Interesting how some small things make a difference, esp. with things like super fast passages. I must add, much respect here for your skills!

Pianistically wrote:

I was curious to see , if pianoteq was regenerating the whole succession of repeated notes when it detects such repetition or was resending the same  calculated wav sound with maybe volume adjustment

Pianoteq I believe generates repeats not in a repetitive way. Each repeat note iirc gets its own complete computation, which takes into consideration all the other params in real time (that was some superb work brought in with iirc ver. 7). This could be likely work older computers harder than in the past?, esp. at the repeat rates you mention.

Definitely would try lower buffer, if poss. That being said, putting up with a higher buffer size with a longer lag can help you decide just how fast you wish to go with lower settings. Give some time wallowing variously within laggy timing and lowest buffers possible. For sure, if your audio isn't capable of lower than 128, it would of course be recommended to use a faster audio unit or PC - but hopefully you don't need that just yet..

Agree also with other ideas above, incl. Damper changes in 'Action' panel. You can instantly facilitate better/cleaner staccato/repeats with a more efficient damper - or loosen up a piano for slower music (and there's an outside chance loosening up might help you too, esp. with other things mentioned below) - I think defaults mostly find a fortuitous line - but for specific things, definitely worth trying alterations there.

Other ideas off-the-cuff..

Check out the presets you use - perhaps altering some things in the FX section. Subtlety can be key:

#
Lower the 'Delay' mix.. or turn it off. (this may ordinarily be OK with much of a repertoire but could get messy with very precise short timings like fast repeated notes for sure.). Milliseconds count, esp. when several are being generated.. continue knocking them down a notch..

##
Evaluate reverb pre-delay and alter its length, or lower it's influence by altering the reverb to focus more on "tails" rather than reflections (to reduce any perceived 'slap' back of attack which might compete with repeated notes' sense of accuracy). Altering also length/size, tone of reverb (either direction, darker/brighter) might help with specific pianos/presets/passages or maybe just repeats.

###
Microphone 'timings'. Within the mic panel, some alteration to the small delays seen in various mics might help. It is simple to reduce that - but it may be best to be subtle at times. Some small moves might really be all you need to help some echoic behaviours get out of your way (crotchets at 132BPM) whilst keeping the overall environment in there the way you want.

But the mics (just like IRL) is an endless pursuit.

If some of those things help your repeated notes, it might be possible to hook them up via MIDI to your middle pedal?? That way, you can press the pedal only for those passages where you wish to clean up some extra echoic sound. Pressing the pedal might be set to just lower the 'mix' of a few things a little - but it may be a nice touch.

It's kind of the other way around here, I will add. Pianoteq, esp. with low low buffer is just so fast that I often prefer to play certain music with a higher buffer. Using synths for a long time and beginning with guitar, I like instant responses from instruments - but for me, maybe because I came late to it, feels 'more real' with a little more latency.. IDK if it is worth your time Pianistically, but never say never.. have you tried a larger buffer size? For example, when I play fast passages with a synth, it's more natural to me to have a low buffer.. but even playing fast with piano (or dpiano) I feel there's a natural space (more lag) than with for example guitar (close on body, closer to ear... feel and sound instant).. a piano has a kind of gap.. there's to me, a kind of 'arc' of electricity somehow, with different latencies and for rythmic purposes I do often love a little more latency - helps physically/haptically.

Whatever it is, I hope you find the best ways to fit the piano around your needs! Cannot say there's another piano software product with such fine controls, so it makes me hopeful you'll get the performance you require. All this typed out pretty quick - trying not to waft on - hoping no er rrrrors - all the best!

I have just tried it . What I did is
- turn off delay effect
- turn off delay in mic settings
- zero delay in reverb

And …….

It works like a charm !! Even better response than on the best sampled VST I have .   So very useful  input from Qexl ! Thank you so much .

Pianistically wrote:

I was curious to see , if pianoteq was regenerating the whole succession of repeated notes when it detects such repetition or was resending the same  calculated wav sound with maybe volume adjustment

Pianoteq I believe generates repeats not in a repetitive way. Each repeat note iirc gets its own complete computation, which takes into consideration all the other params in real time (that was some superb work brought in with iirc ver. 7). This could be likely work older computers harder than in the past?, esp. at the repeat rates you mention.

