pytorch3d/renderer/cameras.py [997:1101]: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - _FIELDS = ( "K", "R", "T", "focal_length", "principal_point", "_in_ndc", # arg is in_ndc but attribute set as _in_ndc "image_size", ) _SHARED_FIELDS = ("_in_ndc",) def __init__( self, focal_length=1.0, principal_point=((0.0, 0.0),), R: torch.Tensor = _R, T: torch.Tensor = _T, K: Optional[torch.Tensor] = None, device: Device = "cpu", in_ndc: bool = True, image_size: Optional[Union[List, Tuple, torch.Tensor]] = None, ) -> None: """ Args: focal_length: Focal length of the camera in world units. A tensor of shape (N, 1) or (N, 2) for square and non-square pixels respectively. principal_point: xy coordinates of the center of the principal point of the camera in pixels. A tensor of shape (N, 2). in_ndc: True if camera parameters are specified in NDC. If camera parameters are in screen space, it must be set to False. R: Rotation matrix of shape (N, 3, 3) T: Translation matrix of shape (N, 3) K: (optional) A calibration matrix of shape (N, 4, 4) If provided, don't need focal_length, principal_point image_size: (height, width) of image size. A tensor of shape (N, 2) or a list/tuple. Required for screen cameras. device: torch.device or string """ # The initializer formats all inputs to torch tensors and broadcasts # all the inputs to have the same batch dimension where necessary. kwargs = {"image_size": image_size} if image_size is not None else {} super().__init__( device=device, focal_length=focal_length, principal_point=principal_point, R=R, T=T, K=K, _in_ndc=in_ndc, **kwargs, # pyre-ignore ) if image_size is not None: if (self.image_size < 1).any(): # pyre-ignore raise ValueError("Image_size provided has invalid values") else: self.image_size = None # When focal length is provided as one value, expand to # create (N, 2) shape tensor if self.focal_length.ndim == 1: # (N,) self.focal_length = self.focal_length[:, None] # (N, 1) self.focal_length = self.focal_length.expand(-1, 2) # (N, 2) def get_projection_transform(self, **kwargs) -> Transform3d: """ Calculate the projection matrix using the multi-view geometry convention. Args: **kwargs: parameters for the projection can be passed in as keyword arguments to override the default values set in __init__. Returns: A `Transform3d` object with a batch of `N` projection transforms. .. code-block:: python fx = focal_length[:, 0] fy = focal_length[:, 1] px = principal_point[:, 0] py = principal_point[:, 1] K = [ [fx, 0, px, 0], [0, fy, py, 0], [0, 0, 0, 1], [0, 0, 1, 0], ] """ K = kwargs.get("K", self.K) if K is not None: if K.shape != (self._N, 4, 4): msg = "Expected K to have shape of (%r, 4, 4)" raise ValueError(msg % (self._N)) else: K = _get_sfm_calibration_matrix( self._N, self.device, kwargs.get("focal_length", self.focal_length), kwargs.get("principal_point", self.principal_point), - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - pytorch3d/renderer/cameras.py [1235:1338]: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - _FIELDS = ( "K", "R", "T", "focal_length", "principal_point", "_in_ndc", "image_size", ) _SHARED_FIELDS = ("_in_ndc",) def __init__( self, focal_length=1.0, principal_point=((0.0, 0.0),), R: torch.Tensor = _R, T: torch.Tensor = _T, K: Optional[torch.Tensor] = None, device: Device = "cpu", in_ndc: bool = True, image_size: Optional[Union[List, Tuple, torch.Tensor]] = None, ) -> None: """ Args: focal_length: Focal length of the camera in world units. A tensor of shape (N, 1) or (N, 2) for square and non-square pixels respectively. principal_point: xy coordinates of the center of the principal point of the camera in pixels. A tensor of shape (N, 2). in_ndc: True if camera parameters are specified in NDC. If False, then camera parameters are in screen space. R: Rotation matrix of shape (N, 3, 3) T: Translation matrix of shape (N, 3) K: (optional) A calibration matrix of shape (N, 4, 4) If provided, don't need focal_length, principal_point, image_size image_size: (height, width) of image size. A tensor of shape (N, 2) or list/tuple. Required for screen cameras. device: torch.device or string """ # The initializer formats all inputs to torch tensors and broadcasts # all the inputs to have the same batch dimension where necessary. kwargs = {"image_size": image_size} if image_size is not None else {} super().__init__( device=device, focal_length=focal_length, principal_point=principal_point, R=R, T=T, K=K, _in_ndc=in_ndc, **kwargs, # pyre-ignore ) if image_size is not None: if (self.image_size < 1).any(): # pyre-ignore raise ValueError("Image_size provided has invalid values") else: self.image_size = None # When focal length is provided as one value, expand to # create (N, 2) shape tensor if self.focal_length.ndim == 1: # (N,) self.focal_length = self.focal_length[:, None] # (N, 1) self.focal_length = self.focal_length.expand(-1, 2) # (N, 2) def get_projection_transform(self, **kwargs) -> Transform3d: """ Calculate the projection matrix using the multi-view geometry convention. Args: **kwargs: parameters for the projection can be passed in as keyword arguments to override the default values set in __init__. Returns: A `Transform3d` object with a batch of `N` projection transforms. .. code-block:: python fx = focal_length[:,0] fy = focal_length[:,1] px = principal_point[:,0] py = principal_point[:,1] K = [ [fx, 0, 0, px], [0, fy, 0, py], [0, 0, 1, 0], [0, 0, 0, 1], ] """ K = kwargs.get("K", self.K) if K is not None: if K.shape != (self._N, 4, 4): msg = "Expected K to have shape of (%r, 4, 4)" raise ValueError(msg % (self._N)) else: K = _get_sfm_calibration_matrix( self._N, self.device, kwargs.get("focal_length", self.focal_length), kwargs.get("principal_point", self.principal_point), - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -