Hello, dear friends!
I read here on this forum about importing "reverb pulses" in WAV format into Pianoteq Standard. However, I would like to know if the reverb presets that are already native to Pianoteq are also based on samples, captured in environments. In this case, could we say that Pianoteq is not 100% Physical Modeling, since the reverb is based on sampling?
Hello, dear friends!
I read here on this forum about importing "reverb pulses" in WAV format into Pianoteq Standard. However, I would like to know if the reverb presets that are already native to Pianoteq are also based on samples, captured in environments. In this case, could we say that Pianoteq is not 100% Physical Modeling, since the reverb is based on sampling?
The reverb is on top of the modelling, not sure why you want to have this discussion anyway.
The actual piano sound is modelled.
Professor Leandro Duarte wrote:Hello, dear friends!
I read here on this forum about importing "reverb pulses" in WAV format into Pianoteq Standard. However, I would like to know if the reverb presets that are already native to Pianoteq are also based on samples, captured in environments. In this case, could we say that Pianoteq is not 100% Physical Modeling, since the reverb is based on sampling?The reverb is on top of the modelling, not sure why you want to have this discussion anyway.
The actual piano sound is modelled.
I don't see such a discussion as a "problem". Would this be a prohibited topic here on the forum?
Finally, is the Pianoteq's reverb based on sampling or not?
dikrek wrote:Professor Leandro Duarte wrote:Hello, dear friends!
I read here on this forum about importing "reverb pulses" in WAV format into Pianoteq Standard. However, I would like to know if the reverb presets that are already native to Pianoteq are also based on samples, captured in environments. In this case, could we say that Pianoteq is not 100% Physical Modeling, since the reverb is based on sampling?The reverb is on top of the modelling, not sure why you want to have this discussion anyway.
The actual piano sound is modelled.
I don't see such a discussion as a "problem". Would this be a prohibited topic here on the forum?
Finally, is the Pianoteq's reverb based on sampling or not?
It’s a convolution reverb. It says so right on the product page.
Why do you ask though?
BTW I tend to use my own reverb.
The built-in reverb uses impulse responses, so yes, they're sampled spaces.
The piano sound itself is entirely modelled -- and anechoic. This is distinct from the acoustic environment in which a piano is placed. A Steinway sound obviously has no acoustic environment baked in, so in modelling it, you can't include information about space.
There's perhaps a slight nuance here, in that Pianoteq does model the way a piano sound propagates through an anechoic space but this expressly excludes information about the characteristics of the specific space a piano happens to find itself in. If Pianoteq were to use an algorithmic reverb instead, this would indeed mean that it also uses modeled spaces. But this has nothing to do with the piano sound in itself.
Professor Leandro Duarte wrote:dikrek wrote:The reverb is on top of the modelling, not sure why you want to have this discussion anyway.
The actual piano sound is modelled.
I don't see such a discussion as a "problem". Would this be a prohibited topic here on the forum?
Finally, is the Pianoteq's reverb based on sampling or not?
It’s a convolution reverb. It says so right on the product page.
Why do you ask though?
BTW I tend to use my own reverb.
I ask because, in a way, this disappoints me a little.
I am an almost fanatical enthusiast for pure Physical Modeling and I have always seen Pianoteq as a materialization of my searches.
Please forgive me if I seem a little rude, but I confess that I "hate sampler". The simple fact of thinking that something was recorded to be played back later makes me sad and makes me want to vomit.
dikrek wrote:Professor Leandro Duarte wrote:I don't see such a discussion as a "problem". Would this be a prohibited topic here on the forum?
Finally, is the Pianoteq's reverb based on sampling or not?
It’s a convolution reverb. It says so right on the product page.
Why do you ask though?
BTW I tend to use my own reverb.
I ask because, in a way, this disappoints me a little.
I am an almost fanatical enthusiast for pure Physical Modeling and I have always seen Pianoteq as a materialization of my searches.Please forgive me if I seem a little rude, but I confess that I "hate sampler". The simple fact of thinking that something was recorded to be played back later makes me sad and makes me want to vomit.
Then you can rest assured the piano modelling isn’t sampled at all. If you don’t want to use the built-in reverb then don’t
(I assume then you use only algorithmic reverbs in your DAW).
BTW an advanced production technique is to use a high quality convolution reverb to put the sounds in a natural space, followed by an algorithmic one that has fancy modulation.
It’s all just tools.
Professor Leandro Duarte wrote:dikrek wrote:It’s a convolution reverb. It says so right on the product page.
Why do you ask though?
BTW I tend to use my own reverb.
I ask because, in a way, this disappoints me a little.
I am an almost fanatical enthusiast for pure Physical Modeling and I have always seen Pianoteq as a materialization of my searches.Please forgive me if I seem a little rude, but I confess that I "hate sampler". The simple fact of thinking that something was recorded to be played back later makes me sad and makes me want to vomit.
Then you can rest assured the piano modelling isn’t sampled at all. If you don’t want to use the built-in reverb then don’t
(I assume then you use only algorithmic reverbs in your DAW).
BTW an advanced production technique is to use a high quality convolution reverb to put the sounds in a natural space, followed by an algorithmic one that has fancy modulation.
It’s all just tools.
I completely agree with this response. But there is some irony in Modartt's choosing to implement a "sampled" reverb in a modeled instrument. I'd not really thought about it before.
I'm not so sure that the internal reverbs, especially the legacy reverb presets, aren't algorithmic. If you look at the options for fine-tuning the reverbs, wav impulses have different options than the built-ins and exclude settings like "resample" and lack the readout of Hz rate, duration, channels and bit-depth. So I'd suspect that the sampled reverb is only a function that's active if the user loads a sample--which also can take time for the engine to parse which isn't the case for the defaults. So as long as you're using a built-in reverb, I'm rather confident it's algorithmic.
Also, if memory serves, the wav impulse support was only added around PTQ 5 or 6, before that it could only have been algorithmic. Also, since PTQ is still roughly 40-50MB of data, the impulses would need to be baked into the executable, as there isn't anything in the preset files or user data that sidecars or embeds the impulse files (which means if you use an impulse file, the preset retains the reference to the impulse file but when sharing it the other user must have the same impulse file). As there are over 20 presets, if each were a 1MB file which is a reasonable average for impulse files (my personal collection has files ranging from 700kB for stereo 44.1kHz up to 6MB for long quad 48kHz files), that's already over half of the executable's binary data, which seems highly unlikely to me.
So, I'd say that if you're not manually loading an impulse file, you rest assured that you're not playing a hybrid piano, you're playing something entirely modeled from start-to-finish.
I'm not so sure that the internal reverbs, especially the legacy reverb presets, aren't algorithmic. If you look at the options for fine-tuning the reverbs, wav impulses have different options than the built-ins and exclude settings like "resample" and lack the readout of Hz rate, duration, channels and bit-depth. So I'd suspect that the sampled reverb is only a function that's active if the user loads a sample--which also can take time for the engine to parse which isn't the case for the defaults. So as long as you're using a built-in reverb, I'm rather confident it's algorithmic.
Also, if memory serves, the wav impulse support was only added around PTQ 5 or 6, before that it could only have been algorithmic. Also, since PTQ is still roughly 40-50MB of data, the impulses would need to be baked into the executable, as there isn't anything in the preset files or user data that sidecars or embeds the impulse files (which means if you use an impulse file, the preset retains the reference to the impulse file but when sharing it the other user must have the same impulse file). As there are over 20 presets, if each were a 1MB file which is a reasonable average for impulse files (my personal collection has files ranging from 700kB for stereo 44.1kHz up to 6MB for long quad 48kHz files), that's already over half of the executable's binary data, which seems highly unlikely to me.
So, I'd say that if you're not manually loading an impulse file, you rest assured that you're not playing a hybrid piano, you're playing something entirely modeled from start-to-finish.
That may well be the case since the executable is so small. But in any case, convolution reverbs aren't necessarily "static" anyway. The initial impulse is one thing, but what the software does with it is another.
Case in point: https://www.liquidsonics.com/software/s...fessional/
You can combine algorithmic and convolution in a single reverb.
Hello, dear friends!
I read here on this forum about importing "reverb pulses" in WAV format into Pianoteq Standard. However, I would like to know if the reverb presets that are already native to Pianoteq are also based on samples, captured in environments. In this case, could we say that Pianoteq is not 100% Physical Modeling, since the reverb is based on sampling?
Looking at the package content on Mac OS Big Sur shows:
Pianoteq Unix executable 22.2 Mb
Presources.dat 31.6 Mb
The overall size of pianoteq8.app is 54.7 Mb
The documentation clearly states that the Reverberation is based on Convolution. The only exception seems to be the legacy reverb preset. The samples are presumably included in the Presources. dat files.
I don't think it is shocking, the core business of Moddart is to model string instruments, not to model spaces and many people use external plugins within DAW or their own impulse files. Moreover , Convolution is not equivalent to sampling as it makes use of an impulse file to transform a sound file using an algorithm .
I'm not so sure that the internal reverbs, especially the legacy reverb presets, aren't algorithmic. If you look at the options for fine-tuning the reverbs, wav impulses have different options than the built-ins and exclude settings like "resample" and lack the readout of Hz rate, duration, channels and bit-depth. So I'd suspect that the sampled reverb is only a function that's active if the user loads a sample--which also can take time for the engine to parse which isn't the case for the defaults. So as long as you're using a built-in reverb, I'm rather confident it's algorithmic.
Also, if memory serves, the wav impulse support was only added around PTQ 5 or 6, before that it could only have been algorithmic. Also, since PTQ is still roughly 40-50MB of data, the impulses would need to be baked into the executable, as there isn't anything in the preset files or user data that sidecars or embeds the impulse files (which means if you use an impulse file, the preset retains the reference to the impulse file but when sharing it the other user must have the same impulse file). As there are over 20 presets, if each were a 1MB file which is a reasonable average for impulse files (my personal collection has files ranging from 700kB for stereo 44.1kHz up to 6MB for long quad 48kHz files), that's already over half of the executable's binary data, which seems highly unlikely to me.
So, I'd say that if you're not manually loading an impulse file, you rest assured that you're not playing a hybrid piano, you're playing something entirely modeled from start-to-finish.
I liked your answer. You brought a plausible and comforting explanation. I hope it is confirmed.
The instruments are 100% modelled. The additional reverb however (at least since I've used it) is made up of impulse responses (sampling).
To me it's actually a somewhat complex question because the sound from the physical models is changed by virtual microphone positioning. IOW the in room sound is already modelled before it gets to our chosen reverb. However it's more like the modelled microphones are working in an anechoic chamber - they aren't picking up room reverberations - though the model changes with wall positioning, and moving microphones closer to the floor.
The legacy reverb option is an algorithm. You could use a more sophisticated positional reverb like ARVerb.
Modern piano recordings have been recorded with multiple microphones and mixed to taste for longer than I've been alive on this planet. Close positioned microphones mixed with open positioned microphones. This then presented through stereo fakery.
Recordings are inherently artificial anyway.
https://www.modartt.com/user_manual?pro...mp;lang=en
"15.6. Reverberation
A new convolution reverberation unit simulates acoustic surroundings ranging from Studio up to very long reverbs like Cathedral or Taj Mahal. It includes other types of reverbs such as Plate, Spring, Speakers, Broceliande... In PIANOTEQ STANDARD or PIANOTEQ PRO, you can even load your own reverb impulses via Load WAV impulse... in the reverb menu."
I don't see any indication that there is an algorithmic reverb besides the convolution reverb.
tmyoung wrote:So, I'd say that if you're not manually loading an impulse file, you rest assured that you're not playing a hybrid piano, you're playing something entirely modeled from start-to-finish.
I liked your answer. You brought a plausible and comforting explanation. I hope it is confirmed.
I'm afraid it's not accurate, though. The spaces in Pianoteq are sampled impulse responses. There's no getting around it.
If you really are allergic to sampling, just turn off the built-in reverb and insert an algorithmic reverb after Pianoteq in your DAW. I use Cinematic Rooms Pro, which is a really excellent and realistic, fully algorithmic reverb. I prefer it to Pianoteq's built-in reverb for most scenarios.
Further to my last post here in this thread-
I wonder if at some point in the future there will be an offline rendering option for superior spatial rendering. Extending the microphone modelling to actually capturing the audio with the full room reverberation based on room shape, diffraction, absorption, occlusion, amount of listeners in the room and so on.
This may be extremely CPU intensive for very little gain - in other words it's completely possible we would simply prefer the results we are already getting with our idealised version of a virtual room anyway
It's probably too much to do real time- unless it's something that could be offloaded to the GPU?
Again however I think those are better suited to calculations that involve impulse responses.
I think being "allergic to sampling" is funny. Convolution reverbs oftentimes sound way more realistic than algorithmic reverbs. It's just a fact of life, live with it.
