Hi! In today's lecture, I find it hard understanding what is the texture map (p47) and why the method we learned today can produce high-quality texture maps (p56)? Could you elaborate a little bit? Thanks!
Without going into technical details of the optimization procedure, the explanation is literal. It is basically "painting" the mesh in colors, which images provide.
Assume that the reconstruction of 3D mesh is already done, given a bunch of images (slide 46). Then, there is an established correspondence between mesh triangles and images pixels, as this is what 3D optimization does. Then, nothing prevents us from coloring all mesh triangles in corresponding pixel colors (just average all colors that correspond to the particular mesh triangle).
The better the 3D optimization is done, the finer the structure of the model, the better quality of texture maps can be produced. You can look at the slide 48. The rightmost is the texture mapped to a mesh, this is not a real photo.
This is very clear, thanks!
