The biggest show stopper is the increased delay.
I installed the standard edition of 5.5.0 (trial), and there, I don't notice the delay. Even with 10.7 ms latency, the delay is acceptable.
With stage (purchased) the delay is noticeable at 2.7 ms latency and unbearable at 10.7 ms latency.
How comes that there's a noticeable difference in delay between stage and standard?

With respect to the changed instrument sound, I think you (Qexl and Amen Ptah Ra) are right: I have to get used to it and maybe change a few settings. (b.t.w. on the recordings, I was not playing myself (I wish I could). It was a midi-file).

I also notice some sort of latency? is this by design?

aandrmusic wrote:

EvilDragon wrote:

Pianoteq does utilize the physical model indeed. But that model is tweaked via many, many parameters (which we don't see on the interface) in order to get close to a reference, which is obviously going to be a recording of some kind.

For Ant. Petrof, they used samples recorded in an anechoic chamber, which gave the cleanest possible reference of that particular piano.

For all sorts of practical reasons, yes a recording (of some kind) would have to be used (at some point), at least to calibrate the model.

One of the objections to sample libraries has been that they include all the artifacts and limitations of the recording/reproduction process.
I guess those are constants that both camps are simply stuck with...

As you say, samples are indeed needed to calibrate the model, they serve as reference. But of course recording artifacts such as background noise or embedded reverberation don't go through to Pianoteq.

I'd guess you basically "subtract" the sound of the model from the recording - if what remains sounds like something significant figure it out, repeat until only noise and recording artifacts remain.

Yep, I could set up a lab to do that <grin>

Just a note to say "Thanks!" Pianoteq Pro is by far the best purchase I ever made. I'm eagerly awaiting Organteq, and I'd buy pretty much everything from Moddart sight unseen - keep modeling!!

Which leads me to a question - I'd like to buy all of the sound packs, however I see no 'upgrade' path from pro - is there one? I would prefer to save a bit of money if possible while still supporting them, so does Moddart ever run any specials on all sound upgrades?

,

The only "everything" bundle I know of is the Pro Studio - the list price of that is approx twice the price of Pro (which already includes 4 instrument packs).

Historically ...There have been Black Friday sales - in some years an early summer sale (close to now).
As they say in investing past performance is not necessarily and indicator of future performance (something like that).
On the other hand, once a company starts a tradition of discounts at particular times of year their customers tend to wait for it - so there is pent up demand just WAITING and watching the calendar.

I think some decent discount for "everything I don't already have" would be nice.

Which leads me to a question - I'd like to buy all of the sound packs, however I see no 'upgrade' path from pro - is there one? I would prefer to save a bit of money if possible while still supporting them, so does Moddart ever run any specials on all sound upgrades?

EvilDragon wrote:

Pianoteq does utilize the physical model indeed. But that model is tweaked via many, many parameters (which we don't see on the interface) in order to get close to a reference, which is obviously going to be a recording of some kind.

For Ant. Petrof, they used samples recorded in an anechoic chamber, which gave the cleanest possible reference of that particular piano.

For all sorts of practical reasons, yes a recording (of some kind) would have to be used (at some point), at least to calibrate the model.

One of the objections to sample libraries has been that they include all the artifacts and limitations of the recording/reproduction process.
I guess those are constants that both camps are simply stuck with...

As you say, samples are indeed needed to calibrate the model, they serve as reference. But of course recording artifacts such as background noise or embedded reverberation don't go through to Pianoteq.

Pianoteq does utilize the physical model indeed. But that model is tweaked via many, many parameters (which we don't see on the interface) in order to get close to a reference, which is obviously going to be a recording of some kind.

For Ant. Petrof, they used samples recorded in an anechoic chamber, which gave the cleanest possible reference of that particular piano.

For all sorts of practical reasons, yes a recording (of some kind) would have to be used (at some point), at least to calibrate the model.

One of the objections to sample libraries has been that they include all the artifacts and limitations of the recording/reproduction process.
I guess those are constants that both camps are simply stuck with...

OK, I can live with that<grin>  - PTQ still has smaller footprint and other advantages.

For Ant. Petrof, they used samples recorded in an anechoic chamber, which gave the cleanest possible reference of that particular piano.

It seems to imply that pianoteq (at least in some cases) doesn't model physical pianos, but instead imitates sampled sound libraries.

http://forum.pianoworld.com/ubbthreads....844/2.html

Scroll down to an entry by member C

"Hm, I just noticed in another thread the “Bechstein Digital Grand” sample based VST made by Bechstein themselves. I thought that particular DG reference was a fancy name invented by Pianoteq but a second read of their description actually confirms they modeled their sound after the samples of an actual piano and not after an actual piano. Which actually confirms once again a previous statement some of us have made: Pianoteq synthesizes the sound in a way that recreates a pre-recorded sound. Whether they can also model the physical object which we call “piano” rather than approaching recorded waves is a matter of dispute. Or maybe they support both? Even simultaneously? But since Bechstein DG is to my ears the least synthetic Pianoteq model so far (although I can still detect some hints of it), I’m wondering if it’s the way Pianoteq should utilize most in future."

I "bought into" pianoteq in the belief that it was 100% "modeled",
i.e. mathematically derived from the physics of how stretched strings vibrate and pass their vibrations on to wood, how the sound board wood selectively filters some frequencies, as does the cabinet, etc.
I later discovered that some of the incidental noises are basically samples, which is probably OK although the builders of wooden pianos would probably eliminate those entirely if only they could find "perfect" sound deadening felt.

So I'm wondering how much "better" modeling can be than sampling - - if it is a model of a sample library ?!?
Does it "improve on" the sample library ?, or are there losses ?

Just acquired (d/l or u/l?) this latest 6.51.  All seem to have improved.  Some of the "home" settings seem exquisite, really do.  But Bluthner seems as always, the best of the best.  So playable. I'm totally stumped for a second choice.  I'll never get any work done at this rate . . .

Yeah, that Blüthner sure is something special.

Just posting this in case anyone else runs into the same issue.

I did also noticed that the performance has been improved in 6.5.1. I wasn't able to play the modell D Jazz anymore on my Odroid Xu4, I guess because it uses 4 mics. I quickly encountered cpu overloads.
Now, this has all been resolved and everything seems to be fine.
Only thing that's still not working is the "Auto(optimistic)" polyphony setting. Auto will not even give me 32 note polyphony while I can easily get 64 notes if I manually select it.

But again, thanks for the quick 6.5.1 update and all improvements to the piano sounds. Really good job guys.

Just copy the download link, and change 651 into 641 (or what other version you want to have).
At least in Linux, you can have multiple instances of Pianoteq installed.