Predict SDF and deformation for tetrahedron vertices :param planes: triplane feature map for the geometry

Predict Texture given triplanes :param planes: the triplane feature map :param tex_pos: Position we want to query the texture

Get the input camera parameters.

This method allows to interpolate the pre-trained position encodings, to be able to use the model on higher resolution images.

Load configuration from YAML file

Read image as grayscale and resize to img_size. Inputs impath: Path to input image. Returns grayim: uint8 num

Loads an image and returns it as a linearized float Mitsuba Tensor

Perform Differentiable Optimal Transport in Log-space for stability

Perform Sinkhorn Normalization in Log-space for stability

img: m * n * 8, [R, G, B, alpha, depth, normal_x, normal_y, normal_z] ref_img: m * n * 3, black background

Prepare mesh by generating normals and a UV mapping

Project 2D image points to 3D space using the depth map, camera intrinsics, and pose. :param image_points: Nx2 array of image points :pa

Does a projection of a 3D point onto a batch of 2D planes, returning 2D plane coordinates. Takes plane axes of shape n_planes, 3, 3

Removes keypoints too close to the border

Function to render a generated mesh with nvdiffrast :param mesh_v: List of vertices for the mesh :param mesh_f: List of faces