When to (not) use normal maps

Question

Learning to model and I've got a question about what I've surmised about normals mapping so far (correct me if I make a mistake):

We create a normal map from a high poly model, in order to export the detailing into a texture to be used on a low poly model. This normal map contains information for the lighting of bumps, dents, etc. so that these do not need to actually be rendered and use no geometry, making your model take up less resources which is important in games.

So my question is: say I am modelling for a still image render and would like to experiment with the positions of lights or I am modelling to create a 3d-printed miniature, would it then be wise to skip the normals entirely and have the details sit as geometry in the model?

From what I understand the low poly model is still needed to UV-unwrap for texturing, as high poly counts seem to be impossible to unwrap, but in the mentioned examples would there be any other changes to a general workflow of base mesh->retopology into lowpoly->unwrap->high poly detailed model->bake normals->paint ?

Answer 1

Not only do normal maps increase performance in games, but they also increase the speed and performance of renders. Normal maps are only supposed to be used for small details such as small dents, scratches, marks, etc. For small details, you should probably use a normal map.

However, normal maps do not modify the actual geometry of the object. They only change the direction light hits the mesh in certain areas, known as a normal (hence the name). To modify the geometry of a mesh with a texture, use a displacement texture.

So, for 3D printing, do not use a normal map as they only affect the shading of an object and not its geometry (3D printing only uses geometry for slicing the model). Use a displacement texture with the Displacement Modifier instead and apply it.