The only way to get good acoustic sound is with a microphone. But, I don't know how that would sound with a semi-hollow body. My friend Danny Kalb (Blues Project) just got a Martin Ellipse. That's a small-bodied martin acoustic with a Fishman Ellipse system. It's a very convincing acoustic sound through a guitar amp. However this particular system is only for round-hole acoustics. See: http://www.fishman.com/products/details.asp?id=86
Here's a Highlander system. It might be a challenge to rig such a thing in a semi-hollow body, but perhaps it cold be done: http://www.highlanderpickups.com/install/images/hmic/hmic.pdf
So now we're probably talking about brewing up a blend of bridge piezzo & internal mike / with the mag pickups. It would sure help if the guitar had a cavity cover in the back.
The thing to always keep in mind with any acoustic pickup is that they are a compromise. Or, more precisely, a series of compromises. On one end of the spectrum you have the best possible tone (a really great microphone), and at the other end you have volume before feedback, mobility of the player, and separation from the rest of the band.
Microphones have a lot of problems. First of all, if you don't stand absolutely still, the sound is all over the place. Next, they pickup everything on the stage; and if you have a loud drummer or a bass player with an SVT, that is what you are going to hear through the mic. I've had shows where the guitar player insisted on a mic, and when it was soloed the drums were louder than the guitar. Finally, the internal mics (such as on the various Fishman Blender systems) don't sound much like a guitar, much less your particular guitar. Maybe it's just me, but I've never much liked the sound of my guitar when my head was sticking through the soundhole, and that is what internal mics sound like to me. For all of these reasons, I consider microphones to be a very bad idea for most acoustic musicians. They work fine for guys doing small duo or solo gigs, or really any gig with no electric instruments (or banjos...;)), but beyond that they become more problematic.
Piezos are, for most guys, the best set of compromises. Most of the pickups we install are for guys who are playing in the church band on Sundays, and in that situation a good undersaddle (i.e., the Fishman Matrix Infinity, the D-TAR Timbreline, the basic Highlander pickup, or any of a number of other pickups) is ideal. Because of the pressure of the strings on the pickup, you get more output from the pickup, which gives you more volume before feedback, better isolation (piezos turn the whole top of the guitar into a microphone - on Ani DiFranco's live album "Living in Clip," the audience applause at the end of each track was captured completely through the pickup on her guitar), they are easy to setup (which is important with most of the sound guys you have working in churches, who are usually not that experienced), and the sound is actually an advantage in a busy band situation, because it doesn't take up too much space which helps to cut through the mix.
The archtop pickup I linked to is a passive system, which is very problematic in my mind. Piezo pickups put out a ridiculously high impedance signal (tens of millions of ohms, or there abouts), which means they are way more prone to damage from stray RF and cable capacitance. They also sound really thin and weak if they are not buffered asap. That is why active systems are so popular, but they are a real pain in the ass on F-hole instruments. If you are putting a pickup on a archtop guitar, a mandolin, or a resonator, what we usually suggest is a passive pickup with a good quality standalone preamp or DI box with as short of a cable as you can bear - certainly no longer than ten feet. The most popular preamp/DI on the market right now (in my experience) is the L.R. Baggs Paracoustic DI, and I recommend it frequently as it is a very good sounding unit. (I'd suggest the one I'm designing, but I haven't actually made any yet so that would seem premature at this point :)). The Control X would be a fine alternative to the Para DI, though it has less control of the signal.