def main(argv):
parser = argparse.ArgumentParser(
description="Generate ridge-like fBm noise.")
parser.add_argument(
"-o",
"--output",
help="Output file name (without file extension). If not specified then \
the default file name will be used.")
parser.add_argument("--png",
action="store_true",
help="Automatically save a png of the noise.")
args = parser.parse_args()
my_dir = os.path.dirname(argv[0])
output_dir = os.path.join(my_dir, 'output')
if args.output:
output_path = os.path.join(output_dir, args.output)
else:
output_path = os.path.join(output_dir, 'ridge')
shape = (512, ) * 2
values = np.zeros(shape)
for p in range(1, 10):
a = 2**p
values += np.abs(noise_octave(shape, a) - 0.5) / a
result = (1.0 - util.normalize(values))**2
np.save(output_path, result)
# Optionally save out an image as well.
if args.png:
util.save_as_png(result, output_path + '_gray.png')
util.save_as_png(util.hillshaded(result), output_path + '_hill.png')
def main(argv):
parser = argparse.ArgumentParser(
description="Generate domain-warped fBm noise.")
parser.add_argument("-o", "--output", help="Output file name (without file \
extension). If not specified then the default file name will be used.")
parser.add_argument("--png", action="store_true", help="Automatically save \
a png of the noise.")
args = parser.parse_args()
my_dir = os.path.dirname(argv[0])
output_dir = os.path.join(my_dir, 'output')
if args.output:
output_path = os.path.join(output_dir, args.output)
else:
output_path = os.path.join(output_dir, 'domain_warping')
shape = (512,) * 2
values = util.fbm(shape, -2, lower=2.0)
offsets = 150 * (util.fbm(shape, -2, lower=1.5) +
1j * util.fbm(shape, -2, lower=1.5))
result = util.sample(values, offsets)
np.save(output_path, result)
# Optionally save out an image as well.
if args.png:
util.save_as_png(result, output_path + '_gray.png')
util.save_as_png(util.hillshaded(result), output_path + '_hill.png')
def main(argv):
my_dir = os.path.dirname(argv[0])
source_array_dir = os.path.join(my_dir, 'array_files')
training_samples_dir = os.path.join(my_dir, 'training_samples')
sample_dim = 512
sample_shape = (sample_dim, ) * 2
sample_area = np.prod(sample_shape)
# Create the training sample directory, if it doesn't already exist.
try:
os.mkdir(training_samples_dir)
except:
pass
source_array_paths = [
os.path.join(source_array_dir, path)
for path in os.listdir(source_array_dir)
]
training_id = 0
for (index, source_array_path) in enumerate(source_array_paths):
print('(%d / %d) Created %d samples so far' %
(index + 1, len(source_array_paths), training_id))
data = np.load(source_array_path)
# Load heightmap and correct for latitude (to an approximation)
source_array_raw = data['height']
latitude_deg = (data['minY'] + data['maxY']) / 2
latitude_correction = np.cos(np.radians(latitude_deg))
source_array_shape = (int(
np.round(source_array_raw.shape[0] * latitude_correction)),
source_array_raw.shape[1])
source_array = cv2.resize(source_array_raw, source_array_shape)
# Determine the number of samples to use per source array.
sampleable_area = np.subtract(source_array_shape, sample_shape).prod()
samples_per_array = int(np.ceil(sampleable_area / sample_area))
if len(source_array.shape) == 0:
print('Invalid array at %s' % source_array_path)
continue
for _ in range(samples_per_array):
# Select a sample from the source array.
row = np.random.randint(source_array.shape[0] - sample_shape[0])
col = np.random.randint(source_array.shape[1] - sample_shape[1])
sample = source_array[row:(row + sample_shape[0]),
col:(col + sample_shape[1])]
# Scale and clean the sample
sample = clean_sample(sample)
# Write the sample to a file
if sample is not None:
for variant in get_variants(sample):
output_path = os.path.join(training_samples_dir,
str(training_id) + '.png')
util.save_as_png(variant, output_path)
training_id += 1
def main(argv):
if len(argv) != 3:
print('Usage: %s <input_array.np[yz]> <output_image.png>' % (argv[0],))
sys.exit(-1)
input_path = argv[1]
output_path = argv[2]
height, land_mask = util.load_from_file(input_path)
util.save_as_png(util.hillshaded(height, land_mask=land_mask), output_path)
def main(argv):
parser = argparse.ArgumentParser(
description=
"Generates a PNG containing the terrain height in grayscale.")
