PLY load normals

Summary: Add ability to load normals when they are present in a PLY file.

Reviewed By: nikhilaravi

Differential Revision: D26458971

fbshipit-source-id: 658270b611f7624eab4f5f62ff438038e1d25723

pytorch3d/io/ply_io.py

@@ -780,9 +780,9 @@ def _load_ply_raw(f, path_manager: PathManager) -> Tuple[_PlyHeader, dict]:
 
 def _get_verts_column_indices(
     vertex_head: _PlyElementType,
-) -> Tuple[List[int], Optional[List[int]], float]:
+) -> Tuple[List[int], Optional[List[int]], float, Optional[List[int]]]:
     """
-    Get the columns of verts and verts_colors in the vertex
+    Get the columns of verts, verts_colors, and verts_normals in the vertex
     element of a parsed ply file, together with a color scale factor.
     When the colors are in byte format, they are scaled from 0..255 to [0,1].
     Otherwise they are not scaled.
@@ -793,11 +793,14 @@ def _get_verts_column_indices(
         property double x
         property double y
         property double z
+        property double nx
+        property double ny
+        property double nz
         property uchar red
         property uchar green
         property uchar blue
 
-    then the return value will be ([0,1,2], [6,7,8], 1.0/255)
+    then the return value will be ([0,1,2], [6,7,8], 1.0/255, [3,4,5])
 
     Args:
         vertex_head: as returned from load_ply_raw.
@@ -807,9 +810,12 @@ def _get_verts_column_indices(
         color_idxs: List[int] of 3 color columns if they are present,
                     otherwise None.
         color_scale: value to scale colors by.
+        normal_idxs: List[int] of 3 normals columns if they are present,
+                    otherwise None.
     """
     point_idxs: List[Optional[int]] = [None, None, None]
     color_idxs: List[Optional[int]] = [None, None, None]
+    normal_idxs: List[Optional[int]] = [None, None, None]
     for i, prop in enumerate(vertex_head.properties):
         if prop.list_size_type is not None:
             raise ValueError("Invalid vertices in file: did not expect list.")
@@ -819,6 +825,9 @@ def _get_verts_column_indices(
         for j, name in enumerate(["red", "green", "blue"]):
             if prop.name == name:
                 color_idxs[j] = i
+        for j, name in enumerate(["nx", "ny", "nz"]):
+            if prop.name == name:
+                normal_idxs[j] = i
     if None in point_idxs:
         raise ValueError("Invalid vertices in file.")
     color_scale = 1.0
@@ -831,21 +840,23 @@ def _get_verts_column_indices(
         point_idxs,
         None if None in color_idxs else cast(List[int], color_idxs),
         color_scale,
+        None if None in normal_idxs else cast(List[int], normal_idxs),
     )
 
 
 def _get_verts(
     header: _PlyHeader, elements: dict
-) -> Tuple[torch.Tensor, Optional[torch.Tensor]]:
+) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[torch.Tensor]]:
     """
-    Get the vertex locations and colors from a parsed ply file.
+    Get the vertex locations, colors and normals from a parsed ply file.
 
     Args:
         header, elements: as returned from load_ply_raw.
 
     Returns:
         verts: FloatTensor of shape (V, 3).
         vertex_colors: None or FloatTensor of shape (V, 3).
+        vertex_normals: None or FloatTensor of shape (V, 3).
     """
 
     vertex = elements.get("vertex", None)
@@ -854,14 +865,16 @@ def _get_verts(
     if not isinstance(vertex, list):
         raise ValueError("Invalid vertices in file.")
     vertex_head = next(head for head in header.elements if head.name == "vertex")
-    point_idxs, color_idxs, color_scale = _get_verts_column_indices(vertex_head)
+    point_idxs, color_idxs, color_scale, normal_idxs = _get_verts_column_indices(
+        vertex_head
+    )
 
     # Case of no vertices
     if vertex_head.count == 0:
         verts = torch.zeros((0, 3), dtype=torch.float32)
         if color_idxs is None:
-            return verts, None
-        return verts, torch.zeros((0, 3), dtype=torch.float32)
+            return verts, None, None
+        return verts, torch.zeros((0, 3), dtype=torch.float32), None
 
