def estimate_radiance_fn(point_light_position, surface_point_position,
surface_point_normal, observation_point):
return point_light.estimate_radiance(point_light_radiance_init,
point_light_position,
surface_point_position,
surface_point_normal,
observation_point, fake_brdf)
def test_estimate_radiance_value_exceptions_raised(self):
"""Tests that the value exceptions are raised correctly."""
point_light_radiance = random_tensor(tensor_shape=(1, 1))
point_light_position = random_tensor(tensor_shape=(1, 3))
surface_point_position = random_tensor(tensor_shape=(3,))
surface_point_normal = random_tensor(tensor_shape=(3,))
observation_point = random_tensor(tensor_shape=(3,))
# Verify that an InvalidArgumentError is raised as the given
# surface_point_normal is not normalized.
with self.assertRaises(tf.errors.InvalidArgumentError):
self.evaluate(
point_light.estimate_radiance(point_light_radiance,
point_light_position,
surface_point_position,
surface_point_normal, observation_point,
returning_zeros_brdf))
def test_estimate_radiance_preset(self,
light_radiance,
light_pos,
observation_pos,
expected_result,
reflected_light_fall_off=False):
"""Tests the output of estimate radiance function with various parameters.
In this test the point on the surface is always [0, 0, 0] ,the surface
normal is [0, 0, 1] and the fake brdf function returns the (normalized)
direction of the outgoing light as its output.
Args:
light_radiance: An array of size K representing the point light radiances.
light_pos: An array of size [3,] representing the point light positions.
observation_pos: An array of size [3,] representing the observation point.
expected_result: An array of size [3,] representing the expected result of
the estimated reflected radiance function.
reflected_light_fall_off: A boolean specifying whether or not to include
the fall off of the reflected light in the calculation. Defaults to
False.
"""
tensor_size = np.random.randint(3) + 1
tensor_shape = np.random.randint(1, 10, size=(tensor_size)).tolist()
lights_tensor_size = np.random.randint(3) + 1
lights_tensor_shape = np.random.randint(
1, 10, size=(lights_tensor_size)).tolist()
point_light_radiance = np.tile(light_radiance,
lights_tensor_shape + [1])
point_light_position = np.tile(light_pos, lights_tensor_shape + [1])
surface_point_normal = np.tile([0.0, 0.0, 1.0], tensor_shape + [1])
surface_point_position = np.tile([0.0, 0.0, 0.0], tensor_shape + [1])
observation_point = np.tile(observation_pos, tensor_shape + [1])
expected = np.tile(expected_result,
tensor_shape + lights_tensor_shape + [1])
pred = point_light.estimate_radiance(
point_light_radiance,
point_light_position,
surface_point_position,
surface_point_normal,
observation_point,
fake_brdf,
reflected_light_fall_off=reflected_light_fall_off)
self.assertAllClose(expected, pred)