def test_default_image_optim_loop_logging_smoke(caplog, optim_asset_loader):
asset = optim_asset_loader("default_image_optim_loop")
num_steps = 1
optim_logger = optim.OptimLogger()
log_fn = optim.default_image_optim_log_fn(optim_logger, log_freq=1)
with asserts.assert_logs(caplog, logger=optim_logger):
optim.default_image_optim_loop(
asset.input.image,
asset.input.criterion,
num_steps=num_steps,
log_fn=log_fn,
)
def test_default_image_optim_loop_logging_smoke(self):
asset = self.load_asset(path.join("optim", "default_image_optim_loop"))
num_steps = 1
optim_logger = optim.OptimLogger()
log_fn = optim.default_image_optim_log_fn(optim_logger, log_freq=1)
with self.assertLogs(optim_logger.logger, "INFO"):
optim.default_image_optim_loop(
asset.input.image,
asset.input.criterion,
num_steps=num_steps,
log_fn=log_fn,
)
def test_default_image_optim_loop(self):
asset = self.load_asset(path.join("optim", "default_image_optim_loop"))
actual = optim.default_image_optim_loop(
asset.input.image,
asset.input.criterion,
get_optimizer=asset.params.get_optimizer,
num_steps=asset.params.num_steps,
quiet=True,
)
desired = asset.output.image
self.assertTensorAlmostEqual(actual, desired, rtol=1e-4)
def test_default_image_optim_loop(optim_asset_loader):
asset = optim_asset_loader("default_image_optim_loop")
actual = optim.default_image_optim_loop(
asset.input.image,
asset.input.criterion,
get_optimizer=asset.params.get_optimizer,
num_steps=asset.params.num_steps,
quiet=True,
)
desired = asset.output.image
ptu.assert_allclose(actual, desired, rtol=1e-4)
def gatys_ecker_bethge_2015_nst(
content_image: torch.Tensor,
style_image: torch.Tensor,
num_steps: int = 500,
impl_params: bool = True,
criterion: Optional[PerceptualLoss] = None,
quiet: bool = False,
logger: Optional[logging.Logger] = None,
log_fn: Optional[Callable[[int, Union[torch.Tensor, pystiche.LossDict]],
None]] = None,
) -> torch.Tensor:
if criterion is None:
criterion = gatys_ecker_bethge_2015_perceptual_loss(
impl_params=impl_params)
device = content_image.device
criterion = criterion.to(device)
starting_point = "content" if impl_params else "random"
input_image = get_input_image(starting_point=starting_point,
content_image=content_image)
preprocessor = gatys_ecker_bethge_2015_preprocessor().to(device)
postprocessor = gatys_ecker_bethge_2015_postprocessor().to(device)
criterion.set_content_image(preprocessor(content_image))
criterion.set_style_image(preprocessor(style_image))
return default_image_optim_loop(
input_image,
criterion,
get_optimizer=gatys_ecker_bethge_2015_optimizer,
num_steps=num_steps,
preprocessor=preprocessor,
postprocessor=postprocessor,
quiet=quiet,
logger=logger,
log_fn=log_fn,
)
print(criterion)
########################################################################################
# We set the target images for the optimization ``criterion``.
criterion.set_content_image(content_image)
criterion.set_style_image(style_image)
########################################################################################
# We perform the unguided NST and show the result.
starting_point = "content"
input_image = get_input_image(starting_point, content_image=content_image)
output_image = default_image_optim_loop(input_image,
criterion,
num_steps=500,
logger=demo_logger())
########################################################################################
show_image(output_image)
########################################################################################
# While the result is not completely unreasonable, the building has a strong blueish
# cast that looks unnatural. Since the optimization was unconstrained the color of the
# sky was used for building. In the remainder of this example we will solve this by
# dividing the images in multiple separate regions.
########################################################################################
# Guided image optimization
# -------------------------
# ----------------------------------
#
# As a baseline we use a standard image optimization without pyramid.
starting_point = "content"
input_image = get_input_image(starting_point, content_image=content_image)
show_image(input_image, title="Input image")
########################################################################################
# We time the NST performed by :func:`~pystiche.optim.default_image_optim_loop` and
# show the result.
start_without_pyramid = time.time()
output_image = optim.default_image_optim_loop(
input_image, criterion, num_steps=400, logger=demo.logger()
)
stop_without_pyramid = time.time()
show_image(output_image, title="Output image without pyramid")
########################################################################################
elapsed_time_without_pyramid = stop_without_pyramid - start_without_pyramid
print(
f"Without pyramid the optimization took {elapsed_time_without_pyramid:.0f} seconds."
)
########################################################################################
# As you can see the small blurry branches on the left side of the image were picked up
# by the style transfer. They distort the mosaic pattern, which minders the quality of
