remove visibility when rendering if not a target. also colors for train vs. eval images

nerfstudio/data/dataparsers/nerfstudio_dataparser.py

@@ -61,7 +61,7 @@ class NerfstudioDataParserConfig(DataParserConfig):
     """The method to use to center the poses."""
     auto_scale_poses: bool = True
     """Whether to automatically scale the poses to fit in +/- 1 bounding box."""
-    eval_mode: Literal["train-split-fraction", "eval-frame-index"] = "train-split-fraction"
+    eval_mode: Literal["train-split-fraction", "eval-frame-index", "eval-interval"] = "train-split-fraction"
     """The method to use for splitting the dataset into train and eval."""
     train_split_fraction: float = 0.9
     """The fraction of images to use for training. The remaining images are for eval."""
@@ -69,6 +69,8 @@ class NerfstudioDataParserConfig(DataParserConfig):
     """The index of the frames to use for train."""
     eval_frame_indices: Tuple[int, ...] = (1,)
     """The index of the frames to use for eval."""
+    eval_interval: int = 8
+    """The interval between frames to use for eval."""
     depth_unit_scale_factor: float = 1e-3
     """Scales the depth values to meters. Default value is 0.001 for a millimeter to meter conversion."""
 
@@ -213,6 +215,8 @@ class Nerfstudio(DataParser):
                 indices = [i for i, path in enumerate(image_filenames) if path in split_filenames]
                 CONSOLE.log(f"[yellow] Dataset is overriding {split}_indices to {indices}")
                 indices = np.array(indices, dtype=np.int32)
+                i_train = indices
+                i_eval = indices
             elif has_split_files_spec:
                 raise RuntimeError(f"The dataset's list of filenames for split {split} is missing.")
             else:
@@ -226,12 +230,6 @@ class Nerfstudio(DataParser):
                 )  # equally spaced training images starting and ending at 0 and num_images-1
                 i_eval = np.setdiff1d(i_all, i_train)  # eval images are the remaining images
                 assert len(i_eval) == num_eval_images
-                if split == "train":
-                    indices = i_train
-                elif split in ["val", "test"]:
-                    indices = i_eval
-                else:
-                    raise ValueError(f"Unknown dataparser split {split}")
         elif self.config.eval_mode == "eval-frame-index":
             # keep around some metadata
             eval_frame_index_0_metadata = []
@@ -256,12 +254,24 @@ class Nerfstudio(DataParser):
                         eval_frame_index_0_metadata.append(0)
                 if frame_index in self.config.eval_frame_indices:
                     i_eval.append(idx)
-            if split == "train":
-                indices = i_train
-            elif split in ["val", "test"]:
-                indices = i_eval
-            else:
-                raise ValueError(f"Unknown dataparser split {split}")
+        elif self.config.eval_mode == "eval-interval":
+            # filter image_filenames and poses based on a specified interval for eval
+            # this chunk of code is very similar to the mipnerf360 code
+            num_images = len(image_filenames)
+            all_indices = np.arange(num_images)
+            train_indices = all_indices[all_indices % self.config.eval_interval != 0]
+            eval_indices = all_indices[all_indices % self.config.eval_interval == 0]
+            i_train = train_indices
+            i_eval = eval_indices
+        else:
+            raise ValueError(f"Unknown eval mode {self.config.eval_mode}")
+
+        if split == "train":
+            indices = i_train
+        elif split in ["val", "test"]:
+            indices = i_eval
+        else:
+            raise ValueError(f"Unknown dataparser split {split}")
 
         if "orientation_override" in meta:
             orientation_method = meta["orientation_override"]

nerfstudio/viewer/app/src/modules/Scene/Scene.jsx

@@ -271,7 +271,7 @@ export function get_scene_tree() {
         if (!prev.has(key)) {
           // keys_valid.push(key);
           const json = current[key];
-          const camera = drawCamera(json, key);
+          const camera = drawCamera(json, key, 0x000000, 0xffffff);
           sceneTree.set_object_from_path([CAMERAS_NAME, key], camera);
         }
       }
@@ -308,7 +308,7 @@ export function get_scene_tree() {
         if (!prev.has(key)) {
           // keys_valid.push(key);
           const json = current[key];
-          const camera = drawCamera(json, key);
+          const camera = drawCamera(json, key, 0xff0000, 0xff5555);
           sceneTree.set_object_from_path([EVAL_CAMERAS_NAME, key], camera);
         }
       }

nerfstudio/viewer/app/src/modules/Scene/drawing.js

@@ -50,6 +50,7 @@ export function getCameraWireframe(
   focalLength = 4,
   w = 1.5,
   h = 2,
+  color = 0x000000,
 ) {
   // Returns a wireframe of a 3D line-plot of a camera symbol.
   // A wireframe is a frustum.
@@ -93,17 +94,18 @@ export function getCameraWireframe(
 
   const geometry = new THREE.BufferGeometry().setFromPoints(points);
   const material = new THREE.LineBasicMaterial({
-    color: 0x000000,
+    color: color,
     linewidth: 1,
   });
   const lines = new THREE.LineSegments(geometry, material);
   return lines;
 }
 
-export function drawCameraImagePlane(width, height, imageString, name) {
+export function drawCameraImagePlane(width, height, imageString, name, color = 0xffffff) {
   // imageString is the texture as a base64 string
   const geometry = new THREE.PlaneGeometry(width, height);
   const material = new THREE.MeshBasicMaterial({
+    color: color,
     side: THREE.DoubleSide,
   });
   const texture = new THREE.TextureLoader().load(imageString);
@@ -117,7 +119,7 @@ function transpose(matrix) {
   return matrix[0].map((col, i) => matrix.map((row) => row[i]));
 }
 
-export function drawCamera(camera, name): THREE.Object3D {
+export function drawCamera(camera, name, wireframe_color = 0x000000, image_plane_color = 0xffffff): THREE.Object3D {
   const group = new THREE.Group();
 
   console.assert(
@@ -133,6 +135,7 @@ export function drawCamera(camera, name): THREE.Object3D {
     displayedFocalLength,
     width,
     height,
+    wireframe_color,
   );
   cameraWireframeObject.translateZ(displayedFocalLength); // move the wireframe frustum back
   group.add(cameraWireframeObject);
@@ -141,6 +144,7 @@ export function drawCamera(camera, name): THREE.Object3D {
     height * 2,
     camera.image,
     name,
+    image_plane_color
   );
   group.add(cameraImagePlaneObject);
 
@@ -156,11 +160,3 @@ export function drawCamera(camera, name): THREE.Object3D {
 
   return group;
 }
-
-export function drawCameras(cameras): Record<number, THREE.Object3D> {
-  const cameraObjects = {};
-  for (const [key, camera] of Object.entries(cameras)) {
-    cameraObjects[key] = drawCamera(camera);
-  }
-  return cameraObjects;
-}

scripts/render.py

@@ -376,6 +376,10 @@ class RenderTrajectory:
         else:
             assert_never(self.traj)
 
+        # check if visibility is a desired output. if not, disable it for fast rendering
+        if "visibility" not in self.rendered_output_names:
+            pipeline.model.visibility_field = None
+
         _render_trajectory_video(
             pipeline,
             camera_path,