1 | /* |
2 | * Copyright 2011 Google Inc. |
3 | * |
4 | * Use of this source code is governed by a BSD-style license that can be |
5 | * found in the LICENSE file. |
6 | */ |
7 | |
8 | #include "src/gpu/geometry/GrPathUtils.h" |
9 | |
10 | #include "include/gpu/GrTypes.h" |
11 | #include "src/core/SkMathPriv.h" |
12 | #include "src/core/SkPointPriv.h" |
13 | |
14 | static const SkScalar gMinCurveTol = 0.0001f; |
15 | |
16 | SkScalar GrPathUtils::scaleToleranceToSrc(SkScalar devTol, |
17 | const SkMatrix& viewM, |
18 | const SkRect& pathBounds) { |
19 | // In order to tesselate the path we get a bound on how much the matrix can |
20 | // scale when mapping to screen coordinates. |
21 | SkScalar stretch = viewM.getMaxScale(); |
22 | |
23 | if (stretch < 0) { |
24 | // take worst case mapRadius amoung four corners. |
25 | // (less than perfect) |
26 | for (int i = 0; i < 4; ++i) { |
27 | SkMatrix mat; |
28 | mat.setTranslate((i % 2) ? pathBounds.fLeft : pathBounds.fRight, |
29 | (i < 2) ? pathBounds.fTop : pathBounds.fBottom); |
30 | mat.postConcat(viewM); |
31 | stretch = std::max(stretch, mat.mapRadius(SK_Scalar1)); |
32 | } |
33 | } |
34 | SkScalar srcTol = 0; |
35 | if (stretch <= 0) { |
36 | // We have degenerate bounds or some degenerate matrix. Thus we set the tolerance to be the |
37 | // max of the path pathBounds width and height. |
38 | srcTol = std::max(pathBounds.width(), pathBounds.height()); |
39 | } else { |
40 | srcTol = devTol / stretch; |
41 | } |
42 | if (srcTol < gMinCurveTol) { |
43 | srcTol = gMinCurveTol; |
44 | } |
45 | return srcTol; |
46 | } |
47 | |
48 | uint32_t GrPathUtils::quadraticPointCount(const SkPoint points[], SkScalar tol) { |
49 | // You should have called scaleToleranceToSrc, which guarantees this |
50 | SkASSERT(tol >= gMinCurveTol); |
51 | |
52 | SkScalar d = SkPointPriv::DistanceToLineSegmentBetween(points[1], points[0], points[2]); |
53 | if (!SkScalarIsFinite(d)) { |
54 | return kMaxPointsPerCurve; |
55 | } else if (d <= tol) { |
56 | return 1; |
57 | } else { |
58 | // Each time we subdivide, d should be cut in 4. So we need to |
59 | // subdivide x = log4(d/tol) times. x subdivisions creates 2^(x) |
60 | // points. |
61 | // 2^(log4(x)) = sqrt(x); |
62 | SkScalar divSqrt = SkScalarSqrt(d / tol); |
63 | if (((SkScalar)SK_MaxS32) <= divSqrt) { |
64 | return kMaxPointsPerCurve; |
65 | } else { |
66 | int temp = SkScalarCeilToInt(divSqrt); |
67 | int pow2 = GrNextPow2(temp); |
68 | // Because of NaNs & INFs we can wind up with a degenerate temp |
69 | // such that pow2 comes out negative. Also, our point generator |
70 | // will always output at least one pt. |
71 | if (pow2 < 1) { |
72 | pow2 = 1; |
73 | } |
74 | return std::min(pow2, kMaxPointsPerCurve); |
75 | } |
76 | } |
77 | } |
78 | |
79 | uint32_t GrPathUtils::generateQuadraticPoints(const SkPoint& p0, |
80 | const SkPoint& p1, |
81 | const SkPoint& p2, |
82 | SkScalar tolSqd, |
83 | SkPoint** points, |
84 | uint32_t pointsLeft) { |
85 | if (pointsLeft < 2 || |
86 | (SkPointPriv::DistanceToLineSegmentBetweenSqd(p1, p0, p2)) < tolSqd) { |
87 | (*points)[0] = p2; |
88 | *points += 1; |
89 | return 1; |
90 | } |
91 | |
92 | SkPoint q[] = { |
93 | { SkScalarAve(p0.fX, p1.fX), SkScalarAve(p0.fY, p1.fY) }, |
94 | { SkScalarAve(p1.fX, p2.fX), SkScalarAve(p1.fY, p2.fY) }, |
95 | }; |
96 | SkPoint r = { SkScalarAve(q[0].fX, q[1].fX), SkScalarAve(q[0].fY, q[1].fY) }; |
97 | |
98 | pointsLeft >>= 1; |
99 | uint32_t a = generateQuadraticPoints(p0, q[0], r, tolSqd, points, pointsLeft); |
100 | uint32_t b = generateQuadraticPoints(r, q[1], p2, tolSqd, points, pointsLeft); |
101 | return a + b; |
102 | } |
103 | |
104 | uint32_t GrPathUtils::cubicPointCount(const SkPoint points[], |
105 | SkScalar tol) { |
106 | // You should have called scaleToleranceToSrc, which guarantees this |
107 | SkASSERT(tol >= gMinCurveTol); |
108 | |
109 | SkScalar d = std::max( |
110 | SkPointPriv::DistanceToLineSegmentBetweenSqd(points[1], points[0], points[3]), |
111 | SkPointPriv::DistanceToLineSegmentBetweenSqd(points[2], points[0], points[3])); |
