Libav
utils.c
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
3  *
4  * This file is part of Libav.
5  *
6  * Libav is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * Libav is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with Libav; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include "config.h"
22 
23 #define _SVID_SOURCE // needed for MAP_ANONYMOUS
24 #include <assert.h>
25 #include <inttypes.h>
26 #include <math.h>
27 #include <stdio.h>
28 #include <string.h>
29 #if HAVE_SYS_MMAN_H
30 #include <sys/mman.h>
31 #if defined(MAP_ANON) && !defined(MAP_ANONYMOUS)
32 #define MAP_ANONYMOUS MAP_ANON
33 #endif
34 #endif
35 #if HAVE_VIRTUALALLOC
36 #define WIN32_LEAN_AND_MEAN
37 #include <windows.h>
38 #endif
39 
40 #include "libavutil/attributes.h"
41 #include "libavutil/avutil.h"
42 #include "libavutil/bswap.h"
43 #include "libavutil/cpu.h"
44 #include "libavutil/intreadwrite.h"
45 #include "libavutil/mathematics.h"
46 #include "libavutil/opt.h"
47 #include "libavutil/pixdesc.h"
48 #include "libavutil/ppc/cpu.h"
49 #include "libavutil/x86/asm.h"
50 #include "libavutil/x86/cpu.h"
51 #include "rgb2rgb.h"
52 #include "swscale.h"
53 #include "swscale_internal.h"
54 
55 unsigned swscale_version(void)
56 {
58 }
59 
60 const char *swscale_configuration(void)
61 {
62  return LIBAV_CONFIGURATION;
63 }
64 
65 const char *swscale_license(void)
66 {
67 #define LICENSE_PREFIX "libswscale license: "
68  return LICENSE_PREFIX LIBAV_LICENSE + sizeof(LICENSE_PREFIX) - 1;
69 }
70 
71 #define RET 0xC3 // near return opcode for x86
72 
73 typedef struct FormatEntry {
77 } FormatEntry;
78 
80  [AV_PIX_FMT_YUV420P] = { 1, 1 },
81  [AV_PIX_FMT_YUYV422] = { 1, 1 },
82  [AV_PIX_FMT_RGB24] = { 1, 1 },
83  [AV_PIX_FMT_BGR24] = { 1, 1 },
84  [AV_PIX_FMT_YUV422P] = { 1, 1 },
85  [AV_PIX_FMT_YUV444P] = { 1, 1 },
86  [AV_PIX_FMT_YUV410P] = { 1, 1 },
87  [AV_PIX_FMT_YUV411P] = { 1, 1 },
88  [AV_PIX_FMT_GRAY8] = { 1, 1 },
89  [AV_PIX_FMT_MONOWHITE] = { 1, 1 },
90  [AV_PIX_FMT_MONOBLACK] = { 1, 1 },
91  [AV_PIX_FMT_PAL8] = { 1, 0 },
92  [AV_PIX_FMT_YUVJ420P] = { 1, 1 },
93  [AV_PIX_FMT_YUVJ422P] = { 1, 1 },
94  [AV_PIX_FMT_YUVJ444P] = { 1, 1 },
95  [AV_PIX_FMT_YVYU422] = { 1, 1 },
96  [AV_PIX_FMT_UYVY422] = { 1, 1 },
97  [AV_PIX_FMT_UYYVYY411] = { 0, 0 },
98  [AV_PIX_FMT_BGR8] = { 1, 1 },
99  [AV_PIX_FMT_BGR4] = { 0, 1 },
100  [AV_PIX_FMT_BGR4_BYTE] = { 1, 1 },
101  [AV_PIX_FMT_RGB8] = { 1, 1 },
102  [AV_PIX_FMT_RGB4] = { 0, 1 },
103  [AV_PIX_FMT_RGB4_BYTE] = { 1, 1 },
104  [AV_PIX_FMT_NV12] = { 1, 1 },
105  [AV_PIX_FMT_NV21] = { 1, 1 },
106  [AV_PIX_FMT_ARGB] = { 1, 1 },
107  [AV_PIX_FMT_RGBA] = { 1, 1 },
108  [AV_PIX_FMT_ABGR] = { 1, 1 },
109  [AV_PIX_FMT_BGRA] = { 1, 1 },
110  [AV_PIX_FMT_GRAY16BE] = { 1, 1 },
111  [AV_PIX_FMT_GRAY16LE] = { 1, 1 },
112  [AV_PIX_FMT_YUV440P] = { 1, 1 },
113  [AV_PIX_FMT_YUVJ440P] = { 1, 1 },
114  [AV_PIX_FMT_YUVA420P] = { 1, 1 },
115  [AV_PIX_FMT_YUVA422P] = { 1, 1 },
116  [AV_PIX_FMT_YUVA444P] = { 1, 1 },
117  [AV_PIX_FMT_YUVA420P9BE] = { 1, 1 },
118  [AV_PIX_FMT_YUVA420P9LE] = { 1, 1 },
119  [AV_PIX_FMT_YUVA422P9BE] = { 1, 1 },
120  [AV_PIX_FMT_YUVA422P9LE] = { 1, 1 },
121  [AV_PIX_FMT_YUVA444P9BE] = { 1, 1 },
122  [AV_PIX_FMT_YUVA444P9LE] = { 1, 1 },
123  [AV_PIX_FMT_YUVA420P10BE]= { 1, 1 },
124  [AV_PIX_FMT_YUVA420P10LE]= { 1, 1 },
125  [AV_PIX_FMT_YUVA422P10BE]= { 1, 1 },
126  [AV_PIX_FMT_YUVA422P10LE]= { 1, 1 },
127  [AV_PIX_FMT_YUVA444P10BE]= { 1, 1 },
128  [AV_PIX_FMT_YUVA444P10LE]= { 1, 1 },
129  [AV_PIX_FMT_YUVA420P16BE]= { 1, 1 },
130  [AV_PIX_FMT_YUVA420P16LE]= { 1, 1 },
131  [AV_PIX_FMT_YUVA422P16BE]= { 1, 1 },
132  [AV_PIX_FMT_YUVA422P16LE]= { 1, 1 },
133  [AV_PIX_FMT_YUVA444P16BE]= { 1, 1 },
134  [AV_PIX_FMT_YUVA444P16LE]= { 1, 1 },
135  [AV_PIX_FMT_RGB48BE] = { 1, 1 },
136  [AV_PIX_FMT_RGB48LE] = { 1, 1 },
137  [AV_PIX_FMT_RGBA64BE] = { 0, 0, 1 },
138  [AV_PIX_FMT_RGBA64LE] = { 0, 0, 1 },
139  [AV_PIX_FMT_RGB565BE] = { 1, 1 },
140  [AV_PIX_FMT_RGB565LE] = { 1, 1 },
141  [AV_PIX_FMT_RGB555BE] = { 1, 1 },
142  [AV_PIX_FMT_RGB555LE] = { 1, 1 },
143  [AV_PIX_FMT_BGR565BE] = { 1, 1 },
144  [AV_PIX_FMT_BGR565LE] = { 1, 1 },
145  [AV_PIX_FMT_BGR555BE] = { 1, 1 },
146  [AV_PIX_FMT_BGR555LE] = { 1, 1 },
147  [AV_PIX_FMT_YUV420P16LE] = { 1, 1 },
148  [AV_PIX_FMT_YUV420P16BE] = { 1, 1 },
149  [AV_PIX_FMT_YUV422P16LE] = { 1, 1 },
150  [AV_PIX_FMT_YUV422P16BE] = { 1, 1 },
151  [AV_PIX_FMT_YUV444P16LE] = { 1, 1 },
152  [AV_PIX_FMT_YUV444P16BE] = { 1, 1 },
153  [AV_PIX_FMT_RGB444LE] = { 1, 1 },
154  [AV_PIX_FMT_RGB444BE] = { 1, 1 },
155  [AV_PIX_FMT_BGR444LE] = { 1, 1 },
156  [AV_PIX_FMT_BGR444BE] = { 1, 1 },
157  [AV_PIX_FMT_YA8] = { 1, 0 },
158  [AV_PIX_FMT_YA16BE] = { 1, 0 },
159  [AV_PIX_FMT_YA16LE] = { 1, 0 },
160  [AV_PIX_FMT_BGR48BE] = { 1, 1 },
161  [AV_PIX_FMT_BGR48LE] = { 1, 1 },
162  [AV_PIX_FMT_BGRA64BE] = { 0, 0, 1 },
163  [AV_PIX_FMT_BGRA64LE] = { 0, 0, 1 },
164  [AV_PIX_FMT_YUV420P9BE] = { 1, 1 },
165  [AV_PIX_FMT_YUV420P9LE] = { 1, 1 },
166  [AV_PIX_FMT_YUV420P10BE] = { 1, 1 },
167  [AV_PIX_FMT_YUV420P10LE] = { 1, 1 },
168  [AV_PIX_FMT_YUV422P9BE] = { 1, 1 },
169  [AV_PIX_FMT_YUV422P9LE] = { 1, 1 },
170  [AV_PIX_FMT_YUV422P10BE] = { 1, 1 },
171  [AV_PIX_FMT_YUV422P10LE] = { 1, 1 },
172  [AV_PIX_FMT_YUV444P9BE] = { 1, 1 },
173  [AV_PIX_FMT_YUV444P9LE] = { 1, 1 },
174  [AV_PIX_FMT_YUV444P10BE] = { 1, 1 },
175  [AV_PIX_FMT_YUV444P10LE] = { 1, 1 },
176  [AV_PIX_FMT_GBRP] = { 1, 1 },
177  [AV_PIX_FMT_GBRP9LE] = { 1, 1 },
178  [AV_PIX_FMT_GBRP9BE] = { 1, 1 },
179  [AV_PIX_FMT_GBRP10LE] = { 1, 1 },
180  [AV_PIX_FMT_GBRP10BE] = { 1, 1 },
181  [AV_PIX_FMT_GBRP16LE] = { 1, 0 },
182  [AV_PIX_FMT_GBRP16BE] = { 1, 0 },
183  [AV_PIX_FMT_GBRAP] = { 1, 1 },
184  [AV_PIX_FMT_GBRAP16LE] = { 1, 0 },
185  [AV_PIX_FMT_GBRAP16BE] = { 1, 0 },
186  [AV_PIX_FMT_XYZ12BE] = { 0, 0, 1 },
187  [AV_PIX_FMT_XYZ12LE] = { 0, 0, 1 },
188  [AV_PIX_FMT_P010LE] = { 1, 0 },
189  [AV_PIX_FMT_P010BE] = { 1, 0 },
190 };
191 
193 {
194  return (unsigned)pix_fmt < AV_PIX_FMT_NB ?
195  format_entries[pix_fmt].is_supported_in : 0;
196 }
197 
199 {
200  return (unsigned)pix_fmt < AV_PIX_FMT_NB ?
201  format_entries[pix_fmt].is_supported_out : 0;
202 }
203 
205 {
206  return (unsigned)pix_fmt < AV_PIX_FMT_NB ?
207  format_entries[pix_fmt].is_supported_endianness : 0;
208 }
209 
211 {
213  if (desc)
214  return desc->name;
215  else
216  return "Unknown format";
217 }
218 
219 static double getSplineCoeff(double a, double b, double c, double d,
220  double dist)
221 {
222  if (dist <= 1.0)
223  return ((d * dist + c) * dist + b) * dist + a;
224  else
225  return getSplineCoeff(0.0,
226  b + 2.0 * c + 3.0 * d,
227  c + 3.0 * d,
228  -b - 3.0 * c - 6.0 * d,
229  dist - 1.0);
230 }
231 
232 static av_cold int initFilter(int16_t **outFilter, int32_t **filterPos,
233  int *outFilterSize, int xInc, int srcW,
234  int dstW, int filterAlign, int one,
235  int flags, int cpu_flags,
236  SwsVector *srcFilter, SwsVector *dstFilter,
237  double param[2], int is_horizontal)
238 {
239  int i;
240  int filterSize;
241  int filter2Size;
242  int minFilterSize;
243  int64_t *filter = NULL;
244  int64_t *filter2 = NULL;
245  const int64_t fone = 1LL << 54;
246  int ret = -1;
247 
248  emms_c(); // FIXME should not be required but IS (even for non-MMX versions)
249 
250  // NOTE: the +3 is for the MMX(+1) / SSE(+3) scaler which reads over the end
251  FF_ALLOC_OR_GOTO(NULL, *filterPos, (dstW + 3) * sizeof(**filterPos), fail);
252 
253  if (FFABS(xInc - 0x10000) < 10) { // unscaled
254  int i;
255  filterSize = 1;
256  FF_ALLOCZ_OR_GOTO(NULL, filter,
257  dstW * sizeof(*filter) * filterSize, fail);
258 
259  for (i = 0; i < dstW; i++) {
260  filter[i * filterSize] = fone;
261  (*filterPos)[i] = i;
262  }
263  } else if (flags & SWS_POINT) { // lame looking point sampling mode
264  int i;
265  int xDstInSrc;
266  filterSize = 1;
267  FF_ALLOC_OR_GOTO(NULL, filter,
268  dstW * sizeof(*filter) * filterSize, fail);
269 
270  xDstInSrc = xInc / 2 - 0x8000;
271  for (i = 0; i < dstW; i++) {
272  int xx = (xDstInSrc - ((filterSize - 1) << 15) + (1 << 15)) >> 16;
273 
274  (*filterPos)[i] = xx;
275  filter[i] = fone;
276  xDstInSrc += xInc;
277  }
278  } else if ((xInc <= (1 << 16) && (flags & SWS_AREA)) ||
279  (flags & SWS_FAST_BILINEAR)) { // bilinear upscale
280  int i;
281  int xDstInSrc;
282  filterSize = 2;
283  FF_ALLOC_OR_GOTO(NULL, filter,
284  dstW * sizeof(*filter) * filterSize, fail);
285 
286  xDstInSrc = xInc / 2 - 0x8000;
287  for (i = 0; i < dstW; i++) {
288  int xx = (xDstInSrc - ((filterSize - 1) << 15) + (1 << 15)) >> 16;
289  int j;
290 
291  (*filterPos)[i] = xx;
292  // bilinear upscale / linear interpolate / area averaging
293  for (j = 0; j < filterSize; j++) {
294  int64_t coeff = fone - FFABS((xx << 16) - xDstInSrc) *
295  (fone >> 16);
296  if (coeff < 0)
297  coeff = 0;
298  filter[i * filterSize + j] = coeff;
299  xx++;
300  }
301  xDstInSrc += xInc;
302  }
303  } else {
304  int64_t xDstInSrc;
305  int sizeFactor;
306 
307  if (flags & SWS_BICUBIC)
308  sizeFactor = 4;
309  else if (flags & SWS_X)
310  sizeFactor = 8;
311  else if (flags & SWS_AREA)
312  sizeFactor = 1; // downscale only, for upscale it is bilinear
313  else if (flags & SWS_GAUSS)
314  sizeFactor = 8; // infinite ;)
315  else if (flags & SWS_LANCZOS)
316  sizeFactor = param[0] != SWS_PARAM_DEFAULT ? ceil(2 * param[0]) : 6;
317  else if (flags & SWS_SINC)
318  sizeFactor = 20; // infinite ;)
319  else if (flags & SWS_SPLINE)
320  sizeFactor = 20; // infinite ;)
321  else if (flags & SWS_BILINEAR)
322  sizeFactor = 2;
323  else {
324  sizeFactor = 0; // GCC warning killer
325  assert(0);
326  }
327 
328  if (xInc <= 1 << 16)
329  filterSize = 1 + sizeFactor; // upscale
330  else
331  filterSize = 1 + (sizeFactor * srcW + dstW - 1) / dstW;
332 
333  filterSize = FFMIN(filterSize, srcW - 2);
334  filterSize = FFMAX(filterSize, 1);
335 
336  FF_ALLOC_OR_GOTO(NULL, filter,
337  dstW * sizeof(*filter) * filterSize, fail);
338 
339  xDstInSrc = xInc - 0x10000;
340  for (i = 0; i < dstW; i++) {
341  int xx = (xDstInSrc - ((int64_t)(filterSize - 2) << 16)) / (1 << 17);
342  int j;
343  (*filterPos)[i] = xx;
344  for (j = 0; j < filterSize; j++) {
345  int64_t d = (FFABS(((int64_t)xx << 17) - xDstInSrc)) << 13;
346  double floatd;
347  int64_t coeff;
348 
349  if (xInc > 1 << 16)
350  d = d * dstW / srcW;
351  floatd = d * (1.0 / (1 << 30));
352 
353  if (flags & SWS_BICUBIC) {
354  int64_t B = (param[0] != SWS_PARAM_DEFAULT ? param[0] : 0) * (1 << 24);
355  int64_t C = (param[1] != SWS_PARAM_DEFAULT ? param[1] : 0.6) * (1 << 24);
356 
357  if (d >= 1LL << 31) {
358  coeff = 0.0;
359  } else {
360  int64_t dd = (d * d) >> 30;
361  int64_t ddd = (dd * d) >> 30;
362 
363  if (d < 1LL << 30)
364  coeff = (12 * (1 << 24) - 9 * B - 6 * C) * ddd +
365  (-18 * (1 << 24) + 12 * B + 6 * C) * dd +
366  (6 * (1 << 24) - 2 * B) * (1 << 30);
367  else
368  coeff = (-B - 6 * C) * ddd +
369  (6 * B + 30 * C) * dd +
370  (-12 * B - 48 * C) * d +
371  (8 * B + 24 * C) * (1 << 30);
372  }
373  coeff *= fone >> (30 + 24);
374  }
375  else if (flags & SWS_X) {
376  double A = param[0] != SWS_PARAM_DEFAULT ? param[0] : 1.0;
377  double c;
378 
379  if (floatd < 1.0)
380  c = cos(floatd * M_PI);
381  else
382  c = -1.0;
383  if (c < 0.0)
384  c = -pow(-c, A);
385  else
386  c = pow(c, A);
387  coeff = (c * 0.5 + 0.5) * fone;
388  } else if (flags & SWS_AREA) {
389  int64_t d2 = d - (1 << 29);
390  if (d2 * xInc < -(1LL << (29 + 16)))
391  coeff = 1.0 * (1LL << (30 + 16));
392  else if (d2 * xInc < (1LL << (29 + 16)))
393  coeff = -d2 * xInc + (1LL << (29 + 16));
394  else
395  coeff = 0.0;
396  coeff *= fone >> (30 + 16);
397  } else if (flags & SWS_GAUSS) {
398  double p = param[0] != SWS_PARAM_DEFAULT ? param[0] : 3.0;
399  coeff = (pow(2.0, -p * floatd * floatd)) * fone;
400  } else if (flags & SWS_SINC) {
401  coeff = (d ? sin(floatd * M_PI) / (floatd * M_PI) : 1.0) * fone;
402  } else if (flags & SWS_LANCZOS) {
403  double p = param[0] != SWS_PARAM_DEFAULT ? param[0] : 3.0;
404  coeff = (d ? sin(floatd * M_PI) * sin(floatd * M_PI / p) /
405  (floatd * floatd * M_PI * M_PI / p) : 1.0) * fone;
406  if (floatd > p)
407  coeff = 0;
408  } else if (flags & SWS_BILINEAR) {
409  coeff = (1 << 30) - d;
410  if (coeff < 0)
411  coeff = 0;
412  coeff *= fone >> 30;
413  } else if (flags & SWS_SPLINE) {
414  double p = -2.196152422706632;
415  coeff = getSplineCoeff(1.0, 0.0, p, -p - 1.0, floatd) * fone;
416  } else {
417  coeff = 0.0; // GCC warning killer
418  assert(0);
419  }
420 
421  filter[i * filterSize + j] = coeff;
422  xx++;
423  }
424  xDstInSrc += 2 * xInc;
425  }
426  }
427 
428  /* apply src & dst Filter to filter -> filter2
429  * av_free(filter);
430  */
431  assert(filterSize > 0);
432  filter2Size = filterSize;
433  if (srcFilter)
434  filter2Size += srcFilter->length - 1;
435  if (dstFilter)
436  filter2Size += dstFilter->length - 1;
437  assert(filter2Size > 0);
438  FF_ALLOCZ_OR_GOTO(NULL, filter2, filter2Size * dstW * sizeof(*filter2), fail);
439 
440  for (i = 0; i < dstW; i++) {
441  int j, k;
442 
443  if (srcFilter) {
444  for (k = 0; k < srcFilter->length; k++) {
445  for (j = 0; j < filterSize; j++)
446  filter2[i * filter2Size + k + j] +=
447  srcFilter->coeff[k] * filter[i * filterSize + j];
448  }
449  } else {
450  for (j = 0; j < filterSize; j++)
451  filter2[i * filter2Size + j] = filter[i * filterSize + j];
452  }
453  // FIXME dstFilter
454 
455  (*filterPos)[i] += (filterSize - 1) / 2 - (filter2Size - 1) / 2;
456  }
457  av_freep(&filter);
458 
459  /* try to reduce the filter-size (step1 find size and shift left) */
460  // Assume it is near normalized (*0.5 or *2.0 is OK but * 0.001 is not).
461  minFilterSize = 0;
462  for (i = dstW - 1; i >= 0; i--) {
463  int min = filter2Size;
464  int j;
465  int64_t cutOff = 0.0;
466 
467  /* get rid of near zero elements on the left by shifting left */
468  for (j = 0; j < filter2Size; j++) {
469  int k;
470  cutOff += FFABS(filter2[i * filter2Size]);
471 
472  if (cutOff > SWS_MAX_REDUCE_CUTOFF * fone)
473  break;
474 
475  /* preserve monotonicity because the core can't handle the
476  * filter otherwise */
477  if (i < dstW - 1 && (*filterPos)[i] >= (*filterPos)[i + 1])
478  break;
479 
480  // move filter coefficients left
481  for (k = 1; k < filter2Size; k++)
482  filter2[i * filter2Size + k - 1] = filter2[i * filter2Size + k];
483  filter2[i * filter2Size + k - 1] = 0;
484  (*filterPos)[i]++;
485  }
486 
487  cutOff = 0;
488  /* count near zeros on the right */
489  for (j = filter2Size - 1; j > 0; j--) {
490  cutOff += FFABS(filter2[i * filter2Size + j]);
491 
492  if (cutOff > SWS_MAX_REDUCE_CUTOFF * fone)
493  break;
494  min--;
495  }
496 
497  if (min > minFilterSize)
498  minFilterSize = min;
499  }
500 
501  if (PPC_ALTIVEC(cpu_flags)) {
502  // we can handle the special case 4, so we don't want to go the full 8
503  if (minFilterSize < 5)
504  filterAlign = 4;
505 
506  /* We really don't want to waste our time doing useless computation, so
507  * fall back on the scalar C code for very small filters.
508  * Vectorizing is worth it only if you have a decent-sized vector. */
509  if (minFilterSize < 3)
510  filterAlign = 1;
511  }
512 
513  if (INLINE_MMX(cpu_flags)) {
514  // special case for unscaled vertical filtering
515  if (minFilterSize == 1 && filterAlign == 2)
516  filterAlign = 1;
517  }
518 
519  assert(minFilterSize > 0);
520  filterSize = (minFilterSize + (filterAlign - 1)) & (~(filterAlign - 1));
521  assert(filterSize > 0);
522  filter = av_malloc(filterSize * dstW * sizeof(*filter));
523  if (filterSize >= MAX_FILTER_SIZE * 16 /
524  ((flags & SWS_ACCURATE_RND) ? APCK_SIZE : 16) || !filter)
525  goto fail;
526  *outFilterSize = filterSize;
527 
528  if (flags & SWS_PRINT_INFO)
530  "SwScaler: reducing / aligning filtersize %d -> %d\n",
531  filter2Size, filterSize);
532  /* try to reduce the filter-size (step2 reduce it) */
533  for (i = 0; i < dstW; i++) {
534  int j;
535 
536  for (j = 0; j < filterSize; j++) {
537  if (j >= filter2Size)
538  filter[i * filterSize + j] = 0;
539  else
540  filter[i * filterSize + j] = filter2[i * filter2Size + j];
541  if ((flags & SWS_BITEXACT) && j >= minFilterSize)
542  filter[i * filterSize + j] = 0;
543  }
544  }
545 
546  // FIXME try to align filterPos if possible
547 
548  // fix borders
549  if (is_horizontal) {
550  for (i = 0; i < dstW; i++) {
551  int j;
552  if ((*filterPos)[i] < 0) {
553  // move filter coefficients left to compensate for filterPos
554  for (j = 1; j < filterSize; j++) {
555  int left = FFMAX(j + (*filterPos)[i], 0);
556  filter[i * filterSize + left] += filter[i * filterSize + j];
557  filter[i * filterSize + j] = 0;
558  }
559  (*filterPos)[i] = 0;
560  }
561 
562  if ((*filterPos)[i] + filterSize > srcW) {
563  int shift = (*filterPos)[i] + filterSize - srcW;
564  // move filter coefficients right to compensate for filterPos
565  for (j = filterSize - 2; j >= 0; j--) {
566  int right = FFMIN(j + shift, filterSize - 1);
567  filter[i * filterSize + right] += filter[i * filterSize + j];
568  filter[i * filterSize + j] = 0;
569  }
570  (*filterPos)[i] = srcW - filterSize;
571  }
572  }
573  }
574 
575  // Note the +1 is for the MMX scaler which reads over the end
576  /* align at 16 for AltiVec (needed by hScale_altivec_real) */
577  FF_ALLOCZ_OR_GOTO(NULL, *outFilter,
578  *outFilterSize * (dstW + 3) * sizeof(int16_t), fail);
579 
580  /* normalize & store in outFilter */
581  for (i = 0; i < dstW; i++) {
582  int j;
583  int64_t error = 0;
584  int64_t sum = 0;
585 
586  for (j = 0; j < filterSize; j++) {
587  sum += filter[i * filterSize + j];
588  }
589  sum = (sum + one / 2) / one;
590  for (j = 0; j < *outFilterSize; j++) {
591  int64_t v = filter[i * filterSize + j] + error;
592  int intV = ROUNDED_DIV(v, sum);
593  (*outFilter)[i * (*outFilterSize) + j] = intV;
594  error = v - intV * sum;
595  }
596  }
597 
598  (*filterPos)[dstW + 0] =
599  (*filterPos)[dstW + 1] =
600  (*filterPos)[dstW + 2] = (*filterPos)[dstW - 1]; /* the MMX/SSE scaler will
601  * read over the end */
602  for (i = 0; i < *outFilterSize; i++) {
603  int k = (dstW - 1) * (*outFilterSize) + i;
604  (*outFilter)[k + 1 * (*outFilterSize)] =
605  (*outFilter)[k + 2 * (*outFilterSize)] =
606  (*outFilter)[k + 3 * (*outFilterSize)] = (*outFilter)[k];
607  }
608 
609  ret = 0;
610 
611 fail:
612  av_free(filter);
613  av_free(filter2);
614  return ret;
615 }
616 
617 #if HAVE_MMXEXT_INLINE
618 static av_cold int init_hscaler_mmxext(int dstW, int xInc, uint8_t *filterCode,
619  int16_t *filter, int32_t *filterPos,
620  int numSplits)
621 {
622  uint8_t *fragmentA;
623  x86_reg imm8OfPShufW1A;
624  x86_reg imm8OfPShufW2A;
625  x86_reg fragmentLengthA;
626  uint8_t *fragmentB;
627  x86_reg imm8OfPShufW1B;
628  x86_reg imm8OfPShufW2B;
629  x86_reg fragmentLengthB;
630  int fragmentPos;
631 
632  int xpos, i;
633 
634  // create an optimized horizontal scaling routine
635  /* This scaler is made of runtime-generated MMXEXT code using specially tuned
636  * pshufw instructions. For every four output pixels, if four input pixels
637  * are enough for the fast bilinear scaling, then a chunk of fragmentB is
638  * used. If five input pixels are needed, then a chunk of fragmentA is used.
639  */
640 
641  // code fragment
642 
643  __asm__ volatile (
644  "jmp 9f \n\t"
645  // Begin
646  "0: \n\t"
647  "movq (%%"FF_REG_d", %%"FF_REG_a"), %%mm3 \n\t"
648  "movd (%%"FF_REG_c", %%"FF_REG_S"), %%mm0 \n\t"
649  "movd 1(%%"FF_REG_c", %%"FF_REG_S"), %%mm1 \n\t"
650  "punpcklbw %%mm7, %%mm1 \n\t"
651  "punpcklbw %%mm7, %%mm0 \n\t"
652  "pshufw $0xFF, %%mm1, %%mm1 \n\t"
653  "1: \n\t"
654  "pshufw $0xFF, %%mm0, %%mm0 \n\t"
655  "2: \n\t"
656  "psubw %%mm1, %%mm0 \n\t"
657  "movl 8(%%"FF_REG_b", %%"FF_REG_a"), %%esi \n\t"
658  "pmullw %%mm3, %%mm0 \n\t"
659  "psllw $7, %%mm1 \n\t"
660  "paddw %%mm1, %%mm0 \n\t"
661 
662  "movq %%mm0, (%%"FF_REG_D", %%"FF_REG_a") \n\t"
663 
664  "add $8, %%"FF_REG_a" \n\t"
665  // End
666  "9: \n\t"
667  // "int $3 \n\t"
668  "lea " LOCAL_MANGLE(0b) ", %0 \n\t"
669  "lea " LOCAL_MANGLE(1b) ", %1 \n\t"
670  "lea " LOCAL_MANGLE(2b) ", %2 \n\t"
671  "dec %1 \n\t"
672  "dec %2 \n\t"
673  "sub %0, %1 \n\t"
674  "sub %0, %2 \n\t"
675  "lea " LOCAL_MANGLE(9b) ", %3 \n\t"
676  "sub %0, %3 \n\t"
677 
678 
679  : "=r" (fragmentA), "=r" (imm8OfPShufW1A), "=r" (imm8OfPShufW2A),
680  "=r" (fragmentLengthA)
681  );
682 
683  __asm__ volatile (
684  "jmp 9f \n\t"
685  // Begin
686  "0: \n\t"
687  "movq (%%"FF_REG_d", %%"FF_REG_a"), %%mm3 \n\t"
688  "movd (%%"FF_REG_c", %%"FF_REG_S"), %%mm0 \n\t"
689  "punpcklbw %%mm7, %%mm0 \n\t"
690  "pshufw $0xFF, %%mm0, %%mm1 \n\t"
691  "1: \n\t"
692  "pshufw $0xFF, %%mm0, %%mm0 \n\t"
693  "2: \n\t"
694  "psubw %%mm1, %%mm0 \n\t"
695  "movl 8(%%"FF_REG_b", %%"FF_REG_a"), %%esi \n\t"
696  "pmullw %%mm3, %%mm0 \n\t"
697  "psllw $7, %%mm1 \n\t"
698  "paddw %%mm1, %%mm0 \n\t"
699 
700  "movq %%mm0, (%%"FF_REG_D", %%"FF_REG_a") \n\t"
701 
702  "add $8, %%"FF_REG_a" \n\t"
703  // End
704  "9: \n\t"
705  // "int $3 \n\t"
706  "lea " LOCAL_MANGLE(0b) ", %0 \n\t"
707  "lea " LOCAL_MANGLE(1b) ", %1 \n\t"
708  "lea " LOCAL_MANGLE(2b) ", %2 \n\t"
709  "dec %1 \n\t"
710  "dec %2 \n\t"
711  "sub %0, %1 \n\t"
712  "sub %0, %2 \n\t"
713  "lea " LOCAL_MANGLE(9b) ", %3 \n\t"
714  "sub %0, %3 \n\t"
715 
716 
717  : "=r" (fragmentB), "=r" (imm8OfPShufW1B), "=r" (imm8OfPShufW2B),
718  "=r" (fragmentLengthB)
719  );
720 
721  xpos = 0; // lumXInc/2 - 0x8000; // difference between pixel centers
722  fragmentPos = 0;
723 
724  for (i = 0; i < dstW / numSplits; i++) {
725  int xx = xpos >> 16;
726 
727  if ((i & 3) == 0) {
728  int a = 0;
729  int b = ((xpos + xInc) >> 16) - xx;
730  int c = ((xpos + xInc * 2) >> 16) - xx;
731  int d = ((xpos + xInc * 3) >> 16) - xx;
732  int inc = (d + 1 < 4);
733  uint8_t *fragment = (d + 1 < 4) ? fragmentB : fragmentA;
734  x86_reg imm8OfPShufW1 = (d + 1 < 4) ? imm8OfPShufW1B : imm8OfPShufW1A;
735  x86_reg imm8OfPShufW2 = (d + 1 < 4) ? imm8OfPShufW2B : imm8OfPShufW2A;
736  x86_reg fragmentLength = (d + 1 < 4) ? fragmentLengthB : fragmentLengthA;
737  int maxShift = 3 - (d + inc);
738  int shift = 0;
739 
740  if (filterCode) {
741  filter[i] = ((xpos & 0xFFFF) ^ 0xFFFF) >> 9;
742  filter[i + 1] = (((xpos + xInc) & 0xFFFF) ^ 0xFFFF) >> 9;
743  filter[i + 2] = (((xpos + xInc * 2) & 0xFFFF) ^ 0xFFFF) >> 9;
744  filter[i + 3] = (((xpos + xInc * 3) & 0xFFFF) ^ 0xFFFF) >> 9;
745  filterPos[i / 2] = xx;
746 
747  memcpy(filterCode + fragmentPos, fragment, fragmentLength);
748 
749  filterCode[fragmentPos + imm8OfPShufW1] = (a + inc) |
750  ((b + inc) << 2) |
751  ((c + inc) << 4) |
752  ((d + inc) << 6);
753  filterCode[fragmentPos + imm8OfPShufW2] = a | (b << 2) |
754  (c << 4) |
755  (d << 6);
756 
757  if (i + 4 - inc >= dstW)
758  shift = maxShift; // avoid overread
759  else if ((filterPos[i / 2] & 3) <= maxShift)
760  shift = filterPos[i / 2] & 3; // align
761 
762  if (shift && i >= shift) {
763  filterCode[fragmentPos + imm8OfPShufW1] += 0x55 * shift;
764  filterCode[fragmentPos + imm8OfPShufW2] += 0x55 * shift;
765  filterPos[i / 2] -= shift;
766  }
767  }
768 
769  fragmentPos += fragmentLength;
770 
771  if (filterCode)
772  filterCode[fragmentPos] = RET;
773  }
774  xpos += xInc;
775  }
776  if (filterCode)
777  filterPos[((i / 2) + 1) & (~1)] = xpos >> 16; // needed to jump to the next part
778 
779  return fragmentPos + 1;
780 }
781 #endif /* HAVE_MMXEXT_INLINE */
782 
783 static void getSubSampleFactors(int *h, int *v, enum AVPixelFormat format)
784 {
786  *h = desc->log2_chroma_w;
787  *v = desc->log2_chroma_h;
788 }
789 
790 int sws_setColorspaceDetails(struct SwsContext *c, const int inv_table[4],
791  int srcRange, const int table[4], int dstRange,
792  int brightness, int contrast, int saturation)
793 {
794  const AVPixFmtDescriptor *desc_dst = av_pix_fmt_desc_get(c->dstFormat);
795  const AVPixFmtDescriptor *desc_src = av_pix_fmt_desc_get(c->srcFormat);
796  memcpy(c->srcColorspaceTable, inv_table, sizeof(int) * 4);
797  memcpy(c->dstColorspaceTable, table, sizeof(int) * 4);
798 
799  c->brightness = brightness;
800  c->contrast = contrast;
801  c->saturation = saturation;
802  c->srcRange = srcRange;
803  c->dstRange = dstRange;
804  if (isYUV(c->dstFormat) || isGray(c->dstFormat))
805  return -1;
806 
807  c->dstFormatBpp = av_get_bits_per_pixel(desc_dst);
808  c->srcFormatBpp = av_get_bits_per_pixel(desc_src);
809 
810  ff_yuv2rgb_c_init_tables(c, inv_table, srcRange, brightness,
811  contrast, saturation);
812  // FIXME factorize
813 
814  if (ARCH_PPC)
815  ff_yuv2rgb_init_tables_ppc(c, inv_table, brightness,
816  contrast, saturation);
817  return 0;
818 }
819 
820 int sws_getColorspaceDetails(struct SwsContext *c, int **inv_table,
821  int *srcRange, int **table, int *dstRange,
822  int *brightness, int *contrast, int *saturation)
823 {
824  if (isYUV(c->dstFormat) || isGray(c->dstFormat))
825  return -1;
826 
827  *inv_table = c->srcColorspaceTable;
828  *table = c->dstColorspaceTable;
829  *srcRange = c->srcRange;
830  *dstRange = c->dstRange;
