yfeng95 / PRNet

Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network (ECCV 2018)
http://openaccess.thecvf.com/content_ECCV_2018/papers/Yao_Feng_Joint_3D_Face_ECCV_2018_paper.pdf
MIT License
4.95k stars 947 forks source link

Compute transformation with no skew #73

Closed MaxBazik closed 5 years ago

MaxBazik commented 5 years ago

Hey, I noticed you allow for skew when computing your camera. This is not so good and introduces error when extracting the rotation matrix from your transform.

The code I give you is more exact and will convert better to a scaled orthographic camera representation.

[1] Multiple View Geometry in Computer Vision pg. 170-172 [2]http://nghiaho.com/?page_id=671

yfeng95 commented 5 years ago

Hi, Good job! I've found the problem very early, but haven't time to fix it. Thanks a lot for your contribution!