112 | d = SkScalarSqrt(d); |
113 | if (!SkScalarIsFinite(d)) { |
114 | return kMaxPointsPerCurve; |
115 | } else if (d <= tol) { |
116 | return 1; |
117 | } else { |
118 | SkScalar divSqrt = SkScalarSqrt(d / tol); |
119 | if (((SkScalar)SK_MaxS32) <= divSqrt) { |
120 | return kMaxPointsPerCurve; |
121 | } else { |
122 | int temp = SkScalarCeilToInt(SkScalarSqrt(d / tol)); |
123 | int pow2 = GrNextPow2(temp); |
124 | // Because of NaNs & INFs we can wind up with a degenerate temp |
125 | // such that pow2 comes out negative. Also, our point generator |
126 | // will always output at least one pt. |
127 | if (pow2 < 1) { |
128 | pow2 = 1; |
129 | } |
130 | return std::min(pow2, kMaxPointsPerCurve); |
131 | } |
132 | } |
133 | } |
134 | |
135 | uint32_t GrPathUtils::generateCubicPoints(const SkPoint& p0, |
136 | const SkPoint& p1, |
137 | const SkPoint& p2, |
138 | const SkPoint& p3, |
139 | SkScalar tolSqd, |
140 | SkPoint** points, |
141 | uint32_t pointsLeft) { |
142 | if (pointsLeft < 2 || |
143 | (SkPointPriv::DistanceToLineSegmentBetweenSqd(p1, p0, p3) < tolSqd && |
144 | SkPointPriv::DistanceToLineSegmentBetweenSqd(p2, p0, p3) < tolSqd)) { |
145 | (*points)[0] = p3; |
146 | *points += 1; |
147 | return 1; |
148 | } |
149 | SkPoint q[] = { |
150 | { SkScalarAve(p0.fX, p1.fX), SkScalarAve(p0.fY, p1.fY) }, |
151 | { SkScalarAve(p1.fX, p2.fX), SkScalarAve(p1.fY, p2.fY) }, |
152 | { SkScalarAve(p2.fX, p3.fX), SkScalarAve(p2.fY, p3.fY) } |
153 | }; |
154 | SkPoint r[] = { |
155 | { SkScalarAve(q[0].fX, q[1].fX), SkScalarAve(q[0].fY, q[1].fY) }, |
156 | { SkScalarAve(q[1].fX, q[2].fX), SkScalarAve(q[1].fY, q[2].fY) } |
157 | }; |
158 | SkPoint s = { SkScalarAve(r[0].fX, r[1].fX), SkScalarAve(r[0].fY, r[1].fY) }; |
159 | pointsLeft >>= 1; |
160 | uint32_t a = generateCubicPoints(p0, q[0], r[0], s, tolSqd, points, pointsLeft); |
161 | uint32_t b = generateCubicPoints(s, r[1], q[2], p3, tolSqd, points, pointsLeft); |
162 | return a + b; |
163 | } |
164 | |
165 | int GrPathUtils::worstCasePointCount(const SkPath& path, int* subpaths, SkScalar tol) { |
166 | // You should have called scaleToleranceToSrc, which guarantees this |
167 | SkASSERT(tol >= gMinCurveTol); |
168 | |
169 | int pointCount = 0; |
170 | *subpaths = 1; |
171 | |
172 | bool first = true; |
173 | |
174 | SkPath::Iter iter(path, false); |
175 | SkPath::Verb verb; |
176 | |
177 | SkPoint pts[4]; |
178 | while ((verb = iter.next(pts)) != SkPath::kDone_Verb) { |
179 | |
180 | switch (verb) { |
181 | case SkPath::kLine_Verb: |
182 | pointCount += 1; |
183 | break; |
184 | case SkPath::kConic_Verb: { |
185 | SkScalar weight = iter.conicWeight(); |
186 | SkAutoConicToQuads converter; |
187 | const SkPoint* quadPts = converter.computeQuads(pts, weight, tol); |
188 | for (int i = 0; i < converter.countQuads(); ++i) { |
189 | pointCount += quadraticPointCount(quadPts + 2*i, tol); |
190 | } |
191 | } |
192 | case SkPath::kQuad_Verb: |
193 | pointCount += quadraticPointCount(pts, tol); |
194 | break; |
195 | case SkPath::kCubic_Verb: |
196 | pointCount += cubicPointCount(pts, tol); |
197 | break; |
198 | case SkPath::kMove_Verb: |
199 | pointCount += 1; |
200 | if (!first) { |
201 | ++(*subpaths); |
202 | } |
203 | break; |
204 | default: |
205 | break; |
206 | } |
207 | first = false; |
208 | } |
209 | return pointCount; |
210 | } |
211 | |
212 | void GrPathUtils::QuadUVMatrix::set(const SkPoint qPts[3]) { |
213 | SkMatrix m; |
214 | // We want M such that M * xy_pt = uv_pt |
215 | // We know M * control_pts = [0 1/2 1] |
216 | // [0 0 1] |
217 | // [1 1 1] |
218 | // And control_pts = [x0 x1 x2] |
219 | // [y0 y1 y2] |
220 | // [1 1 1 ] |
221 | // We invert the control pt matrix and post concat to both sides to get M. |
222 | // Using the known form of the control point matrix and the result, we can |
223 | // optimize and improve precision. |
224 | |
225 | double x0 = qPts[0].fX; |
226 | double y0 = qPts[0].fY; |
227 | double x1 = qPts[1].fX; |
228 | double y1 = qPts[1].fY; |
