Hi

I think I know what you are refering to. Check out this video.

I believe the technique you are refering to is when they used really high poly models in a lit environment to generate the textures which are then applied to low poly models, making them look as though they have alot more detail than they actually do.

I can not answer your question here about what you can do with lighting to make the model look better. That is a huge area of discussion of which there are many tutorials and research papers online.

You should not change the position of the light for each mesh unless you have reason too. I might create a strange effect, however if you are looking for a realistic look this would definately be the wrong thing to do.

Ciarán

I believe you are referring to normal mapping.

This technique extracts normal information from a high polygon model and stores it in a special texture (normal map). When mapped to a low polygon model, this normal map then makes lighting calculations look more detailed (it is necessary to compute lighting on per-pixel basis).