def fov(self, value):
with tf.device('/device:cpu:' + str(pyredner.get_cpu_device_id())):
self._fov = tf.identity(value).cpu()
fov_factor = 1.0 / tf.tan(transform.radians(0.5 * self._fov))
o = tf.convert_to_tensor(np.ones([1], dtype=np.float32),
dtype=tf.float32)
diag = tf.concat([fov_factor, fov_factor, o], 0)
self._cam_to_ndc = tf.linalg.tensor_diag(diag)
self.ndc_to_cam = tf.linalg.inv(self._cam_to_ndc)
def parse_transform(node):
ret = tf.eye(4)
for child in node:
if child.tag == 'matrix':
value = tf.convert_to_tensor(
np.reshape(
np.fromstring(child.attrib['value'],
dtype=np.float32,
sep=' '), (4, 4)))
ret = value @ ret
elif child.tag == 'translate':
x = float(child.attrib['x'])
y = float(child.attrib['y'])
z = float(child.attrib['z'])
value = transform.gen_translate_matrix(tf.constant([x, y, z]))
ret = value @ ret
elif child.tag == 'scale':
x = float(child.attrib['x'])
y = float(child.attrib['y'])
z = float(child.attrib['z'])
value = transform.gen_scale_matrix(tf.constant([x, y, z]))
ret = value @ ret
elif child.tag == 'rotate':
x = float(child.attrib['x']) if 'x' in child.attrib else 0.0
y = float(child.attrib['y']) if 'y' in child.attrib else 0.0
z = float(child.attrib['z']) if 'z' in child.attrib else 0.0
angle = transform.radians(float(child.attrib['angle']))
axis = np.array([x, y, z])
axis = axis / np.linalg.norm(axis)
cos_theta = math.cos(angle)
sin_theta = math.sin(angle)
mat = np.zeros([4, 4], dtype=np.float32)
mat[0,
0] = axis[0] * axis[0] + (1.0 - axis[0] * axis[0]) * cos_theta
mat[0, 1] = axis[0] * axis[1] * (1.0 -
cos_theta) - axis[2] * sin_theta
mat[0, 2] = axis[0] * axis[2] * (1.0 -
cos_theta) + axis[1] * sin_theta
mat[1, 0] = axis[0] * axis[1] * (1.0 -
cos_theta) + axis[2] * sin_theta
mat[1,
1] = axis[1] * axis[1] + (1.0 - axis[1] * axis[1]) * cos_theta
mat[1, 2] = axis[1] * axis[2] * (1.0 -
cos_theta) - axis[0] * sin_theta
mat[2, 0] = axis[0] * axis[2] * (1.0 -
cos_theta) - axis[1] * sin_theta
mat[2, 1] = axis[1] * axis[2] * (1.0 -
cos_theta) + axis[0] * sin_theta
mat[2,
2] = axis[2] * axis[2] + (1.0 - axis[2] * axis[2]) * cos_theta
mat[3, 3] = 1.0
ret = tf.convert_to_tensor(mat) @ ret
return ret
def fov(self, value):
if value is not None:
self._fov = value
with tf.device('/device:cpu:' + str(pyredner.get_cpu_device_id())):
fov_factor = 1.0 / tf.tan(transform.radians(0.5 * self._fov))
o = tf.ones([1], dtype=tf.float32)
diag = tf.concat([fov_factor, fov_factor, o], 0)
self._intrinsic_mat = tf.linalg.tensor_diag(diag)
self.intrinsic_mat_inv = tf.linalg.inv(self._intrinsic_mat)
else:
self._fov = None
def __init__(self,
position: Optional[tf.Tensor] = None,
look_at: Optional[tf.Tensor] = None,
up: Optional[tf.Tensor] = None,
fov: Optional[tf.Tensor] = None,
clip_near: float = 1e-4,
resolution: Tuple[int] = (256, 256),
cam_to_world: Optional[tf.Tensor] = None,
intrinsic_mat: Optional[tf.Tensor] = None,
camera_type = pyredner.camera_type.perspective,
fisheye: bool = False):
assert(tf.executing_eagerly())
if position is not None:
assert(position.dtype == tf.float32)
assert(len(position.shape) == 1 and position.shape[0] == 3)
if look_at is not None:
assert(look_at.dtype == tf.float32)
assert(len(look_at.shape) == 1 and look_at.shape[0] == 3)
if up is not None:
assert(up.dtype == tf.float32)
assert(len(up.shape) == 1 and up.shape[0] == 3)
if fov is not None:
assert(fov.dtype == tf.float32)
assert(len(fov.shape) == 1 and fov.shape[0] == 1)
assert(isinstance(clip_near, float))
if position is None and look_at is None and up is None:
assert(cam_to_world is not None)
self.position = position
self.look_at = look_at
self.up = up
self.fov = fov
with tf.device('/device:cpu:' + str(pyredner.get_cpu_device_id())):
if cam_to_world is not None:
self.cam_to_world = cam_to_world
else:
self.cam_to_world = None
if intrinsic_mat is None:
if camera_type == redner.CameraType.perspective:
fov_factor = 1.0 / tf.tan(transform.radians(0.5 * fov))
o = tf.ones([1], dtype=tf.float32)
diag = tf.concat([fov_factor, fov_factor, o], 0)
self._intrinsic_mat = tf.linalg.tensor_diag(diag)
else:
self._intrinsic_mat = tf.eye(3, dtype=tf.float32)
else:
self._intrinsic_mat = intrinsic_mat
self.intrinsic_mat_inv = tf.linalg.inv(self._intrinsic_mat)
self.clip_near = clip_near
self.resolution = resolution
self.camera_type = camera_type
if fisheye:
self.camera_type = redner.CameraType.fisheye
def __init__(self,
position: tf.Tensor,
look_at: tf.Tensor,
up: tf.Tensor,
fov: tf.Tensor,
clip_near: float,
resolution: Tuple[int],
cam_to_ndc: tf.Tensor = None,
camera_type=redner.CameraType.perspective,
fisheye: bool = False):
assert (tf.executing_eagerly())
assert (position.dtype == tf.float32)
assert (len(position.shape) == 1 and position.shape[0] == 3)
assert (look_at.dtype == tf.float32)
assert (len(look_at.shape) == 1 and look_at.shape[0] == 3)
assert (up.dtype == tf.float32)
assert (len(up.shape) == 1 and up.shape[0] == 3)
if fov is not None:
assert (fov.dtype == tf.float32)
assert (len(fov.shape) == 1 and fov.shape[0] == 1)
assert (isinstance(clip_near, float))
with tf.device('/device:cpu:' + str(pyredner.get_cpu_device_id())):
self.position = tf.identity(position).cpu()
self.look_at = tf.identity(look_at).cpu()
self.up = tf.identity(up).cpu()
self.fov = tf.identity(fov).cpu()
if cam_to_ndc is None:
if camera_type == redner.CameraType.perspective:
fov_factor = 1.0 / tf.tan(transform.radians(0.5 * fov))
o = tf.convert_to_tensor(np.ones([1], dtype=np.float32),
dtype=tf.float32)
diag = tf.concat([fov_factor, fov_factor, o], 0)
self._cam_to_ndc = tf.linalg.tensor_diag(diag)
else:
self._cam_to_ndc = tf.eye(3, dtype=tf.float32)
else:
self._cam_to_ndc = tf.identity(cam_to_ndc).cpu()
self.ndc_to_cam = tf.linalg.inv(self.cam_to_ndc)
self.clip_near = clip_near
self.resolution = resolution
self.camera_type = camera_type
if fisheye:
self.camera_type = redner.CameraType.fisheye