Apply the distortion loss to the nerf dataset

[qihangGH]

Hi @sunset1995 ! Thank you for your great work! May I ask if it is possible to add the distortion loss to the nerf dataset? Since the nerf dataset contains bounded scenes. Could we just let s be uniformly normalized in the Euclidean space? If so, is that correct to set

's': (step_id + 0.5) / N_steps[ray_id.long()],
'n_max': N_steps[ray_id.long()]

in the forward method of the class DVGO? Here, the N_steps is calculated by the sample_ray method at the begining of the forward method.