def ComputeNormal(self):
if len(self.Faces) == 0:
return
Face_Normal = [0.0, 0.0, 0.0]
for f in self.Faces:
Face_Normal = vecAdd(Face_Normal, f.normal)
# Integrity Check
self.Vert.Normal = Face_Normal
if vecNorm(self.Vert.Normal) < 0.001:
return
self.Vert.Normal = vecNormalized(self.Vert.Normal)
return
def ComputeCost(self, v):
edgelength = 1.0
if self.parent.Settings.UseEdgelength:
length = vecSub(v.Vert.Position, self.Vert.Position)
edgelength = vecNorm(length)
if (len(self.Neighbors) == len(v.Neighbors)):
same_neighbors = True
for neighbor in self.Neighbors:
if neighbor == v:
continue
if not v.IsNeighbor(neighbor):
same_neighbors = False
break
if same_neighbors:
# raw_input("ERROR: ComputeCost() same neighbors detected.")
return 999999.9
curvature = 0.001
sides = list()
for f in self.Faces:
if v.IsFace(f):
sides.append(f)
if self.parent.Settings.UseCurvature:
for f in self.Faces:
mincurv = 1.0
for s in sides:
dotproduct = vecDotProduct(f.normal, s.normal)
mincurv = min([mincurv, (1.002 - dotproduct) / 2.0])
curvature = max([curvature, mincurv])
if self.IsBorder():
WEIGHT_BORDER = 100
curvature = curvature * WEIGHT_BORDER
if self.parent.Settings.ProtectTexture:
if not self.IsSameUV(v):
WEIGHT_UV = 1.5
curvature = curvature * WEIGHT_UV
if self.parent.Settings.KeepBorder and self.IsBorder():
# raw_input("KEEP BORDER activated, Press ENTER to Continue.")
curvature = 999999.9
cost = edgelength * curvature
# print "DEBUG: ComputeCost() v[%d] to v[%d], c=%f" % (self.ID, v.ID, cost)
return cost
def ComputeNormal(self):
# v = [[], [], []]
v0 = self.vertex[0].Vert.Position
v1 = self.vertex[1].Vert.Position
v2 = self.vertex[2].Vert.Position
# v.append(self.vertex[0].Vert.Position)
# v.append(self.vertex[1].Vert.Position)
# v.append(self.vertex[2].Vert.Position)
# a = vecSub(v[1], v[0])
# b = vecSub(v[2], v[0])
a = vecSub(v1, v0)
b = vecSub(v2, v0)
_normal = self.normal
self.normal = vecCrossProduct(a, b)
if vecNorm(self.normal) < 0.001:
return
self.normal = vecNormalized(self.normal)
length = vecDotProduct(self.normal, _normal)
if length < 0:
self.normal = vecscalarMul(self.normal, -1)
return
def branchentry(self, branch):
if not isinstance(branch, CgfFormat.MeshChunk):
# keep recursing
return True
# get tangents and normals
if not (branch.normals_data and branch.tangents_data):
return True
oldtangents = [tangent for tangent in branch.tangents_data.tangents]
self.toaster.msg("recalculating new tangent space")
branch.update_tangent_space()
newtangents = [tangent for tangent in branch.tangents_data.tangents]
self.toaster.msgblockbegin("validating and checking old with new")
for norm, oldtangent, newtangent in zip(branch.normals_data.normals,
oldtangents, newtangents):
# self.toaster.msg("*** %s ***" % (norm,))
# check old
norm = (norm.x, norm.y, norm.z)
tan_norm = tuple(x / 32767.0
for x in (oldtangent[0].x, oldtangent[0].y,
oldtangent[0].z))
bin_norm = tuple(x / 32767.0
for x in (oldtangent[1].x, oldtangent[1].y,
oldtangent[1].z))
if abs(mathutils.vecNorm(norm) - 1) > self.SENSITIVITY:
self.toaster.logger.warn("normal has non-unit norm")
if abs(mathutils.vecNorm(tan_norm) - 1) > self.SENSITIVITY:
self.toaster.logger.warn("oldtangent has non-unit norm")
if abs(mathutils.vecNorm(bin_norm) - 1) > self.SENSITIVITY:
self.toaster.logger.warn("oldbinormal has non-unit norm")
if (oldtangent[0].w != oldtangent[1].w):
raise ValueError(
"inconsistent oldtangent w coordinate (%i != %i)" %
(oldtangent[0].w, oldtangent[1].w))
if not (oldtangent[0].w in (-32767, 32767)):
raise ValueError("invalid oldtangent w coordinate (%i)" %
oldtangent[0].w)
if oldtangent[0].w > 0:
cross = mathutils.vecCrossProduct(tan_norm, bin_norm)
else:
cross = mathutils.vecCrossProduct(bin_norm, tan_norm)
crossnorm = mathutils.vecNorm(cross)
if abs(crossnorm - 1) > self.SENSITIVITY:
# a lot of these...
self.toaster.logger.warn(
"tan_norm and bin_norm not orthogonal")
self.toaster.logger.warn("%s %s" % (tan_norm, bin_norm))
self.toaster.logger.warn("(error is %f)" % abs(crossnorm - 1))
cross = mathutils.vecscalarMul(cross, 1.0 / crossnorm)
if mathutils.vecDistance(norm, cross) > self.SENSITIVITY:
self.toaster.logger.warn(
"norm not cross product of tangent and binormal")
# self.toaster.logger.warn("norm = %s" % (norm,))
# self.toaster.logger.warn("tan_norm = %s" % (tan_norm,))
# self.toaster.logger.warn("bin_norm = %s" % (bin_norm,))
# self.toaster.logger.warn("tan_norm bin_norm cross prod = %s" % (cross,))
self.toaster.logger.warn("(error is %f)" %
mathutils.vecDistance(norm, cross))
# compare old with new
if sum((abs(oldtangent[0].x - newtangent[0].x),
abs(oldtangent[0].y - newtangent[0].y),
abs(oldtangent[0].z - newtangent[0].z),
abs(oldtangent[0].w - newtangent[0].w),
abs(oldtangent[1].x - newtangent[1].x),
abs(oldtangent[1].y - newtangent[1].y),
abs(oldtangent[1].z - newtangent[1].z),
abs(oldtangent[1].w -
newtangent[1].w))) > self.SENSITIVITY * 32767.0:
ntan = tuple(x / 32767.0
for x in (newtangent[0].x, newtangent[0].y,
newtangent[0].z))
nbin = tuple(x / 32767.0
for x in (newtangent[1].x, newtangent[1].y,
newtangent[1].z))
self.toaster.logger.warn(
"old and new tangents differ substantially")
self.toaster.logger.warn("old tangent")
self.toaster.logger.warn("%s %s" % (tan_norm, bin_norm))
self.toaster.logger.warn("new tangent")
self.toaster.logger.warn("%s %s" % (ntan, nbin))
self.toaster.msgblockend()
def getTangentSpace(vertices=None,
normals=None,
uvs=None,
triangles=None,
orientation=False,
orthogonal=True):
"""Calculate tangent space data.
>>> vertices = [(0,0,0), (0,1,0), (1,0,0)]
>>> normals = [(0,0,1), (0,0,1), (0,0,1)]
>>> uvs = [(0,0), (0,1), (1,0)]
>>> triangles = [(0,1,2)]
>>> getTangentSpace(vertices = vertices, normals = normals, uvs = uvs, triangles = triangles)
([(0.0, 1.0, 0.0), (0.0, 1.0, 0.0), (0.0, 1.0, 0.0)], [(1.0, 0.0, 0.0), (1.0, 0.0, 0.0), (1.0, 0.0, 0.0)])
:param vertices: A list of vertices (triples of floats/ints).
:param normals: A list of normals (triples of floats/ints).
:param uvs: A list of uvs (pairs of floats/ints).
:param triangles: A list of triangle indices (triples of ints).
:param orientation: Set to ``True`` to return orientation (this is used by
for instance Crysis).
:return: Two lists of vectors, tangents and binormals. If C{orientation}
is ``True``, then returns an extra list with orientations (containing
floats which describe the total signed surface of all faces sharing
the particular vertex).
"""
# validate input
if len(vertices) != len(normals) or len(vertices) != len(uvs):
raise ValueError(
"lists of vertices, normals, and uvs must have the same length")
bin_norm = [(0, 0, 0) for i in range(len(vertices))]
tan_norm = [(0, 0, 0) for i in range(len(vertices))]
orientations = [0 for i in range(len(vertices))]
# calculate tangents and binormals from vertex and texture coordinates
for t1, t2, t3 in triangles:
# skip degenerate triangles
if t1 == t2 or t2 == t3 or t3 == t1:
continue
# get vertices, uvs, and directions of the triangle
v1 = vertices[t1]
v2 = vertices[t2]
v3 = vertices[t3]
w1 = uvs[t1]
w2 = uvs[t2]
w3 = uvs[t3]
v2v1 = mathutils.vecSub(v2, v1)
v3v1 = mathutils.vecSub(v3, v1)
w2w1 = mathutils.vecSub(w2, w1)
w3w1 = mathutils.vecSub(w3, w1)
# surface of triangle in texture space
r = w2w1[0] * w3w1[1] - w3w1[0] * w2w1[1]
# sign of surface
r_sign = (1 if r >= 0 else -1)
# contribution of this triangle to tangents and binormals
sdir = (r_sign * (w3w1[1] * v2v1[0] - w2w1[1] * v3v1[0]),
r_sign * (w3w1[1] * v2v1[1] - w2w1[1] * v3v1[1]),
r_sign * (w3w1[1] * v2v1[2] - w2w1[1] * v3v1[2]))
try:
sdir = mathutils.vecNormalized(sdir)
except ZeroDivisionError: # catches zero vector
continue # skip triangle
except ValueError: # catches invalid data
continue # skip triangle
tdir = (r_sign * (w2w1[0] * v3v1[0] - w3w1[0] * v2v1[0]),
r_sign * (w2w1[0] * v3v1[1] - w3w1[0] * v2v1[1]),
r_sign * (w2w1[0] * v3v1[2] - w3w1[0] * v2v1[2]))
try:
tdir = mathutils.vecNormalized(tdir)
except ZeroDivisionError: # catches zero vector
continue # skip triangle
except ValueError: # catches invalid data
continue # skip triangle
# vector combination algorithm could possibly be improved
for i in (t1, t2, t3):
tan_norm[i] = mathutils.vecAdd(tan_norm[i], tdir)
bin_norm[i] = mathutils.vecAdd(bin_norm[i], sdir)
orientations[i] += r
# convert into orthogonal space
xvec = (1, 0, 0)
yvec = (0, 1, 0)
for i, norm in enumerate(normals):
if abs(1 - mathutils.vecNorm(norm)) > 0.01:
raise ValueError(
"tangentspace: unnormalized normal in list of normals (%s, norm is %f)"
% (norm, mathutils.vecNorm(norm)))
try:
# turn norm, bin_norm, tan_norm into a base via Gram-Schmidt
bin_norm[i] = mathutils.vecSub(
bin_norm[i],
mathutils.vecscalarMul(
norm, mathutils.vecDotProduct(norm, bin_norm[i])))
bin_norm[i] = mathutils.vecNormalized(bin_norm[i])
tan_norm[i] = mathutils.vecSub(
tan_norm[i],
mathutils.vecscalarMul(
norm, mathutils.vecDotProduct(norm, tan_norm[i])))
tan_norm[i] = mathutils.vecSub(
tan_norm[i],
mathutils.vecscalarMul(
bin_norm[i], mathutils.vecDotProduct(norm, bin_norm[i])))
tan_norm[i] = mathutils.vecNormalized(tan_norm[i])
except ZeroDivisionError:
# insuffient data to set tangent space for this vertex
# in that case pick a space
bin_norm[i] = mathutils.vecCrossProduct(xvec, norm)
try:
bin_norm[i] = mathutils.vecNormalized(bin_norm[i])
except ZeroDivisionError:
bin_norm[i] = mathutils.vecCrossProduct(yvec, norm)
bin_norm[i] = mathutils.vecNormalized(bin_norm[i])
tan_norm[i] = mathutils.vecCrossProduct(norm, bin_norm[i])
# return result
if orientation:
return tan_norm, bin_norm, orientations
else:
return tan_norm, bin_norm
def get_mass_center_inertia_polyhedron(vertices,
triangles,
density=1,
solid=True):
"""Return mass, center of gravity, and inertia matrix for a polyhedron.
