The circuit I did is intended to boost 1 X Li-Ion (= 3 X NiMH's in voltage) to drive an LV at decent current. It's efficiency is probably about the same as BB. I don't have a BB to compare, but a BB won't drive a LV at decent current (>rated) from less than about 4V from information I have gained, this circuit will. I'm guessing on BB performance here.
The circuit I have is not suitable for 2 X C NiMH. A 3 C NiMH is OK.
An LIII can be driven from 2 X NiMH from a BadBoy quite nicely, but don't expect 1A, perhaps less.
Of your choices,
Others will differ. It's a case of different strokes for different folks:
2C NIMH x Lux III (Try to get a T-bin) >> Average performance (MadMax).
2C NIMH x Lux V (Try to get a T-bin) >> VERY average performance, no circuit recommendations.
3C NIMH x Lux III (Try to get a T-bin) >> Now you're talking. Bright spot, the best throw. Up to 2 A possible (circuit I have), or BB/MM for average current/brightness.
3C NIMH x Lux V (Try to get X-bin) >> Again, now you're talking. Perhaps over 1 A (circuit I have), or BB for 700mA odd down to battery voltage of about 4V, dropping from there.
Given you're probably talking a Mag, I'd personally go the 3C NiMH x Lux V boosted (regulated) to anything over 1A from NiMH. This of course will debated depending on size/weight/performance etc. It would be be my personal choice though from those suggested for a distance throwing flashlight. A PM6 mod for closer illumination.
Note, you can DD an LV from 3 X Li, in fact a 2C, but then you'd *only* have an a Space Needle. It's OK as it's simple to do, cheap, but non-recargeable. I use my flashlight quite a bit so a 123 config is a no-go, otherwise I'd have gone a Mr Bulk SNII config. A damn nice light in it's time....until electronics finally caught up with it.
Note, for an LIII solution at better efficiency, try Jarheads solution in his efficiency post. I don't have samples of similar chips to tinker with just yet.
Chris