Definitely would try lower buffer, if poss. That being said, putting up with a higher buffer size with a longer lag can help you decide just how fast you wish to go with lower settings. Give some time wallowing variously within laggy timing and lowest buffers possible. For sure, if your audio isn't capable of lower than 128, it would of course be recommended to use a faster audio unit or PC - but hopefully you don't need that just yet..

Agree also with other ideas above, incl. Damper changes in 'Action' panel. You can instantly facilitate better/cleaner staccato/repeats with a more efficient damper - or loosen up a piano for slower music (and there's an outside chance loosening up might help you too, esp. with other things mentioned below) - I think defaults mostly find a fortuitous line - but for specific things, definitely worth trying alterations there.

Other ideas off-the-cuff..

Check out the presets you use - perhaps altering some things in the FX section. Subtlety can be key:

#
Lower the 'Delay' mix.. or turn it off. (this may ordinarily be OK with much of a repertoire but could get messy with very precise short timings like fast repeated notes for sure.). Milliseconds count, esp. when several are being generated.. continue knocking them down a notch..

##
Evaluate reverb pre-delay and alter its length, or lower it's influence by altering the reverb to focus more on "tails" rather than reflections (to reduce any perceived 'slap' back of attack which might compete with repeated notes' sense of accuracy). Altering also length/size, tone of reverb (either direction, darker/brighter) might help with specific pianos/presets/passages or maybe just repeats.

###
Microphone 'timings'. Within the mic panel, some alteration to the small delays seen in various mics might help. It is simple to reduce that - but it may be best to be subtle at times. Some small moves might really be all you need to help some echoic behaviours get out of your way (crotchets at 132BPM) whilst keeping the overall environment in there the way you want.

But the mics (just like IRL) is an endless pursuit.

If some of those things help your repeated notes, it might be possible to hook them up via MIDI to your middle pedal?? That way, you can press the pedal only for those passages where you wish to clean up some extra echoic sound. Pressing the pedal might be set to just lower the 'mix' of a few things a little - but it may be a nice touch.

It's kind of the other way around here, I will add. Pianoteq, esp. with low low buffer is just so fast that I often prefer to play certain music with a higher buffer. Using synths for a long time and beginning with guitar, I like instant responses from instruments - but for me, maybe because I came late to it, feels 'more real' with a little more latency.. IDK if it is worth your time Pianistically, but never say never.. have you tried a larger buffer size? For example, when I play fast passages with a synth, it's more natural to me to have a low buffer.. but even playing fast with piano (or dpiano) I feel there's a natural space (more lag) than with for example guitar (close on body, closer to ear... feel and sound instant).. a piano has a kind of gap.. there's to me, a kind of 'arc' of electricity somehow, with different latencies and for rythmic purposes I do often love a little more latency - helps physically/haptically.

Whatever it is, I hope you find the best ways to fit the piano around your needs! Cannot say there's another piano software product with such fine controls, so it makes me hopeful you'll get the performance you require. All this typed out pretty quick - trying not to waft on - hoping no er rrrrors - all the best!

thank you for that. Lot of interesting options to explore here. I will report back after trying.

I was curious to see , if pianoteq was regenerating the whole succession of repeated notes when it detects such repetition or was resending the same  calculated wav sound with maybe volume adjustment

Pianoteq I believe generates repeats not in a repetitive way. Each repeat note iirc gets its own complete computation, which takes into consideration all the other params in real time (that was some superb work brought in with iirc ver. 7). This could be likely work older computers harder than in the past?, esp. at the repeat rates you mention.

Definitely would try lower buffer, if poss. That being said, putting up with a higher buffer size with a longer lag can help you decide just how fast you wish to go with lower settings. Give some time wallowing variously within laggy timing and lowest buffers possible. For sure, if your audio isn't capable of lower than 128, it would of course be recommended to use a faster audio unit or PC - but hopefully you don't need that just yet..

