> the 28V or so have to be dropped in the current limiting resistor...
You have 30V. You need 1.7V.
You CAN just waste 94% of your power. It's bad for the Earth, and your power bill; but as you say, it's not like a BIG load.
You can ignore that available 30V supply, add a 120V:6V power transformer to make something closer to what you want. Ah, but do you want idiot Americans tapping their 120V wall-wires? And what for the European and other customers: stock a second 240V:6V transformer? Try to show how to tangle four oddly colored wires to work on 120V or 240V?
You can take that 30VDC, chop it through a teeny transformer, rectify to a low DC voltage efficiently. Complicated, noisy, over $5, and either a tight design or it will have standby losses of a large part of a Watt.
My pencil leans to just taking the hit.
Yes, 10mA per LED may be more than you need. Get efficient LEDs, with just the beam-width you need (is operator on-axis or off to the side?), mock it up with a 9V battery and 1K and 10K resistors while sitting in various typical studio lighting, next to other LED-ed boxes.
For showing the bad idea, assume 10mA per LED.
You say you need four.
Parallel is simple: 40mA maximum demand. Potentially 0mA-40mA or 10mA-40mA change of load.
Series is trickier, but you can run four (or ten) 10mA LEDs at a "constant" 10mA load. A quarter of the parallel scheme, much less heat. If you can stand the complex-ier switching, this is very clever.
Use a resistor whenever possible. They are cheap, reliable, and have fewer legs to solder/trim than most alternatives.
Neglecting LEDs, 10mA at 30V is 3K resistor. With 1 to 4 LEDs under, the string current varies from 9.4mA to 7.7mA, 20% sag, less than 2dB pressure. I'm not sure what the eye will sense, but I think 20% is subtle. Also notice that from 1 to 2 LEDs is not a 6% drop, it is an 88% increase of light! "Ideally" it could be 100% increase; but 88% more total light will mask the 6% drop on the first light.
You could even argue that four lights should NOT be four times brighter: if the first is bright-enough, four is on the way to overwhelming.
Taking a generous ~~5mA goal for a good LED, four in series, the worst-case total dissipation is 0.150 Watts, all in the resistor (no lights). A 12-cent 1/2W resistor will never cause a warranty claim. The savings from this dumb "un-constant current regulator" can be put toward your added transistors to level-shift into the series string.