229 | double x2 = qPts[2].fX; |
230 | double y2 = qPts[2].fY; |
231 | double det = x0*y1 - y0*x1 + x2*y0 - y2*x0 + x1*y2 - y1*x2; |
232 | |
233 | if (!sk_float_isfinite(det) |
234 | || SkScalarNearlyZero((float)det, SK_ScalarNearlyZero * SK_ScalarNearlyZero)) { |
235 | // The quad is degenerate. Hopefully this is rare. Find the pts that are |
236 | // farthest apart to compute a line (unless it is really a pt). |
237 | SkScalar maxD = SkPointPriv::DistanceToSqd(qPts[0], qPts[1]); |
238 | int maxEdge = 0; |
239 | SkScalar d = SkPointPriv::DistanceToSqd(qPts[1], qPts[2]); |
240 | if (d > maxD) { |
241 | maxD = d; |
242 | maxEdge = 1; |
243 | } |
244 | d = SkPointPriv::DistanceToSqd(qPts[2], qPts[0]); |
245 | if (d > maxD) { |
246 | maxD = d; |
247 | maxEdge = 2; |
248 | } |
249 | // We could have a tolerance here, not sure if it would improve anything |
250 | if (maxD > 0) { |
251 | // Set the matrix to give (u = 0, v = distance_to_line) |
252 | SkVector lineVec = qPts[(maxEdge + 1)%3] - qPts[maxEdge]; |
253 | // when looking from the point 0 down the line we want positive |
254 | // distances to be to the left. This matches the non-degenerate |
255 | // case. |
256 | lineVec = SkPointPriv::MakeOrthog(lineVec, SkPointPriv::kLeft_Side); |
257 | // first row |
258 | fM[0] = 0; |
259 | fM[1] = 0; |
260 | fM[2] = 0; |
261 | // second row |
262 | fM[3] = lineVec.fX; |
263 | fM[4] = lineVec.fY; |
264 | fM[5] = -lineVec.dot(qPts[maxEdge]); |
265 | } else { |
266 | // It's a point. It should cover zero area. Just set the matrix such |
267 | // that (u, v) will always be far away from the quad. |
268 | fM[0] = 0; fM[1] = 0; fM[2] = 100.f; |
269 | fM[3] = 0; fM[4] = 0; fM[5] = 100.f; |
270 | } |
271 | } else { |
272 | double scale = 1.0/det; |
273 | |
274 | // compute adjugate matrix |
275 | double a2, a3, a4, a5, a6, a7, a8; |
276 | a2 = x1*y2-x2*y1; |
277 | |
278 | a3 = y2-y0; |
279 | a4 = x0-x2; |
280 | a5 = x2*y0-x0*y2; |
281 | |
282 | a6 = y0-y1; |
283 | a7 = x1-x0; |
284 | a8 = x0*y1-x1*y0; |
285 | |
286 | // this performs the uv_pts*adjugate(control_pts) multiply, |
287 | // then does the scale by 1/det afterwards to improve precision |
288 | m[SkMatrix::kMScaleX] = (float)((0.5*a3 + a6)*scale); |
289 | m[SkMatrix::kMSkewX] = (float)((0.5*a4 + a7)*scale); |
290 | m[SkMatrix::kMTransX] = (float)((0.5*a5 + a8)*scale); |
291 | |
292 | m[SkMatrix::kMSkewY] = (float)(a6*scale); |
293 | m[SkMatrix::kMScaleY] = (float)(a7*scale); |
294 | m[SkMatrix::kMTransY] = (float)(a8*scale); |
295 | |
296 | // kMPersp0 & kMPersp1 should algebraically be zero |
297 | m[SkMatrix::kMPersp0] = 0.0f; |
298 | m[SkMatrix::kMPersp1] = 0.0f; |
299 | m[SkMatrix::kMPersp2] = (float)((a2 + a5 + a8)*scale); |
300 | |
301 | // It may not be normalized to have 1.0 in the bottom right |
302 | float m33 = m.get(SkMatrix::kMPersp2); |
303 | if (1.f != m33) { |
304 | m33 = 1.f / m33; |
305 | fM[0] = m33 * m.get(SkMatrix::kMScaleX); |
306 | fM[1] = m33 * m.get(SkMatrix::kMSkewX); |
307 | fM[2] = m33 * m.get(SkMatrix::kMTransX); |
308 | fM[3] = m33 * m.get(SkMatrix::kMSkewY); |
309 | fM[4] = m33 * m.get(SkMatrix::kMScaleY); |
310 | fM[5] = m33 * m.get(SkMatrix::kMTransY); |
311 | } else { |
312 | fM[0] = m.get(SkMatrix::kMScaleX); |
313 | fM[1] = m.get(SkMatrix::kMSkewX); |
314 | fM[2] = m.get(SkMatrix::kMTransX); |
315 | fM[3] = m.get(SkMatrix::kMSkewY); |
316 | fM[4] = m.get(SkMatrix::kMScaleY); |
317 | fM[5] = m.get(SkMatrix::kMTransY); |
318 | } |
319 | } |
320 | } |
321 | |
322 | //////////////////////////////////////////////////////////////////////////////// |
323 | |
324 | // k = (y2 - y0, x0 - x2, x2*y0 - x0*y2) |
325 | // l = (y1 - y0, x0 - x1, x1*y0 - x0*y1) * 2*w |
326 | // m = (y2 - y1, x1 - x2, x2*y1 - x1*y2) * 2*w |
327 | void GrPathUtils::getConicKLM(const SkPoint p[3], const SkScalar weight, SkMatrix* out) { |
328 | SkMatrix& klm = *out; |
329 | const SkScalar w2 = 2.f * weight; |
330 | klm[0] = p[2].fY - p[0].fY; |
331 | klm[1] = p[0].fX - p[2].fX; |
332 | klm[2] = p[2].fX * p[0].fY - p[0].fX * p[2].fY; |
333 | |
334 | klm[3] = w2 * (p[1].fY - p[0].fY); |
335 | klm[4] = w2 * (p[0].fX - p[1].fX); |
336 | klm[5] = w2 * (p[1].fX * p[0].fY - p[0].fX * p[1].fY); |
337 | |
338 | klm[6] = w2 * (p[2].fY - p[1].fY); |
339 | klm[7] = w2 * (p[1].fX - p[2].fX); |
340 | klm[8] = w2 * (p[2].fX * p[1].fY - p[1].fX * p[2].fY); |
341 | |
342 | // scale the max absolute value of coeffs to 10 |
343 | SkScalar scale = 0.f; |
344 | for (int i = 0; i < 9; ++i) { |
345 | scale = std::max(scale, SkScalarAbs(klm[i])); |
346 | } |
347 | SkASSERT(scale > 0.f); |
348 | scale = 10.f / scale; |
349 | for (int i = 0; i < 9; ++i) { |
350 | klm[i] *= scale; |
351 | } |
352 | } |
353 | |
354 | //////////////////////////////////////////////////////////////////////////////// |
355 | |
356 | namespace { |
357 | |
358 | // a is the first control point of the cubic. |
359 | // ab is the vector from a to the second control point. |
360 | // dc is the vector from the fourth to the third control point. |
361 | // d is the fourth control point. |
362 | // p is the candidate quadratic control point. |
363 | // this assumes that the cubic doesn't inflect and is simple |
364 | bool is_point_within_cubic_tangents(const SkPoint& a, |
365 | const SkVector& ab, |
366 | const SkVector& dc, |
367 | const SkPoint& d, |
368 | SkPathPriv::FirstDirection dir, |
369 | const SkPoint p) { |
370 | SkVector ap = p - a; |
371 | SkScalar apXab = ap.cross(ab); |
372 | if (SkPathPriv::kCW_FirstDirection == dir) { |
373 | if (apXab > 0) { |
374 | return false; |
375 | } |
376 | } else { |
377 | SkASSERT(SkPathPriv::kCCW_FirstDirection == dir); |
378 | if (apXab < 0) { |
379 | return false; |
380 | } |
381 | } |
382 | |
383 | SkVector dp = p - d; |
384 | SkScalar dpXdc = dp.cross(dc); |
385 | if (SkPathPriv::kCW_FirstDirection == dir) { |
386 | if (dpXdc < 0) { |
387 | return false; |
388 | } |
389 | } else { |
390 | SkASSERT(SkPathPriv::kCCW_FirstDirection == dir); |
391 | if (dpXdc > 0) { |
392 | return false; |
393 | } |
394 | } |
395 | return true; |
396 | } |
397 | |
398 | void convert_noninflect_cubic_to_quads(const SkPoint p[4], |
399 | SkScalar toleranceSqd, |
400 | SkTArray<SkPoint, true>* quads, |
401 | int sublevel = 0, |
402 | bool preserveFirstTangent = true, |
403 | bool preserveLastTangent = true) { |
404 | // Notation: Point a is always p[0]. Point b is p[1] unless p[1] == p[0], in which case it is |
405 | // p[2]. Point d is always p[3]. Point c is p[2] unless p[2] == p[3], in which case it is p[1]. |
406 | SkVector ab = p[1] - p[0]; |
407 | SkVector dc = p[2] - p[3]; |
408 | |
409 | if (SkPointPriv::LengthSqd(ab) < SK_ScalarNearlyZero) { |
410 | if (SkPointPriv::LengthSqd(dc) < SK_ScalarNearlyZero) { |
411 | SkPoint* degQuad = quads->push_back_n(3); |
412 | degQuad[0] = p[0]; |
413 | degQuad[1] = p[0]; |
414 | degQuad[2] = p[3]; |
415 | return; |
416 | } |
417 | ab = p[2] - p[0]; |
418 | } |
419 | if (SkPointPriv::LengthSqd(dc) < SK_ScalarNearlyZero) { |
420 | dc = p[1] - p[3]; |
421 | } |
422 | |
423 | static const SkScalar kLengthScale = 3 * SK_Scalar1 / 2; |
424 | static const int kMaxSubdivs = 10; |
425 | |
426 | ab.scale(kLengthScale); |
427 | dc.scale(kLengthScale); |
428 | |
429 | // c0 and c1 are extrapolations along vectors ab and dc. |
430 | SkPoint c0 = p[0] + ab; |
431 | SkPoint c1 = p[3] + dc; |
432 | |
433 | SkScalar dSqd = sublevel > kMaxSubdivs ? 0 : SkPointPriv::DistanceToSqd(c0, c1); |
434 | if (dSqd < toleranceSqd) { |
435 | SkPoint newC; |
436 | if (preserveFirstTangent == preserveLastTangent) { |
437 | // We used to force a split when both tangents need to be preserved and c0 != c1. |
438 | // This introduced a large performance regression for tiny paths for no noticeable |
439 | // quality improvement. However, we aren't quite fulfilling our contract of guaranteeing |
440 | // the two tangent vectors and this could introduce a missed pixel in |
441 | // GrAAHairlinePathRenderer. |
442 | newC = (c0 + c1) * 0.5f; |
443 | } else if (preserveFirstTangent) { |
444 | newC = c0; |
445 | } else { |