parser.add_argument("input_array",
help="<input_array.np[yz]> (include file extension)")
parser.add_argument("output_image",
help="<output_image.png> (include file extension)")
args = parser.parse_args()
my_dir = os.path.dirname(argv[0])
output_dir = os.path.join(my_dir, 'output')
input_path = args.input_array
output_path = os.path.join(output_dir, args.output_image)
height, _ = util.load_from_file(input_path)
util.save_as_png(height, output_path)
def main(argv):
parser = argparse.ArgumentParser(
description="Generate fractional Brownian motion (fBm) noise.")
parser.add_argument(
"-s",
"--seed",
type=int,
help="Noise generator seed. If not specified then a random seed will be \
used. SEED MUST BE AN INTEGER.")
parser.add_argument(
"-o",
"--output",
help="Output noise file name (without file extension). If not specified \
then the default file name will be used.")
parser.add_argument("--png",
action="store_true",
help="Automatically save a png of the noise.")
args = parser.parse_args()
my_dir = os.path.dirname(argv[0])
output_dir = os.path.join(my_dir, 'output')
if args.output:
output_path = os.path.join(output_dir, args.output)
else:
output_path = os.path.join(output_dir, 'plain_fbm')
shape = (512, ) * 2
if args.seed:
input_seed = args.seed
else:
input_seed = None
# Generate the noise.
fbm_noise = util.fbm(shape, -2, lower=2.0, seed=input_seed)
np.save(output_path, fbm_noise)
# Optionally save out an image as well.
if args.png:
util.save_as_png(fbm_noise, output_path + '_gray.png')
util.save_as_png(util.hillshaded(fbm_noise), output_path + '_hill.png')
def main(argv):
parser = argparse.ArgumentParser(
description="Run a terrain erosion simulation.")
group = parser.add_mutually_exclusive_group()
group.add_argument(
"-f",
"--file",
help="Run simulation using an input image file instead of generating a \
new fBm noise. (Only works with a square image.) If not specified then \
noise will be generated.")
group.add_argument(
"-s",
"--seed",
type=int,
help="Noise generator seed. If not specified then a random seed will be \
used. SEED MUST BE AN INTEGER.")
parser.add_argument(
"-o",
"--output",
help="Output simulation file name (without file extension). If not \
specified then the default file name will be used.")
parser.add_argument("--snapshot",
action="store_true",
help="Save a numbered image of every iteration.")
parser.add_argument("--png",
action="store_true",
help="Automatically save a png of the simulation.")
args = parser.parse_args()
my_dir = os.path.dirname(argv[0])
output_dir = os.path.join(my_dir, 'output')
try:
os.mkdir(output_dir)
except:
pass
if args.output:
output_path = os.path.join(output_dir, args.output)
else:
output_path = os.path.join(output_dir, 'simulation')
if args.seed:
input_seed = args.seed
else:
input_seed = None
# Grid dimension constants if using fBm noise
full_width = 200
dim = 512
shape = [dim] * 2
cell_width = full_width / dim
cell_area = cell_width**2
# `terrain` represents the actual terrain height we're interested in
if not args.file:
terrain = util.fbm(shape, -2.0, seed=input_seed)
else:
terrain = util.image_to_array(args.file)
dim = terrain.shape[0]
shape = terrain.shape
cell_width = full_width / dim
cell_area = cell_width**2
# Snapshotting parameters. Only needed for generating the simulation
# timelapse.
if args.snapshot:
snapshot_dir = os.path.join(output_dir, 'sim_snaps')
snapshot_file_template = 'sim-%05d.png'
try:
os.mkdir(snapshot_dir)
except:
pass
# Water-related constants
rain_rate = 0.0008 * cell_area
evaporation_rate = 0.0005
# Slope constants
min_height_delta = 0.05
repose_slope = 0.03
gravity = 30.0
gradient_sigma = 0.5
# Sediment constants
sediment_capacity_constant = 50.0
dissolving_rate = 0.25
deposition_rate = 0.001
# The numer of iterations is proportional to the grid dimension. This is to
# allow changes on one side of the grid to affect the other side.
iterations = int(1.4 * dim)
# `sediment` is the amount of suspended "dirt" in the water. Terrain will be
# transfered to/from sediment depending on a number of different factors.
sediment = np.zeros_like(terrain)
# The amount of water. Responsible for carrying sediment.
water = np.zeros_like(terrain)
# The water velocity.
velocity = np.zeros_like(terrain)
