Pianoteq 4 foretaste (Page 2 of 3)

Hi,
I believe the "freeze parameters" feature is your answer (it's the check-mark left of the "random" button).
I wonder why people keep missing this great feature. These questions keep coming up again and again. It's in the FAQ, by the way...
Perhaps there should be a Q/A section in the forum like "Yahoo! Answers" where users can ask other users, and rate the best answers... just a thought. (Of course, I wouldn't like it to diverge Modartt's development, but it might reduce support in the long run).

Best regards,
Eran

Those clips are fairly shorts, but i think it's the first time i don't hear the pianoteq "signature" in them (that's a good thing IMHO).

This "foretaste" thread is getting a little long-in-the-tooth isn't it ?

Is there any indication of when (or IF) version 4 is near ?

..and never forget the RECEPTOR-users... like me :-)))

"Thin" is Pianoteq 2. Pianoteq 3 is a LOOOOT better. Does not sound thin nor artificial at all. It has presence, it has the decay, it has the timbre. Everything much better than Pianoteq 2 ever was. People are asking me which magical samples I'm using for that awesome piano sound, and I just say "hey, it's not samples at all!"

So, if the advancement that happened from v2 to v3 is any sign, you should brace yourself because you're gonna crap your pants, in the most positive sense of that saying!

It's funny you should say, I was never really amazed by the sound and behavior of Roland's pianos... "Supernatural" isn't quite the word that describes it well IMHO. Not even the V-Piano, which is absolutely terribly overpriced for what it is.

But OK. Opinions are like assholes, we all have one.

I think red is right, Pianoteq has major problem with attack, which is perhaps the most complicated thing to get right with modeling. I am not a technician or psychoacoustical expert but I can imagine two very basic things: acoustic phenomenon at the very first moment is very complex and also our perceptual system is adapted to hear many many nuances in the beginning of sounds.

I've been resently wondering that perhaps the hybrid technology (combining sampling in the attack and modeling with sustain and resonances) might be the easiest way to get the most realistic pianosound. In fact Roland supernatural pianos are example of this.

Pianoteq has improved a lot and probably v4 will continue this development. When I read this forum, I cannot help seeing same thing again and again: classical oriented players favor this sound which is something close what you can hear in classical recording; acoustic piano sound recorded in natural acoustics from distance. In fact it's something you won't get from most sample libraries. This piano sound is more about whole instrument, resonances and acoustic space and NOT so much about the attack. On the other hand most PTQ critics (which are often new users or demoing this software) are after closely recorded pop/rock/jazz -piano sound in which the attack and hammer sound is very much present. Think this is the kind of a problem here: PTQ is modeled sound, (almost) everything can be adjusted, so it promises to give you infinite amount of different piano sounds (which it does indeed), but still it cannot convincingly provide you something which is very familiar to many ears: closely recorded modern grand with punchy attack. Now, some people here will say, that this is not realistic piano sound, it's not something you'll hear in front of a REAL grand. Still I'd say that this is realism to many and in fact in my opinion some Yamaha grand in dry acoustics in players perspective can be very close to it...

I had Roland RD-700gx with Supernatural expansion, which was a great improvement over gx without it. Maybe even better than V-piano. Still I felt that sound was kind of muddy and not enough present especially on stage. So I changed my RD to Nord Piano which I haven't regret. Nord sounds very close recorded, maybe even too much, but for sure it will cut through mix. Recently I have experimented with mixing Nord with PTQ. Nord gives me very punchy attack, PTQ gives me more body to the bottom, more sustain and all the resonances. Best of both worlds...? Maybe I'll do my next recording with this combination...

This been said, I still love PTQ and its uncompromising philosophy.

I hope not, the ones I listened to sounded terrible!

That sounds like Pianoteq 2 there.

I wonder about the use of the word "attack." Is it the attack--the moment that the hammer hits the strings--that is the worry? Sometimes I do worry about the attack needing to be more the sound of a long felt hammer hitting the strings.

But I also sometimes wonder if I want more of the bloom right after the attack. It may be a standing wave, or sets of waves, created when the reverberated note (the set of partials) meets the vibrations still coming from the soundboard and strings.

