Band III is absolutely full! There are a few gaps, but Band I has the "wide open spaces". The downside of Band I is that the aerials are huge!
TBH, I have always preferred Band IV and V - it's easy and cheap to get high gain aerials (albeit matched to 75Ω), but this permits the use of cheap TV coax, really high gain (and cheap) Yagis, and there's plenty of room in the TV bands if you look around. Also (and most important), a TV aerial doesn't ever look out of place on a rooftop!
The earliest Band IV links used hacked Pye Pocketfone PF1 pairs. We used to disable the transmit phase modulator, add a simple ~90MHz PLL as the FM source, and peak them up for a couple of hundred milliwatts on the output frequency. The receivers required a bit more work - the crystal oscillator and the multiplier chain was still used, along with the front end, but the IF, squelch and audio parts were chopped off the board (with a hacksaw!). A couple of ceramic filters. a transistor IF amplifier stage, and a CA3089 IF / discriminator IC, and you had the receiver. The transmitter was matched to 75Ω, as was the receiver input, and it was possible to get fully noise-quietening links over tens of miles! We used to buy the PF1s from "Garex" for about £8/pair!
Later on, I developed a really nifty PLL TX for Band IV and V that would deliver either 1 or 5 Watts into 75Ω. There was a matching RX that used a TV tunerhead and a nice PCB for the 35MHz to 10.7 MHz IF down-mixing and a fairly conventional 10.7 MHz IF and demodulator (though the last run of these included a PLL demodulator to improve the receiver's useful sensitivity by more than 10dB). These link set-ups were used by many stations in the 80s, and only one of them was ever tracked by the DTI! The distance record for one of the Band IV links was a little over 45 miles!
The very final Band IV / V link system used a receiver that mixed the incoming signal down to the required output frequency, so it never actually received the modulation, it just took ~500MHz in, and gave ~100MHz out, using a PLL controlled local oscillator. This made it possible to (slightly) move the output frequency of the main rig from the studio end! One set-up like this was used in the Hollywood Hills in the 80s, with great success. The output rig was powered by a couple of big "gel-cells", and these were charged by a mixture of solar cells and a little wind turbine. We were putting out around 140 Watts, and with the directional transmit aerial, we used to have an ERP in the region of 500 Watts - enough to get a good, solid signal into much of the greater Los Angeles area.
"Why is my rig humming?"
"Because it doesn't know the words!"