help of pred_camera parameter

[zhangzhenyu-zzy]

hello,i find in this work,we can predict camera parameter as [z,tx,ty],i want to take the differentiable rendering in trannning,i use the pytorch3D to replace the original renderer,but pytorch3D need the R and T parameter to set the camera,how can i use the [z,tx,ty] or [tx,ty,focal-length/img-size×z] to get the R,T parameter?