446 | newC = c1; |
447 | } |
448 | |
449 | SkPoint* pts = quads->push_back_n(3); |
450 | pts[0] = p[0]; |
451 | pts[1] = newC; |
452 | pts[2] = p[3]; |
453 | return; |
454 | } |
455 | SkPoint choppedPts[7]; |
456 | SkChopCubicAtHalf(p, choppedPts); |
457 | convert_noninflect_cubic_to_quads( |
458 | choppedPts + 0, toleranceSqd, quads, sublevel + 1, preserveFirstTangent, false); |
459 | convert_noninflect_cubic_to_quads( |
460 | choppedPts + 3, toleranceSqd, quads, sublevel + 1, false, preserveLastTangent); |
461 | } |
462 | |
463 | void convert_noninflect_cubic_to_quads_with_constraint(const SkPoint p[4], |
464 | SkScalar toleranceSqd, |
465 | SkPathPriv::FirstDirection dir, |
466 | SkTArray<SkPoint, true>* quads, |
467 | int sublevel = 0) { |
468 | // Notation: Point a is always p[0]. Point b is p[1] unless p[1] == p[0], in which case it is |
469 | // p[2]. Point d is always p[3]. Point c is p[2] unless p[2] == p[3], in which case it is p[1]. |
470 | |
471 | SkVector ab = p[1] - p[0]; |
472 | SkVector dc = p[2] - p[3]; |
473 | |
474 | if (SkPointPriv::LengthSqd(ab) < SK_ScalarNearlyZero) { |
475 | if (SkPointPriv::LengthSqd(dc) < SK_ScalarNearlyZero) { |
476 | SkPoint* degQuad = quads->push_back_n(3); |
477 | degQuad[0] = p[0]; |
478 | degQuad[1] = p[0]; |
479 | degQuad[2] = p[3]; |
480 | return; |
481 | } |
482 | ab = p[2] - p[0]; |
483 | } |
484 | if (SkPointPriv::LengthSqd(dc) < SK_ScalarNearlyZero) { |
485 | dc = p[1] - p[3]; |
486 | } |
487 | |
488 | // When the ab and cd tangents are degenerate or nearly parallel with vector from d to a the |
489 | // constraint that the quad point falls between the tangents becomes hard to enforce and we are |
490 | // likely to hit the max subdivision count. However, in this case the cubic is approaching a |
491 | // line and the accuracy of the quad point isn't so important. We check if the two middle cubic |
492 | // control points are very close to the baseline vector. If so then we just pick quadratic |
493 | // points on the control polygon. |
494 | |
495 | SkVector da = p[0] - p[3]; |
496 | bool doQuads = SkPointPriv::LengthSqd(dc) < SK_ScalarNearlyZero || |
497 | SkPointPriv::LengthSqd(ab) < SK_ScalarNearlyZero; |
498 | if (!doQuads) { |
499 | SkScalar invDALengthSqd = SkPointPriv::LengthSqd(da); |
500 | if (invDALengthSqd > SK_ScalarNearlyZero) { |
501 | invDALengthSqd = SkScalarInvert(invDALengthSqd); |
502 | // cross(ab, da)^2/length(da)^2 == sqd distance from b to line from d to a. |
503 | // same goes for point c using vector cd. |
504 | SkScalar detABSqd = ab.cross(da); |
505 | detABSqd = SkScalarSquare(detABSqd); |
506 | SkScalar detDCSqd = dc.cross(da); |
507 | detDCSqd = SkScalarSquare(detDCSqd); |
508 | if (detABSqd * invDALengthSqd < toleranceSqd && |
509 | detDCSqd * invDALengthSqd < toleranceSqd) { |
510 | doQuads = true; |
511 | } |
512 | } |
513 | } |
514 | if (doQuads) { |
515 | SkPoint b = p[0] + ab; |
516 | SkPoint c = p[3] + dc; |
517 | SkPoint mid = b + c; |
518 | mid.scale(SK_ScalarHalf); |
519 | // Insert two quadratics to cover the case when ab points away from d and/or dc |
520 | // points away from a. |
521 | if (SkVector::DotProduct(da, dc) < 0 || SkVector::DotProduct(ab, da) > 0) { |
522 | SkPoint* qpts = quads->push_back_n(6); |
523 | qpts[0] = p[0]; |
524 | qpts[1] = b; |
525 | qpts[2] = mid; |
526 | qpts[3] = mid; |
527 | qpts[4] = c; |
528 | qpts[5] = p[3]; |
529 | } else { |
530 | SkPoint* qpts = quads->push_back_n(3); |
531 | qpts[0] = p[0]; |
532 | qpts[1] = mid; |
533 | qpts[2] = p[3]; |
534 | } |
535 | return; |
536 | } |
537 | |
538 | static const SkScalar kLengthScale = 3 * SK_Scalar1 / 2; |
539 | static const int kMaxSubdivs = 10; |
540 | |
541 | ab.scale(kLengthScale); |
542 | dc.scale(kLengthScale); |
543 | |
544 | // c0 and c1 are extrapolations along vectors ab and dc. |
545 | SkVector c0 = p[0] + ab; |
546 | SkVector c1 = p[3] + dc; |
547 | |
548 | SkScalar dSqd = sublevel > kMaxSubdivs ? 0 : SkPointPriv::DistanceToSqd(c0, c1); |
549 | if (dSqd < toleranceSqd) { |
550 | SkPoint cAvg = (c0 + c1) * 0.5f; |
551 | bool subdivide = false; |
552 | |
553 | if (!is_point_within_cubic_tangents(p[0], ab, dc, p[3], dir, cAvg)) { |
554 | // choose a new cAvg that is the intersection of the two tangent lines. |
555 | ab = SkPointPriv::MakeOrthog(ab); |
556 | SkScalar z0 = -ab.dot(p[0]); |
557 | dc = SkPointPriv::MakeOrthog(dc); |
558 | SkScalar z1 = -dc.dot(p[3]); |
559 | cAvg.fX = ab.fY * z1 - z0 * dc.fY; |
560 | cAvg.fY = z0 * dc.fX - ab.fX * z1; |
561 | SkScalar z = ab.fX * dc.fY - ab.fY * dc.fX; |
562 | z = SkScalarInvert(z); |
563 | cAvg.fX *= z; |
564 | cAvg.fY *= z; |
565 | if (sublevel <= kMaxSubdivs) { |
566 | SkScalar d0Sqd = SkPointPriv::DistanceToSqd(c0, cAvg); |
567 | SkScalar d1Sqd = SkPointPriv::DistanceToSqd(c1, cAvg); |
568 | // We need to subdivide if d0 + d1 > tolerance but we have the sqd values. We know |
569 | // the distances and tolerance can't be negative. |
570 | // (d0 + d1)^2 > toleranceSqd |
571 | // d0Sqd + 2*d0*d1 + d1Sqd > toleranceSqd |
572 | SkScalar d0d1 = SkScalarSqrt(d0Sqd * d1Sqd); |
573 | subdivide = 2 * d0d1 + d0Sqd + d1Sqd > toleranceSqd; |
574 | } |
575 | } |
576 | if (!subdivide) { |
577 | SkPoint* pts = quads->push_back_n(3); |
578 | pts[0] = p[0]; |
579 | pts[1] = cAvg; |
580 | pts[2] = p[3]; |
581 | return; |
582 | } |
583 | } |
584 | SkPoint choppedPts[7]; |
585 | SkChopCubicAtHalf(p, choppedPts); |
586 | convert_noninflect_cubic_to_quads_with_constraint( |
587 | choppedPts + 0, toleranceSqd, dir, quads, sublevel + 1); |
588 | convert_noninflect_cubic_to_quads_with_constraint( |
589 | choppedPts + 3, toleranceSqd, dir, quads, sublevel + 1); |
590 | } |
591 | } |
592 | |
593 | void GrPathUtils::convertCubicToQuads(const SkPoint p[4], |
594 | SkScalar tolScale, |
595 | SkTArray<SkPoint, true>* quads) { |
596 | if (!p[0].isFinite() || !p[1].isFinite() || !p[2].isFinite() || !p[3].isFinite()) { |
597 | return; |
598 | } |
599 | if (!SkScalarIsFinite(tolScale)) { |
600 | return; |
601 | } |
602 | SkPoint chopped[10]; |
603 | int count = SkChopCubicAtInflections(p, chopped); |
604 | |
605 | const SkScalar tolSqd = SkScalarSquare(tolScale); |
606 | |
607 | for (int i = 0; i < count; ++i) { |
608 | SkPoint* cubic = chopped + 3*i; |
609 | convert_noninflect_cubic_to_quads(cubic, tolSqd, quads); |
610 | } |
611 | } |
612 | |
613 | void GrPathUtils::convertCubicToQuadsConstrainToTangents(const SkPoint p[4], |
614 | SkScalar tolScale, |
615 | SkPathPriv::FirstDirection dir, |
616 | SkTArray<SkPoint, true>* quads) { |
617 | if (!p[0].isFinite() || !p[1].isFinite() || !p[2].isFinite() || !p[3].isFinite()) { |
618 | return; |
619 | } |
620 | if (!SkScalarIsFinite(tolScale)) { |
621 | return; |
622 | } |
623 | SkPoint chopped[10]; |
624 | int count = SkChopCubicAtInflections(p, chopped); |
625 | |
626 | const SkScalar tolSqd = SkScalarSquare(tolScale); |
627 | |
628 | for (int i = 0; i < count; ++i) { |
629 | SkPoint* cubic = chopped + 3*i; |
630 | convert_noninflect_cubic_to_quads_with_constraint(cubic, tolSqd, dir, quads); |
631 | } |
632 | } |
633 | |
634 | //////////////////////////////////////////////////////////////////////////////// |
635 | |
636 | using ExcludedTerm = GrPathUtils::ExcludedTerm; |
637 | |
638 | ExcludedTerm GrPathUtils::calcCubicInverseTransposePowerBasisMatrix(const SkPoint p[4], |
639 | SkMatrix* out) { |
640 | static_assert(SK_SCALAR_IS_FLOAT); |
641 | |
642 | // First convert the bezier coordinates p[0..3] to power basis coefficients X,Y(,W=[0 0 0 1]). |
643 | // M3 is the matrix that does this conversion. The homogeneous equation for the cubic becomes: |
644 | // |
645 | // | X Y 0 | |
646 | // C(t,s) = [t^3 t^2*s t*s^2 s^3] * | . . 0 | |
647 | // | . . 0 | |
648 | // | . . 1 | |
649 | // |
650 | const Sk4f M3[3] = {Sk4f(-1, 3, -3, 1), |
651 | Sk4f(3, -6, 3, 0), |
652 | Sk4f(-3, 3, 0, 0)}; |
653 | // 4th col of M3 = Sk4f(1, 0, 0, 0)}; |
654 | Sk4f X(p[3].x(), 0, 0, 0); |
655 | Sk4f Y(p[3].y(), 0, 0, 0); |
656 | for (int i = 2; i >= 0; --i) { |
657 | X += M3[i] * p[i].x(); |
658 | Y += M3[i] * p[i].y(); |
