Hello,
Thank you for your great work CameraCtrl. I’d like to ask: when computing the quantitative metrics RotErr and TransErr in "CameraCtrl", are the camera parameters R and T for both the generated video and the gt video obtained using COLMAP? Or are the R and T for the gt video taken directly from the annotations in RealEstate10K, and only the generated video's parameters are estimated using COLMAP?
Thank you!