831  *brightness = c->brightness;
832  *contrast = c->contrast;
833  *saturation = c->saturation;
834 
835  return 0;
836 }
837 
839 {
840  switch (*format) {
841  case AV_PIX_FMT_YUVJ420P:
842  *format = AV_PIX_FMT_YUV420P;
843  return 1;
844  case AV_PIX_FMT_YUVJ422P:
845  *format = AV_PIX_FMT_YUV422P;
846  return 1;
847  case AV_PIX_FMT_YUVJ444P:
848  *format = AV_PIX_FMT_YUV444P;
849  return 1;
850  case AV_PIX_FMT_YUVJ440P:
851  *format = AV_PIX_FMT_YUV440P;
852  return 1;
853  default:
854  return 0;
855  }
856 }
857 
859 {
860  SwsContext *c = av_mallocz(sizeof(SwsContext));
861 
862  if (c) {
865  }
866 
867  return c;
868 }
869 
871  SwsFilter *dstFilter)
872 {
873  int i;
874  int usesVFilter, usesHFilter;
875  int unscaled;
876  SwsFilter dummyFilter = { NULL, NULL, NULL, NULL };
877  int srcW = c->srcW;
878  int srcH = c->srcH;
879  int dstW = c->dstW;
880  int dstH = c->dstH;
881  int dst_stride = FFALIGN(dstW * sizeof(int16_t) + 16, 16);
882  int dst_stride_px = dst_stride >> 1;
883  int flags, cpu_flags;
884  enum AVPixelFormat srcFormat = c->srcFormat;
885  enum AVPixelFormat dstFormat = c->dstFormat;
886  const AVPixFmtDescriptor *desc_src = av_pix_fmt_desc_get(srcFormat);
887  const AVPixFmtDescriptor *desc_dst = av_pix_fmt_desc_get(dstFormat);
888 
889  cpu_flags = av_get_cpu_flags();
890  flags = c->flags;
891  emms_c();
892  if (!rgb15to16)
893  ff_rgb2rgb_init();
894 
895  unscaled = (srcW == dstW && srcH == dstH);
896 
897  if (!(unscaled && sws_isSupportedEndiannessConversion(srcFormat) &&
898  av_pix_fmt_swap_endianness(srcFormat) == dstFormat)) {
899  if (!sws_isSupportedInput(srcFormat)) {
900  av_log(c, AV_LOG_ERROR, "%s is not supported as input pixel format\n",
901  sws_format_name(srcFormat));
902  return AVERROR(EINVAL);
903  }
904  if (!sws_isSupportedOutput(dstFormat)) {
905  av_log(c, AV_LOG_ERROR, "%s is not supported as output pixel format\n",
906  sws_format_name(dstFormat));
907  return AVERROR(EINVAL);
908  }
909  }
910 
911  i = flags & (SWS_POINT |
912  SWS_AREA |
913  SWS_BILINEAR |
915  SWS_BICUBIC |
916  SWS_X |
917  SWS_GAUSS |
918  SWS_LANCZOS |
919  SWS_SINC |
920  SWS_SPLINE |
921  SWS_BICUBLIN);
922 
923  /* provide a default scaler if not set by caller */
924  if (!i) {
925  if (dstW < srcW && dstH < srcH)
926  flags |= SWS_GAUSS;
927  else if (dstW > srcW && dstH > srcH)
928  flags |= SWS_SINC;
929  else
930  flags |= SWS_LANCZOS;
931  c->flags = flags;
932  } else if (i & (i - 1)) {
933  av_log(c, AV_LOG_ERROR,
934  "Exactly one scaler algorithm must be chosen\n");
935  return AVERROR(EINVAL);
936  }
937  /* sanity check */
938  if (srcW < 4 || srcH < 1 || dstW < 8 || dstH < 1) {
939  /* FIXME check if these are enough and try to lower them after
940  * fixing the relevant parts of the code */
941  av_log(c, AV_LOG_ERROR, "%dx%d -> %dx%d is invalid scaling dimension\n",
942  srcW, srcH, dstW, dstH);
943  return AVERROR(EINVAL);
944  }
945 
946  if (!dstFilter)
947  dstFilter = &dummyFilter;
948  if (!srcFilter)
949  srcFilter = &dummyFilter;
950 
951  c->lumXInc = (((int64_t)srcW << 16) + (dstW >> 1)) / dstW;
952  c->lumYInc = (((int64_t)srcH << 16) + (dstH >> 1)) / dstH;
953  c->dstFormatBpp = av_get_bits_per_pixel(desc_dst);
954  c->srcFormatBpp = av_get_bits_per_pixel(desc_src);
955  c->vRounder = 4 * 0x0001000100010001ULL;
956 
957  usesVFilter = (srcFilter->lumV && srcFilter->lumV->length > 1) ||
958  (srcFilter->chrV && srcFilter->chrV->length > 1) ||
959  (dstFilter->lumV && dstFilter->lumV->length > 1) ||
960  (dstFilter->chrV && dstFilter->chrV->length > 1);
961  usesHFilter = (srcFilter->lumH && srcFilter->lumH->length > 1) ||
962  (srcFilter->chrH && srcFilter->chrH->length > 1) ||
963  (dstFilter->lumH && dstFilter->lumH->length > 1) ||
964  (dstFilter->chrH && dstFilter->chrH->length > 1);
965 
968 
969  if (isPlanarRGB(dstFormat)) {
970  if (!(flags & SWS_FULL_CHR_H_INT)) {
971  av_log(c, AV_LOG_DEBUG,
972  "%s output is not supported with half chroma resolution, switching to full\n",
973  av_get_pix_fmt_name(dstFormat));
974  flags |= SWS_FULL_CHR_H_INT;
975  c->flags = flags;
976  }
977  }
978 
979  /* reuse chroma for 2 pixels RGB/BGR unless user wants full
980  * chroma interpolation */
981  if (flags & SWS_FULL_CHR_H_INT &&
982  isAnyRGB(dstFormat) &&
983  !isPlanarRGB(dstFormat) &&
984  dstFormat != AV_PIX_FMT_RGBA &&
985  dstFormat != AV_PIX_FMT_ARGB &&
986  dstFormat != AV_PIX_FMT_BGRA &&
987  dstFormat != AV_PIX_FMT_ABGR &&
988  dstFormat != AV_PIX_FMT_RGB24 &&
989  dstFormat != AV_PIX_FMT_BGR24) {
990  av_log(c, AV_LOG_ERROR,
991  "full chroma interpolation for destination format '%s' not yet implemented\n",
992  sws_format_name(dstFormat));
993  flags &= ~SWS_FULL_CHR_H_INT;
994  c->flags = flags;
995  }
996  if (isAnyRGB(dstFormat) && !(flags & SWS_FULL_CHR_H_INT))
997  c->chrDstHSubSample = 1;
998 
999  // drop some chroma lines if the user wants it
1000  c->vChrDrop = (flags & SWS_SRC_V_CHR_DROP_MASK) >>
1002  c->chrSrcVSubSample += c->vChrDrop;
1003 
1004  /* drop every other pixel for chroma calculation unless user
1005  * wants full chroma */
1006  if (isAnyRGB(srcFormat) && !(flags & SWS_FULL_CHR_H_INP) &&
1007  srcFormat != AV_PIX_FMT_RGB8 && srcFormat != AV_PIX_FMT_BGR8 &&
1008  srcFormat != AV_PIX_FMT_RGB4 && srcFormat != AV_PIX_FMT_BGR4 &&
1009  srcFormat != AV_PIX_FMT_RGB4_BYTE && srcFormat != AV_PIX_FMT_BGR4_BYTE &&
1010  srcFormat != AV_PIX_FMT_GBRP9BE && srcFormat != AV_PIX_FMT_GBRP9LE &&
1011  srcFormat != AV_PIX_FMT_GBRP10BE && srcFormat != AV_PIX_FMT_GBRP10LE &&
1012  srcFormat != AV_PIX_FMT_GBRP16BE && srcFormat != AV_PIX_FMT_GBRP16LE &&
1013  ((dstW >> c->chrDstHSubSample) <= (srcW >> 1) ||
1014  (flags & SWS_FAST_BILINEAR)))
1015  c->chrSrcHSubSample = 1;
1016 
1017  // Note the AV_CEIL_RSHIFT is so that we always round toward +inf.
