pianoteq but also to others, including software synths such as those by Arturia, for example). Each instrument is modelled and using a set of internal parameters and user adjustable parameters . While you can play and modify the user parameters , you cannot change the internal parameters for a particular instrument , so at best by modifying string length and tuning parameters you

So here’s my question:
Could we, starting from one model, get close (a little or a lot) to another one by reprogramming each parameter note by note?

My apologies in advance if this sounds naïve or too simple — they may have already been discussed here. But as a practical musician (though with many years of experience using and programming synthesizers), these questions naturally come to mind, because the technical barrier — the physical/mathematical complexity of physical modeling — has always seemed a bit mysterious, even arcane to me (unlike virtual synths, which are much simpler and more accessible).

Thanks,

Could we, starting from one model, get close (a little or a lot) to another one by reprogramming each parameter note by note?

Well, you can change settings to make closer until some point, but also depends of how far it is by the initial state of nature of the piano model you take compared to the target piano you want to try to recreate.
And it demands patience and skills.

It's like make-up in movies, which allow some actor to look somehow closer to a historic character they will portrait, but not like identical twins.

Maybe one day Modartt create a A.I version of pianoteq, connected to a data banks of hundred piano brands/model characteristics plus dozens of possible tuning&voicing  variations of each model, and we would says to pianoteq what we want, and pianoteq A.I would sellect the best starting modelled piano as starting point and do the settings that are more adequate to try to reach the desired sound.

jmanrique wrote:

Hi everyone,

I think I understand what physical modeling is, the simulation of the behavior of strings, hammers, soundboard, etc. But I’m not a physicist, and I don’t really know how far that “tuning” of the model goes until it becomes difficult, if not impossible, to distinguish it from its physical “analog” counterpart (and this could apply not only to Pianoteq but also to others, including software synths such as those by Arturia, for example).

So here’s my question:
Could we, starting from one model, get close (a little or a lot) to another one by reprogramming each parameter note by note?

My apologies in advance if this sounds naïve or too simple — they may have already been discussed here. But as a practical musician (though with many years of experience using and programming synthesizers), these questions naturally come to mind, because the technical barrier — the physical/mathematical complexity of physical modeling — has always seemed a bit mysterious, even arcane to me (unlike virtual synths, which are much simpler and more accessible).

Thanks,

daniel_r328 wrote:

You can manipulate some of these hidden variables using the Morphing feature which combines the parameters of multiple models, which gets you one step closer but not all the way.

Indeed, I was about to mention the Morphing too!

jmanrique, Morphing allows you to change the sound continuously from on instrument to another. Here is an example:
https://youtu.be/k2oZlMueJCM?t=19
(video is a bit old but shows quite well what it can do I find)