Take a look at the HW schematic inside…
From PV voltage (120-450V) there is a boost converter up to the HV DC bus voltage (about 450V).
And from that HV DC bus, there is an isolated high frequency DC/DC converter. 450V/48V.
The “isolated” part was a big marketing thing at the time… 
They keep a big part of the Multi RS topology, where from that HV DC bus there is also generated the 230V AC, through a bidirectional DC/AC converter.
Ah interesting. Still requires higher vPV than vBat though - I think of a “boost” solar controller as one that can take lower vPV and boost it to output higher vBat, like the little Genasun controllers that have been around forever, can take ~18vPV and boost to 48vBat output.
Indeed, they require 120V and up and also that “no more than 8 times float” because of the winding ratio of the high frequency transformer and also the PWM limits for it to function withing optimum parameters.
And I also suspect that some of these limitations are imposed by the fact that instead of a full H-bridge topology on the HV side (as all manufacturers use), they (Victron) used a strange half bridge topology with an additional transformer for balancing the sides of the bridge.
Makes sense guess that’s where the efficiency losses arise (96% for rs450s versus 98-99% for the 150 and 250V units).
Indeed, a simple step down (buck) converter is always more efficient than a transformer based converter.
On the transformer based one you also have the loses in the transformer core.