You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We used poses for AFHQ from EG3D for a fair comparison.
See Sec. 4.2 in the appendix of EG3D for how to obtain poses.
One thing that I need to mention: EG3D uses perspective-n-point (PnP) to obtain poses. PnP requires 3D points. I asked the authors before. They said that they manually measured a cat to obtain a (very rough) set of 3D points.
Very nice work. How did you generate the camera poses for AFHQ dataset, can you give me a more detailed method?
The text was updated successfully, but these errors were encountered: