Has Pianoteq reached its limits? (Page 1 of 2)

I have been involved in research for much of my life.  Pianoteq is a new concept and it is also a research project.  While it does something similar to the many sampling libraries out there, it does it in a totally different way.  Hence, the issues and problems to solve are all new.  The opportunities and the potential are also new.  I have learned it is impossible to schedule innovation and creativity.  I believe this to be true for music and it is also true for science.  This is part of the art in both disciplines.  Other management concepts like adding more resources and people to an effort are not often effective, because you have to find people with the right understanding and passion.  I must agree with sigasa that the improvements will take place in the time that it takes.  As such, it requires patience.  It is also clear that many improvements have already been made, which shows a good track record.  I would not mind if research could be completed faster or on demand, but I have never observed this in the real world.  As Billy Crystal once said, if you rush miracles, you get rotten miracles (paraphrase from The Princess Bride).  If Pianoteq cannot meet your needs, it certainly does not bother me if you make use of other tools and software.  Pianoteq meets my needs quite nicely and has for the last 2-3 years, so I will keep using it.

As with any program there's always a group of followers and non-followers, which is actually not important. One can either use or not use a program as he wishes and have and state an opinion and have all right to do so...
But apart from those obvious remarks I think there are so many parameters involved in piano simulation that probably the main limitation now is computer power. Where in the past a Cray supercomputer had to be used to do very complex calculations, one can easily do those on Intel or Mac nowadays. But at the same time the complexity of what we want grows as well. Thus the need for faster CPUs.
For the time being probably sampling which captures all nuances of a sound will be the best deal, problem being that because of the size of the samples and the amount of data that has to be put through, sample makers make the decision of using less possible steps per key. But at least they have all nuances in every step.
Pianoteq has the far better playabilty but since everything has to be made 'on the fly' it will surely need a lot of power for it to make every nuance, if the model even allows that or if they will even be able to program all of those extra quirky sounds that a piano might make and that can be captured in a recording for a sample library. I can imagine that having more horsepower will help and it feels like Modartt have enough 'drive' to improve on its model upto hitting the next limitation.
And I know this is just another opinion - no spokesman here too...
Just a matter of ventilating a thought..  :-)

cheers
Hans

Philippe, thank for your reply! That's exactly the answer I expected to hear and I will not give up trying and evaluating Pianoteq It's a tremendous achievement what you have done so far and you deserve a hgh regard for that. Thanks again!

Just curious: Can you point us this "middle" more spesific?

I wonder if "real world" oscillating strings have a bit of random frequency drift for each cycle, and possibly that is not being captured in the model?

Does this produce a more woody sound Jake?

Indeed, and MANY people are saying they want more of this in Pianoteq!

Does "woodiness" mean "mellowness"?  As the timbre becomes brighter, does that mean less wood and more metal?

I still think Pianoteq lacks metal, not wood. By this, I mean that it lacks something in the very high harmonics that I hear in a) many real recordings, and b) many sampled pianos. 

Greg.

The "roar" of the wood.  The distortion the cabinet imparts upon the strings - vibrations feeding back upon the strings and then back into the wood again... rumbling, creaking, stresses of the seams -like nailing into a rafter.  And if you don't want that, if you only want the attack and sustain characteristics of an imaginary perfect piano note, then just turn the "wood" parameter knob down...

Seems like we're branching off from the original thread topic. Maybe that's good. But don't we need references to better define the terms that we're using? Which piano emulations or recordings have wood? Which have bell?

I think "woodness" is mostly associated with attack, the moment when hammers hit strings. IMO Roland's V-piano has succeeded to make it quite well, so it's possible. Of course who knows, maybe Roland is fooling us: perhaps they are using sampling instead of complete modeling

I had a V-piano over a weekend and I compared it to PTQ, Roland RD700GX (sampled), Yamaha S90XS (sampled), Clavia Nord (sampled) and Ivory (sampled). I was surprised how close V-piano was to these sampled competitors. In fact it was so close to others that I decided not start to save money for it. Compared to others, PTQ was sounding clearly different. Somebody might say it's the best and most natural; I agree that in some respects and in some registers it's maybe the best. But same time it has some qualities which makes it little artificial sounding; you might call this lack of "woodness". I have located these most problematic qualities under middle C.

Very strange... Cause I played it in the music shop few days ago and then compared it to new Yamaha CP1 and Nord Piano. To my fingers V-piano was clearly a winner. Obviously these opinions are matter of taste and habit. Like sound. Soundwise those professional manufacturers are closer to each other than before; IMO Roland has improved their quality when Yamaha hasn't done much anything...

You're maybe right: many digital pianos have big problems with fast repetition. I've heard that new Korg SV-1 is the worst case here. On the other hand Roland has tried to improve it with their V-piano: you don't have to release keys to all the way up to make it sound and this should make it easier to play repetitions, I don't know.

I was speaking more about overall feeling of it: to my fingers V-pianos touch is good and this escapement is also very nice.

(now were're slipping off-road again....)

Sorry, one more question: what digital piano has a good action? Just curious...

1+ for the Kawai keyboards. To me Kawai is hands down making the best keyboard instruments these days in terms of feel. Yamaha and Roland, with each new keyboard, seem locked into slightly improving upon their previous model.

Maybe I'm being too harsh--I haven't played all of the latest Rolands and Yamahas. But I do find that when I go into a music store, I experiment with the Rolands and Yamahas and ask myself the extent to which they feel like a piano. With the Kawai instruments, I usually find myself playing more--the keys just immediately feel right, even on their "lower end" keyboards, like the EP3 and the MP5. Haven't tried one with PianoTeq yet, however.

if you play the v-piano through a really good set of closed headphones, you will be amazed at how real it sounds, and how broad the spectrum of timbres you can achieve with the different pianos and hammer hardness, key touch sensitivity, and EQ settings.  I emphasize the headphones comment- because i think playing it through external speakers sounds much more synthetic- but i find that true of all sampled and modeled pianos... I own one, its not perfect, but i love playing it- and i play it all the time, and if nothing else all the playing is improving my piano skills. I used to spend all my time screwing around with the setting of my sampled pianos, now, i'm comfortable just playing.... And i sort of understand what Evil Dragon is saying about the action- i actually like it because to me it is very fast.  but it also has a shallow "throw" (is that the term?) vs a real piano- and it makes it seems somewhat "hard" at the bottom of the stroke.  so to some, it might lack cushion and seem unreal.  I haven't used it as a controller, i would like to eventually run PTQ threw it someday.  i wonder if that would introduce more latency than i could tolerate, because that is also something i no longer think about...   btw,  Roland sucks at communications, i've sent their support people emails and never had one answered. They could use someone from PTQ to manage their user sites

It's interesting about the perceived shallow throw of the Roland - this is not the first time I've seen this comment.  I've actually measured the white key travel of a RD700GX at 9.5mm (although with an uncontrolled down pressure)  and this is commensurate with some concert grand piano actions. I wonder whether there is something about the feel that makes it feel shallow, even though it is not shallow. (not particularly shallow, in any case, in my opinion). I'm assuming the V-Piano has the same throw as the RD700GX. (I realise the V has the PHAIII and the GX has the PHAII with escapement)

When I measure a Yamaha GH action, it's 10mm, which is only a very small increase over the Roland.

Greg.

I'd agree about the longer decay in Pianoteq - I'll repeat what I've noted previously - I find the decay too long.  But I'll admit it may be somewhat a result of my use of the damper pedal (having always been a bit lazy, I use it too much I suppose).

However, I've always wished that the higher notes on an AP did sustain longer, but I suppose years of playing an AP has conditioned my perception of what the sustain/decay "should" be like.

The beauty of Pianoteq is that we can design our own pianos - we're no longer stuck with the limitations that piano builders have struggled with for a few hundred years.

Visiting my piano re-builder friend yesterday (he has a gorgeous seven foot 1915 Heintzman to die for, but I digress), he pointed out that the Heintzman used a high tension scaling.  I asked him what he meant.  "High tension scaling uses 180 pounds tension in each string - low tension pianos like the Yamahas, use 170 lbs).  You learn something everyday if you listen.

While 180 lb doesn't seem terribly high, when one multiplies this by the number of strings, one wonders how they managed to build pianos without an iron frame (or one realize why the first pianos were so weak).

Glenn

Higher order differential equations, my friend.

I was just thinking about the original question and wondering, which part of such equations create that real piano sound, or that woody effect everyone keeps talking about? I wouldn't have thought that this was unobtainable. It seem like it would only take a few relative adjustments, a shade here, or colouring of the sound there. It seems to me that the core work has already been achieved.

I haven't visited this topic which I have created since my last vist because I thought it was already agreed that Pianoteq is constantly evolving product and may (will) eventually reach a better sound in future.

I see there are some discussions regarding the meaning of "woody". Unfortunately I am currently without a piano and waiting for Kawai CA-63 to arrive (I am intending to use it with Ivory) and after that I will make some comparisons of short piano phrases and sequences in the middle range of the keyboard where Pianoteq creates the feeling of you playing something of metal origin and another with Synthogy Ivory which has the captured sound of real piano. I suppose "wood" is not actually a description of the real piano sound but is used mostly to imply the opposite of "metal". Probably because a real piano is made of wood and metal and there is subconscious relation between its sound and the materials it's made of.

I've heard the description "metallic sound" even for real pianos and I think the other side of spectrum where "non-metallic" sound resides is simply called "woody" as a reflection of one's perception on balance between wood and metal in a piano.

"Pianos are constructed extremely precisely, but they're still made of materials that always have imperfections"

That's why I'm actually building my own virtual piano. I've try the pianoteq demo and really, in my opinion, the sound is lifeless, and pretty much horrible, especially if you listen it with an headphone. Stereo is strange, there is a gap in the middle range and I think modartt use a little bit of samples ( for the hammer and resonance things). In my opinion, pianoteq is build with a lot of compromise and is far from reality although they talk about "true modelling". Pianoteq sound like a piano but is not warm and generous like a real one. Electrics add-ons sounds better. 

If you are interesting, there is some demo from my virtual piano (piano sonata KV 576 -I can join the midi file for those who want to compare) to listen to on my website : http://www.myosismusic.net/   

Sorry for my english.

Only for string noise in note attacks. Resonances are COMPLETELY MODELLED. Do your research before posting false claims.

If stereo sounds weird to you, you have to tweak microphone positions a little. If you're using headphones, then you switch the output mode to BINAURAL, and suddenly the sound is vastly better!

Seems to me like you didn't read manual at all.

And sorry, Pianoteq sounds much better than your piano by hundreds of miles. You need a LOT, and I really mean A FREAKING LOT more to do to make it sound as real as Pianoteq is. My 15 year old Korg X5D has better piano sound than this!

What's saddest is that you registered to this forum only to bash Pianoteq. Shame on you! 

TO EVIL DRAGON:

-Only for string noise in note attacks.

That's what I would say. So there is samples, and that's why the sound is strange for me. It's a mixture of differents elements, and this is the first thing I have heard. 

-If stereo sounds weird to you, you have to tweak microphone positions a little. If you're using headphones, then you switch the output mode to BINAURAL, and suddenly the sound is vastly better!

Probably.

-Seems to me like you didn't read manual at all.

I've read the pianoteq 2 manual. But I've read much more on piano on the web. and I'm not a scientist. I'm just in love with sound.

-And sorry, Pianoteq sounds much better than your piano by hundreds of miles. You need a LOT, and I really mean A FREAKING LOT more to do to make it sound as real as Pianoteq is. My 15 year old Korg X5D has better piano sound than this!

Thanks for your compliment. But this is YOUR opinion. What about your own piano model design ? Does it sound better? I'd like to hear it. About mine, it's a personal wish rather than a commercial one. And I don't pretend to make a better model, and I'm not making a MODEL. It only will be different, it will sound different, and he will probably have those things pianoteq doesn't have and mine will not have those things pianoteq have ! So , i'm not making a war, I'm just answering to my ear. I don't like how pianoteq sound probably because of the synthesis method they use. However I love the concept, and Modartt have all my respect. They push me to the direction I'm taking.

-What's saddest is that you registered to this forum only to bash Pianoteq. Shame on you!

I'm just curious, and I want to share my obsession and interest about piano emulation. I've no shame. I feel free. If you  don't, that's your problem . I will post in other topic if I want regardless of negative opinion about me. Pianoteq is a reference and I think it's one of the best digital musical idea although I don't like the sound. I don't "bash" pianoteq.

The saddest thing is the tone of your answer. Talking like you canno't make things happens. That's my opinion.

Regards

-->>TO DONSMITH.

Thanks for your comments. Glad you find it interesting. As I don't speak english very well, I'm not sure to understand very well your question. I' think you imagine that I am a physician ! I'm not so my answer is : if pianoteq is exactly the reproduction of physical phenomenon (?) , the answer is yes, physical modelling have reach it's limit. But, there is a lot of phenomenon that don't exist in pianoteq because of polyphony. They made a lot of compromise , it's an evidence. If they don't you could play a maximum of two notes at the same time . Acoustic is very complex and CPU intensive to emulate. I think we have to wait the 10th generation of piano and for more powerfull  computers. 

So the question should be : does your ear appreciate this sound or not? Sometimes my ear are much more pleased with a simple pianotoy than a sophisticated Harpsichord.

About my plugin, I don't use digital waveguide for the strings. I use traditional  synthesis and if my plug doesn't sound as a reallly real piano, or emulate a particular one, it will have a lot of different color and tonal character.

Regards.

I remember a great BBC documentary on synthesizers way back in the 70s.  It ended with the interviewer asking about the possibility of synthesizing a piano. The synthesist laughed, saying it was not possible. I do remember him commenting that "lots of oscillators and lots of phasers would be needed".

Greg.