Agree also with other ideas above, incl. Damper changes in 'Action' panel. You can instantly facilitate better/cleaner staccato/repeats with a more efficient damper - or loosen up a piano for slower music (and there's an outside chance loosening up might help you too, esp. with other things mentioned below) - I think defaults mostly find a fortuitous line - but for specific things, definitely worth trying alterations there.

Other ideas off-the-cuff..

Check out the presets you use - perhaps altering some things in the FX section. Subtlety can be key:

#
Lower the 'Delay' mix.. or turn it off. (this may ordinarily be OK with much of a repertoire but could get messy with very precise short timings like fast repeated notes for sure.). Milliseconds count, esp. when several are being generated.. continue knocking them down a notch..

##
Evaluate reverb pre-delay and alter its length, or lower it's influence by altering the reverb to focus more on "tails" rather than reflections (to reduce any perceived 'slap' back of attack which might compete with repeated notes' sense of accuracy). Altering also length/size, tone of reverb (either direction, darker/brighter) might help with specific pianos/presets/passages or maybe just repeats.

###
Microphone 'timings'. Within the mic panel, some alteration to the small delays seen in various mics might help. It is simple to reduce that - but it may be best to be subtle at times. Some small moves might really be all you need to help some echoic behaviours get out of your way (crotchets at 132BPM) whilst keeping the overall environment in there the way you want.

But the mics (just like IRL) is an endless pursuit.

If some of those things help your repeated notes, it might be possible to hook them up via MIDI to your middle pedal?? That way, you can press the pedal only for those passages where you wish to clean up some extra echoic sound. Pressing the pedal might be set to just lower the 'mix' of a few things a little - but it may be a nice touch.

It's kind of the other way around here, I will add. Pianoteq, esp. with low low buffer is just so fast that I often prefer to play certain music with a higher buffer. Using synths for a long time and beginning with guitar, I like instant responses from instruments - but for me, maybe because I came late to it, feels 'more real' with a little more latency.. IDK if it is worth your time Pianistically, but never say never.. have you tried a larger buffer size? For example, when I play fast passages with a synth, it's more natural to me to have a low buffer.. but even playing fast with piano (or dpiano) I feel there's a natural space (more lag) than with for example guitar (close on body, closer to ear... feel and sound instant).. a piano has a kind of gap.. there's to me, a kind of 'arc' of electricity somehow, with different latencies and for rythmic purposes I do often love a little more latency - helps physically/haptically.

Whatever it is, I hope you find the best ways to fit the piano around your needs! Cannot say there's another piano software product with such fine controls, so it makes me hopeful you'll get the performance you require. All this typed out pretty quick - trying not to waft on - hoping no er rrrrors - all the best!

It seems to be an overall latency  that can be felt with with fingers  but not with the ears as there are no dropped notes, it is more about 'haptic' feeling than auditive feeling and overall response.

Okay, that's pretty much what I expected. What's not expected is that it's any different with Pianoteq than with any other piano VSTi. In my experience the real-time response of most VSTis is on the order of a millisecond with not much variation, and Pianoteq is as fast as anything. The majority of latency you hear/feel is in the MIDI transmission and total audio output latency to which all VSTis are subject equally.

The only explanation I can think of is that a piano sound with a very fast and bright attack might feel more reponsive than one with a slower, darker one. You might experiment with different presets in Pianoteq and see if some feel better than others.

EDIT: It occurs to me that a shorter decay might play into that as well - faster damping might feel more reponsive.

Pianistically wrote:

well, the problem doesn’t exist in offline rendering , because there is no latency issue in offline sound rendering .

I did a quick test driving Pianoteq via recorded MIDI looped back over a physical MIDI cable so there is no possibility of the DAW or Pianoteq "looking ahead" to buffer up the audio in advance. Pianoteq seems perfectly capable of rendering repeated 32nd notes of 1/64th duration at 132bpm in real time with this setup.

So I would ask what exactly is the symptom of Pianoteq not keeping up with your live repetition? Are notes actually being dropped or is it maybe just that your feel for the repetition response is thrown off by the high latency of a 128-sample audio buffer plus other audio and MIDI hardware latencies?

