It is really good job! But I want to know why the center_depth is 1.0 in inference. Does the model have generalization ability for larger-scale movement trajectories?
And have you tried circular trajectories on the horizontal plane in inference or training? Can you tell me how you handle the target camera trajectory during training？
It is really good job! But I want to know why the center_depth is 1.0 in inference. Does the model have generalization ability for larger-scale movement trajectories?
And have you tried circular trajectories on the horizontal plane in inference or training? Can you tell me how you handle the target camera trajectory during training？