Sampleled libraries vs Physically Modeled libraries

New here, love PT(9)

Bought Steinberg The Grand3 to find it cannot handle Half Damper sustain (full CC 0-127) pedal, only on off.

Stumbled across PT trial to find it does handle HD sustain. Immediately purchased PRO for this and all it's other goodness.

Emailed Spectrasonics about Keyscape. Keyscape, Halion, and SINE (Novation) does not handle HD sustain.

Googled "does sampled music libraries not support half damper sustain while physically modeled libraries do"? Answer according to AI (ugh!) "It is not strictly true that all sampled libraries lack half-damper sustain, while all physically modeled libraries have it, but the statement reflects a fundamental difference in how they handle articulation. Physically modeled libraries are naturally designed to support half-pedaling continuously, while sampled libraries often require complex scripting to simulate this effect, with varying degrees of success."

I do not yet fully understand all the diffs / foibles about this topic subject. Then I hear Gamechanger PLUS sustain pedal on Youtube. Sounds nice at least for guitar. I believe it works with some looping technique. I could not find a VST using the same technique / algorhythem. Reverb is not sustain.

Questions: Is there a fundamental differance about this topic sublect that may prevent or make algorithms difficult for sampled libraries to allow HD sustain? Would PLUS type algorithm based VST be a viable work around or sound too fake for polyphonic systems?

As mentioned love PT and encourage Modartt to contnue with other acoustic instruments.

daniel_r328 wrote:

For half-pedaling in particular, blending samples together can give perfectly acceptable results, as Pianistically said.

I'd just add that the differences between sampled and modelled really come to the fore in fast repeats (trills, ostinatos, etc), where sampled instruments inevitably sound quite robotic or "too stable".

For more sustained music, there's a bit of a battle of tastes of whether you prefer the mastered, "curated" sound of samples or the raw (and perhaps a bit too-quantised) sound of models. Opinions are strong and (because internet) often stated as absolutes

Btw you mentioned guitar pedals vs sustain in pianos, so I just wanted to point out that they are different: those guitar pedals are a form of reverb to the dry sound, whereas sustains on a piano is a continuation of the dry sound plus sympathetic resonance from the other strings, leading to a more sophisticated sound evolution. There's a new crop of hybrid libraries that use samples for the dry sound plus a modal resonator model for the sympathetic resonances, but they can feel a bit detached imo. So it's not impossible at all - just pretty hard. One could write a whole PhD thesis about modelling piano resonances and launch a whole vi company with the findings

regarding repeated notes and the machine gun effect at fast speed above 12 notes/sec you are right , pianoteq excels . Most sampled libraries get the repeated sound in cache ( except ViLabs which has an option to simulate the lack of amplitude of the hammer in repeated notes and don’t get the samples from the cache) .
What is paradoxal here is that composer’s intention when they write sequences of repeated notes expect the volume to be the same apart specific accents on given notes and don’t take into consideration that that the first note of the sequence is going to be louder . So in a nutshell , we are replicating an imperfection of the action

I was talking about the fact that repeated notes (unsustained) sound too similar in sampled libraries, because they always use the same sample (or a finite rotation of samples) at a given velocity. In a real instrument as well as models like Pianoteq, every keystroke is informed by the string motion already in flight, so there is a cumulative sound evolution when you have multiple key strikes. A pianist would listen for the evolving sound and subtly adjust their playing to get the desired swell, so this effect is interpretation-relevant. But you can't do this very well with sampled libraries, because the multiple keystrokes interact nonlinearly, so you can't get the same effect by (say) playing multiple samples on top of each other. So that's kind of a separate thing to all the pedal stuff

Pianistically wrote:

I disagree here

1) the  machine gun issue with repeated notes happens at high repetition speed > 10 notes / sec . At that speed the pianist doesn’t adjust and relies  essentially on the action .

2) again sampled libraries like Vi labs have solved the issue regarding variable velocity . Sure they cannot take into consideration the fact that the string was already vibrating when re-stoke but in the grand scheme of things this is a drop in the ocean .

Hmm I don't know about that. Sure, at a certain playing tempo the pianist adds involuntary variation to their playing, leading among other things to velocity fluctuations - but involuntary doesn't equal random. Everybody's fluctuations will be slightly different in a recognisable way; call it a fingerprint of their playing style. In my opinion this subtlety is relevant: during practice, you want to hear your fingerprint and mould it over time, and during performance, you want your audience to hear it. Replacing it with stochastic variation would feel like playing with someone else's fingers. I guess it depends on the genre how important that is... for just for me repeated notes in sampled libraries are already a problem at much lower rates than 10 notes/sec; ymmv

The reason II is extremely difficult to have total control of the dynamics with such speed , technique and limitation of piano actions is. :
1) different fingers used for each note which in itself is a real challenge as each finger is different from a physical lever standpoint . This is why we use same fingers in portato , ro make sure of equality of volume
2) reduced distance between hammer - string fights against you if the notes need to be played at mf level

Shows I am still amature.

It’s curious that you’re discussing the ribatutto using an example from Scarlatti (who wrote that for the harpsichord, not the piano), along with the accordion example, another instrument entirely, a different world.

Daniel, regarding that 432 fingering, your 'self-taught' background really shows; any more or less formal teacher would have imposed 321 on you (which, by the way, is the best for the piano, the 4th finger is the 'poor relative' of the hand, yet you’re using it on the strong beat and doing it over and over again, haha).

But generally, I agree with you: we already add enough variation ourselves, whether intentionally or not, without the software adding yet another layer of it, one that is, indeed, completely out of our control.

Daniel, regarding that 432 fingering, your 'self-taught' background really shows; any more or less formal teacher would have imposed 321 on you