659 | } |
660 | |
661 | // The matrix is 3x4. In order to invert it, we first need to make it square by throwing out one |
662 | // of the middle two rows. We toss the row that leaves us with the largest absolute determinant. |
663 | // Since the right column will be [0 0 1], the respective determinants reduce to x0*y2 - y0*x2 |
664 | // and x0*y1 - y0*x1. |
665 | SkScalar dets[4]; |
666 | Sk4f D = SkNx_shuffle<0,0,2,1>(X) * SkNx_shuffle<2,1,0,0>(Y); |
667 | D -= SkNx_shuffle<2,3,0,1>(D); |
668 | D.store(dets); |
669 | ExcludedTerm skipTerm = SkScalarAbs(dets[0]) > SkScalarAbs(dets[1]) ? |
670 | ExcludedTerm::kQuadraticTerm : ExcludedTerm::kLinearTerm; |
671 | SkScalar det = dets[ExcludedTerm::kQuadraticTerm == skipTerm ? 0 : 1]; |
672 | if (0 == det) { |
673 | return ExcludedTerm::kNonInvertible; |
674 | } |
675 | SkScalar rdet = 1 / det; |
676 | |
677 | // Compute the inverse-transpose of the power basis matrix with the 'skipRow'th row removed. |
678 | // Since W=[0 0 0 1], it follows that our corresponding solution will be equal to: |
679 | // |
680 | // | y1 -x1 x1*y2 - y1*x2 | |
681 | // 1/det * | -y0 x0 -x0*y2 + y0*x2 | |
682 | // | 0 0 det | |
683 | // |
684 | SkScalar x[4], y[4], z[4]; |
685 | X.store(x); |
686 | Y.store(y); |
687 | (X * SkNx_shuffle<3,3,3,3>(Y) - Y * SkNx_shuffle<3,3,3,3>(X)).store(z); |
688 | |
689 | int middleRow = ExcludedTerm::kQuadraticTerm == skipTerm ? 2 : 1; |
690 | out->setAll( y[middleRow] * rdet, -x[middleRow] * rdet, z[middleRow] * rdet, |
691 | -y[0] * rdet, x[0] * rdet, -z[0] * rdet, |
692 | 0, 0, 1); |
693 | |
694 | return skipTerm; |
695 | } |
696 | |
697 | inline static void calc_serp_kcoeffs(SkScalar tl, SkScalar sl, SkScalar tm, SkScalar sm, |
698 | ExcludedTerm skipTerm, SkScalar outCoeffs[3]) { |
699 | SkASSERT(ExcludedTerm::kQuadraticTerm == skipTerm || ExcludedTerm::kLinearTerm == skipTerm); |
700 | outCoeffs[0] = 0; |
701 | outCoeffs[1] = (ExcludedTerm::kLinearTerm == skipTerm) ? sl*sm : -tl*sm - tm*sl; |
702 | outCoeffs[2] = tl*tm; |
703 | } |
704 | |
705 | inline static void calc_serp_lmcoeffs(SkScalar t, SkScalar s, ExcludedTerm skipTerm, |
706 | SkScalar outCoeffs[3]) { |
707 | SkASSERT(ExcludedTerm::kQuadraticTerm == skipTerm || ExcludedTerm::kLinearTerm == skipTerm); |
708 | outCoeffs[0] = -s*s*s; |
709 | outCoeffs[1] = (ExcludedTerm::kLinearTerm == skipTerm) ? 3*s*s*t : -3*s*t*t; |
710 | outCoeffs[2] = t*t*t; |
711 | } |
712 | |
713 | inline static void calc_loop_kcoeffs(SkScalar td, SkScalar sd, SkScalar te, SkScalar se, |
714 | SkScalar tdse, SkScalar tesd, ExcludedTerm skipTerm, |
715 | SkScalar outCoeffs[3]) { |
716 | SkASSERT(ExcludedTerm::kQuadraticTerm == skipTerm || ExcludedTerm::kLinearTerm == skipTerm); |
717 | outCoeffs[0] = 0; |
718 | outCoeffs[1] = (ExcludedTerm::kLinearTerm == skipTerm) ? sd*se : -tdse - tesd; |
719 | outCoeffs[2] = td*te; |
720 | } |
721 | |
722 | inline static void calc_loop_lmcoeffs(SkScalar t2, SkScalar s2, SkScalar t1, SkScalar s1, |
723 | SkScalar t2s1, SkScalar t1s2, ExcludedTerm skipTerm, |
724 | SkScalar outCoeffs[3]) { |
725 | SkASSERT(ExcludedTerm::kQuadraticTerm == skipTerm || ExcludedTerm::kLinearTerm == skipTerm); |
726 | outCoeffs[0] = -s2*s2*s1; |
727 | outCoeffs[1] = (ExcludedTerm::kLinearTerm == skipTerm) ? s2 * (2*t2s1 + t1s2) |
728 | : -t2 * (t2s1 + 2*t1s2); |
729 | outCoeffs[2] = t2*t2*t1; |
730 | } |
731 | |
732 | // For the case when a cubic bezier is actually a quadratic. We duplicate k in l so that the |
733 | // implicit becomes: |
734 | // |
735 | // k^3 - l*m == k^3 - l*k == k * (k^2 - l) |
736 | // |
737 | // In the quadratic case we can simply assign fixed values at each control point: |
738 | // |
739 | // | ..K.. | | pts[0] pts[1] pts[2] pts[3] | | 0 1/3 2/3 1 | |
740 | // | ..L.. | * | . . . . | == | 0 0 1/3 1 | |
741 | // | ..K.. | | 1 1 1 1 | | 0 1/3 2/3 1 | |
742 | // |
743 | static void calc_quadratic_klm(const SkPoint pts[4], double d3, SkMatrix* klm) { |
744 | SkMatrix klmAtPts; |
745 | klmAtPts.setAll(0, 1.f/3, 1, |
746 | 0, 0, 1, |
747 | 0, 1.f/3, 1); |
748 | |
749 | SkMatrix inversePts; |
750 | inversePts.setAll(pts[0].x(), pts[1].x(), pts[3].x(), |
751 | pts[0].y(), pts[1].y(), pts[3].y(), |
752 | 1, 1, 1); |
753 | SkAssertResult(inversePts.invert(&inversePts)); |
754 | |
755 | klm->setConcat(klmAtPts, inversePts); |
756 | |
757 | // If d3 > 0 we need to flip the orientation of our curve |
758 | // This is done by negating the k and l values |
759 | if (d3 > 0) { |
760 | klm->postScale(-1, -1); |
761 | } |
762 | } |
763 | |
764 | // For the case when a cubic bezier is actually a line. We set K=0, L=1, M=-line, which results in |
765 | // the following implicit: |
766 | // |
767 | // k^3 - l*m == 0^3 - 1*(-line) == -(-line) == line |
768 | // |
769 | static void calc_line_klm(const SkPoint pts[4], SkMatrix* klm) { |
770 | SkScalar ny = pts[0].x() - pts[3].x(); |
771 | SkScalar nx = pts[3].y() - pts[0].y(); |
772 | SkScalar k = nx * pts[0].x() + ny * pts[0].y(); |
773 | klm->setAll( 0, 0, 0, |
774 | 0, 0, 1, |
775 | -nx, -ny, k); |
776 | } |
777 | |
778 | SkCubicType GrPathUtils::getCubicKLM(const SkPoint src[4], SkMatrix* klm, double tt[2], |
779 | double ss[2]) { |
780 | double d[4]; |
781 | SkCubicType type = SkClassifyCubic(src, tt, ss, d); |
782 | |
783 | if (SkCubicType::kLineOrPoint == type) { |
784 | calc_line_klm(src, klm); |
785 | return SkCubicType::kLineOrPoint; |
786 | } |
787 | |
788 | if (SkCubicType::kQuadratic == type) { |
789 | calc_quadratic_klm(src, d[3], klm); |
790 | return SkCubicType::kQuadratic; |
791 | } |
792 | |
793 | SkMatrix CIT; |
794 | ExcludedTerm skipTerm = calcCubicInverseTransposePowerBasisMatrix(src, &CIT); |
795 | if (ExcludedTerm::kNonInvertible == skipTerm) { |
796 | // This could technically also happen if the curve were quadratic, but SkClassifyCubic |
797 | // should have detected that case already with tolerance. |
798 | calc_line_klm(src, klm); |
799 | return SkCubicType::kLineOrPoint; |
800 | } |
801 | |
802 | const SkScalar t0 = static_cast<SkScalar>(tt[0]), t1 = static_cast<SkScalar>(tt[1]), |
803 | s0 = static_cast<SkScalar>(ss[0]), s1 = static_cast<SkScalar>(ss[1]); |
804 | |
805 | SkMatrix klmCoeffs; |
806 | switch (type) { |
807 | case SkCubicType::kCuspAtInfinity: |
808 | SkASSERT(1 == t1 && 0 == s1); // Infinity. |
809 | // fallthru. |
810 | case SkCubicType::kLocalCusp: |
811 | case SkCubicType::kSerpentine: |
812 | calc_serp_kcoeffs(t0, s0, t1, s1, skipTerm, &klmCoeffs[0]); |
813 | calc_serp_lmcoeffs(t0, s0, skipTerm, &klmCoeffs[3]); |
814 | calc_serp_lmcoeffs(t1, s1, skipTerm, &klmCoeffs[6]); |
815 | break; |
816 | case SkCubicType::kLoop: { |
817 | const SkScalar tdse = t0 * s1; |
818 | const SkScalar tesd = t1 * s0; |
819 | calc_loop_kcoeffs(t0, s0, t1, s1, tdse, tesd, skipTerm, &klmCoeffs[0]); |
820 | calc_loop_lmcoeffs(t0, s0, t1, s1, tdse, tesd, skipTerm, &klmCoeffs[3]); |
821 | calc_loop_lmcoeffs(t1, s1, t0, s0, tesd, tdse, skipTerm, &klmCoeffs[6]); |
822 | break; |
823 | } |
824 | default: |
825 | SK_ABORT("Unexpected cubic type." ); |
826 | break; |
827 | } |
828 | |
829 | klm->setConcat(klmCoeffs, CIT); |
830 | return type; |
831 | } |
832 | |
833 | int GrPathUtils::chopCubicAtLoopIntersection(const SkPoint src[4], SkPoint dst[10], SkMatrix* klm, |
834 | int* loopIndex) { |
835 | SkSTArray<2, SkScalar> chops; |
836 | *loopIndex = -1; |
837 | |
838 | double t[2], s[2]; |
839 | if (SkCubicType::kLoop == GrPathUtils::getCubicKLM(src, klm, t, s)) { |
840 | SkScalar t0 = static_cast<SkScalar>(t[0] / s[0]); |
841 | SkScalar t1 = static_cast<SkScalar>(t[1] / s[1]); |
842 | SkASSERT(t0 <= t1); // Technically t0 != t1 in a loop, but there may be FP error. |
843 | |
844 | if (t0 < 1 && t1 > 0) { |
845 | *loopIndex = 0; |
846 | if (t0 > 0) { |
847 | chops.push_back(t0); |
848 | *loopIndex = 1; |
849 | } |
850 | if (t1 < 1) { |
851 | chops.push_back(t1); |
852 | *loopIndex = chops.count() - 1; |
853 | } |
854 | } |
855 | } |
856 | |
857 | SkChopCubicAt(src, dst, chops.begin(), chops.count()); |
858 | return chops.count() + 1; |
859 | } |
860 | |