>>> from pyffi.utils.quickhull import qhull3d
>>> box = [(0,0,0),(1,0,0),(0,2,0),(0,0,3),(1,2,0),(0,2,3),(1,0,3),(1,2,3)]
>>> vertices, triangles = qhull3d(box)
>>> mass, center, inertia = get_mass_center_inertia_polyhedron(
... vertices, triangles, density = 4)
>>> mass
24.0
>>> center
(0.5, 1.0, 1.5)
>>> inertia
((26.0, 0.0, 0.0), (0.0, 20.0, 0.0), (0.0, 0.0, 10.0))
>>> poly = [(3,0,0),(0,3,0),(-3,0,0),(0,-3,0),(0,0,3),(0,0,-3)] # very rough approximation of a sphere of radius 2
>>> vertices, triangles = qhull3d(poly)
>>> mass, center, inertia = get_mass_center_inertia_polyhedron(
... vertices, triangles, density = 3)
>>> mass
108.0
>>> center
(0.0, 0.0, 0.0)
>>> abs(inertia[0][0] - 194.4) < 0.0001
True
>>> abs(inertia[1][1] - 194.4) < 0.0001
True
>>> abs(inertia[2][2] - 194.4) < 0.0001
True
>>> abs(inertia[0][1]) < 0.0001
True
>>> abs(inertia[0][2]) < 0.0001
True
>>> abs(inertia[1][2]) < 0.0001
True
>>> sphere = []
>>> N = 10
>>> for j in range(-N+1, N):
... theta = j * 0.5 * math.pi / N
... st, ct = math.sin(theta), math.cos(theta)
... M = max(3, int(ct * 2 * N + 0.5))
... for i in range(0, M):
... phi = i * 2 * math.pi / M
... s, c = math.sin(phi), math.cos(phi)
... sphere.append((2*s*ct, 2*c*ct, 2*st)) # construct sphere of radius 2
>>> sphere.append((0,0,2))
>>> sphere.append((0,0,-2))
>>> vertices, triangles = qhull3d(sphere)
>>> mass, center, inertia = get_mass_center_inertia_polyhedron(
... vertices, triangles, density = 3, solid = True)
>>> abs(mass - 100.53) < 10 # 3*(4/3)*pi*2^3 = 100.53
True
>>> sum(abs(x) for x in center) < 0.01 # is center at origin?
