there has to be some connection to calculus here in these diagrams
I remember in class professor mentioned that even though the points are still on the surface, the normals point in crazy directions but how do you make normals any different if you don't move the points off the surface a little bit? Might be a dumb question but can someone explain what's going on here?
I find the way we connect vertex matters. Take this for example, if we connect two vertexes in same vertical line, and pick another adjacent vertex, the normal will looks good.
It seems that if we use the normal of the vertices to interpolate the normal of every point on the surface, instead of using the position of the triangle vertices to calculate the normal, we would get the correct results. There must be different ways to approximate and calculate the normal of each vertex based on only the position of the neighbor vertices. The easiest way might be to use the weighted average the the calculated normal of neighboring faces. Are there "better" ways to do this?