# Optionally save the unmodified starting noise if we're not using file input.
if args.snapshot and args.png and not args.file:
fbm_path = output_path + '_fbm.png'
util.save_as_png(terrain, fbm_path)
for i in range(0, iterations):
print('Iteration: %d / %d' % (i + 1, iterations))
# Set a deterministic seed for our random number generator
rng = np.random.default_rng(i)
# Add precipitation. This is done via simple uniform random distribution,
# although other models use a raindrop model
water += rng.random(shape) * rain_rate
# Use a different RNG seed for the next step
rng = np.random.default_rng(i + 3)
# Compute the normalized gradient of the terrain height to determine where
# water and sediment will be moving.
gradient = np.zeros_like(terrain, dtype='complex')
gradient = util.simple_gradient(terrain)
gradient = np.select([np.abs(gradient) < 1e-10],
[np.exp(2j * np.pi * rng.random(shape))],
gradient)
gradient /= np.abs(gradient)
# Compute the difference between the current height the height offset by
# `gradient`.
neighbor_height = util.sample(terrain, -gradient)
height_delta = terrain - neighbor_height
# The sediment capacity represents how much sediment can be suspended in
# water. If the sediment exceeds the quantity, then it is deposited,
# otherwise terrain is eroded.
sediment_capacity = (
(np.maximum(height_delta, min_height_delta) / cell_width) *
velocity * water * sediment_capacity_constant)
deposited_sediment = np.select(
[
height_delta < 0,
sediment > sediment_capacity,
],
[
np.minimum(height_delta, sediment),
deposition_rate * (sediment - sediment_capacity),
],
# If sediment <= sediment_capacity
dissolving_rate * (sediment - sediment_capacity))
# Don't erode more sediment than the current terrain height.
deposited_sediment = np.maximum(-height_delta, deposited_sediment)
# Update terrain and sediment quantities.
sediment -= deposited_sediment
terrain += deposited_sediment
sediment = util.displace(sediment, gradient)
water = util.displace(water, gradient)
# Smooth out steep slopes.
terrain = apply_slippage(terrain, repose_slope, cell_width)
# Update velocity
velocity = gravity * height_delta / cell_width
# Apply evaporation
water *= 1 - evaporation_rate
# Snapshot, if applicable.
if args.snapshot:
snapshot_path = os.path.join(snapshot_dir,
snapshot_file_template % i)
util.save_as_png(terrain, snapshot_path)
# Normalize terrain values before saving.
result = util.normalize(terrain)
np.save(output_path, result)
# Optionally save out an image as well.
if args.png:
util.save_as_png(result, output_path + '_gray.png')
util.save_as_png(util.hillshaded(result), output_path + '_hill.png')
def erosion(terrain, enable_snapshotting=False, dirname='snapshots'):
# Grid dimension constants
full_width = 200
shape = terrain.shape
dim = shape[1]
cell_width = full_width / dim
cell_area = cell_width ** 2
# Snapshotting parameters. Only needed for generating the simulation
# timelapse.
enable_snapshotting = False
my_dir = os.path.dirname(dirname)
snapshot_dir = os.path.join(my_dir, 'sim_snaps')
snapshot_file_template = 'sim-%05d.png'
if enable_snapshotting:
try:
os.mkdir(snapshot_dir)
except:
pass
# Water-related constants
rain_rate = 0.0008 * cell_area
evaporation_rate = 0.0005
# Slope constants
min_height_delta = 0.05
repose_slope = 0.03
gravity = 30.0
gradient_sigma = 0.5
# Sediment constants
sediment_capacity_constant = 50.0
dissolving_rate = 0.25
deposition_rate = 0.001
# The numer of iterations is proportional to the grid dimension. This is to
# allow changes on one side of the grid to affect the other side.
iterations = int(1.4 * dim)
# `terrain` represents the actual terrain height we're interested in
# terrain = util.fbm(shape, -2.0)
# `sediment` is the amount of suspended "dirt" in the water. Terrain will be
# transfered to/from sediment depending on a number of different factors.
sediment = np.zeros_like(terrain)
# The amount of water. Responsible for carrying sediment.
water = np.zeros_like(terrain)
# The water velocity.
velocity = np.zeros_like(terrain)
for i in range(0, iterations):
print('%d / %d' % (i + 1, iterations))
