Is physically modeled stuff actually worth the hype?

[CYKLVM]

I learned about pianoteq recently, is this physical modeled stuff actually that good in your opinion (compared to sampled instruments), and also, is there any similar stuff for guitars yet?

[Dirk Diggler]

Yes to me its a huge improvement over samples. Size and performance are the biggest differences.

[DaveClark]

Just beware of the fact that most instruments and effects that claim to perform physical modelling are not actually doing true physical modelling; instead they are performing phenomenological modelling. There is nothing wrong with this other than people claiming that they are doing physical modelling. If drums were physically modelled, for example, there would not be a simple pitch shifting knob for tuning the drums. You would have to virtually tighten and loosen virtual lugs in particular sequences, virtually straighten out virtual wrinkles in the virtual skins, etc. All materials would have associated physical parameters.

When I queried a developer of a number of very well-known instruments that are touted as being physically modelled as to whether they were actually using physical models or phenomenological models, he admitted frankly that they were, in fact, phenomenological models, but also expressed his opinion that we were getting closer to the point where we could indeed solve partial differential equations in real time.

Again, there is nothing wrong with employing phenomenological models. We are lucky to have them.

Why this matters: People have a tendency to believe that if an instrument or effect is physically modelled, then nothing remains to be done. “It’s all there,” to quote someone who believes that MSoundFactory is the best physical modelling program out there and apparently that MDrummer is a physical modelling program. Quite to the contrary: MDrummer, for example, uses a pitch-shifting knob to tune the drums. When you tune a real drum set, you are doing far more than just changing the pitch, as explained recently by someone from Native Instruments.

[zerocrossing]

As I like to say, it’s rarely useful to talk about the underlying technology behind an instrument, besides as an academic discussion. If it works well for you, then that’s it. I have SWAM Cello, and I like it a lot, but I’ll also use samples when they make sense.

[ugo]

I really enjoy making sounds that have acoustic-like attributes but do not actually exist as real instruments so, for me, the big advantage of these kinds of synths is their flexibility rather than accuracy.

[Seafire Mk2]

A lot of new topics from a new member...hmmm

[DCrown]

Size is the only advantage imo, playability and response is terrible in most cases, that' s why I prefer samples.
Pianoteq is the best example, I programm some weird sounds sometimes, thats what I use it for sometimes, but when I play piano I can perform on a sampled piano plugin a lot better, the feel, playability, response is superior

[tapper mike]

A lot of new topics from a new member...hmmm

Good don't scare the new user away. We could use fresh content from a fresh face.

[CapnLockheed]

A lot of new topics from a new member...hmmm

Says the guy that's only been a member since January and only has 271 posts...hrmnn...LOL!!

[aMUSEd]

I learned about pianoteq recently, is this physical modeled stuff actually that good in your opinion (compared to sampled instruments), and also, is there any similar stuff for guitars yet?

It's great for Pianos, harp, keys and some percussion, but I'm less convinced by the classical guitar emu - that one needs further work.

My favourite PM plugin for guitar is Chromaphone (even though String Studio is meant to be for strings I prefer the denser, more well defined plucked tones I can get from Chromaphone)

[CandelaAudio]

Some Physical Modelling is much harder to emulate than others. Sustained tones in general are harder than plucks, etc. For example I am yet to hear any convincing PM for bowed string instruments simply because bowed string instruments require emulation of a huge range of different articulation types, playing techniques and organic imperfections. A real player can produce the same one note in so many different ways, on different strings, in different positions, bowed at different speeds and pressure, with different contact points with the string, etc. I haven't seen anything that comes anywhere close to modelling half of that convincingly. It only sounds convincing for those not truly familiar with the real thing, in my opinion.

However something I really love about Physical Modelling is that in general it does sound more organic than other types of synthesis, and it allows you to take a sound which may offer an illusion of being "real" and "acoustic", and turn it into something "un-real" and synthetic. The seamless morphing between the "real" and the "weird".

[jamcat]

SWAM strings can be quite convincing, but it takes a lot of work from the user. The new Expressive E Soliste also sounds quite good.

With these types of physically modeled string instruments, various articulations aren’t directly modeled. Instead, the physical response is modeled, and the various articulations naturally arise from that.

SWAM:



Soliste:

[CandelaAudio]

They are good instruments, I agree, and you can do a lot with them, but I would never mistake them for the real thing. I find them more interesting as some sort of hybrid pseudo acoustic instruments than as emulations of the real thing. They are still too perfect and precise, and to my ears have still an unmistakable electronic tone. The staccato attack still sounds synthetic to me, and the transition from white tone, no vibrato, to vibrato does not sound to me anywhere near as organic as a real player would make it.

[jamcat]

I will spend a lot of time creating constant variation within the performance when I work with SWAM strings. When you use artificial sounds, you tend to be very aware and self-conscious of them sounding synthetic. If you have too stagnant of a performance, the strings start sounding more like flutes.

When you’re playing real instruments, you’re trying to perform as flawlessly as possible. But with artificial sounds, you end up doing the opposite, adding flaws to try to make it sound more organic. It’s a peculiar thing.

Ultimately, I think the best approach (other than hiring real string players and booking a studio) is to capture the best SWAM performance you can, and then use AI to transform it into a “real” string performance. I’m not sure if we’re there yet for a full quartet, but there are some solo violin AI models out there, like ACE Studio.

[Seafire Mk2]

A lot of new topics from a new member...hmmm

Says the guy that's only been a member since January and only has 271 posts...hrmnn...LOL!!

Not quite