There are always caveats when it comes to modeling.
awsum, that's actually just about completely wrong, though it's a common misconception. Digital audio is perfectly capable of perfectly reproducing sound so long as the clock is steady and the sampling frequency is appropriate to the material. The later is trivial today. The former isn't completely trivial, but even pretty bad modern clocks introduce jitter that's almost inaudible if you don't know how to listen for it or if you're not in a very good listening environment.
The "problem" with guitars & tubes is that the distortion introduced by tubes is both nonlinear and coherent, which makes replicating them much harder that replicating something where either of those is not the case. They've gotten
very good in recent years. There's just one real problem with replicating a wall of sound in software: volume.
Your ears respond differently to quiet sounds than they do to loud ones. I don't just mean that one sounds louder and might hurt and the other doesn't…I mean the frequency response of human ears depends on the volume of the sounds you're hearing. This effect is described in Fletcher-Munson curves.
Unfortunately, compensating for that in software/modeling is just about impossible because it would have to "listen" to the output and adjust filters to match. There's stuff that can do it, but not much. And it's generally not worth it.
The other effect of volume is that guitars respond to sound waves near them. The strings do it a lot. The wood does it some. Playing a guitar in front of a wall of sound will just not sound the same as playing into a modeling of a wall of sound where the guitar can't "hear" it and respond to it. Part of that is just feedback, but it also alters how both the wood and the strings vibrate and alters the overtones that the guitar strings generate before they even get to anything else.
I'm honestly kind of the opposite of an analog snob. I've done subjective and objective listening tests. In a good enough listening environment, I can tell the difference between something that's been run
through an analog console and something that hasn't (the board in question was a Neve 88RS…listening was done on genelecs in a pretty good control room)…I can hear the difference between well-encoded 320kbps mp3s and uncompressed audio (same environment the first time, but I can do it on Sennheiser HD-25s and prosumer level converters as well). Vinyl vs. uncompressed digital is easy. Tape (1/4" 2-track) vs. uncompressed digital is easy.
I honestly prefer the way digital sounds. It introduces fewer artifacts. It has an audibly lower noise floor. It has a
much wider usable dynamic range.
Tube saturation/distortion (and to a lesser extent tape saturation) is the only place where I don't believe that digital audio holds up. Unfortunately……I play blues. That edge b/t natural tube compression and slightly broken just barely clipping is where I live. Modeling doesn't really stand up compared to tube amps. But Amplitube 3 is free. And I live in an apartment. Even if I could justify the expense of a Fuchs ODS (my favorite amps…hands down) I couldn't make it sound good at apartment volumes anyway. And I'm nowhere near a good enough guitarist to blow 3 grand on a head.
)
