update docs with new binding

nerfstudio-project · Oct 6, 2023 · c32e869 · c32e869
1 parent 0eecd29
commit c32e869
Show file tree

Hide file tree

Showing 5 changed files with 9 additions and 4 deletions.
diff --git a/diff_rast/__init__.py b/diff_rast/__init__.py
@@ -6,6 +6,7 @@
 from .get_tile_bin_edges import GetTileBinEdges
 from .map_gaussian_to_intersects import MapGaussiansToIntersects
 from .sh import SphericalHarmonics
+from .rasterize_forward_kernel import RasterizeForwardKernel
 from .version import __version__
 
 __all__ = [
@@ -18,4 +19,5 @@
     "GetTileBinEdges",
     "MapGaussiansToIntersects",
     "SphericalHarmonics",
+    "RasterizeForwardKernel",
 ]
diff --git a/diff_rast/rasterize_forward_kernel.py b/diff_rast/rasterize_forward_kernel.py
@@ -1,4 +1,4 @@
-"""Python bindings for mapping gaussians to interset IDs"""
+"""Python bindings for forward rasterization"""
 
 from typing import Tuple, Any, Optional
 
@@ -11,7 +11,7 @@
 
 
 class RasterizeForwardKernel(Function):
-    """Kernel function for rasterizing each tile.
+    """Kernel function for rasterizing and alpha-composing each tile.
 
     Args:
         tile_bounds (Tuple): tile dimensions as a len 3 tuple (tiles.x , tiles.y, 1).

diff --git a/docs/source/apis/utils.rst b/docs/source/apis/utils.rst
@@ -15,4 +15,6 @@ In addition to the main projection and rasterization functions, a few CUDA kerne
 
 .. autoclass:: MapGaussiansToIntersects
 
-.. autoclass:: ComputeCumulativeIntersects
+.. autoclass:: ComputeCumulativeIntersects
+
+.. autoclass:: RasterizeForwardKernel
diff --git a/docs/source/tests/tests.rst b/docs/source/tests/tests.rst
@@ -18,5 +18,5 @@ The tests include:
     ./test_get_tile_bin_edges.py
     ./test_map_gaussians.py
     ./test_project_gaussians
+    ./test_rasterize_forward_kernel.py
     ./test_sh.py
-
diff --git a/tests/test_rasterize_forward_kernel.py b/tests/test_rasterize_forward_kernel.py
@@ -13,6 +13,7 @@ def test_rasterize_forward_kernel():
     torch.manual_seed(42)
 
     num_points = 100
+
     means3d = torch.randn((num_points, 3), device=device, requires_grad=True)
     scales = torch.randn((num_points, 3), device=device)
     colors = torch.randn((num_points, 3), device=device)