I think being "allergic to sampling" is funny. Convolution reverbs oftentimes sound way more realistic than algorithmic reverbs. It's just a fact of life, live with it.
I agree. I used to prefer convolution reverbs to algorithmic ones but Cinematic Rooms (especially the Pro version) has changed that.
EvilDragon wrote:I think being "allergic to sampling" is funny. Convolution reverbs oftentimes sound way more realistic than algorithmic reverbs. It's just a fact of life, live with it.
I agree. I used to prefer convolution reverbs to algorithmic ones but Cinematic Rooms (especially the Pro version) has changed that.
I like many subtly artificial sounding reverbs and more fantastical ones, along with those that can mix with convolution.
Often they create an idealised perfect room or infinite space - arguably more enjoyable than real environment impulse responses which can recreate room imperfections on certain material - inevitably room convolution is going to be more realistic than say "Cinematic Rooms" but also warts and all - so it's nice to have both!
There is also nostalgic sound qualities to certain reverbs such as plate reverbs, spring reverbs, early digital reverbs.
Pianophile wrote:EvilDragon wrote:I think being "allergic to sampling" is funny. Convolution reverbs oftentimes sound way more realistic than algorithmic reverbs. It's just a fact of life, live with it.
I agree. I used to prefer convolution reverbs to algorithmic ones but Cinematic Rooms (especially the Pro version) has changed that.
I like many subtly artificial sounding reverbs and more fantastical ones, along with those that can mix with convolution.
Often they create an idealised perfect room or infinite space - arguably more enjoyable than real environment impulse responses which can recreate room imperfections on certain material - inevitably room convolution is going to be more realistic than say "Cinematic Rooms" but also warts and all - so it's nice to have both!There is also nostalgic sound qualities to certain reverbs such as plate reverbs, spring reverbs, early digital reverbs.
Plate reverbs are nice
Hi all,
Convolution is the name of the process which applies a reverb impulse to a dry signal. But it does not mean that the impulse has been recorded from a real room, the impulse could be computed via physical modelling of the room. We do not claim to have a physically modeled reverb because it would be an overstatement. Still, the impulses in Pianoteq are computed from the parameters, they are not pre-recorded impulses.
Hi all,
Convolution is the name of the process which applies a reverb impulse to a dry signal. But it does not mean that the impulse has been recorded from a real room, the impulse could be computed via physical modelling of the room. We do not claim to have a physically modeled reverb because it would be an overstatement. Still, the impulses in Pianoteq are computed from the parameters, they are not pre-recorded impulses.
Probably the lion's share of the convolution reverb plugins are based on impulse responses captured from hardware reverbs anyway, as opposed to captured via microphones in real rooms.
Then there are those that use self generated impulse responses, rather than those directly captured from hardware or real room reverb. It sounds like you are saying it is one of those.
Hi all,
Convolution is the name of the process which applies a reverb impulse to a dry signal. But it does not mean that the impulse has been recorded from a real room, the impulse could be computed via physical modelling of the room. We do not claim to have a physically modeled reverb because it would be an overstatement. Still, the impulses in Pianoteq are computed from the parameters, they are not pre-recorded impulses.
thank you for clarification . In fact , given all the mathematical transformations that happen in the convolution/ deconvolution process I wonder if there is any audible advantage using a real recorded impulse file vs a virtual one . The comparison has nothing to do when one compares the pro’s and con’s of instrument modelling vs instrument sampling which I think are clear . In the case of reverberation , you are just looking to enrich the sound of the instrument and have a nice effect which is , no matter what the source impulse file is , essentially calculated. Comparing algo vs convo is more meaningful as the mathematical transformations of the instrument sound file are different . That’s obviously my own opinion and I trust Modartt is best placed to chose the appropriate method to enrich dry sounds .
Self-generated impulse responses needn't be the cheap inferior option, they will use substantially less space.
Here is a premium convolution using a synthesis engine:
https://www.liquidsonics.com/software/illusion/
I haven't got that one but I tried it some years ago and it's good.
I think being "allergic to sampling" is funny. Convolution reverbs oftentimes sound way more realistic than algorithmic reverbs. It's just a fact of life, live with it.
Even if that were true, Mario, and I disagree that it is, one could also argue that is equally funny, if not a lot funnier, to make that distinction if the sound source is a virtual piano (or any other virtual instrument, for that matter) — modelled or sampled, that makes little difference as none of them sound entirely realistic. In fact, the difference between a virtual piano and a real one is MUCH greater than whatever difference you assume there to be between the perceived realism of a convolution-based reverb and that of a good algorithmic one.
Differences have to have musical relevance and need to be judged in context, to be worth being considered as meaningful, in my opinion. In a musical universe that is solely populated by virtual instruments, whatever difference there may be between convolution and algorithmic reverbs, has no such relevance whatsoever.
_
Virtual pianos can get pretty damned close to the real thing - at least compared to those as captured on a stereo recording, though not as experienced in person - either from the position as an audience listener or as the performer.
The most artificial element of virtual instruments remaining is the spatial representation. This is a limitation of the playback systems rather than the plugins themselves.
Whether regular stereo, binaural, surround and alternative formats non are truly representative of the way that real instruments sound in a real environment.
Combine that with real world compromised loudspeaker positioning and limited dynamics of playback systems it's probably a limitation that could remain permanently.
Algorithms Vs convolution it's more about picking a pleasing sound palette. Depending on the impulse responses and the algorithms involved one can be more realistic than the other and vice versa.
Virtual pianos can get pretty damned close to the real thing - at least compared to those as captured on a stereo recording, though not as experienced in person - either from the position as an audience listener or as the performer.
The most artificial element of virtual instruments remaining is the spatial representation. This is a limitation of the playback systems rather than the plugins themselves.
Whether regular stereo, binaural, surround and alternative formats non are truly representative of the way that real instruments sound in a real environment.Combine that with real world compromised loudspeaker positioning and limited dynamics of playback systems it's probably a limitation that could remain permanently.
Algorithms Vs convolution it's more about picking a pleasing sound palette. Depending on the impulse responses and the algorithms involved one can be more realistic than the other and vice versa.
true .Also speakers no matter how good they are don’t diffuse the sound like an acoustic grand piano soundboard and as you say this is not anymore a plug-in or VST problem.
I am sorry but don’t agree. If speakers or playback systems are such a big problem, I wouldn’t be able to enjoy the many glorious and sublime sounding piano recordings I have in my music collection on my computer, would I?
Besides, I (and many other people) can recognize a good, convincing pianosound, and differentiate between a real and a virtual piano, on a single Auratone or even on a speaker of inferior quality. Only to say that I’m of the opinion that speakers and playback systems don’t and shouldn’t even enter into this whole equation.
Good sound, and the realism of real instruments captured therein, is remarkably invulnerable, I’ve always found. It can withstand being converted to lowres mp3’s, it survives AM and FM modulation, it emerges almost intact after being submitted to Soundcloud or Youtube compression algorithms, you can EQ it badly and it still sounds believable, … in short … you can seriously mess with it in many ways, and it will still come out sounding instantly recognizeable as what it is: a recording of reality.
(In marked contrast: the illusion of realism clumsily suggested by virtual instruments is often amazingly and frustratingly vulnerable and fragile, and even the slightest tampering with it will upset it. A virtual instrument may fool the ear for a moment that it sounds pretty convincing, but add other instruments and sounds to the mix, and all of sudden, the instrument’s many weaknesses are revealed. I’ve had that happening more times than I can remember. Sometimes even adding reverb will expose the flaws in a virtual instrument.
And talking of virtual pianos getting “pretty damn close” to the real thing: I’m afraid I don’t agree with that either. If I were to say that I’m Martha Argerich’s equal on the piano because I can play a C major scale as well as she can, everyone would immediately dismiss that statement, and its logic, as utterly ridiculous nonsense. And rightly so. And yet, that very same reasoning is used here.
Virtual pianos are nowhere near close to the real thing, in my opinion. At best, they can come pleasingly close — and, luckily, usefully so — to a tiny fraction of the infinite chaotic sonic complexity that is still the exclusive domain of the real thing.
_
Good sound, and the realism of real instruments captured therein, is remarkably invulnerable, I’ve always found. It can withstand being converted to lowres mp3’s, it survives AM and FM modulation, it emerges almost intact after being submitted to Soundcloud or Youtube compression algorithms, you can EQ it badly and it still sounds believable, … in short … you can seriously mess with it in many ways, and it will still come out sounding instantly recognizeable as what it is: a recording of reality.
(In marked contrast: the illusion of realism clumsily suggested by virtual instruments is often amazingly and frustratingly vulnerable and fragile, and even the slightest tampering with it will upset it. A virtual instrument may fool the ear for a moment that it sounds pretty convincing, but add other instruments and sounds to the mix, and all of sudden, the instrument’s many weaknesses are revealed. I’ve had that happening more times than I can remember. Sometimes even adding reverb will expose the flaws in a virtual instrument.
I will agree here - digital artifice is easily exposed whereas adding extreme modifications to a live recording just makes it sound like a bad live recording, not something originating from a synthetic source.
This is similar to the way that I have found that digital CGI movie graphics look more realistic the higher the resolution of the video format. Drop the resolution down and suddenly they look far more fake. On the other hand digitized real footage continues to look like real footage, just more blocky, even at extremely low resolutions.
And talking of virtual pianos getting “pretty damn close” to the real thing: I’m afraid I don’t agree with that either. If I were to say that I’m Martha Argerich’s equal on the piano because I can play a C major scale as well as she can, everyone would immediately dismiss that statement, and its logic, as utterly ridiculous nonsense. And rightly so. And yet, that very same reasoning is used here.
_
Oddball analogy I have to say. I think they would call that a straw man argument these days.
Yes you have to paint over the cracks to make a virtual digital representation more convincing in a recording. It's too easy to see the wizard behind the curtain - but the tools for the wizard are getting a lot better.
Then again perhaps my suspension of disbelief is too strong!
I am sorry but don’t agree. If speakers or playback systems are such a big problem, I wouldn’t be able to enjoy the many glorious and sublime sounding piano recordings I have in my music collection on my computer, would I?
Besides, I (and many other people) can recognize a good, convincing pianosound, and differentiate between a real and a virtual piano, on a single Auratone or even on a speaker of inferior quality. Only to say that I’m of the opinion that speakers and playback systems don’t and shouldn’t even enter into this whole equation.Good sound, and the realism of real instruments captured therein, is remarkably invulnerable, I’ve always found. It can withstand being converted to lowres mp3’s, it survives AM and FM modulation, it emerges almost intact after being submitted to Soundcloud or Youtube compression algorithms, you can EQ it badly and it still sounds believable, … in short … you can seriously mess with it in many ways, and it will still come out sounding instantly recognizeable as what it is: a recording of reality.
(In marked contrast: the illusion of realism clumsily suggested by virtual instruments is often amazingly and frustratingly vulnerable and fragile, and even the slightest tampering with it will upset it. A virtual instrument may fool the ear for a moment that it sounds pretty convincing, but add other instruments and sounds to the mix, and all of sudden, the instrument’s many weaknesses are revealed. I’ve had that happening more times than I can remember. Sometimes even adding reverb will expose the flaws in a virtual instrument.And talking of virtual pianos getting “pretty damn close” to the real thing: I’m afraid I don’t agree with that either. If I were to say that I’m Martha Argerich’s equal on the piano because I can play a C major scale as well as she can, everyone would immediately dismiss that statement, and its logic, as utterly ridiculous nonsense. And rightly so. And yet, that very same reasoning is used here.
Virtual pianos are nowhere near close to the real thing, in my opinion. At best, they can come pleasingly close — and, luckily, usefully so — to a tiny fraction of the infinite chaotic sonic complexity that is still the exclusive domain of the real thing.
_
Just compare your recordings with the real thing when you are in the room where the grand piano is played and recorded. Analog all the way with no conversion absolutely lossless and authentic . I beg to differ with your opinion , I am afraid .
Virtual pianos can get pretty damned close to the real thing - at least compared to those as captured on a stereo recording, though not as experienced in person - either from the position as an audience listener or as the performer.
The most artificial element of virtual instruments remaining is the spatial representation. This is a limitation of the playback systems rather than the plugins themselves.
Whether regular stereo, binaural, surround and alternative formats non are truly representative of the way that real instruments sound in a real environment.Combine that with real world compromised loudspeaker positioning and limited dynamics of playback systems it's probably a limitation that could remain permanently.
Algorithms Vs convolution it's more about picking a pleasing sound palette. Depending on the impulse responses and the algorithms involved one can be more realistic than the other and vice versa.
I think it more has to do with the fact that headphones don't currently correlate with the head's position relative to the piano. If you turn your head or move around, the piano sound doesn't change as it would if we weren't in a virtual environment. If that's the case, then with more and more AR/VR tech becoming commonplace--if the Apple VR gamble pays off in iPod/iPhone/iPad proportions (though I'll believe it when I see it), we'll likely see better spatial controls for headphones being relatively easily added to Mac-compatible software within the next 3-ish years if widespread adoption occurs. Speakers present a different set of considerations, but in practice, I prefer being able to hear the audience perspective as I play--as it saves a lot of mental calculus for translating what I hear as a performer to what the audience hears. In a word, I'm lazy...
That said, both algorithms and convolvers aren't quite to the point of perfect environment emulation--especially for large multisource sound generators like virtual orchestras, but I suspect as tools like AI are applied to larger and larger audio dataset questions like improving reverb, we could finally reach a true *and simple to implement* "can't tell the difference" type of solution.
Piet De Ridder wrote:I am sorry but don’t agree. If speakers or playback systems are such a big problem, I wouldn’t be able to enjoy the many glorious and sublime sounding piano recordings I have in my music collection on my computer, would I?
Besides, I (and many other people) can recognize a good, convincing pianosound, and differentiate between a real and a virtual piano, on a single Auratone or even on a speaker of inferior quality. Only to say that I’m of the opinion that speakers and playback systems don’t and shouldn’t even enter into this whole equation.Good sound, and the realism of real instruments captured therein, is remarkably invulnerable, I’ve always found. It can withstand being converted to lowres mp3’s, it survives AM and FM modulation, it emerges almost intact after being submitted to Soundcloud or Youtube compression algorithms, you can EQ it badly and it still sounds believable, … in short … you can seriously mess with it in many ways, and it will still come out sounding instantly recognizeable as what it is: a recording of reality.
(In marked contrast: the illusion of realism clumsily suggested by virtual instruments is often amazingly and frustratingly vulnerable and fragile, and even the slightest tampering with it will upset it. A virtual instrument may fool the ear for a moment that it sounds pretty convincing, but add other instruments and sounds to the mix, and all of sudden, the instrument’s many weaknesses are revealed. I’ve had that happening more times than I can remember. Sometimes even adding reverb will expose the flaws in a virtual instrument.And talking of virtual pianos getting “pretty damn close” to the real thing: I’m afraid I don’t agree with that either. If I were to say that I’m Martha Argerich’s equal on the piano because I can play a C major scale as well as she can, everyone would immediately dismiss that statement, and its logic, as utterly ridiculous nonsense. And rightly so. And yet, that very same reasoning is used here.
Virtual pianos are nowhere near close to the real thing, in my opinion. At best, they can come pleasingly close — and, luckily, usefully so — to a tiny fraction of the infinite chaotic sonic complexity that is still the exclusive domain of the real thing.
_
Just compare your recordings with the real thing when you are in the room where the grand piano is played and recorded. Analog all the way with no conversion absolutely lossless and authentic . I beg to differ with your opinion , I am afraid .
The way you put your reply I'm guessing you are actually in agreement with Piet rather than disagreeing with him like you apparently said?
He's arguing the virtual instrument representation is nowhere even remotely close to realistic.
I think you meant to say you are disagreeing with me rather than Piet.
Further to my last comment Piet isn't talking about analogue Vs digital recording, only virtual vs real instruments in a real room.
I'm pretty confident his answer still stands whatever recording format - digital or analogue. He's not going down the audiophile rabbit hole.
Perhaps I didn’t word my views very well. What I was trying to say, is this: people tend to think too quickly, in my view, that we’re there, or nearly there, simply because a virtual instrument manages, in some production or other, to suggest a particular presence of the real thing to an effective degree. You get that a lot with orchestral sounds too. And with drums. Well, with all instruments, really. And, not surprisingly but quite puzzling to me nonetheless, with pianos too.
But it’s not because a capably handled orchestral library can replace what a real orchestra would contribute to a particular musical situation — which, in specific situations, is perfectly possible these days — that we can conclude that samples are ready to replace orchestras in every possible situation. Samples, at this point in time, are still only capable of simulating only the tiniest slice of what an orchestra is capable of, to a more or less believable degree. That tiny slice already covers quite some ground, sure, but it’s still nothing more than a miniscule fraction of a real orchestra’s sonic universe.
Given what most people like or have to use samples for, it’s no suprise that samples can be used most effectively to simulate what may be called the cinematic variant of the orchestral idiom. That’s also the idiom which nearly all developers aim for when creating orchestral sample libraries. But move outside that cinematic world, and you quickly find that samples are next to impotent to render most of the timbres, articulations, textures, blends, expressions, dynamics and overal presence of an orchestra. Try, for example, mocking up this recording, even with the best libraries and mock-up tools available these days, and see how far you get. Nowhere is where, as will (or should, anyway) already become painfully clear halfway through the first bar.
Sampled drums: same thing. They can do many things very well and they suit many contemporary stylings wonderfully. No discussion. But try to summon the ghost of, say, a Max Roach, a Dannie Richmond or a Tony Williams with even the best virtual drums, and you’re — again — nowhere. Absolutely nowhere. Sampled drums simply can not sound like (a recording of) a real kit played with talent, skill and passion. That’s still totally impossible today.
And I am of the strongest conviction that the same holds true for virtual pianos as well. They’re a great and musically wholly satisfying solution in many circumstances — partly because, to begin with, most of these circumstances only require one or two of the many personalities that a (real) piano can assume — and the finest ones can give us countless hours of profound musical pleasure, I don’t dispute any of that, but even so, I’m nowhere near ready or prepared to call them “pretty damn close” to the real thing. And by ‘real thing’ I mean a recording of the real thing, obviously.
And it’s not just a matter of the right timbres, mechanically and/or physically correct resonances or the right dynamic colours and levels at the right (= desired) time, there’s also a rather wide and very problematic range of more technical aspects that virtual instruments struggle unsuccessfully with: frequency build-ups, imaging issues, dynamic differentiations, phase issues and — finally returning to the original topic — spatial issues. To name just a few.
Here’s some audio. Listen to this on whatever playback system you happen to have available — audophile speakers, earbuds, a smartphone, your studio monitors, an iPad, … it doesn’t matter — and you should always hear what can’t be mistaken for anything other than a great sounding recording of a(n equally great sounding) real piano. This is a soundworld that is a million miles beyond the territory available to us through virtual means.
As is this. It’s unfortunately impossible to list everything I would like to point out in this recording — and this post is already long enough as it is anyway —, let me just say that nearly every note, every attack, every release, every chord, every consonance and dissonance, every pedal move (take particular note of the pedal action in this performance), and the almost tangible presence of all the materials that go into the building of a grand piano, is something I haven’t heard yet in any virtualization of the instrument.
