Which is faster, modeling or normal map?

Say I want to add some indentations or something to a model, would using a Normal Map to it be faster, or would modeling it be faster?

:smiley: what a question. Normalmapping has been invented to reduce the poly count and to increase the performance.

Depends on the depth of the dents. If they’re pretty substantial, then you’d have to model it, but normal mapping helps with light stuff. There are other types of maps to help with slightly deeper stuff (self-occlusion or some such), but their aren’t many shaders that support that. BTW Check out Crazy Bump (google it), and also there’s a good video tutorial about how to use Blender to create good normal maps for low-poly high-quality models.