1018  c->chrSrcW = AV_CEIL_RSHIFT(srcW, c->chrSrcHSubSample);
1019  c->chrSrcH = AV_CEIL_RSHIFT(srcH, c->chrSrcVSubSample);
1020  c->chrDstW = AV_CEIL_RSHIFT(dstW, c->chrDstHSubSample);
1021  c->chrDstH = AV_CEIL_RSHIFT(dstH, c->chrDstVSubSample);
1022 
1023  /* unscaled special cases */
1024  if (unscaled && !usesHFilter && !usesVFilter &&
1025  (c->srcRange == c->dstRange || isAnyRGB(dstFormat))) {
1027 
1028  if (c->swscale) {
1029  if (flags & SWS_PRINT_INFO)
1030  av_log(c, AV_LOG_INFO,
1031  "using unscaled %s -> %s special converter\n",
1032  sws_format_name(srcFormat), sws_format_name(dstFormat));
1033  return 0;
1034  }
1035  }
1036 
1037  c->srcBpc = desc_src->comp[0].depth;
1038  if (c->srcBpc < 8)
1039  c->srcBpc = 8;
1040  c->dstBpc = desc_dst->comp[0].depth;
1041  if (c->dstBpc < 8)
1042  c->dstBpc = 8;
1043  if (c->dstBpc == 16)
1044  dst_stride <<= 1;
1046  (FFALIGN(srcW, 16) * 2 * FFALIGN(c->srcBpc, 8) >> 3) + 16,
1047  fail);
1048  if (INLINE_MMXEXT(cpu_flags) && c->srcBpc == 8 && c->dstBpc <= 10) {
1049  c->canMMXEXTBeUsed = (dstW >= srcW && (dstW & 31) == 0 &&
1050  (srcW & 15) == 0) ? 1 : 0;
1051  if (!c->canMMXEXTBeUsed && dstW >= srcW && (srcW & 15) == 0
1052  && (flags & SWS_FAST_BILINEAR)) {
1053  if (flags & SWS_PRINT_INFO)
1054  av_log(c, AV_LOG_INFO,
1055  "output width is not a multiple of 32 -> no MMXEXT scaler\n");
1056  }
1057  if (usesHFilter)
1058  c->canMMXEXTBeUsed = 0;
1059  } else
1060  c->canMMXEXTBeUsed = 0;
1061 
1062  c->chrXInc = (((int64_t)c->chrSrcW << 16) + (c->chrDstW >> 1)) / c->chrDstW;
1063  c->chrYInc = (((int64_t)c->chrSrcH << 16) + (c->chrDstH >> 1)) / c->chrDstH;
1064 
1065  /* Match pixel 0 of the src to pixel 0 of dst and match pixel n-2 of src
1066  * to pixel n-2 of dst, but only for the FAST_BILINEAR mode otherwise do
1067  * correct scaling.
1068  * n-2 is the last chrominance sample available.
1069  * This is not perfect, but no one should notice the difference, the more
1070  * correct variant would be like the vertical one, but that would require
1071  * some special code for the first and last pixel */
1072  if (flags & SWS_FAST_BILINEAR) {
1073  if (c->canMMXEXTBeUsed) {
1074  c->lumXInc += 20;
1075  c->chrXInc += 20;
1076  }
1077  // we don't use the x86 asm scaler if MMX is available
1078  else if (INLINE_MMX(cpu_flags)) {
1079  c->lumXInc = ((int64_t)(srcW - 2) << 16) / (dstW - 2) - 20;
1080  c->chrXInc = ((int64_t)(c->chrSrcW - 2) << 16) / (c->chrDstW - 2) - 20;
1081  }
1082  }
1083 
1084 #define USE_MMAP (HAVE_MMAP && HAVE_MPROTECT && defined MAP_ANONYMOUS)
1085 
1086  /* precalculate horizontal scaler filter coefficients */
1087  {
1088 #if HAVE_MMXEXT_INLINE
1089 // can't downscale !!!
1090  if (c->canMMXEXTBeUsed && (flags & SWS_FAST_BILINEAR)) {
1091  c->lumMmxextFilterCodeSize = init_hscaler_mmxext(dstW, c->lumXInc, NULL,
1092  NULL, NULL, 8);
1093  c->chrMmxextFilterCodeSize = init_hscaler_mmxext(c->chrDstW, c->chrXInc,
1094  NULL, NULL, NULL, 4);
1095 
1096 #if USE_MMAP
1097  c->lumMmxextFilterCode = mmap(NULL, c->lumMmxextFilterCodeSize,
1098  PROT_READ | PROT_WRITE,
1099  MAP_PRIVATE | MAP_ANONYMOUS,
1100  -1, 0);
1101  c->chrMmxextFilterCode = mmap(NULL, c->chrMmxextFilterCodeSize,
1102  PROT_READ | PROT_WRITE,
1103  MAP_PRIVATE | MAP_ANONYMOUS,
1104  -1, 0);
1105 #elif HAVE_VIRTUALALLOC
1106  c->lumMmxextFilterCode = VirtualAlloc(NULL,
1108  MEM_COMMIT,
1109  PAGE_EXECUTE_READWRITE);
1110  c->chrMmxextFilterCode = VirtualAlloc(NULL,
1112  MEM_COMMIT,
1113  PAGE_EXECUTE_READWRITE);
1114 #else
1117 #endif
1118 
1120  return AVERROR(ENOMEM);
1121  FF_ALLOCZ_OR_GOTO(c, c->hLumFilter, (dstW / 8 + 8) * sizeof(int16_t), fail);
1122  FF_ALLOCZ_OR_GOTO(c, c->hChrFilter, (c->chrDstW / 4 + 8) * sizeof(int16_t), fail);
1123  FF_ALLOCZ_OR_GOTO(c, c->hLumFilterPos, (dstW / 2 / 8 + 8) * sizeof(int32_t), fail);
1124  FF_ALLOCZ_OR_GOTO(c, c->hChrFilterPos, (c->chrDstW / 2 / 4 + 8) * sizeof(int32_t), fail);
1125 
1126  init_hscaler_mmxext(dstW, c->lumXInc, c->lumMmxextFilterCode,
1127  c->hLumFilter, c->hLumFilterPos, 8);
1128  init_hscaler_mmxext(c->chrDstW, c->chrXInc, c->chrMmxextFilterCode,
1129  c->hChrFilter, c->hChrFilterPos, 4);
1130 
1131 #if USE_MMAP
1132  mprotect(c->lumMmxextFilterCode, c->lumMmxextFilterCodeSize, PROT_EXEC | PROT_READ);
1133  mprotect(c->chrMmxextFilterCode, c->chrMmxextFilterCodeSize, PROT_EXEC | PROT_READ);
1134 #endif
1135  } else
1136 #endif /* HAVE_MMXEXT_INLINE */
1137  {
1138  const int filterAlign = X86_MMX(cpu_flags) ? 4 :
1139  PPC_ALTIVEC(cpu_flags) ? 8 : 1;
1140 
1141  if (initFilter(&c->hLumFilter, &c->hLumFilterPos,
1142  &c->hLumFilterSize, c->lumXInc,
1143  srcW, dstW, filterAlign, 1 << 14,
1144  (flags & SWS_BICUBLIN) ? (flags | SWS_BICUBIC) : flags,
1145  cpu_flags, srcFilter->lumH, dstFilter->lumH,
1146  c->param, 1) < 0)
1147  goto fail;
1148  if (initFilter(&c->hChrFilter, &c->hChrFilterPos,
1149  &c->hChrFilterSize, c->chrXInc,
1150  c->chrSrcW, c->chrDstW, filterAlign, 1 << 14,
1151  (flags & SWS_BICUBLIN) ? (flags | SWS_BILINEAR) : flags,
1152  cpu_flags, srcFilter->chrH, dstFilter->chrH,
1153  c->param, 1) < 0)
1154  goto fail;
1155  }
1156  } // initialize horizontal stuff
1157 
1158  /* precalculate vertical scaler filter coefficients */
1159  {
1160  const int filterAlign = X86_MMX(cpu_flags) ? 2 :
1161  PPC_ALTIVEC(cpu_flags) ? 8 : 1;
1162 
1164  c->lumYInc, srcH, dstH, filterAlign, (1 << 12),
1165  (flags & SWS_BICUBLIN) ? (flags | SWS_BICUBIC) : flags,
1166  cpu_flags, srcFilter->lumV, dstFilter->lumV,
1167  c->param, 0) < 0)
1168  goto fail;
1170  c->chrYInc, c->chrSrcH, c->chrDstH,
1171  filterAlign, (1 << 12),
1172  (flags & SWS_BICUBLIN) ? (flags | SWS_BILINEAR) : flags,
1173  cpu_flags, srcFilter->chrV, dstFilter->chrV,
1174  c->param, 0) < 0)
1175  goto fail;
1176 
1177 #if HAVE_ALTIVEC
1178  FF_ALLOC_OR_GOTO(c, c->vYCoeffsBank, sizeof(vector signed short) * c->vLumFilterSize * c->dstH, fail);
1179  FF_ALLOC_OR_GOTO(c, c->vCCoeffsBank, sizeof(vector signed short) * c->vChrFilterSize * c->chrDstH, fail);
1180 
1181  for (i = 0; i < c->vLumFilterSize * c->dstH; i++) {
1182  int j;
1183  short *p = (short *)&c->vYCoeffsBank[i];
1184  for (j = 0; j < 8; j++)
1185  p[j] = c->vLumFilter[i];
1186  }
1187 
1188  for (i = 0; i < c->vChrFilterSize * c->chrDstH; i++) {
1189  int j;
1190  short *p = (short *)&c->vCCoeffsBank[i];
1191  for (j = 0; j < 8; j++)
1192  p[j] = c->vChrFilter[i];
1193  }
1194 #endif
1195  }
1196 
1197  // calculate buffer sizes so that they won't run out while handling these damn slices
1198  c->vLumBufSize = c->vLumFilterSize;
1199  c->vChrBufSize = c->vChrFilterSize;
1200  for (i = 0; i < dstH; i++) {
1201  int chrI = (int64_t)i * c->chrDstH / dstH;
1202  int nextSlice = FFMAX(c->vLumFilterPos[i] + c->vLumFilterSize - 1,
1203  ((c->vChrFilterPos[chrI] + c->vChrFilterSize - 1)
1204  << c->chrSrcVSubSample));
1205 
1206  nextSlice >>= c->chrSrcVSubSample;
1207  nextSlice <<= c->chrSrcVSubSample;
1208  if (c->vLumFilterPos[i] + c->vLumBufSize < nextSlice)
1209  c->vLumBufSize = nextSlice - c->vLumFilterPos[i];
1210  if (c->vChrFilterPos[chrI] + c->vChrBufSize <
1211  (nextSlice >> c->chrSrcVSubSample))
1212  c->vChrBufSize = (nextSlice >> c->chrSrcVSubSample) -
1213  c->vChrFilterPos[chrI];
1214  }
1215 
1216  /* Allocate pixbufs (we use dynamic allocation because otherwise we would
1217  * need to allocate several megabytes to handle all possible cases) */
1218  FF_ALLOC_OR_GOTO(c, c->lumPixBuf, c->vLumBufSize * 3 * sizeof(int16_t *), fail);
1219  FF_ALLOC_OR_GOTO(c, c->chrUPixBuf, c->vChrBufSize * 3 * sizeof(int16_t *), fail);
1220  FF_ALLOC_OR_GOTO(c, c->chrVPixBuf, c->vChrBufSize * 3 * sizeof(int16_t *), fail);
1222  FF_ALLOCZ_OR_GOTO(c, c->alpPixBuf, c->vLumBufSize * 3 * sizeof(int16_t *), fail);
1223  /* Note we need at least one pixel more at the end because of the MMX code
1224  * (just in case someone wants to replace the 4000/8000). */
1225  /* align at 16 bytes for AltiVec */
1226  for (i = 0; i < c->vLumBufSize; i++) {
1227  FF_ALLOCZ_OR_GOTO(c, c->lumPixBuf[i + c->vLumBufSize],
1228  dst_stride + 16, fail);
1229  c->lumPixBuf[i] = c->lumPixBuf[i + c->vLumBufSize];
1230  }
1231  // 64 / (c->dstBpc & ~7) is the same as 16 / sizeof(scaling_intermediate)
1232  c->uv_off_px = dst_stride_px + 64 / (c->dstBpc & ~7);
1233  c->uv_off_byte = dst_stride + 16;
1234  for (i = 0; i < c->vChrBufSize; i++) {
1235  FF_ALLOC_OR_GOTO(c, c->chrUPixBuf[i + c->vChrBufSize],
1236  dst_stride * 2 + 32, fail);
1237  c->chrUPixBuf[i] = c->chrUPixBuf[i + c->vChrBufSize];
1238  c->chrVPixBuf[i] = c->chrVPixBuf[i + c->vChrBufSize]
1239  = c->chrUPixBuf[i] + (dst_stride >> 1) + 8;
1240  }
1241  if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1242  for (i = 0; i < c->vLumBufSize; i++) {
1243  FF_ALLOCZ_OR_GOTO(c, c->alpPixBuf[i + c->vLumBufSize],
1244  dst_stride + 16, fail);
1245  c->alpPixBuf[i] = c->alpPixBuf[i + c->vLumBufSize];
1246  }
1247 
1248  // try to avoid drawing green stuff between the right end and the stride end
1249  for (i = 0; i < c->vChrBufSize; i++)
1250  memset(c->chrUPixBuf[i], 64, dst_stride * 2 + 1);
1251 
1252  assert(c->chrDstH <= dstH);
1253 
1254  if (flags & SWS_PRINT_INFO) {
1255  if (flags & SWS_FAST_BILINEAR)
1256  av_log(c, AV_LOG_INFO, "FAST_BILINEAR scaler, ");
1257  else if (flags & SWS_BILINEAR)
1258  av_log(c, AV_LOG_INFO, "BILINEAR scaler, ");
1259  else if (flags & SWS_BICUBIC)
1260  av_log(c, AV_LOG_INFO, "BICUBIC scaler, ");
1261  else if (flags & SWS_X)
1262  av_log(c, AV_LOG_INFO, "Experimental scaler, ");
1263  else if (flags & SWS_POINT)
1264  av_log(c, AV_LOG_INFO, "Nearest Neighbor / POINT scaler, ");
1265  else if (flags & SWS_AREA)
1266  av_log(c, AV_LOG_INFO, "Area Averaging scaler, ");
1267  else if (flags & SWS_BICUBLIN)
1268  av_log(c, AV_LOG_INFO, "luma BICUBIC / chroma BILINEAR scaler, ");
1269  else if (flags & SWS_GAUSS)
1270  av_log(c, AV_LOG_INFO, "Gaussian scaler, ");
1271  else if (flags & SWS_SINC)
1272  av_log(c, AV_LOG_INFO, "Sinc scaler, ");
1273  else if (flags & SWS_LANCZOS)
1274  av_log(c, AV_LOG_INFO, "Lanczos scaler, ");
1275  else if (flags & SWS_SPLINE)
1276  av_log(c, AV_LOG_INFO, "Bicubic spline scaler, ");
1277  else
1278  av_log(c, AV_LOG_INFO, "ehh flags invalid?! ");
1279 
1280  av_log(c, AV_LOG_INFO, "from %s to %s%s ",
1281  sws_format_name(srcFormat),
1282 #ifdef DITHER1XBPP
1283  dstFormat == AV_PIX_FMT_BGR555 || dstFormat == AV_PIX_FMT_BGR565 ||
1284  dstFormat == AV_PIX_FMT_RGB444BE || dstFormat == AV_PIX_FMT_RGB444LE ||
1285  dstFormat == AV_PIX_FMT_BGR444BE || dstFormat == AV_PIX_FMT_BGR444LE ?
