I think during lecture, prof. mentioned that by upsampling we have doubled our "resolution" in the mesh. I was a little confused by this term because if we are splitting every triangle into four smaller triangles does that count as doubling or quadrupling our resolution? What does the term resolution refer to when we are talking about meshes?
archshift
Another question: clearly no new information here is being created by the subdivision process, so the complexity (and entropy) of the fine-grained mesh is identical to that of the un-subdivided mesh.
Can we see this perhaps as a process of compression, where we start with the more fine mesh and apply the inverse subdivision function for a more easily stored version?
isaaclee06
I think this slide/concept is very cool because it shows how loop subdivision for 3D models is similar to increasing the pixel sample size of 2D renderings. In both cases the image becomes "smoother" around the edges to essentially show more realistic details
I think during lecture, prof. mentioned that by upsampling we have doubled our "resolution" in the mesh. I was a little confused by this term because if we are splitting every triangle into four smaller triangles does that count as doubling or quadrupling our resolution? What does the term resolution refer to when we are talking about meshes?
Another question: clearly no new information here is being created by the subdivision process, so the complexity (and entropy) of the fine-grained mesh is identical to that of the un-subdivided mesh.
Can we see this perhaps as a process of compression, where we start with the more fine mesh and apply the inverse subdivision function for a more easily stored version?
I think this slide/concept is very cool because it shows how loop subdivision for 3D models is similar to increasing the pixel sample size of 2D renderings. In both cases the image becomes "smoother" around the edges to essentially show more realistic details