def process(self):
if not any((s.is_linked for s in self.outputs)):
return
need_eval = self.outputs['Vertices'].is_linked
curves_s = self.inputs['Curve'].sv_get()
resolution_s = self.inputs['Resolution'].sv_get()
length_s = self.inputs['Length'].sv_get()
samples_s = self.inputs['Samples'].sv_get(default=[[]])
length_s = ensure_nesting_level(length_s, 3)
resolution_s = ensure_nesting_level(resolution_s, 2)
samples_s = ensure_nesting_level(samples_s, 2)
curves_s = ensure_nesting_level(curves_s, 2, data_types=(SvCurve,))
ts_out = []
verts_out = []
for curves, resolutions, input_lengths_i, samples_i in zip_long_repeat(curves_s, resolution_s, length_s, samples_s):
for curve, resolution, input_lengths, samples in zip_long_repeat(curves, resolutions, input_lengths_i, samples_i):
mode = self.mode
accuracy = self.accuracy
if self.id_data.sv_draft:
mode = 'LIN'
accuracy = self.accuracy_draft
if self.specify_accuracy:
tolerance = 10 ** (-accuracy)
else:
tolerance = None
solver = SvCurveLengthSolver(curve)
solver.prepare(mode, resolution, tolerance=tolerance)
if self.eval_mode == 'AUTO':
total_length = solver.get_total_length()
input_lengths = np.linspace(0.0, total_length, num = samples)
else:
input_lengths = np.array(input_lengths)
ts = solver.solve(input_lengths)
ts_out.append(ts.tolist())
if need_eval:
verts = curve.evaluate_array(ts).tolist()
verts_out.append(verts)
self.outputs['T'].sv_set(ts_out)
self.outputs['Vertices'].sv_set(verts_out)
def process(self):
if not any((s.is_linked for s in self.outputs)):
return
need_eval = self.outputs['Vertices'].is_linked
curves = self.inputs['Curve'].sv_get()
resolution_s = self.inputs['Resolution'].sv_get()
length_s = self.inputs['Length'].sv_get()
samples_s = self.inputs['Samples'].sv_get(default=[[]])
length_s = ensure_nesting_level(length_s, 2)
ts_out = []
verts_out = []
for curve, resolution, input_lengths, samples, in zip_long_repeat(
curves, resolution_s, length_s, samples_s):
if self.eval_mode == 'AUTO':
if isinstance(samples, (list, tuple)):
samples = samples[0]
if isinstance(resolution, (list, tuple)):
resolution = resolution[0]
mode = self.mode
if self.id_data.sv_draft:
mode = 'LIN'
solver = SvCurveLengthSolver(curve)
solver.prepare(mode, resolution)
if self.eval_mode == 'AUTO':
total_length = solver.get_total_length()
input_lengths = np.linspace(0.0, total_length, num=samples)
else:
input_lengths = np.array(input_lengths)
ts = solver.solve(input_lengths)
ts_out.append(ts.tolist())
if need_eval:
verts = curve.evaluate_array(ts).tolist()
verts_out.append(verts)
self.outputs['T'].sv_set(ts_out)
self.outputs['Vertices'].sv_set(verts_out)
class SvBendAlongCurveField(SvVectorField):
ZERO = 'ZERO'
FRENET = 'FRENET'
HOUSEHOLDER = 'householder'
TRACK = 'track'
DIFF = 'diff'
TRACK_NORMAL = 'track_normal'
def __init__(self, curve, algorithm, scale_all, axis, t_min, t_max, up_axis=None, resolution=50, length_mode='T'):
self.curve = curve
self.axis = axis
self.t_min = t_min
self.t_max = t_max
self.algorithm = algorithm
self.scale_all = scale_all
self.up_axis = up_axis
self.length_mode = length_mode
if algorithm == SvBendAlongCurveField.ZERO:
self.curve.pre_calc_torsion_integral(resolution)
elif algorithm == SvBendAlongCurveField.TRACK_NORMAL:
self.normal_tracker = SvNormalTrack(curve, resolution)
if length_mode == 'L':
self.length_solver = SvCurveLengthSolver(curve)
self.length_solver.prepare('SPL', resolution)
self.__description__ = "Bend along {}".format(curve)
def get_matrix(self, tangent, scale):
return MathutilsRotationCalculator.get_matrix(tangent, scale, self.axis,
self.algorithm, self.scale_all, self.up_axis)
def get_matrices(self, ts, scale):
n = len(ts)
if self.scale_all:
scale_matrix = np.array([
[scale, 0, 0],
[0, scale, 0],
[0, 0, 1/scale]
])
else:
scale_matrix = np.array([
[1, 0, 0],
[0, 1, 0],
[0, 0, 1/scale]
])
if self.algorithm == SvBendAlongCurveField.FRENET:
frenet, _ , _ = self.curve.frame_array(ts)
return frenet @ scale_matrix
elif self.algorithm == SvBendAlongCurveField.ZERO:
frenet, _ , _ = self.curve.frame_array(ts)
angles = - self.curve.torsion_integral(ts)
zeros = np.zeros((n,))
ones = np.ones((n,))
row1 = np.stack((np.cos(angles), np.sin(angles), zeros)).T # (n, 3)
row2 = np.stack((-np.sin(angles), np.cos(angles), zeros)).T # (n, 3)
row3 = np.stack((zeros, zeros, ones)).T # (n, 3)
rotation_matrices = np.dstack((row1, row2, row3))
return frenet @ rotation_matrices @ scale_matrix
elif self.algorithm == SvBendAlongCurveField.TRACK_NORMAL:
matrices = self.normal_tracker.evaluate_array(ts)
return matrices @ scale_matrix
else:
raise Exception("Unsupported algorithm")
def get_t_value(self, x, y, z):
curve_t_min, curve_t_max = self.curve.get_u_bounds()
t = [x, y, z][self.axis]
if self.length_mode == 'T':
t = (curve_t_max - curve_t_min) * (t - self.t_min) / (self.t_max - self.t_min) + curve_t_min
else:
t = (t - self.t_min) / (self.t_max - self.t_min) # 0 .. 1
t = t * self.length_solver.get_total_length()
t = self.length_solver.solve(np.array([t]))[0]
return t
def get_t_values(self, xs, ys, zs):
curve_t_min, curve_t_max = self.curve.get_u_bounds()
ts = [xs, ys, zs][self.axis]
if self.length_mode == 'T':
ts = (curve_t_max - curve_t_min) * (ts - self.t_min) / (self.t_max - self.t_min) + curve_t_min
else:
ts = (ts - self.t_min) / (self.t_max - self.t_min) # 0 .. 1
ts = ts * self.length_solver.get_total_length()
ts = self.length_solver.solve(ts)
return ts
def get_scale(self):
if self.length_mode == 'T':
curve_t_min, curve_t_max = self.curve.get_u_bounds()
t_range = curve_t_max - curve_t_min
else:
t_range = self.length_solver.get_total_length()
return (self.t_max - self.t_min) / t_range
def evaluate(self, x, y, z):
t = self.get_t_value(x, y, z)
spline_tangent = self.curve.tangent(t)
spline_vertex = self.curve.evaluate(t)
scale = self.get_scale()
if self.algorithm in {SvBendAlongCurveField.ZERO, SvBendAlongCurveField.FRENET, SvBendAlongCurveField.TRACK_NORMAL}:
matrix = self.get_matrices(np.array([t]), scale)
else:
matrix = self.get_matrix(spline_tangent, scale)
src_vector_projection = np.array([x, y, z])
src_vector_projection[self.axis] = 0
new_vertex = np.matmul(matrix, src_vector_projection) + spline_vertex
vector = new_vertex - np.array([x, y, z])
return vector
def evaluate_grid(self, xs, ys, zs):
def multiply(matrices, vectors):
vectors = vectors[np.newaxis]
vectors = np.transpose(vectors, axes=(1,2,0))
r = matrices @ vectors
return r[:,:,0]
ts = self.get_t_values(xs, ys, zs).flatten()
spline_tangents = self.curve.tangent_array(ts)
spline_vertices = self.curve.evaluate_array(ts)
scale = self.get_scale()
if self.algorithm in {SvBendAlongCurveField.ZERO, SvBendAlongCurveField.FRENET, SvBendAlongCurveField.TRACK_NORMAL}:
matrices = self.get_matrices(ts, scale)
else:
matrices = np.vectorize(lambda t : self.get_matrix(t, scale), signature='(3)->(3,3)')(spline_tangents)
src_vectors = np.stack((xs, ys, zs)).T
src_vector_projections = src_vectors.copy()
src_vector_projections[:,self.axis] = 0
#multiply = np.vectorize(lambda m, v: m @ v, signature='(3,3),(3)->(3)')
new_vertices = multiply(matrices, src_vector_projections) + spline_vertices
R = (new_vertices - src_vectors).T
return R[0], R[1], R[2]
class SvBendAlongCurveField(SvVectorField):
def __init__(self,
curve,
algorithm,
scale_all,
axis,
t_min,
t_max,
up_axis=None,
resolution=50,
length_mode='T'):
self.curve = curve
self.axis = axis
self.t_min = t_min
self.t_max = t_max
self.algorithm = algorithm
self.scale_all = scale_all
self.up_axis = up_axis
self.length_mode = length_mode
if algorithm == 'ZERO':
self.curve.pre_calc_torsion_integral(resolution)
if length_mode == 'L':
self.length_solver = SvCurveLengthSolver(curve)
self.length_solver.prepare('SPL', resolution)
self.__description__ = "Bend along {}".format(curve)
def get_matrix(self, tangent, scale):