# Add precipitation. This is done by via simple uniform random distribution,
# although other models use a raindrop model
water += np.random.rand(*shape) * rain_rate
# Compute the normalized gradient of the terrain height to determine where
# water and sediment will be moving.
gradient = np.zeros_like(terrain, dtype='complex')
gradient = util.simple_gradient(terrain)
gradient = np.select([np.abs(gradient) < 1e-10],
[np.exp(2j * np.pi * np.random.rand(*shape))],
gradient)
gradient /= np.abs(gradient)
# Compute the difference between teh current height the height offset by
# `gradient`.
neighbor_height = util.sample(terrain, -gradient)
height_delta = terrain - neighbor_height
# The sediment capacity represents how much sediment can be suspended in
# water. If the sediment exceeds the quantity, then it is deposited,
# otherwise terrain is eroded.
sediment_capacity = (
(np.maximum(height_delta, min_height_delta) / cell_width) * velocity *
water * sediment_capacity_constant)
deposited_sediment = np.select(
[
height_delta < 0,
sediment > sediment_capacity,
], [
np.minimum(height_delta, sediment),
deposition_rate * (sediment - sediment_capacity),
],
# If sediment <= sediment_capacity
dissolving_rate * (sediment - sediment_capacity))
# Don't erode more sediment than the current terrain height.
deposited_sediment = np.maximum(-height_delta, deposited_sediment)
# Update terrain and sediment quantities.
sediment -= deposited_sediment
terrain += deposited_sediment
sediment = util.displace(sediment, gradient)
water = util.displace(water, gradient)
# Smooth out steep slopes.
terrain = apply_slippage(terrain, repose_slope, cell_width)
# Update velocity
velocity = gravity * height_delta / cell_width
# Apply evaporation
water *= 1 - evaporation_rate
# Snapshot, if applicable.
if enable_snapshotting:
output_path = os.path.join(snapshot_dir, snapshot_file_template % i)
util.save_as_png(terrain, output_path)
return terrain
def main(argv):
parser = argparse.ArgumentParser(
description="Generate terrain from a river network.")
parser.add_argument("-o", "--output",
help="Output file name (without file extension). If not specified then \
the default file name will be used.")
parser.add_argument("--png", action="store_true",
help="Automatically save a png of the terrain.")
args = parser.parse_args()
my_dir = os.path.dirname(argv[0])
output_dir = os.path.join(my_dir, 'output')
if args.output:
output_path = os.path.join(output_dir, args.output)
else:
output_path = os.path.join(output_dir, 'river_network')
dim = 512
shape = (dim,) * 2
disc_radius = 1.0
max_delta = 0.05
river_downcutting_constant = 1.3
directional_inertia = 0.4
default_water_level = 1.0
evaporation_rate = 0.2
print ('Generating...')
print(' ...initial terrain shape')
land_mask = remove_lakes(
(util.fbm(shape, -2, lower=2.0) + bump(shape, 0.2 * dim) - 1.1) > 0)
coastal_dropoff = np.tanh(util.dist_to_mask(land_mask) / 80.0) * land_mask
mountain_shapes = util.fbm(shape, -2, lower=2.0, upper=np.inf)
initial_height = (
(util.gaussian_blur(np.maximum(mountain_shapes - 0.40, 0.0), sigma=5.0)
+ 0.1) * coastal_dropoff)
deltas = util.normalize(np.abs(util.gaussian_gradient(initial_height)))
print(' ...sampling points')
points = util.poisson_disc_sampling(shape, disc_radius)
coords = np.floor(points).astype(int)
print(' ...delaunay triangulation')
tri = sp.spatial.Delaunay(points)
(indices, indptr) = tri.vertex_neighbor_vertices
neighbors = [indptr[indices[k]:indices[k + 1]] for k in range(len(points))]
points_land = land_mask[coords[:, 0], coords[:, 1]]
points_deltas = deltas[coords[:, 0], coords[:, 1]]
print(' ...initial height map')
points_height = compute_height(points, neighbors, points_deltas)
print(' ...river network')
(upstream, downstream, volume) = compute_river_network(
points, neighbors, points_height, points_land,
directional_inertia, default_water_level, evaporation_rate)
print(' ...final terrain height')
new_height = compute_final_height(
points, neighbors, points_deltas, volume, upstream,
max_delta, river_downcutting_constant)
terrain_height = render_triangulation(shape, tri, new_height)
np.savez(output_path, height=terrain_height, land_mask=land_mask)
# Optionally save out an image as well.
if args.png:
util.save_as_png(terrain_height, output_path + '_gray.png')
util.save_as_png(util.hillshaded(
terrain_height, land_mask=land_mask), output_path + '_hill.png')
