Read and unpack the next bytes from a binary file. :param fid: :param num_bytes: Sum of combination of {2, 4, 8}, e.g. 2, 6, 16, 30, etc.

Evaluate spherical harmonics at unit directions using hardcoded SH polynomials. Works with torch/np/jnp. ... Can be 0 or more batch d

Render the scene. Background tensor (bg_color) must be on GPU!

Transform points from world space to view space. Args: points (torch.Tensor): Should have shape (n_cameras, N, 3). cameras (List[

Normalize a vector.

r"""Convert discrete signed distance fields encoded on tetrahedral grids to triangle meshes using marching tetrahedra algorithm as described in `

Integrate a new observation into the TSDF. Args: img (torch.Tensor): Image. Has shape (H, W, 3) or (3, H, W).

_summary_ Args: points (torch.Tensor): Tensor with shape (N, 3) cameras (List[Camera]): _description_

Transforms points X by the affine transform (L, T). A matrix M with shape (..., d+1, d+1) can also be provided, in which case L and T are ext

Orients and centers the poses. We provide two methods for orientation: pca and up. pca: Orient the poses so that the principal component of the p

Extract pivots from Gaussians, in a differentiable manner. Each Gaussian will spawn 9 pivots. A list of indices can be provided to generate pi

Render the scene. Background tensor (bg_color) must be on GPU!

Get the tetra points of the Gaussian model. Args: downsample_ratio (float, optional): The ratio to downsample the tetra

Computes the depth order regularization loss. Args: iteration: Current training iteration. rendered_depth: Tensor containing

Compute initial SDF values with binary search and linearization. Args: voronoi_points (torch.Tensor): The voronoi points. (N_vor

Computes the mesh regularization loss and updates the mesh state. Args: iteration: Current training iteration. render_pkg: D

Evaluate vertex colors using Depth Fusion. Vertices that are not visible won't be assigned a color. Args: views (List[Camera]): List

Evaluate vertex colors for all vertices of the mesh, including the ones that are not visible. Useful for avoiding artifacts when rendering the mes

Evaluate the TSDF values at the given points using Depth Fusion. Args: points (torch.Tensor): Points at which to compute the TSDF. Has sh

Finds affine transform (L, T) such that Y = X @ L + T Args: X (torch.Tensor): Has shape (..., N, d) Y (torch.Tensor): Has shape (

Return the k nearest neighbor indices of points in points2. If points2 is None, return the k nearest neighbor indices of points in points.

Returns the parameters of Gaussians. Args: model (GaussianModel): The GaussianModel object. Returns: dict: A dictionary cont

Initializes depth priors using DepthAnythingV2. Args: scene: The scene object containing training cameras. config: Configura

Initializes components required for mesh regularization. Args: scene: The scene object containing training cameras. config:

r"""Function that measures Learned Perceptual Image Patch Similarity (LPIPS). Arguments: x, y (torch.Tensor): the input tensors to co