It’s not that I don’t like virtual instruments. I love ‘em. Seriously. Much of my musical activity revolves around them. But I can only work with them if I accept and even embrace their artificiality and their many limitations. Any other approach, especially one seeking realism, is inevitably doomed to be deeply frustrating and dissatisfying to me. So I most certainly never ever consider them close to the real thing. I like to think of them, and also treat them, as inhabiting a parallel reality. The reality where plastic flowers live.
- - -
Audio example 1: Franz Schubert, Symphony nr. 4 in C minor - IV. Allegro / Freiburger Barockorchester (Harmonia Mundi)
Audio example 2: "Parvaneh" / Thomas Rückert, piano (ECM)
Audio example 3: Igor Stravinsky, "Epitáfio / Erika Ribeiro, piano (Rocinante)
_
Yes, Piet, that's exactly how I feel about it. Pianoteq is itself a wondrous thing, and I'm very glad to have it - especially as there's no way I could afford studio time with a real piano. As I've said before, Pianoteq is a very good simulator. But... 'exactly like a real piano'? 'Taint so, honey, 'taint so!
Piet's links worked horribly for some reason here.
Here's an alternative:
https://youtu.be/yCpiZBUOYuQ?feature=shared
Key Fumbler wrote:Virtual pianos can get pretty damned close to the real thing - at least compared to those as captured on a stereo recording, though not as experienced in person - either from the position as an audience listener or as the performer.
The most artificial element of virtual instruments remaining is the spatial representation. This is a limitation of the playback systems rather than the plugins themselves.
Whether regular stereo, binaural, surround and alternative formats non are truly representative of the way that real instruments sound in a real environment.Combine that with real world compromised loudspeaker positioning and limited dynamics of playback systems it's probably a limitation that could remain permanently.
Algorithms Vs convolution it's more about picking a pleasing sound palette. Depending on the impulse responses and the algorithms involved one can be more realistic than the other and vice versa.
I think it more has to do with the fact that headphones don't currently correlate with the head's position relative to the piano. If you turn your head or move around, the piano sound doesn't change as it would if we weren't in a virtual environment. If that's the case, then with more and more AR/VR tech becoming commonplace--if the Apple VR gamble pays off in iPod/iPhone/iPad proportions (though I'll believe it when I see it), we'll likely see better spatial controls for headphones being relatively easily added to Mac-compatible software within the next 3-ish years if widespread adoption occurs. Speakers present a different set of considerations, but in practice, I prefer being able to hear the audience perspective as I play--as it saves a lot of mental calculus for translating what I hear as a performer to what the audience hears. In a word, I'm lazy...
That said, both algorithms and convolvers aren't quite to the point of perfect environment emulation--especially for large multisource sound generators like virtual orchestras, but I suspect as tools like AI are applied to larger and larger audio dataset questions like improving reverb, we could finally reach a true *and simple to implement* "can't tell the difference" type of solution.
There's another question in there. How the player perceives the realism of the experience in the moment of performance. The experience that fools the brain that you are there in the sonic sense, rather than they were there. .
HRTF technology with headphones is a crude approximation of replicating loudspeaker in front of the listener. Expecting a naturalistic fully 3d experience from that is a tall order. It could be fun trying though.
I think that trying to use headphones for that spatial virtual reality/HRTF truly fooling the brain is probably much, much harder to achieve realistically than getting the audio over the line to be that much closer to a real performance recording in a real room. What Modartt has achieved so far with such computational efficiency is remarkable. It's so much more alive than relatively static samples.
I've never fancied trying one of those head-tracking headphone apps. For me, one of the advantages of headphone listening is that the sound doesn't change when you move your head. But then, we're all different...
I've never fancied trying one of those head-tracking headphone apps. For me, one of the advantages of headphone listening is that the sound doesn't change when you move your head. But then, we're all different...
It might be cool for video games.
The in your head sound is very different to the natural HRTF experience - I can imagine small flat dwellers that always wanted loudspeakers preferring the experience - if it's done really well.
In reality from a commercial perspective I doubt it would ever get beyond special interest.
Gaming with VR - if they can manage convincing vertical height dimensions too it might be cool. Surround from headphones has always been flaky at best.
Perhaps I didn’t word my views very well. What I was trying to say, is this: people tend to think too quickly, in my view, that we’re there, or nearly there, simply because a virtual instrument manages, in some production or other, to suggest a particular presence of the real thing to an effective degree. You get that a lot with orchestral sounds too. And with drums. Well, with all instruments, really. And, not surprisingly but quite puzzling to me nonetheless, with pianos too.
But it’s not because a capably handled orchestral library can replace what a real orchestra would contribute to a particular musical situation — which, in specific situations, is perfectly possible these days — that we can conclude that samples are ready to replace orchestras in every possible situation. Samples, at this point in time, are still only capable of simulating only the tiniest slice of what an orchestra is capable of, to a more or less believable degree. That tiny slice already covers quite some ground, sure, but it’s still nothing more than a miniscule fraction of a real orchestra’s sonic universe.
Given what most people like or have to use samples for, it’s no suprise that samples can be used most effectively to simulate what may be called the cinematic variant of the orchestral idiom. That’s also the idiom which nearly all developers aim for when creating orchestral sample libraries. But move outside that cinematic world, and you quickly find that samples are next to impotent to render most of the timbres, articulations, textures, blends, expressions, dynamics and overal presence of an orchestra. Try, for example, mocking up this recording, even with the best libraries and mock-up tools available these days, and see how far you get. Nowhere is where, as will (or should, anyway) already become painfully clear halfway through the first bar.Sampled drums: same thing. They can do many things very well and they suit many contemporary stylings wonderfully. No discussion. But try to summon the ghost of, say, a Max Roach, a Dannie Richmond or a Tony Williams with even the best virtual drums, and you’re — again — nowhere. Absolutely nowhere. Sampled drums simply can not sound like (a recording of) a real kit played with talent, skill and passion. That’s still totally impossible today.
And I am of the strongest conviction that the same holds true for virtual pianos as well. They’re a great and musically wholly satisfying solution in many circumstances — partly because, to begin with, most of these circumstances only require one or two of the many personalities that a (real) piano can assume — and the finest ones can give us countless hours of profound musical pleasure, I don’t dispute any of that, but even so, I’m nowhere near ready or prepared to call them “pretty damn close” to the real thing. And by ‘real thing’ I mean a recording of the real thing, obviously.
