Been thinking about how much better/more accurate and more stable
sampling has gotten over the last 20 years or so and
wondering if you guys think that the ‘new’ Modeling technology
will do the some over time.
If modeling tech gets to pro levels in its sounds, perhaps
this would really help with file size/latency/ etc etc.
Or do you think Modeling tech can only go ‘so far’ and we
will never be able to completely replace very accurate samples.
The one area where I’ve looked into this issue is piano libraries. While the Roland RD-700 still uses samples, to my ears it sounds better than the more expensive and highly praised V-Piano, which uses modelling. However, I do think modelling has a future, because the V-Piano has a few advantages: the change in tonal character as one strikes the key harder and harder is much more realistic, and sympathetic resonances are easier to control. Right now however to my ears samples still sound closer to the real thing, especially during the sustain portion of a note. Maybe I’m a purist/snob/psycho, but when it comes to piano samples, you get what you pay for; it’s not a coincidence that the most expensive ones, like the Imperial Grand or the Imperfect Piano, are also the best
I’ve been looking into this as well. Can’t give a good conclusion because I haven’t had the chance to compare both at a good level, but on paper modeling should be the winner if they can develop it to perfection. To get samples to sound better you need more samples, it’s that simple.
Sampling has sort of come to the end of the read, it won’t get much better, and it’s old technology from when nobody expected computers to be able to do real time computation needed for something like modeling. Still sampling is great, no doubt! But give modeling another decade …
In my opinion both have future (for some decades at least).
Sampling:
Very authentic reproduction of single notes of the instrument being “emulated”.
Need HUGE amounts of memory (and work during sampling) to cover various playing styles.
Almost impossible to reproduce interaction between notes played at the same time.
Modelling:
Authenticity of modelled sound depends only in quality of the model. In theory, you can create a “perfect” model of any instrument you can imagine.
You can even create instruments, which does not even exist in real life.
Good models of complex instruments require lots of CPU power.
Modelling is of course the ultimate winner here, if we’re going to have faster and faster CPUs in the future. If growth of the CPU power stops for some reason, it still will be better to use sampling for some kinds of instruments for a long time.
In the piano world, I would never have been convinced to any extent by modelling, that is until Pianoteq came along, this has really closed the gap more so than before, at least to my ears. Just sayin’. The arguments for and against are somewhat moot as soon as you take a piano into a mix, when the instrument stands alone, that to me, is the litmus test.
Pianoteq is a great example of modeling.
So is the equally impressive Lounge Lizard and the discontinued B4 from NI.
I can’t for the best of me see why NI stopped supporting that
I use modeled plugs whenever I can to save some precious ram for my sampled drum-kits etc