Added support for 5D input images/latents

Preprocess the image input. Args: image (`PipelineImageInput`): The image input, accepted formats are PI

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

Convert tensor data type from uint8 to float, divide value by 255.0 and permute the dimensions of clip tensor Args: clip (torch.t

Apply the transformation to each frame in the video. Args: video_frames (torch.Tensor): A tensor representing a batch of vide

Args: clip (torch.tensor, dtype=torch.uint8): Size is (T, C, H, W) Return: clip (torch.tensor, dtype=torch.fl

data keys: `action`, `latent`, `depths`, `semantics`, `metainfo` Returns: `actions`: [batch_size, n_frame - 1

Args: image: [[[Image, Image], [Imgage, Image]], [[Image, Image], [Image, Image]], ...], n_batch -> n_view -> n_frame

Initialize the preprocessing class with the target size. Args: size (tuple): The target height and width as a tuple (height,

Input: gray: gray image, tensor/numpy, (H, W) Output: depth: (3, H, W), tensor

Input: semantic: semantic image, tensor/numpy, (N, H, W) Output: depth: (3, H, W), tensor

Convert depth map to 3D points using camera intrinsics. Args: depth_map (npt.NDArray): Depth map of shape (H, W) containing depth va

Convert points(N, >=3) to voxels. Args: points (torch.Tensor): [N, ndim]. Points[:, :3] contain xyz points an