Also, I'm not squeemish about getting into some more technical/mathematical explanations - I just want a firm grasp of the underworkings so I know how to better optimize my models/workflow in the future, so don't be afraid to go into detail.
I'm using xNormal right now, and my questions are below the images:
1. In Maya, you can set the normals to Smooth or Harden. What surprised me was that whatever the Lo-Poly model's normals are set to (smooth or harden) it will radically change the resulting normal map bake. What is going on here? I thought the normal map REPLACED this normals information. What is the relationship between the edge/face normals and the normals read from the Normal Map?
2. For normal map baking, do you generally get better results with a non-distorted UV layout with multiple unconnected shells/pieces, one continuous shell with more distortion, or does it not really make a difference? I was surprised to see that some of the best areas of the bake were in an area where two UV shells weren't connected to each other. Is it generally better to unwrap the UVs into many unconnected pieces?
3. In this bake, on the large rectangular polys (and also on some of the smaller square ones in the base), i'm getting these bizarre 'X' patterns (you can see it in the image above) that doesn't exist in the High-Poly model. Does anyone have an idea what's causing this?
4. in xNormal, what is the difference between using a "cage" and not using one?
5. In general, is it better for the Low-Poly mesh to be completely within the high-poly, completely outside the high-poly, or cut through it as close to the average position of the high-poly as possible?