L1 error and ssim error #24
Description
static_errors = viewspace_point_error_tensor.grad[:static_num]
static_l1_error = static_errors[static_update_filter, 1:2] / static_errors[static_update_filter, 0:1].clamp_min(1e-4)
self.xyz_ssim_error_accum[static_update_filter] += static_errors[static_update_filter, 2:3] / static_errors[static_update_filter, 0:1].clamp_min(1e-4)
Why are they calculated in this way？ How are they related to L1 and SSIM