I hear a lot of people on the boards say that adding dirt and grain to your model will make it more photo-realistic. The concept makes perfect sense to me, but I was wondering if any of you have tips or tricks (or links to other threads) that might explain how to do this efficiently. Thanks.
The way you could obtain this can be approached in many different ways. The one you should probably start out with is UV mapping.
Basically you could apply this method, but use a dirt image to your model, and then I suggest applying the normal to it to make it come out a little:
“Dirt and grain” is probably not the right word for it, unless you are really modeling sandpaper.
I’d suggest that a better observation is that real-world surfaces are almost never “uniformly specular.”
Experiment with using more than one texture … the “texture nodes” facility makes this very easy now. Just a little bit of “noise” can make a lot of difference. But by “noise” I do not necessarily mean “dirty” or “bumpy.”
Take any object reasonably close to you and “look at the light.” Inch by inch, how does the light play across the object? How does it vary? You will notice several different effects in-play at one time. Identify those, implement them, quantify them and mix them . . .