reshow / PRNet-PyTorch

The training and evaluation code for PRNet (《Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network》)
73 stars 16 forks source link

I have the rendered image. How can I convert it to a 3D mesh? #10

Open HOMGH opened 4 years ago

HOMGH commented 4 years ago

Hi @reshow I had a question regarding generating a 3D mesh. For this, we need vertices (N3), textures(N3) and triangle indices. But in a case that the output of our model is just a shape (N3) and texture (224224*3) like https://github.com/tranluan/Nonlinear_Face_3DMM , how can we convert it to a 3D mesh? What triangle would be in this case? Appreciate your help.

reshow commented 3 years ago

Has this problem been solved? I think Nonlinear 3DMM may provide a mapping between N3 and 2242243.