Search:
Submit Search


Physical modeling vs sampling

A discussion forum for anything even marginally Hauptwerk-related.

Physical modeling vs sampling

Postby engrssc » Sat Nov 17, 2018 10:46 pm

Several schools of thoughts/opinions here. Got into quite a discussion recently.

https://www.native-instruments.com/forum/threads/physical-modeling-vs-sampling.320277/

Any wannabe "tweakers" out there? Thoughts? Easier to tweak physical models than sampled.

http://www.church-organs.com/?p=278 This site is hosted by Allen Organ BTW. Personally, I respect Allen for various reasons, but don't appreciate a few of "their" people bashing every other organ including Hauptwerk. OTOH, Allen is in a different category from Hauptwerk.

Starting out my organ instruction was on a Hammond (concert model). Then my organ teacher introduced me to 4M Austin real pipe organ. What a WOW experience, the effect of which has never left me.

Bottom line, in my book, nothing beats real pipes. :)

Rgds,
Ed
User avatar
engrssc
Member
 
Posts: 5628
Joined: Mon Aug 22, 2005 11:12 pm
Location: Roscoe, IL, USA

Re: Physical modeling vs sampling

Postby RichardW » Sun Nov 18, 2018 10:07 am

OK, physical modelling is easier to tweak but it is rather like instead of buying a print of the Mona Lisa in a nice frame you are given a piece of canvas, some oil paints and a few bits of wood.

It is much more flexible but you need a special talent to make anything out of it.

Maybe one day it will be possible to acquire a physically modelled sound engine that will allow a configuration file to be applied that creates a given organ. It will take a lot of development to not sound artificial, though. It would also allow the flexibility to add any other stops you might want.

As someone who is interested in real organs I like the approach of sampling and Martin has made it sound as good as I need so I do not feel the need to explore anything else.

Regards,
Richard
User avatar
RichardW
Member
 
Posts: 788
Joined: Sun Oct 28, 2007 10:16 am
Location: UK

Re: Physical modeling vs sampling

Postby johnstump_organist » Sun Nov 18, 2018 10:35 am

I remember when modeling was first talked about in the early 90's, everyone was saying it would replace sampling as the better way to go. It would supposedly allow attack and decay variable to be varied in real time, which would make it a better way to produce virtual orchestral instruments since the variety of attack and release that wind and string players have on their instruments are a big part of the overall sound. It would also allow for real time simulation of the variants allowed with a good tracker action. I always said pipe organs did well as sampled instruments because the attack and release were pretty much always the same on any electrically controlled chest action. Some variation is possible on a good tracker action, but it is relatively small compared to what a violinist can do.
It was, I think i heard this correctly at the time, developed at Stanford University and they were publishing the details of how to do it and making it available for any and all to use. That may be part of why it didn't take off like it was supposed to - no one could patent it and make exclusive money off of the technology.
Viscount is now saying they use modeling. I haven't heard any of their instruments yet. I know of one person who heard one and said it wasn't very successful but they thought that might be due to low quality amps and speakers used in that installation.
I'm going to go take a look at the discussion now. Hope it doesn't cover a bunch of what I just said.
John
User avatar
johnstump_organist
Member
 
Posts: 466
Joined: Wed Mar 25, 2009 2:15 pm
Location: San Antonio, Texas

Re: Physical modeling vs sampling

Postby telemanr » Sun Nov 18, 2018 4:50 pm

Pianoteq is modelled and folks seem to think highly of it.
Rob Enns
User avatar
telemanr
Member
 
Posts: 1555
Joined: Sat Feb 18, 2006 5:18 pm
Location: Brampton, ON, Canada

Re: Physical modeling vs sampling

Postby engrssc » Sun Nov 18, 2018 6:11 pm

telemanr wrote:Pianoteq is modeled and folks seem to think highly of it.


Esp Pianoteq Version 6. Which applies to primarily pianos and not organs. Quite different instruments. I play both and I appreciate each for what it is. OTOH each is played differently..

