I was just wondering if you still need to have mostly one smoothing group for the object if your normal maps are only coming from a method of generation like the nVidia filter.
I believe that with more than one smoothing group the verts will be split at those faces so if the normal maps display correctly, it would be worth using one smoothing group to keep your model more efficient.(?)
If it helps, this is the model (yet to be optimised) I wish to use the nVidia filter for to generate some normals from some wood reference I have. I wish to retain relatively hard edges and the textures are going to be tiling so do these affect how my smoothing groups should be set up?
Replies
You should split the smoothing groups according to the uv's really. Since split UV's are essentially split verts as well.
Anyway, using 1 smoothing group on this model without a baked normal map to correct the normals of the low resolution model will probably not look right.
Just smooth the model and see how it looks like. The nvidia generated normals will basically "overlay" with that look.
Also talking about efficiency. You could probably optimize that model allot. Unless you have a reason for all of those edgeloops.
Set up your smoothing so that it looks good, that's the only thing you really need to do when doing hand painted height - > normal map conversions. This means adding smoothing groups/hard edges to make flat things appear flat, or adding edge bevels/chamfers.
Using one smoothing group isn't a requirement for baking normals from a high res, and it isn't a requirement for using painted/converted normals, its actually a really, really bad idea, unless your mesh is a simple sphere or otherwise very round object that doesn't have extreme vertex normal changes.