It is more a question of easiness to maintain fluidity in the repetition pattern  when you compare 'sampled' vs 'modelled' with a computer with modest speed. I can also achieve the 132 bpm playing it with pianoteq on the mac book but I feel it is less easy than with the sampled VST I was referring to With a faster computer I have no doubt this difference would not be  perceptible. It seems to be an overall latency  that can be felt with with fingers  but not with the ears as there are no dropped notes, it is more about 'haptic' feeling than auditive feeling and overall response . Appreciate your testing though. thank you.

well, the problem doesn’t exist in offline rendering , because there is no latency issue in offline sound rendering .

I did a quick test driving Pianoteq via recorded MIDI looped back over a physical MIDI cable so there is no possibility of the DAW or Pianoteq "looking ahead" to buffer up the audio in advance. Pianoteq seems perfectly capable of rendering repeated 32nd notes of 1/64th duration at 132bpm in real time with this setup.

So I would ask what exactly is the symptom of Pianoteq not keeping up with your live repetition? Are notes actually being dropped or is it maybe just that your feel for the repetition response is thrown off by the high latency of a 128-sample audio buffer plus other audio and MIDI hardware latencies?

Pianistically wrote:

the issue is only for  repeated notes above 8-10 notes / sec , for instance I can achieve  Kapustin op40/3 repeated notes at tempo with sampled vst such as garritan , but cannot wirh pianoteq the tempo is crotchet = 132 bpm and the repeated notes are 1/16th notes which means they are supposed to be played at  8.8 notes per sec which is the limit on an upright piano and « easy «  to achieve on an acoustic grand .

Just as a test, what happens if you record the MIDI and do an offline render in your DAW of the same MIDI with the sampled vst and with pianoteq?

If the repetition problem goes away with an offline render, you can be reasonably certain it’s your computer’s ability to calculate fast enough that’s the problem.

I would expect Pianoteq to recalculate, because striking a key that rapidly surely produces a bit different sound that just repeating a strike with enough time in between for the strings to be completely dampened.

well, the problem doesn’t exist in offline rendering , because there is no latency issue in offline sound rendering . If the computer will take whatever time it takes to generate the next bit of digital information whichever format you use as an output , so the problem is only related to real time live playing . I guess this is why there are so many options in Logic Pro ( an other daws) for live optimisation.

the issue is only for  repeated notes above 8-10 notes / sec , for instance I can achieve  Kapustin op40/3 repeated notes at tempo with sampled vst such as garritan , but cannot wirh pianoteq the tempo is crotchet = 132 bpm and the repeated notes are 1/16th notes which means they are supposed to be played at  8.8 notes per sec which is the limit on an upright piano and « easy «  to achieve on an acoustic grand .

Just as a test, what happens if you record the MIDI and do an offline render in your DAW of the same MIDI with the sampled vst and with pianoteq?

If the repetition problem goes away with an offline render, you can be reasonably certain it’s your computer’s ability to calculate fast enough that’s the problem.

I would expect Pianoteq to recalculate, because striking a key that rapidly surely produces a bit different sound that just repeating a strike with enough time in between for the strings to be completely dampened.

But I was curious to see , if pianoteq was regenerating the whole succession of repeated notes when it detects such repetition or was resending the same  calculated wav sound with maybe volume adjustment

Yes, AFAIK Pianoteq is *always* recomputing the notes, so if you try to play faster than your computer can compute, you're going to have problems.

Try reducing the sample rate to the minimum (IIRC 11kHz). The sound would be not that good (especially in the treble, but you can use the bass for this test) and see if the situation improves. If it does, you know the culprit for sure, and a more powerful CPU would be needed to solve this problem (assuming you can't bear the lower sample rate)

dikrek wrote:

Pianistically wrote:

Yes , but in this particular case, there is no sustain . Usually , you don't use sustain pedal with repeated notes, as it ruins the effect , similarly to staccatos notes, they are usually payed without sustained pedals. So honestly , I don't think the problems lies with polyphony here. thks for trying to help anyway.

Understood but how many notes is it using without the sustain pedal being used, and repeating the same key at the speed you need? It will show this in the performance page.