And it’s not just a matter of the right timbres, mechanically and/or physically correct resonances or the right dynamic colours and levels at the right (= desired) time, there’s also a rather wide and very problematic range of more technical aspects that virtual instruments struggle unsuccessfully with: frequency build-ups, imaging issues, dynamic differentiations, phase issues and — finally returning to the original topic — spatial issues. To name just a few.
Here’s some audio. Listen to this on whatever playback system you happen to have available — audophile speakers, earbuds, a smartphone, your studio monitors, an iPad, … it doesn’t matter — and you should always hear what can’t be mistaken for anything other than a great sounding recording of a(n equally great sounding) real piano. This is a soundworld that is a million miles beyond the territory available to us through virtual means.
As is this. It’s unfortunately impossible to list everything I would like to point out in this recording — and this post is already long enough as it is anyway —, let me just say that nearly every note, every attack, every release, every chord, every consonance and dissonance, every pedal move (take particular note of the pedal action in this performance), and the almost tangible presence of all the materials that go into the building of a grand piano, is something I haven’t heard yet in any virtualization of the instrument.It’s not that I don’t like virtual instruments. I love ‘em. Seriously. Much of my musical activity revolves around them. But I can only work with them if I accept and even embrace their artificiality and their many limitations. Any other approach, especially one seeking realism, is inevitably doomed to be deeply frustrating and dissatisfying to me. So I most certainly never ever consider them close to the real thing. I like to think of them, and also treat them, as inhabiting a parallel reality. The reality where plastic flowers live.
- - -
Audio example 1: Franz Schubert, Symphony nr. 4 in C minor - IV. Allegro / Freiburger Barockorchester (Harmonia Mundi)
Audio example 2: "Parvaneh" / Thomas Rückert, piano (ECM)
Audio example 3: Igor Stravinsky, "Epitáfio / Erika Ribeiro, piano (Rocinante)_
I want to make it clear that I do not believe virtual pianos are ready to actually replace the real thing. I do however feel they are much, much closer to put it mildly in terms of being able to replace a real piano on a recording compared to the equivalent for a full orchestra - and the intimate nature of expression within a Quartet is probably even harder to fake.
Should virtual pianos do so now or ever if they actually achieve that sonic and performance fidelity? - probably no. People always want the real deal, even if in many circumstances they wouldn't tell the difference. So even if they do achieve that technically it's always going to be treated with suspicion.
However emulating the rigid mechanical nature of a piano (an instrument with extremely close copies within controllers) is completely and utterly different to an orchestra with all the massive amount of articulations, especially amongst the string instruments.
Even if we had perfect virtual instruments these are unlikely to be in the hands of equivalent quality musicians that are simultaneously virtuoso performers and grade A Midi musicians - and certainly not for every instrument.
Hello, dear friends!
I read here on this forum about importing "reverb pulses" in WAV format into Pianoteq Standard. However, I would like to know if the reverb presets that are already native to Pianoteq are also based on samples, captured in environments. In this case, could we say that Pianoteq is not 100% Physical Modeling, since the reverb is based on sampling?
In the subject of convolution reverb, I had lot of free time this week end and tried many free impulse files available online . I have to say the quality varies from ‘awful’ to ‘so so’ . A few of them work ok for some frequencies but completely ruins other frequencies so it seems to me ( just an opinion) that the best strategy is either to use pianoteq convolution presets ( some of them are really nice ) or to use a very good external plugin running pianoteq as VST ( altiverb is apparently well regarded and also Logic Pro Space designer is very good as well with many options to adjust the reverb ) I was for a long time under the false impression than convolution was better than algorithmic reverb, but a bad impulse file or bad sound used for the impulse model could lead to crappy sound .
Professor Leandro Duarte wrote:Hello, dear friends!
I read here on this forum about importing "reverb pulses" in WAV format into Pianoteq Standard. However, I would like to know if the reverb presets that are already native to Pianoteq are also based on samples, captured in environments. In this case, could we say that Pianoteq is not 100% Physical Modeling, since the reverb is based on sampling?In the subject of convolution reverb, I had lot of free time this week end and tried many free impulse files available online . I have to say the quality varies from ‘awful’ to ‘so so’ . A few of them work ok for some frequencies but completely ruins other frequencies so it seems to me ( just an opinion) that the best strategy is either to use pianoteq convolution presets ( some of them are really nice ) or to use a very good external plugin running pianoteq as VST ( altiverb is apparently well regarded and also Logic Pro Space designer is very good as well with many options to adjust the reverb ) I was for a long time under the false impression than convolution was better than algorithmic reverb, but a bad impulse file or bad sound used for the impulse model could lead to crappy sound .
Many good impulses around - here’s this thread
https://gearspace.com/board/music-compu...verbs.html
And this crazy long one
https://www.kvraudio.com/forum/viewtopic.php?t=107337
You can find online free Bricasti M7 impulses, Lexicon 480L… tons of options if one doesn’t want to spend the extra money for Tai Chi, Seventh Heaven or Cinematic Rooms Pro, Symphony 3D, Verberate, etc etc.
Pianistically wrote:Professor Leandro Duarte wrote:Hello, dear friends!
I read here on this forum about importing "reverb pulses" in WAV format into Pianoteq Standard. However, I would like to know if the reverb presets that are already native to Pianoteq are also based on samples, captured in environments. In this case, could we say that Pianoteq is not 100% Physical Modeling, since the reverb is based on sampling?In the subject of convolution reverb, I had lot of free time this week end and tried many free impulse files available online . I have to say the quality varies from ‘awful’ to ‘so so’ . A few of them work ok for some frequencies but completely ruins other frequencies so it seems to me ( just an opinion) that the best strategy is either to use pianoteq convolution presets ( some of them are really nice ) or to use a very good external plugin running pianoteq as VST ( altiverb is apparently well regarded and also Logic Pro Space designer is very good as well with many options to adjust the reverb ) I was for a long time under the false impression than convolution was better than algorithmic reverb, but a bad impulse file or bad sound used for the impulse model could lead to crappy sound .
Many good impulses around - here’s this thread
https://gearspace.com/board/music-compu...verbs.html
And this crazy long one
https://www.kvraudio.com/forum/viewtopic.php?t=107337
You can find online free Bricasti M7 impulses, Lexicon 480L… tons of options if one doesn’t want to spend the extra money for Tai Chi, Seventh Heaven or Cinematic Rooms Pro, Symphony 3D, Verberate, etc etc.
