April 21st, 1999, 10:45 AM
In computing a vertex normal for use in OpenGL, I have been using a simple (unweighted) average of the face normals of the faces that include the vertex. (Jean-Edouard Lachand-Robert's article ("Computing normals to achieve flat and smooth shading in OpenGL") in the OpenGL section of Codeguru provides a very good introduction to this topic). Does anyone have any thoughts as to whether weighting the face normals would bring a measurable quality improvement and how face normal weighting is typically done (I recall hearing that weighting is done based on the angles of the faces at the vertex)? I suspect face normal weighting is overkill for my simple "for fun" project, but I am interested in hearing how its done. Thanks in advance.