But trying to describe such a complex sound is difficult. Would it be better if we posted an mp3 of a note from a recording that sounds good to us and the same note in PT, so that we could all listen to and discuss the specific qualities of a specific note, instead of speaking in generalities?

Well, I like those old hymns, but the reverb on those recordings makes it hard for me to hear the envelope of the notes. I was thinking more along the lines of people posting dry mp's of individual notes that had the quality or qualities that they wanted. From there, we could all do an analysis in the many sound programs that I'm sure that we all have, and possibly edge closer to defining what is wanted.

Not to ignore what you say about dynamics. But could that subject be better explored with, say, an mp3 or wave of a desired soft strike and then a desired hard strike? Given that we are discussing physics, any given phenomenon should be reproducible and open to modelling, but first we all need to isolate precisely what people want to hear.

I must say that I've been down this road before--of asking people on other forums to post notes that have the exact sound that they want. However, the result has been that they never respond or post mp3's of passages that make a note by note analysis impossible. Don, my hope is that you and others will instead post these individual, dry recordings of single notes. Given the number of people interested in this subject here and their combined knowledge of sound and pianos, the tools that we have, and the wonderful expertise of Modartt, the results could be very productive.

If what you say is true, especially as regards current CPU's not having enough capability, then as some have suggested, hybrid sound may be the answer for now, with samples being used for the attack portion and initial onset of tone, and modelling thereafter.

Mostly what I've come across is that samples become static and dead sounding AFTER the attack, as the tone develops. And Pianoteq takes care of tonal development over time very well.

well the problem is either, insufficiently sophisticated algorithms to accurately model every aspect of a pianos behaviour irrespective of the number of cpu cores and clock speeds available. or, if a basic cpu was an ivybridge 8 core running at 5 ghz philippe and his team could throw everything at it and it would sing to every bodies total delight.

hahahahahaha skip. yes record your pianoteq performance as midi data, then press the render button, come back the next day, and it sounds like a steinway model D well that would prove the hardware is the problem and not the software! defo the aliens can wait a bit longer;)

I have the feeling that if Moddart had found the way to compute the extremely complex attack portion of the sound (especially the low notes) they allso would give it to us.
As EvilDragon says, allready now we can run Pianoteq with massive polyphony with no problems, so there is actually spare cpu left for a better attack.
The attack portion in Pianoteq has allway been the weak part (i think) but the most important one for people to believe in the sound they hear.

it's no mystery Roland D50 became so popular using a very short sampled attack part combine with straight "analog" synthesis for the sustain part because the human ear recognize a sound from the very fist part they hear and become more relaxed with the rest (sustain)

And in fact Modart guys have said it in some earlier thread: there's no secret "PTQ supermodel" waiting for us to have faster processors.

Very true. This is why it's so important and challenging also to model it. How much this attack plays role in identifying different piano models also? IMO for example steinway has a special "thunk" in the attack. Could you identify Steinway, Bosendorfer, Yamaha etc. if you listened to sounds without attacks?

Clear, punchy attack gives presence and clarity which is something that most of us are after for. We don't want to have muddy old upright for our first choise piano. IMO somehow Roland engineers have done quite good work with attack with V-piano. So maybe it's not that difficult after all. Although sometimes I've even wondered that maybe they cheat us: perhaps they use samples in the attack (like they use in Supernatural pianos)?! This paranoid hypothesis is difficult to test, cause V-piano cannot moved like PTQ. It's always closely recorded piano sound. If it's completely modeled (like they say), IMO it does this sound quite well.

I can hear the Yamaha signature in the YC5, especially in the bass registers. Unfortunately, I don't particularly like the Yamaha tone (even in real ones) in the bass registers, but hey, it's a start!

Greg.

That describes my preference to a T.  I'm often after a close-miced sound, and the important thing there is 'character'.  Most of the PT presets sound incredibly weak and generic to me.  As you said, this may be perfect for far-miced classical music, but not for what I care about (rock, ballad & jazz type sounds, not necessarily super-powerful or aggressive, but emotive and that requires character).  I also find the presets too 'pure', i.e perfectly tuned and as a result often boring.  'Character' is often found in imperfections, I would love to see much more control over those (eg. more complex detuning and 'aging' type params).

I've managed to get quite a lot of usable sounds out of PT3 with programming, but there's still a slight synthetic thinness to them that I'm never 100% satisfied with.  I love PT for what it's great at - sheer enjoyable & realistic playability - we just need that extra layer of complexity.

I would have no problem if this was achieved with samples in a hybrid model, in fact doesn't PT already use hammer samples, as well as pedal noises etc?  As someone who programs their sounds, I would also really need some way to customise and tweak those portions, for example I find the current hammer sounds a bit too one-dimensional.  If they were more heavily multi-sampled, and you could also tweak them as heavily as other parts (eg. by applying EQ to the samples, using sample crossfading etc.) I'm sure I could get close to perfect sounds even without a massively complex new model.

Can't wait for PT4 : ).

That's fascinating.  At first I was horrified by how lifeless the ring-out is - Pianoteq is miles better, it can really sing (unlike this demo) and I've always loved it for that.  But I also see what you mean about the realistic hammer sound.  It's especially marked in a couple of instances where the note is retriggered in quick succession, it really sounds mechanical & the intermediate partial string dampening and re-excitation is very convincing.  It's not perfect but shows huge potential.

Again fascinating.  I see what you mean by bloom, and your demo is certainly poorer for the lack of it.  But on the other hand, the 'character' is massively improved by your added attack sample.  So from that I'd say you're right, the attack is hugely important, it conveys a lot of the 'body' and construction of the instrument - all that mechanical organic one-of-a-kind stuff I'm looking for.  And the post-strike bloom would also be very welcome.

EDIT: actually listening to it again, I'm no longer sure that it's just about the attack.  It gives you more of the instrument character for sure, but the string ringout probably has to convey the same character to be convincing (after all it's supposed to vibrate in the same 'body').  So it probably does require a more complex model.

Thanks for posting these, without audio references it's hard to know if we are discussing the same things.

Agree.  This is where randomizing a number of parameters is useful.

I don't use a single preset that is not randomized to some extent.

Glenn

In the Pro version, all individually accessible parameters can be randomized note by note (as well as smoothed and rescaled).

Here is the list of available parameters for a piano preset: Detune, Unison Width, Direct Sound Duration, Hammer Hardness Piano Mezzo and Forte, Spectrum Profile, Hammer Noise, Strike Point, Impedance, Cutoff, Q Factor, String Length, Note Resonance, Quadratic Effect, Damper Position, Damping Duration, Mute, Damper Noise, Key Release Noise, Volume.

Automatic "aging" is probably best left to individual choices, but I think the factory C3 Worn-out preset is a pretty good starting point as to what parameters are important.

Nice thanks.

Disagree, I think a generic 'Age' slider would be ideal for the standard version.  Of course even more control would be nice, but PT is clearly designed to keep a clean and simple(ish) UI - there isn't much space left, and an Age slider would be a neat way to expose a complex feature in a simple way.

Of course in Pro you can tweak all you want.

V-Piano and its "great playing feeling"... that's the point. I just recorded a maestro (a real one!) playing Beethoven sonatas on a (as real) grand Yamaha and he complained about nuances he couldn't get as he could on his (own!) Steinway "D"... And he was right. Well, PianoTeq is also made for pianists, real players, and that's the most important thing for me! It makes ALL the difference with sampled pianos. Now, if even the sound (included the attack, of course!) could be improved in v4... ;-)
As it is now, I ALWAYS record with PianoTeq and if the producer is not happy, I switch the MIDI file on whatever sampled piano, I don't care which one, but I couldn't record (= play!) on another one than PianoTeq, except my own grand !
Just my very own opinion, of course!

I agree with most of the previous comments about PianoTeq's lack of convincing attack, and that both a realistic attach with convincing "bloom" would be a very welcomed improvement in PQ 4.

The 2nd part is connection between touch and sound i.e. the feel:

Couldn't agree more! It would be great if Modartt could add velocity settings (tweaked and tested) for various DPs to properly get that connection between keys and sound (similar to Nord's tweaks on their Nord Piano)... I know we can edit the velocity curve in PQ3 but after few attempts (using the demo) i just couldn't get a natural connection between keyboard and PQ3...