Thks I have already the Simplicity files but had issues in the past with the stereo balance , but I will give another go.
dikrek wrote:Pianistically wrote:In the subject of convolution reverb, I had lot of free time this week end and tried many free impulse files available online . I have to say the quality varies from ‘awful’ to ‘so so’ . A few of them work ok for some frequencies but completely ruins other frequencies so it seems to me ( just an opinion) that the best strategy is either to use pianoteq convolution presets ( some of them are really nice ) or to use a very good external plugin running pianoteq as VST ( altiverb is apparently well regarded and also Logic Pro Space designer is very good as well with many options to adjust the reverb ) I was for a long time under the false impression than convolution was better than algorithmic reverb, but a bad impulse file or bad sound used for the impulse model could lead to crappy sound .
Many good impulses around - here’s this thread
https://gearspace.com/board/music-compu...verbs.html
And this crazy long one
https://www.kvraudio.com/forum/viewtopic.php?t=107337
You can find online free Bricasti M7 impulses, Lexicon 480L… tons of options if one doesn’t want to spend the extra money for Tai Chi, Seventh Heaven or Cinematic Rooms Pro, Symphony 3D, Verberate, etc etc.
Thks I have already the Simplicity files but had issues in the past with the stereo balance , but I will give another go.
For free algorithmic try Acon Verberate Basic - same “vivid hall” algo as the full verberate. Not many controls in the free version but still.
https://acondigital.com/products/verberate-basic
Pianistically wrote:dikrek wrote:Many good impulses around - here’s this thread
https://gearspace.com/board/music-compu...verbs.html
And this crazy long one
https://www.kvraudio.com/forum/viewtopic.php?t=107337
You can find online free Bricasti M7 impulses, Lexicon 480L… tons of options if one doesn’t want to spend the extra money for Tai Chi, Seventh Heaven or Cinematic Rooms Pro, Symphony 3D, Verberate, etc etc.
Thks I have already the Simplicity files but had issues in the past with the stereo balance , but I will give another go.
For free algorithmic try Acon Verberate Basic - same “vivid hall” algo as the full verberate. Not many controls in the free version but still.
thanks ! I will try this one too . There are so many plugins and IR files to try . The process is actually very time consuming , as you have to spend a lot of time adjusting the reverb to sound good as I have realised that using an IR file is not a simple plug and play and compare A/B exercise , specifically with piano given the wide range of frequencies the instrument produce , so you need to adjust and tune the reverb quite a lot before the sound is good , find the right mixing parameters and then re- adjust some parameters in Pianoteq , like hardness , unisson , and EQ to make it work .Then what works for a preset , won’t work for another . Bottom line, it’s a full time job !
dikrek wrote:Pianistically wrote:Thks I have already the Simplicity files but had issues in the past with the stereo balance , but I will give another go.
For free algorithmic try Acon Verberate Basic - same “vivid hall” algo as the full verberate. Not many controls in the free version but still.
thanks ! I will try this one too . There are so many plugins and IR files to try . The process is actually very time consuming , as you have to spend a lot of time adjusting the reverb to sound good as I have realised that using an IR file is not a simple plug and play and compare A/B exercise , specifically with piano given the wide range of frequencies the instrument produce , so you need to adjust and tune the reverb quite a lot before the sound is good , find the right mixing parameters and then re- adjust some parameters in Pianoteq , like hardness , unisson , and EQ to make it work .Then what works for a preset , won’t work for another . Bottom line, it’s a full time job !
Samplicity has updated files - BTW I used the quad ones only, and they seemed to work well
Hi all,
Convolution is the name of the process which applies a reverb impulse to a dry signal. But it does not mean that the impulse has been recorded from a real room, the impulse could be computed via physical modelling of the room. We do not claim to have a physically modeled reverb because it would be an overstatement. Still, the impulses in Pianoteq are computed from the parameters, they are not pre-recorded impulses.
Thank you, Julien.
You said exactly what I was hoping to hear.
Pianistically wrote:dikrek wrote:For free algorithmic try Acon Verberate Basic - same “vivid hall” algo as the full verberate. Not many controls in the free version but still.
thanks ! I will try this one too . There are so many plugins and IR files to try . The process is actually very time consuming , as you have to spend a lot of time adjusting the reverb to sound good as I have realised that using an IR file is not a simple plug and play and compare A/B exercise , specifically with piano given the wide range of frequencies the instrument produce , so you need to adjust and tune the reverb quite a lot before the sound is good , find the right mixing parameters and then re- adjust some parameters in Pianoteq , like hardness , unisson , and EQ to make it work .Then what works for a preset , won’t work for another . Bottom line, it’s a full time job !
Samplicity has updated files - BTW I used the quad ones only, and they seemed to work well
I will the new updated files . The quad wav files were the one I have tried in the past . Tank you .
dikrek wrote:Pianistically wrote:thanks ! I will try this one too . There are so many plugins and IR files to try . The process is actually very time consuming , as you have to spend a lot of time adjusting the reverb to sound good as I have realised that using an IR file is not a simple plug and play and compare A/B exercise , specifically with piano given the wide range of frequencies the instrument produce , so you need to adjust and tune the reverb quite a lot before the sound is good , find the right mixing parameters and then re- adjust some parameters in Pianoteq , like hardness , unisson , and EQ to make it work .Then what works for a preset , won’t work for another . Bottom line, it’s a full time job !
Samplicity has updated files - BTW I used the quad ones only, and they seemed to work well
I will the new updated files . The quad wav files were the one I have tried in the past . Tank you .
Maybe they updated the quad ones too / anyway check it out
I can only work with them if I accept and even embrace their artificiality and their many limitations. Any other approach, especially one seeking realism, is inevitably doomed to be deeply frustrating and dissatisfying to me. So I most certainly never ever consider them close to the real thing. I like to think of them, and also treat them, as inhabiting a parallel reality. The reality where plastic flowers live.
Ha! It reminds me a bit of watching computer graphics evolve through the 80s/90s/00s. I remember being blown away by the realism time and again, and now I get a kick out of showing my kids those same movies and they laugh at how poorly they compare by today's standards.
Pianoteq is such a neat intersection of two things that interest me -- pianos and synthesizers. I have the most fun with it when I play it as such. You can morph it and modulate parameters based on expression and do all kinds of things you can't do with either a sample library or a real acoustic grand, and plays so responsively! Really amazing technology by today's standards, and I'm excited to watch it evolve.