MrMark:
I imagine there are all sorts of practical reasons this wouldn't work all that well, but that's a clever insight. I fear you may have ruined my weekend."I learn a bit every time I visit the forum.", indeed.
I think that what to expect can be ballparked.
If you have linear detectors or use lookups to convert readings to linear....
Suppose the source brightness is 1.
If the front and rear detectors are 1 m apart and the source is X meters from the front detector then
Rear reading should be 1 / (X + 1)^2 where front reading should be 1 / X^2.
Suppose I have a circuit that subtracts the rear detector output from the front detector output?
And then if I could ADC that with the front reading for Vref, I'd have the ratio directly.
I bet there's already such a circuit.
Or read the rear detector with the front as Vref?
It would not be hard to lookup table calibrated reads for the ADC reads, and the fetch is 3 cycles per byte.
Calculating the various terms would take far longer.
But really really not to use LDR's. Photo-transistors are like 5 cents or less each in a bag of 50 to 100. Shop for linear response if you can get it (rather than log scale). They are perhaps 10000 times faster than LDR's, look like 5mm leds. If you would work with near-IR then get the black bulb IR detectors and visible light will not interfere.