Open robotNo201 opened 4 months ago
I would also like this @kerrj
Since rendering all multiscale views of even one frame cannot fit in memory, I suppose rendering the dense point cloud just refers to rendering all views and testing? Is there any point cloud sparsification? @kerrj
I wonder if we can extract the highlighted point cloud in some format, such as 3d relevancy map, transformation matrix, or 3d clip embedding? Appreciate your help