Apply windowed scaled dot product cross attention to two sparse tensors. Args: q (SparseTensor): [N, *, H, C] sparse tensor containi

Draw text on an image of the given resolution. The text is automatically wrapped and scaled so that it fits completely within the image while

Draw text on an image of the given resolution. The text is automatically wrapped and scaled so that it fits completely within the image while

Convert a VarLenTensor to a dense representation without for-loop. Returns: dense (torch.Tensor): (N, L, C) dens

Convert a VarLenTensor to a list of tensors.

Convert a SparseTensor to list of tensors.

Compute training losses. Args: vertices (SparseTensor): vertices of each active voxel intersected (SparseTen

Compute training losses. Args: x (SparseTensor): Input sparse tensor for pbr materials. mesh (List[MeshWithP

Compute training losses. Args: ss: The [N x 1 x H x W x D] tensor of binary sparse structure. Returns:

Compute training losses for a single timestep. Args: x_0: The [N x ... x C] sparse tensor of the inputs. con

Compute training losses for a single timestep. Overridden to pass coords from x_0 to get_cond for sparse feature extraction.

Compute training losses for a single timestep. Args: x_0: The [N x C x ...] tensor of noiseless inputs. cond

Compute training losses for a single timestep. Overridden to avoid passing extra kwargs to the model. Args:

Triangulates all meshes in the scene. Returns: None

Unhides all objects in the scene. Returns: None

Unpatchify a tensor. Args: x (torch.Tensor): (N, C, *spatial) tensor patch_size (int): Patch size

Unwrap the model from distributed training.

Concatenate a list of varlen tensors. Args: inputs (List[VarLenTensor]): List of varlen tensors to concatenate.

Visualize the conditioning data.

Visualize the conditioning data.

Visualize the conditioning data.

Visualize the projected 3D grid points on the input image. This is a convenience method that delegates to ProjGrid.visualize

Convert a sample to images, including GT camera view if available. Args: sample: Either a SparseTensor or dict c

Convert a sample to images, including GT camera view if available. Args: sample: Either a tensor or dict contain

Visualize shape samples. Args: x_0: SparseTensor or dict containing 'x_0' camera_angle_x: Optional [

Visualize sparse structure samples. Args: x_0: Latent tensor [B, C, D, H, W] or dict containing 'x_0'

Write a mesh to a PLY file, with the option to save in ASCII or binary format, and optional per-vertex colors. Args: filenam

Write a mesh to a PLY file, with the option to save in ASCII or binary format. Args: filename (str): The filename to write to.

Write a mesh to a PLY file, with the option to save in ASCII or binary format, and optional per-vertex colors. Args: filenam