def test_scalar_3d(benchmark, func, octaves):
if func == "noise.snoise":
benchmark(noise.snoise3, 0.2, 0.5, 0.8, octaves=octaves)
elif func == "noise.pnoise":
benchmark(noise.pnoise3, 0.2, 0.5, 0.8, octaves=octaves)
elif func == "vnoise":
vn = vnoise.Noise()
benchmark(vn.noise3, 0.2, 0.5, 0.8, octaves=octaves)
elif func == "pnoise":
pn = pnoise.Noise()
pn.octaves = octaves
benchmark(pn.perlin, 0.2, 0.5, 0.8)
else:
assert False
def test_vectorized_1d(benchmark, func, octaves):
xx = XX2d.copy()
if func == "noise.snoise":
benchmark(vectorize_func_2d, noise.snoise2, xx, [0.0], octaves=octaves)
elif func == "noise.pnoise":
benchmark(vectorize_func_1d, noise.pnoise1, xx, octaves=octaves)
elif func == "vnoise":
vn = vnoise.Noise()
benchmark(vn.noise1, xx, octaves=octaves)
elif func == "pnoise":
pn = pnoise.Noise()
pn.octaves = octaves
benchmark(pn.perlin, xx, 0.0, 0.0)
else:
assert False
def test_vectorized_3d(benchmark, func, octaves):
xx = XX.copy()
yy = YY.copy()
zz = ZZ.copy()
if func == "noise.snoise":
benchmark(vectorize_func, noise.snoise3, xx, yy, zz, octaves=octaves)
elif func == "noise.pnoise":
benchmark(vectorize_func, noise.pnoise3, xx, yy, zz, octaves=octaves)
elif func == "vnoise":
vn = vnoise.Noise()
benchmark(vn.noise3, xx, yy, zz, octaves=octaves)
elif func == "pnoise":
pn = pnoise.Noise()
pn.octaves = octaves
benchmark(pn.perlin, xx, yy, zz)
else:
assert False
def squiggles(lines: LineCollection, ampl: float, period: float,
quantization: float) -> LineCollection:
"""Apply a squiggle filter to a :class:`LineCollection`.
This filter first densely resample the input lines (based on the ``quantization``
parameter), and then applies a 2D-Perlin-noise-based displacement to all points.
For small values of amplitude (~2px) and period (~12px), this filter gives a "shaky-hand"
style to the lines. Larger values of amplitude (~60px) and period (~400px) result in a
a smoother, liquid-like effect.
Args:
lines: input lines
ampl: squiggle amplitude (px)
period: squiggle period (px)
quantization: quantization (px)
Returns:
filtered lines
"""
# link noise seed to global seed
noise = pnoise.Noise()
noise.seed(random.randint(0, 2**16))
freq = 1.0 / period
new_lines = LineCollection()
for line in lines:
line = interpolate(line, step=quantization)
perlin_x = noise.perlin(freq * line.real,
freq * line.imag,
np.zeros_like(line.real),
grid_mode=False)
perlin_y = noise.perlin(freq * line.real,
freq * line.imag,
1000 * np.ones_like(line.real),
grid_mode=False)
new_lines.append(line + ampl * 2.0 * ((perlin_x - 0.5) +
(perlin_y - 0.5) * 1j))
return new_lines