Rgds,
Ed
User avatar
engrssc
Member
 
Posts: 5628
Joined: Mon Aug 22, 2005 11:12 pm
Location: Roscoe, IL, USA

Re: Physical modeling vs sampling

Postby engrssc » Sun Nov 18, 2018 9:06 pm

Loading appears to go much faster with physical modeling vs sampling. As in less to load?

Rgds,
Ed
User avatar
engrssc
Member
 
Posts: 5628
Joined: Mon Aug 22, 2005 11:12 pm
Location: Roscoe, IL, USA

Re: Physical modeling vs sampling

Postby engrssc » Sun Nov 18, 2018 10:43 pm

telemanr wrote:Pianoteq is modelled and folks seem to think highly of it.


https://www.pianoteq.com/audio/modeld/Jerome%20Kern%20-%20Old%20Man%20River%20-%20Joe%20Felice.mp3

Sounds pretty close to a Steinway Model D if maybe just a little too perfect. (Pianoteq 6 physical modeled.) The Pianoteq 6 permits tuning the model down to a somewhat less perfect but probably more realistic sound.

And that's a piano, not an organ.

Then there's the Garritan Virtual Steinway used by Pierre Fracalanza:

https://www.youtube.com/watch?v=jSGmFtVaizQ&start_radio=1&list=RDjSGmFtVaizQ&t=45

Rgds,
Ed
User avatar
engrssc
Member
 
Posts: 5628
Joined: Mon Aug 22, 2005 11:12 pm
Location: Roscoe, IL, USA

Re: Physical modeling vs sampling

Postby Grant_Youngman » Mon Nov 19, 2018 12:55 am

Hmmm … It appears the Garritan Steinway has gone missing. It's no longer on Garritan's website, and resellers are showing it as "No longer available". Must have been a licensing arrangement that went up in smoke ..
Grant
User avatar
Grant_Youngman
Member
 
Posts: 995
Joined: Sat Jan 20, 2007 7:50 pm
Location: Savannah, Ga

Re: Physical modeling vs sampling

Postby organassist » Mon Nov 19, 2018 4:30 am

I think that it is “horses for courses”. I really like Hauptwerk and think that, currently, sampling produces by far the best results for organ. However, for piano, I think nothing beats Pianoteq 6. It is a huge improvement on Version 5 - which seemed very good until I played 6.

The key difference, in my opinion, is the amount of interaction between notes. Although this is present on a pipe organ, it is much more obvious on a piano. A piano has less than 250 strings and the structure of them (wound or unwound) is very similar, making modeling a practical solution. However, each note can be struck with different velocity and the use of the sustain pedal has a complex impact on the resonance of all notes. These factors cannot be sampled due to the huge number of possible interactions.

Many organs have ten times this number of pipes, constructed of many varied materials. The computer power required to accurately model this complexity is probably beyond most current personal computers. Added to this is the fact that the building acoustic “is the most important stop” - unless this is also modeled accurately the result will be less than realistic.

However, all of this is a personal opinion. I am very happy with both of these respective solutions - in each case they are a joy to play!
organassist
Member
 
Posts: 84
Joined: Sat Mar 12, 2005 4:01 am
Location: Colchester, England

Re: Physical modeling vs sampling

Postby engrssc » Mon Nov 19, 2018 5:42 am

organassist wrote: I am very happy with both of these respective solutions - in each case they are a joy to play!


The joy to play is probably one of the most important factors of any virtual instrument.

Along the way, being able to create a MIDI file in Pianoteq 6 along with a MIDI playlist is cool. 8) and then there is the large number of available pianos represented. Some I didn't know existed. The same holds true with the number of organ sample sets available with Hauptwerk.

Quality is many times mentioned. Here I think a most important component is the quality and accuracy of the sound system(s) used. This must not be overlooked. I spoke with a slightly uninspired Hauptwerk newbee who mentioned he found H/W didn't sound any better than the sound of his 20+ year old XXXX organ. Asking some questions, found he is playing H/W thru the existing amplifiers and speakers of his 20+ year old XXXX organ. Duh. :roll: He never thought about that as making a difference. :o

Rgds,
Ed
User avatar
engrssc
Member
 
Posts: 5628
Joined: Mon Aug 22, 2005 11:12 pm
Location: Roscoe, IL, USA

Re: Physical modeling vs sampling

Postby engrssc » Mon Nov 19, 2018 6:11 am

Grant_Youngman wrote:Hmmm … It appears the Garritan Steinway has gone missing. It's no longer on Garritan's website, and resellers are showing it as "No longer available". Must have been a licensing arrangement that went up in smoke ..


