IMO, the biggest difference between the "photo" look and the "drawing" look comes down to the tonemapping. Just clamping the colors down to 8 bit will clip something, and that gives an effect you see all the time in photos, but very rarely in drawings. People tend to draw things as they would see them rather than as they would be in a photograph, so as a result people almost never draw things like lights clipping to white. (also, a lot of the time the FOV in drawings is very different from a normal camera lens (ex, 25-35mm or so), it's either very wide or very close-up.)
Lux's built in tonemapping algorithms are all global algorithms, meaning they work on the whole image at once. You can get a very different look using a local tonemapper in an app like Luminance/Photoshop/Photomatix/etc. Bright objects will get brought down "on level" with the rest of the scene and it helps keep lights looking well defined. Local tonemapping algorithms can have a habit of overblowing colors or sharpening edges too, which tends to give more of the "hand drawn photo" look. Really, all of this is the same stuff that powers the "painterly HDR photo" thing loved by many and despised by others.

I wrote little tutorial in the wiki of how to use Luminance to do tonemapping to get a more dramatic, but somewhat less photoreal look:
http://www.luxrender.net/wiki/HDR_with_LuminanceWith a lot of my renders, I've taken to putting copies with several different tonemapping algorithms layered together in Photoshop, and blend and mask them to get the look I want in different parts of the image. (for example, using a linear tonemapped base but using a local tonemapped image where the lights are, so they show up cleanly rather than overblown)