diff options
Diffstat (limited to 'scripts/lib/fontbuild/italics2.py')
| -rw-r--r-- | scripts/lib/fontbuild/italics2.py | 148 |
1 files changed, 148 insertions, 0 deletions
diff --git a/scripts/lib/fontbuild/italics2.py b/scripts/lib/fontbuild/italics2.py new file mode 100644 index 0000000..7eee90b --- /dev/null +++ b/scripts/lib/fontbuild/italics2.py @@ -0,0 +1,148 @@ +from curveFitPen import fitGlyph,segmentGlyph +import numpy as np +from numpy.linalg import norm +import math +from scipy.sparse.linalg import cg + +def glyphToMesh(g): + points = [] + edges = {} + offset = 0 + for c in g.contours: + if len(c) < 2: + continue + for i,prev,next in rangePrevNext(len(c)): + points.append((c[i].points[0].x, c[i].points[0].y)) + edges[i + offset] = np.array([prev + offset, next + offset], dtype=int) + offset += len(c) + return np.array(points), edges + +def meshToGlyph(points, g): + g1 = g.copy() + j = 0 + for c in g1.contours: + if len(c) < 2: + continue + for i in range(len(c)): + c[i].points[0].x = points[j][0] + c[i].points[0].y = points[j][1] + j += 1 + return g1 + +def italicize(glyph, angle=12, stemWidth=180, xoffset=-50): + ga,subsegments = segmentGlyph(glyph,25) + va, e = glyphToMesh(ga) + n = len(va) + grad = mapEdges(lambda a,(p,n): normalize(p-a), va, e) + cornerWeights = mapEdges(lambda a,(p,n): normalize(p-a).dot(normalize(a-n)), grad, e)[:,0].reshape((-1,1)) + smooth = np.ones((n,1)) * .02 + smooth[cornerWeights < .6] = 5 + # smooth[cornerWeights >= .9999] = 2 + out = va.copy() + if stemWidth > 100: + out = skewMesh(poisson(skewMesh(out, angle * 2), grad, e, smooth=smooth), -angle * 2) + out = copyMeshDetails(va, out, e, 6) + # return meshToGlyph(out,ga) + normals = edgeNormals(out, e) + center = va + normals * stemWidth * .4 + if stemWidth > 100: + center[:, 0] = va[:, 0] + centerSkew = skewMesh(center.dot(np.array([[.97,0],[0,1]])), angle * .7) + # centerSkew = skewMesh(center, angle * .7) + out = out + (centerSkew - center) + out = copyMeshDetails(skewMesh(va, angle * .7), out, e, 12) + out = skewMesh(out, angle * .3) + out[:,0] += xoffset + # out[:,1] = va[:,1] + gOut = meshToGlyph(out, ga) + # gOut.width *= .97 + gOut.width += 10 + # return gOut + return fitGlyph(glyph, gOut, subsegments) + +def poisson(v, grad, e, smooth=1, P=None, distance=None): + n = len(v) + if distance == None: + distance = mapEdges(lambda a,(p,n): norm(p - a), v, e) + if (P == None): + P = mP(v,e) + P += np.identity(n) * smooth + f = v.copy() + for i,(prev,next) in e.iteritems(): + f[i] = (grad[next] * distance[next] - grad[i] * distance[i]) + out = v.copy() + f += v * smooth + for i in range(len(out[0,:])): + out[:,i] = cg(P, f[:,i])[0] + return out + +def mP(v,e): + n = len(v) + M = np.zeros((n,n)) + for i, edges in e.iteritems(): + w = -2 / float(len(edges)) + for index in edges: + M[i,index] = w + M[i,i] = 2 + return M + +def normalize(v): + n = np.linalg.norm(v) + if n == 0: + return v + return v/n + +def mapEdges(func,v,e,*args): + b = v.copy() + for i, edges in e.iteritems(): + b[i] = func(v[i], [v[j] for j in edges], *args) + return b + +def getNormal(a,b,c): + "Assumes TT winding direction" + p = np.roll(normalize(b - a), 1) + n = -np.roll(normalize(c - a), 1) + p[1] *= -1 + n[1] *= -1 + # print p, n, normalize((p + n) * .5) + return normalize((p + n) * .5) + +def edgeNormals(v,e): + "Assumes a mesh where each vertex has exactly least two edges" + return mapEdges(lambda a,(p,n) : getNormal(a,p,n),v,e) + +def rangePrevNext(count): + c = np.arange(count,dtype=int) + r = np.vstack((c, np.roll(c, 1), np.roll(c, -1))) + return r.T + +def skewMesh(v,angle): + slope = np.tanh([math.pi * angle / 180]) + return v.dot(np.array([[1,0],[slope,1]])) + + +from scipy.ndimage.filters import gaussian_filter1d as gaussian + +def labelConnected(e): + label = 0 + labels = np.zeros((len(e),1)) + for i,(prev,next) in e.iteritems(): + labels[i] = label + if next <= i: + label += 1 + return labels + +def copyGradDetails(a,b,e,scale=15): + n = len(a) + labels = labelConnected(e) + out = a.astype(float).copy() + for i in range(labels[-1]+1): + mask = (labels==i).flatten() + out[mask,:] = gaussian(b[mask,:], scale, mode="wrap", axis=0) + a[mask,:] - gaussian(a[mask,:], scale, mode="wrap", axis=0) + return out + +def copyMeshDetails(va,vb,e,scale=5,smooth=.01): + gradA = mapEdges(lambda a,(p,n): normalize(p-a), va, e) + gradB = mapEdges(lambda a,(p,n): normalize(p-a), vb, e) + grad = copyGradDetails(gradA, gradB, e, scale) + return poisson(vb, grad, e, smooth=smooth)
\ No newline at end of file |