Can normally static parameters become dynamic?

It would be funny, use the piano lid in a pedal, as midi control, making the piano like the mouth of a talking puppet.
:-)

By the way, I wonder why we can't remove the lid, make it more than fully open, like in recording studios when they take the lif off.

The point of mic relatade to the lid, when the mic it's close to the lid, and one more few inches it would pass through it, should have more effect, since the mic after 3 inches would be on the other side and have the lid as a barrier.

I ask people with prime speakers/monitors or a 5.1 system with good speakers, to make the test:

-Place the mic very close to ther lid edge, test the the sound.
-Now place the mic just a little bit more in the direction of the lid, to pass through it, so it will be very close to the lid but in the other side.  And test the sound
-Compare both sound results.

Lifting/lowering the lid reminds me of the desperate attempts to provide dynamics in the late days of the harpsichord.  The "Venetian Swell" was one such device. You lifted a lower lid which resembled a venetian blind. You could either open/close the blinds or lift the whole lid. By foot pedal, but lifting the whole thing took some force. (A good workout for one leg!) You can - or could - see examples at Fenton House in London.

As a harpsichord player for many years, I can allow myself the amusement of such a device trying to stave off the inexorable rise of the fortepiano.

Modeling classic pianos makes so many _other_ instruments possible!
Once you allow straight and crossed strings, I'm pretty sure the internal geometry of the piano is completely arbitrary to the engine. The strings could be placed in a circle, to bring them closer together. It is all a question of surfacing the parameters in a user interface.

Here's a quick way to imagine it in the Pianoteq Midi map interface:
Column 1 accepts velocity and note on in addition to Controller 1-119.
Column 2 accepts Controller 1-119 in addition to all the parameters, etc.
The velocity or note on value (0-127) is inherited by the Controller if they have been linked, and mapped to the appropriate range.

So
Controller 1; (any); Wall Distance; 0-6
Note on; (any); Controller 1

With the above map and the assumptions above:
You can move the wall distance with Controller 1, just like you always could.
Every time you play a note, the note's value is passed to Controller 1.
On the next update cycle, Controller 1 changes the value of wall distance (0-127 is scaled to the interval 0-6). The overall effect is that lower notes are played in larger rooms than higher notes.

I assume that the engine just has a queue of controller changes that it processes as fast as possible. All this does is put an event on that queue, instead of the event coming from the Midi controller. This came up in another thread, where pressing the pedal in the UI caused the engine to react, even though it did not create a Midi event. If the UI can queue up an event, it should be possible to remap velocity to queue an event.

Obviously, there is the possibility of creating feedback loops that will mess you up. It was already possible to make Pianoteq do strange and sonically painful things, so this is par for the course, caveat emptor, yada yada yada.

Changing some parameters can sometimes require a recalculation of the model and may take some time.  Therefore, the application of a parameter may be longer than the time the sound is generated from the key press.  But in version 7, changes in parameters are reflected in the sound rather smoothly.  And in this you can find your charms.

I agree, and I'd like to see more parameters have the "humanize" setting that we have on select action and voicing parameters.  I'd also appreciate a velocity randomizer for things like MIDI Piano Roll playback, since I use the standalone a lot and the only randomize functions that are easily available require a DAW (which I have but it doesn't make it convenient).

At a broader scale, I'd love to see more research going into machine learning to train a MIDI sequencer to reinterpret an input into another style or even genre.  For example, being able to pass a stock, quantize MIDI file of a Chopin waltz and let the software interpret and perform it in the way that Rachmaninoff, Godowsky, or Artur Rubinstein would or even what contemporaries would do with something would be an amazing composition tool.  A tool that gives improv in different styles based on a single input.  Yes, there are some projects that are starting to head in that direction, but there seem to be very few people who have the musical know-how to inform good decisions in building a training database from something like piano rolls--or have the money to commission a database of new MIDI files for the AI training.  (On the one hand, it would give a lot of pianists with nothing to do at the moment some work and income, but it can also scare them away if they decide that it could further put them out of work in the future to have a computer doing what they feel is their unique, human contribution to music.)  As a composer, I'd love the idea, but then getting people to play my music is enough of a challenge that having AI that doesn't quit half-way through a project is an especially appealing option.