And Pianoteq advertises that Steinway has acknowledged the quality and accuracy of the Pianoteq Steinway pianos and thus permits Pianoteq to use the Steinway name, That speaks to the available quality of the Pianoteq physically modeled Steinways. :wink:

So much of what a piano sounds like depends on the (playing) technique of the pianist. In piano contests, the same instrument is used by each of the contestants. Even playing the same composition, there can be quite different results again determined by the technique and skill of the pianist. How hard an organist presses down on the organ keys doesn't affect the resultant sound for the most part. Thinking in terms of second touch and tracker action as being exceptions.

Then there's this thought (from someone else regarding piano vs organ):
:
The pianist has to continue striking keys to continue the sound. ... Another difference is the number of sounds each instrument can generate. A piano sounds like just that- a piano! An organ on the other hand is a bit more versatile; it can be altered to sounds like a variety of different instruments.


Rgds,
Ed
User avatar
engrssc
Member
 
Posts: 5628
Joined: Mon Aug 22, 2005 11:12 pm
Location: Roscoe, IL, USA

Re: Physical modeling vs sampling

Postby IainStinson » Mon Nov 19, 2018 6:57 am

You might find this link to the 1988 organ in the nave of Worcester Cathedral of interest.http://www.churchmusic.org.uk/cathmus/worcorg2.php

Dr Peter Comerford's work at Bradford University (based on Z80 processors I think) has been used by a number of companies as the basis of their instruments. Musicom developed controllers, which were built on this system of additive sythesis, and these were/are used in some of the more expensive digital organs (such as Copemann Hart) others (such as Eminent) developed their own less expensive version of this type of system. These organs do sound rather good if the voicing and settings of the parameters for the synthesis were set by a good "vociers" but setting them up was not a very easy task. Organs using added synthesis tended to need less memory than sampled based organs as they only needed to store the parameters to produce the sound not the wave samples. In the early 2000s I serious looked at buying an Eminent instrument - and had not the Phoenix System appeared (which has more "sample memory" and multiple processors compared to other organs at the time) I think I would have gone ahead with the Eminent. (Interesting that Eminent now sell a Hauptwerk console and in the UK will provide Hauptwerk installation services.)

As memory became cheaper, sample based instruments could use more and higher resolution samples and were able to provide a superior quality sound in a more ecconomiically. (We sometimes forget that early sample based digital organs often only 5-12 samples per stop and "stretched" (interpolated) these samples to produce the whole range of a rank of "pipes".)

Colin Pykett has some intersting articles on his websitehttp://www.pykett.org.uk/ which discus the differences between sampled and sythesised sounds in digital organs.
User avatar
IainStinson
Member
 
Posts: 651
Joined: Tue Dec 29, 2009 7:08 pm
Location: NW England, UK

Re: Physical modeling vs sampling

Postby engrssc » Mon Nov 19, 2018 10:08 am

Indeed, very interesting, Iain. Thanks for sharing.

I can remember back to 1988 which now seems like a long time ago. If that rate of progress keeps pace, just think what the next 30 or so years will produce. 8)

I do think so much of the success of esp a large installation depends on many factors none the less any more important than proper voicing. Having the ability to make these detailed adjustments even 'string by string' appears to be infinite in the Pianoteq physical modeling system. This is beyond the 'use right out the box' approach. If possible, it would be a welcomed addition to Hauptwerk. That is being able to do more extensive voicing (pipe by pipe) than is presently possible. All this in the hands of a skilled "voicer" for sure. This plus improved accuracy of sampling is where the future will go IMHO.

Rgds,
Ed
User avatar
engrssc
Member
 
Posts: 5628
Joined: Mon Aug 22, 2005 11:12 pm
Location: Roscoe, IL, USA


Return to General discussion

Who is online

Users browsing this forum: No registered users and 5 guests