If it's illuminating a scene, then the brighter the ambient light is, the dimmer you'd want to the LED to be (i.e you wouldn't want to run it in daylight).
If it's an indicator light, then the brighter the ambient light is, the brighter the LED should be, in order to be visible.
A large Cd- S light dependent resistor[ LDR ] or PSR can have a light resistance of around 15Ω. My 1/2 in measured 30Ω. Connect a low value pot in series with LDR and connect to LM 317 as constant current regulator. Adjust pot for proper LED current. Connect a high val. pot[ 1k- 10k] in parallel with LDR to adjust dark brightness.
So basically what this drawing I did shows?
I showed this design on a UK electronic site a few days ago, and was told it was not workable, basically because photocells had too high a resistance...
Any suggestions as to a good choice of PSR with low resistance like you seem to have?
Will the PSR have any issue with too much current going thru it?
1 A might a bit much for CDS cell. Found a 1 in. cell but listed at less than 1k- light.or an assortment G14025, 20 pcs at The Electronic Goldmine. Back to drawing board.
Here's an idea, though you'd need to do some experiments to get the right resistor values. I'm pretty sure the brightness would vary in the right direction as the ambient light changes! You might need one more resistor in series with the photocell, if its resistance is low.
You could make it adjustable by making R2 and R3 variable resistors instead of fixed.
As for a threshold, I think that can't be done with a voltage regulator. You'd need to use an op amp or comparator, which would be another very simple design.
Oh well, if that doesn't appeal, you might try boosting the output with a transistor. It seems viable, but I'm not sure what the resistance in parallel with the transistor base ought to be. Making that resistor variable would essentially give you a setpoint for the LED boost.