True
>>> abs(inertia[0][0] - 160.84) < 10
True
>>> mass, center, inertia = get_mass_center_inertia_polyhedron(
... vertices, triangles, density = 3, solid = False)
>>> abs(mass - 150.79) < 10 # 3*4*pi*2^2 = 150.79
True
>>> abs(inertia[0][0] - mass*0.666*4) < 20 # m*(2/3)*2^2
True
"""
# 120 times the covariance matrix of the canonical tetrahedron
# (0,0,0),(1,0,0),(0,1,0),(0,0,1)
# integrate(integrate(integrate(z*z, x=0..1-y-z), y=0..1-z), z=0..1) = 1/120
# integrate(integrate(integrate(y*z, x=0..1-y-z), y=0..1-z), z=0..1) = 1/60
covariance_canonical = ((2, 1, 1), (1, 2, 1), (1, 1, 2))
covariance_correction = 1.0 / 120
covariances = []
masses = []
centers = []
# for each triangle
# construct a tetrahedron from triangle + (0,0,0)
# find its matrix, mass, and center (for density = 1, will be corrected at
# the end of the algorithm)
for triangle in triangles:
# get vertices
vert0, vert1, vert2 = mathutils.operator.itemgetter(
*triangle)(vertices)
# construct a transform matrix that converts the canonical tetrahedron
# into (0,0,0),vert0,vert1,vert2
transform_transposed = (vert0, vert1, vert2)
transform = mathutils.matTransposed(transform_transposed)
# find the covariance matrix of the transformed tetrahedron/triangle
if solid:
# we shall be needing the determinant more than once, so
# precalculate it
determinant = mathutils.matDeterminant(transform)
# C' = det(A) * A * C * A^T
covariances.append(
mathutils.matscalarMul(
mathutils.matMul(
mathutils.matMul(transform, covariance_canonical),
transform_transposed), determinant))
# m = det(A) / 6.0
masses.append(determinant / 6.0)
# find center of gravity of the tetrahedron
centers.append(
tuple(0.25 * sum(vert[i] for vert in (vert0, vert1, vert2))
for i in range(3)))
else:
# find center of gravity of the triangle
centers.append(
tuple(
sum(vert[i] for vert in (vert0, vert1, vert2)) / 3.0
for i in range(3)))
# find mass of triangle
# mass is surface, which is half the norm of cross product
# of two edges
masses.append(
mathutils.vecNorm(
mathutils.vecCrossProduct(mathutils.vecSub(
vert1, vert0), mathutils.vecSub(vert2, vert0))) / 2.0)
# find covariance at center of this triangle
# (this is approximate only as it replaces triangle with point mass
# todo: find better way)
covariances.append(
tuple(
tuple(masses[-1] * x * y for x in centers[-1])
for y in centers[-1]))
# accumulate the results
total_mass = sum(masses)
if total_mass == 0:
# dimension is probably badly chosen
# raise ZeroDivisionError("mass is zero (consider calculating inertia with a lower dimension)")
print("WARNING: mass is nearly zero (%f)" % total_mass)
return 0, (0, 0, 0), ((0, 0, 0), (0, 0, 0), (0, 0, 0))
# weighed average of centers with masses
total_center = (0, 0, 0)
for center, mass in zip(centers, masses):
total_center = mathutils.vecAdd(
total_center, mathutils.vecscalarMul(center, mass / total_mass))
# add covariances, and correct the values
total_covariance = ((0, 0, 0), (0, 0, 0), (0, 0, 0))
for covariance in covariances:
total_covariance = mathutils.matAdd(total_covariance, covariance)
if solid:
total_covariance = mathutils.matscalarMul(total_covariance,
covariance_correction)
# translate covariance to center of gravity:
# C' = C - m * ( x dx^T + dx x^T + dx dx^T )
# with x the translation vector and dx the center of gravity
translate_correction = mathutils.matscalarMul(
tuple(tuple(x * y for x in total_center) for y in total_center),
total_mass)
total_covariance = mathutils.matSub(total_covariance, translate_correction)
# convert covariance matrix into inertia tensor
trace = sum(total_covariance[i][i] for i in range(3))
trace_matrix = tuple(
tuple((trace if i == j else 0) for i in range(3)) for j in range(3))
total_inertia = mathutils.matSub(trace_matrix, total_covariance)
# correct for given density
total_inertia = mathutils.matscalarMul(total_inertia, density)
total_mass *= density
# correct negative mass
if total_mass < 0:
total_mass = -total_mass
total_inertia = tuple(tuple(-x for x in row) for row in total_inertia)
return total_mass, total_center, total_inertia