x = Vector((1.0, 0.0, 0.0))
y = Vector((0.0, 1.0, 0.0))
z = Vector((0.0, 0.0, 1.0))
if self.axis == 0:
ax1, ax2, ax3 = x, y, z
elif self.axis == 1:
ax1, ax2, ax3 = y, x, z
else:
ax1, ax2, ax3 = z, x, y
if self.scale_all:
scale_matrix = Matrix.Scale(1 / scale, 4, ax1) @ Matrix.Scale(
scale, 4, ax2) @ Matrix.Scale(scale, 4, ax3)
else:
scale_matrix = Matrix.Scale(1 / scale, 4, ax1)
scale_matrix = np.array(scale_matrix.to_3x3())
tangent = Vector(tangent)
if self.algorithm == 'householder':
rot = autorotate_householder(ax1, tangent).inverted()
elif self.algorithm == 'track':
axis = "XYZ"[self.axis]
rot = autorotate_track(axis, tangent, self.up_axis)
elif self.algorithm == 'diff':
rot = autorotate_diff(tangent, ax1)
else:
raise Exception("Unsupported algorithm")
rot = np.array(rot.to_3x3())
return np.matmul(rot, scale_matrix)
def get_matrices(self, ts, scale):
frenet, _, _ = self.curve.frame_array(ts)
if self.scale_all:
scale_matrix = np.array([[scale, 0, 0], [0, scale, 0],
[0, 0, 1 / scale]])
else:
scale_matrix = np.array([[1, 0, 0], [0, 1, 0], [0, 0, 1 / scale]])
n = len(ts)
if self.algorithm == 'FRENET':
return frenet @ scale_matrix
elif self.algorithm == 'ZERO':
angles = -self.curve.torsion_integral(ts)
zeros = np.zeros((n, ))
ones = np.ones((n, ))
row1 = np.stack(
(np.cos(angles), np.sin(angles), zeros)).T # (n, 3)
row2 = np.stack(
(-np.sin(angles), np.cos(angles), zeros)).T # (n, 3)
row3 = np.stack((zeros, zeros, ones)).T # (n, 3)
rotation_matrices = np.dstack((row1, row2, row3))
return frenet @ rotation_matrices @ scale_matrix
else:
raise Exception("Unsupported algorithm")
def get_t_value(self, x, y, z):
curve_t_min, curve_t_max = self.curve.get_u_bounds()
t = [x, y, z][self.axis]
if self.length_mode == 'T':
t = (curve_t_max - curve_t_min) * (t - self.t_min) / (
self.t_max - self.t_min) + curve_t_min
else:
t = (t - self.t_min) / (self.t_max - self.t_min) # 0 .. 1
t = t * self.length_solver.get_total_length()
t = self.length_solver.solve(np.array([t]))[0]
return t
def get_t_values(self, xs, ys, zs):
curve_t_min, curve_t_max = self.curve.get_u_bounds()
ts = [xs, ys, zs][self.axis]
if self.length_mode == 'T':
ts = (curve_t_max - curve_t_min) * (ts - self.t_min) / (
self.t_max - self.t_min) + curve_t_min
else:
ts = (ts - self.t_min) / (self.t_max - self.t_min) # 0 .. 1
ts = ts * self.length_solver.get_total_length()
ts = self.length_solver.solve(ts)
return ts
def get_scale(self):
if self.length_mode == 'T':
curve_t_min, curve_t_max = self.curve.get_u_bounds()
t_range = curve_t_max - curve_t_min
else:
t_range = self.length_solver.get_total_length()
return (self.t_max - self.t_min) / t_range
def evaluate(self, x, y, z):
t = self.get_t_value(x, y, z)
spline_tangent = self.curve.tangent(t)
spline_vertex = self.curve.evaluate(t)
scale = self.get_scale()
if self.algorithm in {'ZERO', 'FRENET'}:
matrix = self.get_matrices(np.array([t]), scale)
else:
matrix = self.get_matrix(spline_tangent, scale)
src_vector_projection = np.array([x, y, z])
src_vector_projection[self.axis] = 0
new_vertex = np.matmul(matrix, src_vector_projection) + spline_vertex
vector = new_vertex - np.array([x, y, z])
return vector
def evaluate_grid(self, xs, ys, zs):
ts = self.get_t_values(xs, ys, zs).flatten()
spline_tangents = self.curve.tangent_array(ts)
spline_vertices = self.curve.evaluate_array(ts)
scale = self.get_scale()
if self.algorithm in {'ZERO', 'FRENET'}:
matrices = self.get_matrices(ts, scale)
else:
matrices = np.vectorize(lambda t: self.get_matrix(t, scale),
signature='(3)->(3,3)')(spline_tangents)
src_vectors = np.stack((xs, ys, zs)).T
src_vector_projections = src_vectors.copy()
src_vector_projections[:, self.axis] = 0
#matrices = matrices[np.newaxis][np.newaxis]
multiply = np.vectorize(lambda m, v: m @ v, signature='(3,3),(3)->(3)')
new_vertices = multiply(matrices,
src_vector_projections) + spline_vertices
R = (new_vertices - src_vectors).T
return R[0], R[1], R[2]