Just joined the forum to say thanks for this great post.

I've been following Pianoteq since it came out, and am impressed by its improvements. I am however in the camp that instantly hears the synthetic quality of the instrument and finds it annoying. I add my name to the list of people who hope one day that it will overcome these problems so that its playability is matched by quality of tone. Please don't call me a "hater"... I am a piano player (28 years experience and counting) and studio engineer who has a different opinion from some people here. Your mileage may vary!

Anyhow, I really appreciate this test. I preferred the True Keys overall and am now considering buying it.

Thanks for your time to join and reply. TrueKeys is a lot of fun indeed.

What I find amazing about the many sounds out there, is that each one inspires something different, so they are all worthy as long as you like them. Trying Pianoteq with others fxb is indeed a great idea, try it.

I've been following Pianoteq since it came out, and am impressed by its improvements. I am however in the camp that instantly hears the synthetic quality of the instrument and finds it annoying. I add my name to the list of people who hope one day that it will overcome these problems so that its playability is matched by quality of tone.

Fair enough, and you are of course entitled to those opinions.  I'm curious how extensively you've tried various FXPs and altering parameters.

I actually had (and still have) similar reactions to yours with regard to *most* of the Pianoteq instrument models.  (My main frustration with the particular comparison in this thread was that -- by choosing the Upright -- it compares apples to oranges.  Try as I might, I still haven't warmed to the U4, which sounds/feels extremely synthetic to me.)

But there are a few FXPs that sound/feel dramatically better than most of the others and -- when combined with the range of dynamism in a modeled (as opposed to sampled) sound source -- have made me a very happy and devoted Pianoteq user.  I'd be interested in whether you feel the same way (or at least as strongly) toward, for example, the Dynamic Jazz 3-mic FXP (and please try playing it rather than judging it based on the poor audio demos!).

I'm now wishing I had had the time to make the video I was imagining for the Video Contest -- demonstrating the dramatic effect that some parameter changes can have on the authenticity of the sound...

Your day (perfect molled tone) will come.

I've been following Pianoteq since it came out, and am impressed by its improvements. I am however in the camp that instantly hears the synthetic quality of the instrument and finds it annoying. I add my name to the list of people who hope one day that it will overcome these problems so that its playability is matched by quality of tone. Please don't call me a "hater"... I am a piano player (28 years experience and counting) and studio engineer who has a different opinion from some people here. Your mileage may vary!

Anyhow, I really appreciate this test. I preferred the True Keys overall and am now considering buying it.

They weren't normalized on volume so I guess Pianoteq was easy to spot but I also think the others had chracteristics big enough to identify them.

The pedal down on Ivory was unique, given an openness that wasn't on other samples, when you use the sustain pedal there is some kind of echo and different resonance when you play. Playing bass arpeggios on Ivory was really cool and had a very nice blended sound.

True keys had staccato, only Pianoteq and True keys felt right on it. Contraty to many's belief, the sympathetic resonances on True Keys and Ivory were very clear and audible, which is amazing considering the first one is sampled and the second one is modelled. Between True Keys and Ivory, True Keys is more in your face so feels better when you are playing, some recording sound more beautiful on Ivory, excepto with some resonance on True Keys that sounds really good and above Ivory.

EWQL lacks sympathetic resonance

Also, everyone but EWQL seem to have variable tone when repeating a note.

And about the sound, I think there are two different edges here, one, to sound like a beautiful almost perfectly mic'd recorded clean Piano, and the other to sound like a live acoustic Piano.

I'd put Pianoteq between, also is the less forgiving and you have to be more careful as in a real grand to sound good, Ivory seems the easiest to just play and get a fast beauty sound. EWQL is also between, even when lacking many acoustic features. True Keys may sound the best on particular songs.

Pianoteq sound also seems to move a lot more than the others, which is really cool. Both Pianoteq on defaut options and EWQL seem to decay in more time than the others.

Core is a weeny .MID file.

As far as I know Modartt technology work by searching redundant behavior in the equations of the piano producing sound, and resuming the equation.  The difficult thing it's to find what characteristics are more importante to be left with more fidelity resulting in what our ears perceive as a natural sound.
So if they desire to get a earlier piano model add-on and make it more natural and using more CPU to to a finner result, they would need to rework the entire model from sketchy.

Think in bricks that forms a home. If you have a brick, a model of it, you do not need to make every brick every time, but repeat it.  You arrange the best combnation of bricks to try to get appearance of a home.

We could say that Modartt managed to put 5 elephants in a Volkswagen (a good piano sound with affordable CPU needs).

But still, if they just hog a little more CPU, they could possibly convince even the hardest critical. I think it's possible!

Please, introduce some realtime / offline rendering switch, which makes the offline render a perfect piano! :-)

In this very fine work, computing a single note, repeat, a single note takes 24 hours on a 300 core computer!

Dayummm....!! Are they processing at the molecular or sub-atomic level, which one is it??

My opinion as well!!

since 4.5, pianoteq is very very good. i love it. however, i do agree with the concept that the limitation to further authenticity lies in how many overtones can be modelled, and still make it run on an average laptop. which it does superbly. i think it would be a good idea to be able run more computationally complex algorithms depending on what cpu you have. just like u-he do with their diva synthesizer. also, been able to render in your daw the most complex algorithm irrespective of cpu limitations. which is also what diva does.

I don't think piano modelling scales like that, like, for instance, 3D graphics does. I just saw the movie Gravity on a giant screen and the realism of the CGI is incredible, because all the physics is well known, implemented in hardware, and getting this high quality is just done by increasing processing power (a lot!).

I believe pianoteq doesn't scale like that because the goal being real-time, only the key components of the piano (soundboard, strings, hammer interaction) are modelled using, I'm sure, very, very clever mathematical solutions to hard problems in order to obtain real-time response and starting from the recording and analysis of a real instrument.

This modelling is then incrementally refined with each version, with a slight increase in computational needs following the progress of available hardware, getting closer and closer to the real thing.

All the other published works I have read start from basic physics principles, and are really, really far from real-time.

The best recent example is here, if you read french (or complex equations) :
http://tel.archives-ouvertes.fr/index.p...;version=1

In this very fine work, computing a single note, repeat, a single note takes 24 hours on a 300 core computer!

Of course the goal in this thesis was the analysis of the physics problem of energy dissipation in a piano, not producing sound, but this really shows you that it's not a scalability problem.

This is only my opinion, but the more you read on this subject, the more the achievement Modartt attained becomes incredible and worthy of admiration...

Evil Dragon, since you are very well insight the informatic world, may I ask you if you have some graphic, some statistic, about notebook average power increase (for average users) in the last 5 years?

I have the feeling that pianoteq requirements for CPU changed less than the notebooks evolution in the last years.

That is quite likely a correct assessment.