You can point map out the light, multiply by the albedo, divide by the radiative emissivity of the surface, correcting for temperature variations, and then disentangle the surface roughness, ending with ray tracing the turbo encabulator.
But most people find the math and coordinate transformations in the frequency domain to be too difficult to do in their head, and inputting everything into a slide rule is really slow.
It's probably fastest to use the guess-and-adjust method the other person said.