     # Simple case where the only data is the vertices themselves
     if (
@@ -870,9 +883,10 @@ def _get_verts(
         and vertex[0].ndim == 2
         and vertex[0].shape[1] == 3
     ):
-        return _make_tensor(vertex[0], cols=3, dtype=torch.float32), None
+        return _make_tensor(vertex[0], cols=3, dtype=torch.float32), None, None
 
     vertex_colors = None
+    vertex_normals = None
 
     if len(vertex) == 1:
         # This is the case where the whole vertex element has one type,
@@ -882,6 +896,10 @@ def _get_verts(
             vertex_colors = color_scale * torch.tensor(
                 vertex[0][:, color_idxs], dtype=torch.float32
             )
+        if normal_idxs is not None:
+            vertex_normals = torch.tensor(
+                vertex[0][:, normal_idxs], dtype=torch.float32
+            )
     else:
         # The vertex element is heterogeneous. It was read as several arrays,
         # part by part, where a part is a set of properties with the same type.
@@ -913,13 +931,22 @@ def _get_verts(
                 partnum, col = prop_to_partnum_col[color_idxs[color]]
                 vertex_colors.numpy()[:, color] = vertex[partnum][:, col]
             vertex_colors *= color_scale
+        if normal_idxs is not None:
+            vertex_normals = torch.empty(
+                size=(vertex_head.count, 3), dtype=torch.float32
+            )
+            for axis in range(3):
+                partnum, col = prop_to_partnum_col[normal_idxs[axis]]
+                vertex_normals.numpy()[:, axis] = vertex[partnum][:, col]
 
-    return verts, vertex_colors
+    return verts, vertex_colors, vertex_normals
 
 
 def _load_ply(
     f, *, path_manager: PathManager
-) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[torch.Tensor]]:
+) -> Tuple[
+    torch.Tensor, Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor]
+]:
     """
     Load the data from a .ply file.
 
@@ -935,10 +962,11 @@ def _load_ply(
         verts: FloatTensor of shape (V, 3).
         faces: None or LongTensor of vertex indices, shape (F, 3).
         vertex_colors: None or FloatTensor of shape (V, 3).
+        vertex_normals: None or FloatTensor of shape (V, 3).
     """
     header, elements = _load_ply_raw(f, path_manager=path_manager)
 
-    verts, vertex_colors = _get_verts(header, elements)
+    verts, vertex_colors, vertex_normals = _get_verts(header, elements)
 
     face = elements.get("face", None)
     if face is not None:
@@ -976,7 +1004,7 @@ def _load_ply(
     if faces is not None:
         _check_faces_indices(faces, max_index=verts.shape[0])
 
-    return verts, faces, vertex_colors
+    return verts, faces, vertex_colors, vertex_normals
 
 
 def load_ply(
@@ -1031,7 +1059,7 @@ def load_ply(
 
     if path_manager is None:
         path_manager = PathManager()
-    verts, faces, _ = _load_ply(f, path_manager=path_manager)
+    verts, faces, _, _ = _load_ply(f, path_manager=path_manager)
     if faces is None:
         faces = torch.zeros(0, 3, dtype=torch.int64)
 
@@ -1211,18 +1239,23 @@ def read(
         if not endswith(path, self.known_suffixes):
             return None
 
-        verts, faces, verts_colors = _load_ply(f=path, path_manager=path_manager)
+        verts, faces, verts_colors, verts_normals = _load_ply(
+            f=path, path_manager=path_manager
+        )
         if faces is None:
             faces = torch.zeros(0, 3, dtype=torch.int64)
 
-        textures = None
+        texture = None
         if include_textures and verts_colors is not None:
-            textures = TexturesVertex([verts_colors.to(device)])
+            texture = TexturesVertex([verts_colors.to(device)])
 
+        if verts_normals is not None:
+            verts_normals = [verts_normals]
         mesh = Meshes(
             verts=[verts.to(device)],
             faces=[faces.to(device)],
-            textures=textures,
+            textures=texture,
+            verts_normals=verts_normals,
         )
         return mesh
 
@@ -1286,12 +1319,14 @@ def read(
         if not endswith(path, self.known_suffixes):
             return None
 
-        verts, faces, features = _load_ply(f=path, path_manager=path_manager)
+        verts, faces, features, normals = _load_ply(f=path, path_manager=path_manager)
         verts = verts.to(device)
         if features is not None:
             features = [features.to(device)]
+        if normals is not None:
+            normals = [normals.to(device)]
 
-        pointcloud = Pointclouds(points=[verts], features=features)
+        pointcloud = Pointclouds(points=[verts], features=features, normals=normals)
         return pointcloud
 
     def save(

tests/test_io_ply.py

@@ -216,14 +216,18 @@ def test_save_load_meshes(self):
             [[0, 0, 0], [0, 0, 1], [0, 1, 0], [1, 0, 0]], dtype=torch.float32
         )
         faces = torch.tensor([[0, 1, 2], [0, 2, 3]])
+        normals = torch.tensor(
+            [[0, 1, 0], [1, 0, 0], [1, 4, 1], [1, 0, 0]], dtype=torch.float32
+        )
         vert_colors = torch.rand_like(verts)
         texture = TexturesVertex(verts_features=[vert_colors])
 
-        for do_textures in itertools.product([True, False]):
+        for do_textures, do_normals in itertools.product([True, False], [True, False]):
             mesh = Meshes(
                 verts=[verts],
                 faces=[faces],
                 textures=texture if do_textures else None,
+                verts_normals=[normals] if do_normals else None,
             )
             device = torch.device("cuda:0")
 
@@ -236,12 +240,57 @@ def test_save_load_meshes(self):
             mesh2 = mesh2.cpu()
             self.assertClose(mesh2.verts_padded(), mesh.verts_padded())
             self.assertClose(mesh2.faces_padded(), mesh.faces_padded())
+            if do_normals:
+                self.assertTrue(mesh.has_verts_normals())
+                self.assertTrue(mesh2.has_verts_normals())
+                self.assertClose(
+                    mesh2.verts_normals_padded(), mesh.verts_normals_padded()
+                )
+            else:
+                self.assertFalse(mesh.has_verts_normals())
+                self.assertFalse(mesh2.has_verts_normals())
+                self.assertFalse(torch.allclose(mesh2.verts_normals_padded(), normals))
             if do_textures:
                 self.assertIsInstance(mesh2.textures, TexturesVertex)
                 self.assertClose(mesh2.textures.verts_features_list()[0], vert_colors)
             else:
                 self.assertIsNone(mesh2.textures)
 
+    def test_save_load_with_normals(self):
+        points = torch.tensor(
+            [[0, 0, 0], [0, 0, 1], [0, 1, 0], [1, 0, 0]], dtype=torch.float32
+        )
+        normals = torch.tensor(
+            [[0, 1, 0], [1, 0, 0], [1, 4, 1], [1, 0, 0]], dtype=torch.float32
+        )
+        features = torch.rand_like(points)
+
+        for do_features, do_normals in itertools.product([True, False], [True, False]):
+            cloud = Pointclouds(
+                points=[points],
+                features=[features] if do_features else None,
+                normals=[normals] if do_normals else None,
+            )
+            device = torch.device("cuda:0")
+
+            io = IO()
+            with NamedTemporaryFile(mode="w", suffix=".ply") as f:
+                io.save_pointcloud(cloud.cuda(), f.name)
+                f.flush()
+                cloud2 = io.load_pointcloud(f.name, device=device)
+            self.assertEqual(cloud2.device, device)
+            cloud2 = cloud2.cpu()
+            self.assertClose(cloud2.points_padded(), cloud.points_padded())
+            if do_normals:
+                self.assertClose(cloud2.normals_padded(), cloud.normals_padded())
+            else:
+                self.assertIsNone(cloud.normals_padded())
+                self.assertIsNone(cloud2.normals_padded())
+            if do_features:
+                self.assertClose(cloud2.features_packed(), features)
+            else:
+                self.assertIsNone(cloud2.features_packed())
+
     def test_save_ply_invalid_shapes(self):
         # Invalid vertices shape
         with self.assertRaises(ValueError) as error: