Simplify transforms in point rasterizer

Summary: Update the transform step in the pointcloud rasterizer to use the `update_padded` method on `Pointclouds`.  There was an inefficient step using `offset_points` which went via the packed represntation (and required unecessary additional memory). I think this was before the `update_padded` method was added to `Pointclouds`.

Reviewed By: gkioxari

Differential Revision: D22329166

fbshipit-source-id: 76db8a19654fb2f7807635d4f1c1729debdf3320

pytorch3d/renderer/points/rasterizer.py

@@ -86,22 +86,18 @@ class PointsRasterizer(nn.Module):
             raise ValueError(msg)
 
         pts_world = point_clouds.points_padded()
-        pts_world_packed = point_clouds.points_packed()
-        pts_screen = cameras.transform_points(pts_world, **kwargs)
-
         # NOTE: Retaining view space z coordinate for now.
         # TODO: Remove this line when the convention for the z coordinate in
         # the rasterizer is decided. i.e. retain z in view space or transform
         # to a different range.
-        view_transform = get_world_to_view_transform(R=cameras.R, T=cameras.T)
-        verts_view = view_transform.transform_points(pts_world)
-        pts_screen[..., 2] = verts_view[..., 2]
-
-        # Offset points of input pointcloud to reuse cached padded/packed calculations.
-        pad_to_packed_idx = point_clouds.padded_to_packed_idx()
-        pts_screen_packed = pts_screen.view(-1, 3)[pad_to_packed_idx, :]
-        pts_packed_offset = pts_screen_packed - pts_world_packed
-        point_clouds = point_clouds.offset(pts_packed_offset)
+        pts_view = cameras.get_world_to_view_transform(**kwargs).transform_points(
+            pts_world
+        )
+        pts_screen = cameras.get_projection_transform(**kwargs).transform_points(
+            pts_view
+        )
+        pts_screen[..., 2] = pts_view[..., 2]
+        point_clouds = point_clouds.update_padded(pts_screen)
         return point_clouds
 
     def forward(self, point_clouds, **kwargs) -> PointFragments: