Subversion Repositories Kolibri OS

Rev

Go to most recent revision | Details | Last modification | View Log | RSS feed

Rev Author Line No. Line
1891 serge 1
/* -*- Mode: c; c-basic-offset: 4; tab-width: 8; indent-tabs-mode: t; -*- */
2
/*
3
 *
4
 * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc.
5
 * Copyright © 2000 SuSE, Inc.
6
 *             2005 Lars Knoll & Zack Rusin, Trolltech
7
 * Copyright © 2007 Red Hat, Inc.
8
 *
9
 *
10
 * Permission to use, copy, modify, distribute, and sell this software and its
11
 * documentation for any purpose is hereby granted without fee, provided that
12
 * the above copyright notice appear in all copies and that both that
13
 * copyright notice and this permission notice appear in supporting
14
 * documentation, and that the name of Keith Packard not be used in
15
 * advertising or publicity pertaining to distribution of the software without
16
 * specific, written prior permission.  Keith Packard makes no
17
 * representations about the suitability of this software for any purpose.  It
18
 * is provided "as is" without express or implied warranty.
19
 *
20
 * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
21
 * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
22
 * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
23
 * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
24
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
25
 * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
26
 * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
27
 * SOFTWARE.
28
 */
29
 
30
#ifdef HAVE_CONFIG_H
31
#include 
32
#endif
33
#include 
34
#include 
35
#include "pixman-private.h"
36
 
37
static inline pixman_fixed_32_32_t
38
dot (pixman_fixed_48_16_t x1,
39
     pixman_fixed_48_16_t y1,
40
     pixman_fixed_48_16_t z1,
41
     pixman_fixed_48_16_t x2,
42
     pixman_fixed_48_16_t y2,
43
     pixman_fixed_48_16_t z2)
44
{
45
    /*
46
     * Exact computation, assuming that the input values can
47
     * be represented as pixman_fixed_16_16_t
48
     */
49
    return x1 * x2 + y1 * y2 + z1 * z2;
50
}
51
 
52
static inline double
53
fdot (double x1,
54
      double y1,
55
      double z1,
56
      double x2,
57
      double y2,
58
      double z2)
59
{
60
    /*
61
     * Error can be unbound in some special cases.
62
     * Using clever dot product algorithms (for example compensated
63
     * dot product) would improve this but make the code much less
64
     * obvious
65
     */
66
    return x1 * x2 + y1 * y2 + z1 * z2;
67
}
68
 
69
static uint32_t
70
radial_compute_color (double                    a,
71
		      double                    b,
72
		      double                    c,
73
		      double                    inva,
74
		      double                    dr,
75
		      double                    mindr,
76
		      pixman_gradient_walker_t *walker,
77
		      pixman_repeat_t           repeat)
78
{
79
    /*
80
     * In this function error propagation can lead to bad results:
81
     *  - det can have an unbound error (if b*b-a*c is very small),
82
     *    potentially making it the opposite sign of what it should have been
83
     *    (thus clearing a pixel that would have been colored or vice-versa)
84
     *    or propagating the error to sqrtdet;
85
     *    if det has the wrong sign or b is very small, this can lead to bad
86
     *    results
87
     *
88
     *  - the algorithm used to compute the solutions of the quadratic
89
     *    equation is not numerically stable (but saves one division compared
90
     *    to the numerically stable one);
91
     *    this can be a problem if a*c is much smaller than b*b
92
     *
93
     *  - the above problems are worse if a is small (as inva becomes bigger)
94
     */
95
    double det;
96
 
97
    if (a == 0)
98
    {
99
	double t;
100
 
101
	if (b == 0)
102
	    return 0;
103
 
104
	t = pixman_fixed_1 / 2 * c / b;
105
	if (repeat == PIXMAN_REPEAT_NONE)
106
	{
107
	    if (0 <= t && t <= pixman_fixed_1)
108
		return _pixman_gradient_walker_pixel (walker, t);
109
	}
110
	else
111
	{
112
	    if (t * dr > mindr)
113
		return _pixman_gradient_walker_pixel (walker, t);
114
	}
115
 
116
	return 0;
117
    }
118
 
119
    det = fdot (b, a, 0, b, -c, 0);
120
    if (det >= 0)
121
    {
122
	double sqrtdet, t0, t1;
123
 
124
	sqrtdet = sqrt (det);
125
	t0 = (b + sqrtdet) * inva;
126
	t1 = (b - sqrtdet) * inva;
127
 
128
	if (repeat == PIXMAN_REPEAT_NONE)
129
	{
130
	    if (0 <= t0 && t0 <= pixman_fixed_1)
131
		return _pixman_gradient_walker_pixel (walker, t0);
132
	    else if (0 <= t1 && t1 <= pixman_fixed_1)
133
		return _pixman_gradient_walker_pixel (walker, t1);
134
	}
135
	else
136
	{
137
	    if (t0 * dr > mindr)
138
		return _pixman_gradient_walker_pixel (walker, t0);
139
	    else if (t1 * dr > mindr)
140
		return _pixman_gradient_walker_pixel (walker, t1);
141
	}
142
    }
143
 
144
    return 0;
145
}
146
 
147
static void
148
radial_gradient_get_scanline_32 (pixman_image_t *image,
149
                                 int             x,
150
                                 int             y,
151
                                 int             width,
152
                                 uint32_t *      buffer,
153
                                 const uint32_t *mask)
154
{
155
    /*
156
     * Implementation of radial gradients following the PDF specification.
157
     * See section 8.7.4.5.4 Type 3 (Radial) Shadings of the PDF Reference
158
     * Manual (PDF 32000-1:2008 at the time of this writing).
159
     *
160
     * In the radial gradient problem we are given two circles (c₁,r₁) and
161
     * (c₂,r₂) that define the gradient itself.
162
     *
163
     * Mathematically the gradient can be defined as the family of circles
164
     *
165
     *     ((1-t)·c₁ + t·(c₂), (1-t)·r₁ + t·r₂)
166
     *
167
     * excluding those circles whose radius would be < 0. When a point
168
     * belongs to more than one circle, the one with a bigger t is the only
169
     * one that contributes to its color. When a point does not belong
170
     * to any of the circles, it is transparent black, i.e. RGBA (0, 0, 0, 0).
171
     * Further limitations on the range of values for t are imposed when
172
     * the gradient is not repeated, namely t must belong to [0,1].
173
     *
174
     * The graphical result is the same as drawing the valid (radius > 0)
175
     * circles with increasing t in [-inf, +inf] (or in [0,1] if the gradient
176
     * is not repeated) using SOURCE operatior composition.
177
     *
178
     * It looks like a cone pointing towards the viewer if the ending circle
179
     * is smaller than the starting one, a cone pointing inside the page if
180
     * the starting circle is the smaller one and like a cylinder if they
181
     * have the same radius.
182
     *
183
     * What we actually do is, given the point whose color we are interested
184
     * in, compute the t values for that point, solving for t in:
185
     *
186
     *     length((1-t)·c₁ + t·(c₂) - p) = (1-t)·r₁ + t·r₂
187
     *
188
     * Let's rewrite it in a simpler way, by defining some auxiliary
189
     * variables:
190
     *
191
     *     cd = c₂ - c₁
192
     *     pd = p - c₁
193
     *     dr = r₂ - r₁
194
     *     lenght(t·cd - pd) = r₁ + t·dr
195
     *
196
     * which actually means
197
     *
198
     *     hypot(t·cdx - pdx, t·cdy - pdy) = r₁ + t·dr
199
     *
200
     * or
201
     *
202
     *     ⎷((t·cdx - pdx)² + (t·cdy - pdy)²) = r₁ + t·dr.
203
     *
204
     * If we impose (as stated earlier) that r₁ + t·dr >= 0, it becomes:
205
     *
206
     *     (t·cdx - pdx)² + (t·cdy - pdy)² = (r₁ + t·dr)²
207
     *
208
     * where we can actually expand the squares and solve for t:
209
     *
210
     *     t²cdx² - 2t·cdx·pdx + pdx² + t²cdy² - 2t·cdy·pdy + pdy² =
211
     *       = r₁² + 2·r₁·t·dr + t²·dr²
212
     *
213
     *     (cdx² + cdy² - dr²)t² - 2(cdx·pdx + cdy·pdy + r₁·dr)t +
214
     *         (pdx² + pdy² - r₁²) = 0
215
     *
216
     *     A = cdx² + cdy² - dr²
217
     *     B = pdx·cdx + pdy·cdy + r₁·dr
218
     *     C = pdx² + pdy² - r₁²
219
     *     At² - 2Bt + C = 0
220
     *
221
     * The solutions (unless the equation degenerates because of A = 0) are:
222
     *
223
     *     t = (B ± ⎷(B² - A·C)) / A
224
     *
225
     * The solution we are going to prefer is the bigger one, unless the
226
     * radius associated to it is negative (or it falls outside the valid t
227
     * range).
228
     *
229
     * Additional observations (useful for optimizations):
230
     * A does not depend on p
231
     *
232
     * A < 0 <=> one of the two circles completely contains the other one
233
     *   <=> for every p, the radiuses associated with the two t solutions
234
     *       have opposite sign
235
     */
236
 
237
    gradient_t *gradient = (gradient_t *)image;
238
    source_image_t *source = (source_image_t *)image;
239
    radial_gradient_t *radial = (radial_gradient_t *)image;
240
    uint32_t *end = buffer + width;
241
    pixman_gradient_walker_t walker;
242
    pixman_vector_t v, unit;
243
 
244
    /* reference point is the center of the pixel */
245
    v.vector[0] = pixman_int_to_fixed (x) + pixman_fixed_1 / 2;
246
    v.vector[1] = pixman_int_to_fixed (y) + pixman_fixed_1 / 2;
247
    v.vector[2] = pixman_fixed_1;
248
 
249
    _pixman_gradient_walker_init (&walker, gradient, source->common.repeat);
250
 
251
    if (source->common.transform)
252
    {
253
	if (!pixman_transform_point_3d (source->common.transform, &v))
254
	    return;
255
 
256
	unit.vector[0] = source->common.transform->matrix[0][0];
257
	unit.vector[1] = source->common.transform->matrix[1][0];
258
	unit.vector[2] = source->common.transform->matrix[2][0];
259
    }
260
    else
261
    {
262
	unit.vector[0] = pixman_fixed_1;
263
	unit.vector[1] = 0;
264
	unit.vector[2] = 0;
265
    }
266
 
267
    if (unit.vector[2] == 0 && v.vector[2] == pixman_fixed_1)
268
    {
269
	/*
270
	 * Given:
271
	 *
272
	 * t = (B ± ⎷(B² - A·C)) / A
273
	 *
274
	 * where
275
	 *
276
	 * A = cdx² + cdy² - dr²
277
	 * B = pdx·cdx + pdy·cdy + r₁·dr
278
	 * C = pdx² + pdy² - r₁²
279
	 * det = B² - A·C
280
	 *
281
	 * Since we have an affine transformation, we know that (pdx, pdy)
282
	 * increase linearly with each pixel,
283
	 *
284
	 * pdx = pdx₀ + n·ux,
285
	 * pdy = pdy₀ + n·uy,
286
	 *
287
	 * we can then express B, C and det through multiple differentiation.
288
	 */
289
	pixman_fixed_32_32_t b, db, c, dc, ddc;
290
 
291
	/* warning: this computation may overflow */
292
	v.vector[0] -= radial->c1.x;
293
	v.vector[1] -= radial->c1.y;
294
 
295
	/*
296
	 * B and C are computed and updated exactly.
297
	 * If fdot was used instead of dot, in the worst case it would
298
	 * lose 11 bits of precision in each of the multiplication and
299
	 * summing up would zero out all the bit that were preserved,
300
	 * thus making the result 0 instead of the correct one.
301
	 * This would mean a worst case of unbound relative error or
302
	 * about 2^10 absolute error
303
	 */
304
	b = dot (v.vector[0], v.vector[1], radial->c1.radius,
305
		 radial->delta.x, radial->delta.y, radial->delta.radius);
306
	db = dot (unit.vector[0], unit.vector[1], 0,
307
		  radial->delta.x, radial->delta.y, 0);
308
 
309
	c = dot (v.vector[0], v.vector[1],
310
		 -((pixman_fixed_48_16_t) radial->c1.radius),
311
		 v.vector[0], v.vector[1], radial->c1.radius);
312
	dc = dot (2 * (pixman_fixed_48_16_t) v.vector[0] + unit.vector[0],
313
		  2 * (pixman_fixed_48_16_t) v.vector[1] + unit.vector[1],
314
		  0,
315
		  unit.vector[0], unit.vector[1], 0);
316
	ddc = 2 * dot (unit.vector[0], unit.vector[1], 0,
317
		       unit.vector[0], unit.vector[1], 0);
318
 
319
	while (buffer < end)
320
	{
321
	    if (!mask || *mask++)
322
	    {
323
		*buffer = radial_compute_color (radial->a, b, c,
324
						radial->inva,
325
						radial->delta.radius,
326
						radial->mindr,
327
						&walker,
328
						source->common.repeat);
329
	    }
330
 
331
	    b += db;
332
	    c += dc;
333
	    dc += ddc;
334
	    ++buffer;
335
	}
336
    }
337
    else
338
    {
339
	/* projective */
340
	/* Warning:
341
	 * error propagation guarantees are much looser than in the affine case
342
	 */
343
	while (buffer < end)
344
	{
345
	    if (!mask || *mask++)
346
	    {
347
		if (v.vector[2] != 0)
348
		{
349
		    double pdx, pdy, invv2, b, c;
350
 
351
		    invv2 = 1. * pixman_fixed_1 / v.vector[2];
352
 
353
		    pdx = v.vector[0] * invv2 - radial->c1.x;
354
		    /*    / pixman_fixed_1 */
355
 
356
		    pdy = v.vector[1] * invv2 - radial->c1.y;
357
		    /*    / pixman_fixed_1 */
358
 
359
		    b = fdot (pdx, pdy, radial->c1.radius,
360
			      radial->delta.x, radial->delta.y,
361
			      radial->delta.radius);
362
		    /*  / pixman_fixed_1 / pixman_fixed_1 */
363
 
364
		    c = fdot (pdx, pdy, -radial->c1.radius,
365
			      pdx, pdy, radial->c1.radius);
366
		    /*  / pixman_fixed_1 / pixman_fixed_1 */
367
 
368
		    *buffer = radial_compute_color (radial->a, b, c,
369
						    radial->inva,
370
						    radial->delta.radius,
371
						    radial->mindr,
372
						    &walker,
373
						    source->common.repeat);
374
		}
375
		else
376
		{
377
		    *buffer = 0;
378
		}
379
	    }
380
 
381
	    ++buffer;
382
 
383
	    v.vector[0] += unit.vector[0];
384
	    v.vector[1] += unit.vector[1];
385
	    v.vector[2] += unit.vector[2];
386
	}
387
    }
388
}
389
 
390
static void
391
radial_gradient_property_changed (pixman_image_t *image)
392
{
393
    image->common.get_scanline_32 = radial_gradient_get_scanline_32;
394
    image->common.get_scanline_64 = _pixman_image_get_scanline_generic_64;
395
}
396
 
397
PIXMAN_EXPORT pixman_image_t *
398
pixman_image_create_radial_gradient (pixman_point_fixed_t *        inner,
399
                                     pixman_point_fixed_t *        outer,
400
                                     pixman_fixed_t                inner_radius,
401
                                     pixman_fixed_t                outer_radius,
402
                                     const pixman_gradient_stop_t *stops,
403
                                     int                           n_stops)
404
{
405
    pixman_image_t *image;
406
    radial_gradient_t *radial;
407
 
408
    image = _pixman_image_allocate ();
409
 
410
    if (!image)
411
	return NULL;
412
 
413
    radial = &image->radial;
414
 
415
    if (!_pixman_init_gradient (&radial->common, stops, n_stops))
416
    {
417
	free (image);
418
	return NULL;
419
    }
420
 
421
    image->type = RADIAL;
422
 
423
    radial->c1.x = inner->x;
424
    radial->c1.y = inner->y;
425
    radial->c1.radius = inner_radius;
426
    radial->c2.x = outer->x;
427
    radial->c2.y = outer->y;
428
    radial->c2.radius = outer_radius;
429
 
430
    /* warning: this computations may overflow */
431
    radial->delta.x = radial->c2.x - radial->c1.x;
432
    radial->delta.y = radial->c2.y - radial->c1.y;
433
    radial->delta.radius = radial->c2.radius - radial->c1.radius;
434
 
435
    /* computed exactly, then cast to double -> every bit of the double
436
       representation is correct (53 bits) */
437
    radial->a = dot (radial->delta.x, radial->delta.y, -radial->delta.radius,
438
		     radial->delta.x, radial->delta.y, radial->delta.radius);
439
    if (radial->a != 0)
440
	radial->inva = 1. * pixman_fixed_1 / radial->a;
441
 
442
    radial->mindr = -1. * pixman_fixed_1 * radial->c1.radius;
443
 
444
    image->common.property_changed = radial_gradient_property_changed;
445
 
446
    return image;
447
}
448