I was wondering if there was any downside to having intersecting geometry or whether it's best to have only floating when creating high poly models for normal mapping.
Intersections can create aliasing if the baking tool doesn't multisample when it renders the projection to create the normal map. But usually intersections are the worst on the lowpoly model, because they complicate the projection.
So it's best to create the least amount of separate geometry for a projection?
In the lowpoly more-so than the high. I try to merge as much as possible in the low, and only have separate mesh chunks for areas that need to animate or complex overlaps(in which case you need to explode-bake them).
In the high, generally the more meshes the easier it is to work with, obviously if you're modeling something and in real life the object is made of multiple pieces, there is no reason to try to model it all as one solid chunk. Even with some more complex stuff, it can be a lot easier to model desperately with individual objects, and with anti-aliasing it shouldn't be an issue, or creative geometry along the seams.
Replies
and even in the lowpoly you can get by complications with projection by simpley explodeing the mesh, and bakeing that way.
In the lowpoly more-so than the high. I try to merge as much as possible in the low, and only have separate mesh chunks for areas that need to animate or complex overlaps(in which case you need to explode-bake them).
In the high, generally the more meshes the easier it is to work with, obviously if you're modeling something and in real life the object is made of multiple pieces, there is no reason to try to model it all as one solid chunk. Even with some more complex stuff, it can be a lot easier to model desperately with individual objects, and with anti-aliasing it shouldn't be an issue, or creative geometry along the seams.
Heres an example of "floating" an intersection: