1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
|
/*
* Copyright 2017 Google Inc.
*
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
*/
/*
* This file was autogenerated from GrCircleBlurFragmentProcessor.fp; do not modify.
*/
#include "GrCircleBlurFragmentProcessor.h"
#if SK_SUPPORT_GPU
#include "GrResourceProvider.h"
// Computes an unnormalized half kernel (right side). Returns the summation of all the half
// kernel values.
static float make_unnormalized_half_kernel(float* halfKernel, int halfKernelSize, float sigma) {
const float invSigma = 1.f / sigma;
const float b = -0.5f * invSigma * invSigma;
float tot = 0.0f;
// Compute half kernel values at half pixel steps out from the center.
float t = 0.5f;
for (int i = 0; i < halfKernelSize; ++i) {
float value = expf(t * t * b);
tot += value;
halfKernel[i] = value;
t += 1.f;
}
return tot;
}
// Create a Gaussian half-kernel (right side) and a summed area table given a sigma and number
// of discrete steps. The half kernel is normalized to sum to 0.5.
static void make_half_kernel_and_summed_table(float* halfKernel, float* summedHalfKernel,
int halfKernelSize, float sigma) {
// The half kernel should sum to 0.5 not 1.0.
const float tot = 2.f * make_unnormalized_half_kernel(halfKernel, halfKernelSize, sigma);
float sum = 0.f;
for (int i = 0; i < halfKernelSize; ++i) {
halfKernel[i] /= tot;
sum += halfKernel[i];
summedHalfKernel[i] = sum;
}
}
// Applies the 1D half kernel vertically at points along the x axis to a circle centered at the
// origin with radius circleR.
void apply_kernel_in_y(float* results, int numSteps, float firstX, float circleR,
int halfKernelSize, const float* summedHalfKernelTable) {
float x = firstX;
for (int i = 0; i < numSteps; ++i, x += 1.f) {
if (x < -circleR || x > circleR) {
results[i] = 0;
continue;
}
float y = sqrtf(circleR * circleR - x * x);
// In the column at x we exit the circle at +y and -y
// The summed table entry j is actually reflects an offset of j + 0.5.
y -= 0.5f;
int yInt = SkScalarFloorToInt(y);
SkASSERT(yInt >= -1);
if (y < 0) {
results[i] = (y + 0.5f) * summedHalfKernelTable[0];
} else if (yInt >= halfKernelSize - 1) {
results[i] = 0.5f;
} else {
float yFrac = y - yInt;
results[i] = (1.f - yFrac) * summedHalfKernelTable[yInt] +
yFrac * summedHalfKernelTable[yInt + 1];
}
}
}
// Apply a Gaussian at point (evalX, 0) to a circle centered at the origin with radius circleR.
// This relies on having a half kernel computed for the Gaussian and a table of applications of
// the half kernel in y to columns at (evalX - halfKernel, evalX - halfKernel + 1, ..., evalX +
// halfKernel) passed in as yKernelEvaluations.
static uint8_t eval_at(float evalX, float circleR, const float* halfKernel, int halfKernelSize,
const float* yKernelEvaluations) {
float acc = 0;
float x = evalX - halfKernelSize;
for (int i = 0; i < halfKernelSize; ++i, x += 1.f) {
if (x < -circleR || x > circleR) {
continue;
}
float verticalEval = yKernelEvaluations[i];
acc += verticalEval * halfKernel[halfKernelSize - i - 1];
}
for (int i = 0; i < halfKernelSize; ++i, x += 1.f) {
if (x < -circleR || x > circleR) {
continue;
}
float verticalEval = yKernelEvaluations[i + halfKernelSize];
acc += verticalEval * halfKernel[i];
}
// Since we applied a half kernel in y we multiply acc by 2 (the circle is symmetric about
// the x axis).
return SkUnitScalarClampToByte(2.f * acc);
}
// This function creates a profile of a blurred circle. It does this by computing a kernel for
// half the Gaussian and a matching summed area table. The summed area table is used to compute
// an array of vertical applications of the half kernel to the circle along the x axis. The
// table of y evaluations has 2 * k + n entries where k is the size of the half kernel and n is
// the size of the profile being computed. Then for each of the n profile entries we walk out k
// steps in each horizontal direction multiplying the corresponding y evaluation by the half
// kernel entry and sum these values to compute the profile entry.
static uint8_t* create_circle_profile(float sigma, float circleR, int profileTextureWidth) {
const int numSteps = profileTextureWidth;
uint8_t* weights = new uint8_t[numSteps];
// The full kernel is 6 sigmas wide.
int halfKernelSize = SkScalarCeilToInt(6.0f * sigma);
// round up to next multiple of 2 and then divide by 2
halfKernelSize = ((halfKernelSize + 1) & ~1) >> 1;
// Number of x steps at which to apply kernel in y to cover all the profile samples in x.
int numYSteps = numSteps + 2 * halfKernelSize;
SkAutoTArray<float> bulkAlloc(halfKernelSize + halfKernelSize + numYSteps);
float* halfKernel = bulkAlloc.get();
float* summedKernel = bulkAlloc.get() + halfKernelSize;
float* yEvals = bulkAlloc.get() + 2 * halfKernelSize;
make_half_kernel_and_summed_table(halfKernel, summedKernel, halfKernelSize, sigma);
float firstX = -halfKernelSize + 0.5f;
apply_kernel_in_y(yEvals, numYSteps, firstX, circleR, halfKernelSize, summedKernel);
for (int i = 0; i < numSteps - 1; ++i) {
float evalX = i + 0.5f;
weights[i] = eval_at(evalX, circleR, halfKernel, halfKernelSize, yEvals + i);
}
// Ensure the tail of the Gaussian goes to zero.
weights[numSteps - 1] = 0;
return weights;
}
static uint8_t* create_half_plane_profile(int profileWidth) {
SkASSERT(!(profileWidth & 0x1));
// The full kernel is 6 sigmas wide.
float sigma = profileWidth / 6.f;
int halfKernelSize = profileWidth / 2;
SkAutoTArray<float> halfKernel(halfKernelSize);
uint8_t* profile = new uint8_t[profileWidth];
// The half kernel should sum to 0.5.
const float tot = 2.f * make_unnormalized_half_kernel(halfKernel.get(), halfKernelSize, sigma);
float sum = 0.f;
// Populate the profile from the right edge to the middle.
for (int i = 0; i < halfKernelSize; ++i) {
halfKernel[halfKernelSize - i - 1] /= tot;
sum += halfKernel[halfKernelSize - i - 1];
profile[profileWidth - i - 1] = SkUnitScalarClampToByte(sum);
}
// Populate the profile from the middle to the left edge (by flipping the half kernel and
// continuing the summation).
for (int i = 0; i < halfKernelSize; ++i) {
sum += halfKernel[i];
profile[halfKernelSize - i - 1] = SkUnitScalarClampToByte(sum);
}
// Ensure tail goes to 0.
profile[profileWidth - 1] = 0;
return profile;
}
static sk_sp<GrTextureProxy> create_profile_texture(GrResourceProvider* resourceProvider,
const SkRect& circle, float sigma,
float* solidRadius, float* textureRadius) {
float circleR = circle.width() / 2.0f;
// Profile textures are cached by the ratio of sigma to circle radius and by the size of the
// profile texture (binned by powers of 2).
SkScalar sigmaToCircleRRatio = sigma / circleR;
// When sigma is really small this becomes a equivalent to convolving a Gaussian with a
// half-plane. Similarly, in the extreme high ratio cases circle becomes a point WRT to the
// Guassian and the profile texture is a just a Gaussian evaluation. However, we haven't yet
// implemented this latter optimization.
sigmaToCircleRRatio = SkTMin(sigmaToCircleRRatio, 8.f);
SkFixed sigmaToCircleRRatioFixed;
static const SkScalar kHalfPlaneThreshold = 0.1f;
bool useHalfPlaneApprox = false;
if (sigmaToCircleRRatio <= kHalfPlaneThreshold) {
useHalfPlaneApprox = true;
sigmaToCircleRRatioFixed = 0;
*solidRadius = circleR - 3 * sigma;
*textureRadius = 6 * sigma;
} else {
// Convert to fixed point for the key.
sigmaToCircleRRatioFixed = SkScalarToFixed(sigmaToCircleRRatio);
// We shave off some bits to reduce the number of unique entries. We could probably
// shave off more than we do.
sigmaToCircleRRatioFixed &= ~0xff;
sigmaToCircleRRatio = SkFixedToScalar(sigmaToCircleRRatioFixed);
sigma = circleR * sigmaToCircleRRatio;
*solidRadius = 0;
*textureRadius = circleR + 3 * sigma;
}
static const GrUniqueKey::Domain kDomain = GrUniqueKey::GenerateDomain();
GrUniqueKey key;
GrUniqueKey::Builder builder(&key, kDomain, 1);
builder[0] = sigmaToCircleRRatioFixed;
builder.finish();
sk_sp<GrTextureProxy> blurProfile =
resourceProvider->findOrCreateProxyByUniqueKey(key, kTopLeft_GrSurfaceOrigin);
if (!blurProfile) {
static constexpr int kProfileTextureWidth = 512;
GrSurfaceDesc texDesc;
texDesc.fOrigin = kTopLeft_GrSurfaceOrigin;
texDesc.fWidth = kProfileTextureWidth;
texDesc.fHeight = 1;
texDesc.fConfig = kAlpha_8_GrPixelConfig;
std::unique_ptr<uint8_t[]> profile(nullptr);
if (useHalfPlaneApprox) {
profile.reset(create_half_plane_profile(kProfileTextureWidth));
} else {
// Rescale params to the size of the texture we're creating.
SkScalar scale = kProfileTextureWidth / *textureRadius;
profile.reset(
create_circle_profile(sigma * scale, circleR * scale, kProfileTextureWidth));
}
blurProfile = GrSurfaceProxy::MakeDeferred(resourceProvider, texDesc, SkBudgeted::kYes,
profile.get(), 0);
if (!blurProfile) {
return nullptr;
}
SkASSERT(blurProfile->origin() == kTopLeft_GrSurfaceOrigin);
resourceProvider->assignUniqueKeyToProxy(key, blurProfile.get());
}
return blurProfile;
}
std::unique_ptr<GrFragmentProcessor> GrCircleBlurFragmentProcessor::Make(
GrResourceProvider* resourceProvider, const SkRect& circle, float sigma) {
float solidRadius;
float textureRadius;
sk_sp<GrTextureProxy> profile(
create_profile_texture(resourceProvider, circle, sigma, &solidRadius, &textureRadius));
if (!profile) {
return nullptr;
}
return std::unique_ptr<GrFragmentProcessor>(new GrCircleBlurFragmentProcessor(
circle, textureRadius, solidRadius, std::move(profile), resourceProvider));
}
#include "glsl/GrGLSLColorSpaceXformHelper.h"
#include "glsl/GrGLSLFragmentProcessor.h"
#include "glsl/GrGLSLFragmentShaderBuilder.h"
#include "glsl/GrGLSLProgramBuilder.h"
#include "SkSLCPP.h"
#include "SkSLUtil.h"
class GrGLSLCircleBlurFragmentProcessor : public GrGLSLFragmentProcessor {
public:
GrGLSLCircleBlurFragmentProcessor() {}
void emitCode(EmitArgs& args) override {
GrGLSLFPFragmentBuilder* fragBuilder = args.fFragBuilder;
const GrCircleBlurFragmentProcessor& _outer =
args.fFp.cast<GrCircleBlurFragmentProcessor>();
(void)_outer;
auto circleRect = _outer.circleRect();
(void)circleRect;
auto textureRadius = _outer.textureRadius();
(void)textureRadius;
auto solidRadius = _outer.solidRadius();
(void)solidRadius;
fCircleDataVar = args.fUniformHandler->addUniform(kFragment_GrShaderFlag, kHalf4_GrSLType,
kDefault_GrSLPrecision, "circleData");
fragBuilder->codeAppendf(
"half2 vec = half2(half((sk_FragCoord.x - float(%s.x)) * float(%s.w)), "
"half((sk_FragCoord.y - float(%s.y)) * float(%s.w)));\nhalf dist = "
"float(length(vec)) + (0.5 - float(%s.z)) * float(%s.w);\n%s = %s * texture(%s, "
"float2(half2(dist, 0.5))).%s.w;\n",
args.fUniformHandler->getUniformCStr(fCircleDataVar),
args.fUniformHandler->getUniformCStr(fCircleDataVar),
args.fUniformHandler->getUniformCStr(fCircleDataVar),
args.fUniformHandler->getUniformCStr(fCircleDataVar),
args.fUniformHandler->getUniformCStr(fCircleDataVar),
args.fUniformHandler->getUniformCStr(fCircleDataVar), args.fOutputColor,
args.fInputColor ? args.fInputColor : "half4(1)",
fragBuilder->getProgramBuilder()->samplerVariable(args.fTexSamplers[0]).c_str(),
fragBuilder->getProgramBuilder()->samplerSwizzle(args.fTexSamplers[0]).c_str());
}
private:
void onSetData(const GrGLSLProgramDataManager& data,
const GrFragmentProcessor& _proc) override {
const GrCircleBlurFragmentProcessor& _outer = _proc.cast<GrCircleBlurFragmentProcessor>();
auto circleRect = _outer.circleRect();
(void)circleRect;
auto textureRadius = _outer.textureRadius();
(void)textureRadius;
auto solidRadius = _outer.solidRadius();
(void)solidRadius;
UniformHandle& blurProfileSampler = fBlurProfileSamplerVar;
(void)blurProfileSampler;
UniformHandle& circleData = fCircleDataVar;
(void)circleData;
data.set4f(circleData, circleRect.centerX(), circleRect.centerY(), solidRadius,
1.f / textureRadius);
}
UniformHandle fCircleDataVar;
UniformHandle fBlurProfileSamplerVar;
};
GrGLSLFragmentProcessor* GrCircleBlurFragmentProcessor::onCreateGLSLInstance() const {
return new GrGLSLCircleBlurFragmentProcessor();
}
void GrCircleBlurFragmentProcessor::onGetGLSLProcessorKey(const GrShaderCaps& caps,
GrProcessorKeyBuilder* b) const {}
bool GrCircleBlurFragmentProcessor::onIsEqual(const GrFragmentProcessor& other) const {
const GrCircleBlurFragmentProcessor& that = other.cast<GrCircleBlurFragmentProcessor>();
(void)that;
if (fCircleRect != that.fCircleRect) return false;
if (fTextureRadius != that.fTextureRadius) return false;
if (fSolidRadius != that.fSolidRadius) return false;
if (fBlurProfileSampler != that.fBlurProfileSampler) return false;
return true;
}
GrCircleBlurFragmentProcessor::GrCircleBlurFragmentProcessor(
const GrCircleBlurFragmentProcessor& src)
: INHERITED(kGrCircleBlurFragmentProcessor_ClassID, src.optimizationFlags())
, fCircleRect(src.fCircleRect)
, fTextureRadius(src.fTextureRadius)
, fSolidRadius(src.fSolidRadius)
, fBlurProfileSampler(src.fBlurProfileSampler) {
this->addTextureSampler(&fBlurProfileSampler);
}
std::unique_ptr<GrFragmentProcessor> GrCircleBlurFragmentProcessor::clone() const {
return std::unique_ptr<GrFragmentProcessor>(new GrCircleBlurFragmentProcessor(*this));
}
GR_DEFINE_FRAGMENT_PROCESSOR_TEST(GrCircleBlurFragmentProcessor);
#if GR_TEST_UTILS
std::unique_ptr<GrFragmentProcessor> GrCircleBlurFragmentProcessor::TestCreate(
GrProcessorTestData* testData) {
SkScalar wh = testData->fRandom->nextRangeScalar(100.f, 1000.f);
SkScalar sigma = testData->fRandom->nextRangeF(1.f, 10.f);
SkRect circle = SkRect::MakeWH(wh, wh);
return GrCircleBlurFragmentProcessor::Make(testData->resourceProvider(), circle, sigma);
}
#endif
#endif
|