5 to  6 prior to repetition ( chord) with sustain released during the repetition sequence with only 1 note so number of voices used is absolutely negligible.

And for the last test… if you run the app standalone, with the lowest buffer that doesn’t crackle (try 64), do you still perceive the delay?

Pianistically wrote:

dikrek wrote:

It’s not how many keys you’re hitting at once, as much as all the sustain calculations (that count as polyphony).

Set the poly to minimum and use the sustain pedal and hit the same key over and over, it should count as more than 1 poly.

Yes , but in this particular case, there is no sustain . Usually , you don't use sustain pedal with repeated notes, as it ruins the effect , similarly to staccatos notes, they are usually payed without sustained pedals. So honestly , I don't think the problems lies with polyphony here. thks for trying to help anyway.

Understood but how many notes is it using without the sustain pedal being used, and repeating the same key at the speed you need? It will show this in the performance page.

5 to  6 prior to repetition ( chord) with sustain released during the repetition sequence with only 1 note so number of voices used is absolutely negligible.

dikrek wrote:

Pianistically wrote:

Thks .
I will try that too but I guess in the piece I was referring to , the polyphony is not that big . After all, it might be an area where sampling technology as an advantage as notes attack are loaded in RAM and that probably explains why it’s faster under the fingers , I wonder whether such approach would work with modelling ie load note attacks for a given fxp file . Not sure its possible or not given the modelling workflow that needs to happen before note generation .

It’s not how many keys you’re hitting at once, as much as all the sustain calculations (that count as polyphony).

Set the poly to minimum and use the sustain pedal and hit the same key over and over, it should count as more than 1 poly.

Yes , but in this particular case, there is no sustain . Usually , you don't use sustain pedal with repeated notes, as it ruins the effect , similarly to staccatos notes, they are usually payed without sustained pedals. So honestly , I don't think the problems lies with polyphony here. thks for trying to help anyway.

Understood but how many notes is it using without the sustain pedal being used, and repeating the same key at the speed you need? It will show this in the performance page.

Pianistically wrote:

dikrek wrote:

Then try other extremes:

256 note, or auto pessimistic, or a really small number like 16.

I’m pretty sure it always tries to calculate everything properly.

Thks .
I will try that too but I guess in the piece I was referring to , the polyphony is not that big . After all, it might be an area where sampling technology as an advantage as notes attack are loaded in RAM and that probably explains why it’s faster under the fingers , I wonder whether such approach would work with modelling ie load note attacks for a given fxp file . Not sure its possible or not given the modelling workflow that needs to happen before note generation .

It’s not how many keys you’re hitting at once, as much as all the sustain calculations (that count as polyphony).

Set the poly to minimum and use the sustain pedal and hit the same key over and over, it should count as more than 1 poly.

Yes , but in this particular case, there is no sustain . Usually , you don't use sustain pedal with repeated notes, as it ruins the effect , similarly to staccatos notes, they are usually payed without sustained pedals. So honestly , I don't think the problems lies with polyphony here. thks for trying to help anyway.

dikrek wrote:

Pianistically wrote:

  it is set as auto optimistic . But I guess the issue is probably resolved with a more powerful computer as I use as Mac book pro 2017 , and this limitation doesn’t exist on a silicon Mac . But I was curious to see , if pianoteq was regenerating the whole succession of repeated notes when it detects such repetition or was resending the same  calculated wav sound with maybe volume adjustment

Then try other extremes:

256 note, or auto pessimistic, or a really small number like 16.

I’m pretty sure it always tries to calculate everything properly.

Thks .
I will try that too but I guess in the piece I was referring to , the polyphony is not that big . After all, it might be an area where sampling technology as an advantage as notes attack are loaded in RAM and that probably explains why it’s faster under the fingers , I wonder whether such approach would work with modelling ie load note attacks for a given fxp file . Not sure its possible or not given the modelling workflow that needs to happen before note generation .

It’s not how many keys you’re hitting at once, as much as all the sustain calculations (that count as polyphony).

Set the poly to minimum and use the sustain pedal and hit the same key over and over, it should count as more than 1 poly.