1286  "dithered " : "",
1287 #else
1288  "",
1289 #endif
1290  sws_format_name(dstFormat));
1291 
1292  if (INLINE_MMXEXT(cpu_flags))
1293  av_log(c, AV_LOG_INFO, "using MMXEXT\n");
1294  else if (INLINE_AMD3DNOW(cpu_flags))
1295  av_log(c, AV_LOG_INFO, "using 3DNOW\n");
1296  else if (INLINE_MMX(cpu_flags))
1297  av_log(c, AV_LOG_INFO, "using MMX\n");
1298  else if (PPC_ALTIVEC(cpu_flags))
1299  av_log(c, AV_LOG_INFO, "using AltiVec\n");
1300  else
1301  av_log(c, AV_LOG_INFO, "using C\n");
1302 
1303  av_log(c, AV_LOG_VERBOSE, "%dx%d -> %dx%d\n", srcW, srcH, dstW, dstH);
1304  av_log(c, AV_LOG_DEBUG,
1305  "lum srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n",
1306  c->srcW, c->srcH, c->dstW, c->dstH, c->lumXInc, c->lumYInc);
1307  av_log(c, AV_LOG_DEBUG,
1308  "chr srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n",
1309  c->chrSrcW, c->chrSrcH, c->chrDstW, c->chrDstH,
1310  c->chrXInc, c->chrYInc);
1311  }
1312 
1313  c->swscale = ff_getSwsFunc(c);
1314  return 0;
1315 fail: // FIXME replace things by appropriate error codes
1316  return -1;
1317 }
1318 
1319 SwsContext *sws_getContext(int srcW, int srcH, enum AVPixelFormat srcFormat,
1320  int dstW, int dstH, enum AVPixelFormat dstFormat,
1321  int flags, SwsFilter *srcFilter,
1322  SwsFilter *dstFilter, const double *param)
1323 {
1324  SwsContext *c;
1325 
1326  if (!(c = sws_alloc_context()))
1327  return NULL;
1328 
1329  c->flags = flags;
1330  c->srcW = srcW;
1331  c->srcH = srcH;
1332  c->dstW = dstW;
1333  c->dstH = dstH;
1334  c->srcRange = handle_jpeg(&srcFormat);
1335  c->dstRange = handle_jpeg(&dstFormat);
1336  c->srcFormat = srcFormat;
1337  c->dstFormat = dstFormat;
1338 
1339  if (param) {
1340  c->param[0] = param[0];
1341  c->param[1] = param[1];
1342  }
1344  ff_yuv2rgb_coeffs[SWS_CS_DEFAULT] /* FIXME*/,
1345  c->dstRange, 0, 1 << 16, 1 << 16);
1346 
1347  if (sws_init_context(c, srcFilter, dstFilter) < 0) {
1348  sws_freeContext(c);
1349  return NULL;
1350  }
1351 
1352  return c;
1353 }
1354 
1355 SwsFilter *sws_getDefaultFilter(float lumaGBlur, float chromaGBlur,
1356  float lumaSharpen, float chromaSharpen,
1357  float chromaHShift, float chromaVShift,
1358  int verbose)
1359 {
1360  SwsFilter *filter = av_malloc(sizeof(SwsFilter));
1361  if (!filter)
1362  return NULL;
1363 
1364  if (lumaGBlur != 0.0) {
1365  filter->lumH = sws_getGaussianVec(lumaGBlur, 3.0);
1366  filter->lumV = sws_getGaussianVec(lumaGBlur, 3.0);
1367  } else {
1368  filter->lumH = sws_getIdentityVec();
1369  filter->lumV = sws_getIdentityVec();
1370  }
1371 
1372  if (chromaGBlur != 0.0) {
1373  filter->chrH = sws_getGaussianVec(chromaGBlur, 3.0);
1374  filter->chrV = sws_getGaussianVec(chromaGBlur, 3.0);
1375  } else {
1376  filter->chrH = sws_getIdentityVec();
1377  filter->chrV = sws_getIdentityVec();
1378  }
1379 
1380  if (!filter->lumH || !filter->lumV || !filter->chrH || !filter->chrV)
1381  goto fail;
1382 
1383  if (chromaSharpen != 0.0) {
1384  SwsVector *id = sws_getIdentityVec();
1385  if (!id)
1386  goto fail;
1387  sws_scaleVec(filter->chrH, -chromaSharpen);
1388  sws_scaleVec(filter->chrV, -chromaSharpen);
1389  sws_addVec(filter->chrH, id);
1390  sws_addVec(filter->chrV, id);
1391  sws_freeVec(id);
1392  }
1393 
1394  if (lumaSharpen != 0.0) {
1395  SwsVector *id = sws_getIdentityVec();
1396  if (!id)
1397  goto fail;
1398  sws_scaleVec(filter->lumH, -lumaSharpen);
1399  sws_scaleVec(filter->lumV, -lumaSharpen);
1400  sws_addVec(filter->lumH, id);
1401  sws_addVec(filter->lumV, id);
1402  sws_freeVec(id);
1403  }
1404 
1405  if (chromaHShift != 0.0)
1406  sws_shiftVec(filter->chrH, (int)(chromaHShift + 0.5));
1407 
1408  if (chromaVShift != 0.0)
1409  sws_shiftVec(filter->chrV, (int)(chromaVShift + 0.5));
1410 
1411  sws_normalizeVec(filter->chrH, 1.0);
1412  sws_normalizeVec(filter->chrV, 1.0);
1413  sws_normalizeVec(filter->lumH, 1.0);
1414  sws_normalizeVec(filter->lumV, 1.0);
1415 
1416  if (verbose)
1417  sws_printVec2(filter->chrH, NULL, AV_LOG_DEBUG);
1418  if (verbose)
1419  sws_printVec2(filter->lumH, NULL, AV_LOG_DEBUG);
1420 
1421  return filter;
1422 
1423 fail:
1424  sws_freeVec(filter->lumH);
1425  sws_freeVec(filter->lumV);
1426  sws_freeVec(filter->chrH);
1427  sws_freeVec(filter->chrV);
1428  av_freep(&filter);
1429  return NULL;
1430 }
1431 
1433 {
1434  SwsVector *vec = av_malloc(sizeof(SwsVector));
1435  if (!vec)
1436  return NULL;
1437  vec->length = length;
1438  vec->coeff = av_malloc(sizeof(double) * length);
1439  if (!vec->coeff)
1440  av_freep(&vec);
1441  return vec;
1442 }
1443 
1444 SwsVector *sws_getGaussianVec(double variance, double quality)
1445 {
1446  const int length = (int)(variance * quality + 0.5) | 1;
1447  int i;
1448  double middle = (length - 1) * 0.5;
1449  SwsVector *vec = sws_allocVec(length);
1450 
1451  if (!vec)
1452  return NULL;
1453 
1454  for (i = 0; i < length; i++) {
1455  double dist = i - middle;
1456  vec->coeff[i] = exp(-dist * dist / (2 * variance * variance)) /
1457  sqrt(2 * variance * M_PI);
1458  }
1459 
1460  sws_normalizeVec(vec, 1.0);
1461 
1462  return vec;
1463 }
1464 
1465 SwsVector *sws_getConstVec(double c, int length)
1466 {
1467  int i;
1468  SwsVector *vec = sws_allocVec(length);
1469 
1470  if (!vec)
1471  return NULL;
1472 
1473  for (i = 0; i < length; i++)
1474  vec->coeff[i] = c;
1475 
1476  return vec;
1477 }
1478 
1480 {
1481  return sws_getConstVec(1.0, 1);
1482 }
1483 
1484 static double sws_dcVec(SwsVector *a)
1485 {
1486  int i;
1487  double sum = 0;
1488 
1489  for (i = 0; i < a->length; i++)
1490  sum += a->coeff[i];
1491 
1492  return sum;
1493 }
1494 
1495 void sws_scaleVec(SwsVector *a, double scalar)
1496 {
1497  int i;
1498 
1499  for (i = 0; i < a->length; i++)
1500  a->coeff[i] *= scalar;
1501 }
1502 
1504 {
1505  sws_scaleVec(a, height / sws_dcVec(a));
1506 }
1507 
1509 {
1510  int length = a->length + b->length - 1;
1511  int i, j;
1512  SwsVector *vec = sws_getConstVec(0.0, length);
1513 
1514  if (!vec)
1515  return NULL;
1516 
1517  for (i = 0; i < a->length; i++) {
1518  for (j = 0; j < b->length; j++) {
1519  vec->coeff[i + j] += a->coeff[i] * b->coeff[j];
1520  }
1521  }
1522 
1523  return vec;
1524 }
1525 
1527 {
1528  int length = FFMAX(a->length, b->length);
1529  int i;
1530  SwsVector *vec = sws_getConstVec(0.0, length);
1531 
1532  if (!vec)
1533  return NULL;
1534 
1535  for (i = 0; i < a->length; i++)
1536  vec->coeff[i + (length - 1) / 2 - (a->length - 1) / 2] += a->coeff[i];
1537  for (i = 0; i < b->length; i++)
1538  vec->coeff[i + (length - 1) / 2 - (b->length - 1) / 2] += b->coeff[i];
1539 
1540  return vec;
1541 }
1542 
1544 {
1545  int length = FFMAX(a->length, b->length);
1546  int i;
1547  SwsVector *vec = sws_getConstVec(0.0, length);
1548 
1549  if (!vec)
1550  return NULL;
1551 
1552  for (i = 0; i < a->length; i++)
1553  vec->coeff[i + (length - 1) / 2 - (a->length - 1) / 2] += a->coeff[i];
1554  for (i = 0; i < b->length; i++)
1555  vec->coeff[i + (length - 1) / 2 - (b->length - 1) / 2] -= b->coeff[i];
1556 
1557  return vec;
1558 }
1559 
1560 /* shift left / or right if "shift" is negative */
1562 {
1563  int length = a->length + FFABS(shift) * 2;
1564  int i;
1565  SwsVector *vec = sws_getConstVec(0.0, length);
1566 
1567  if (!vec)
1568  return NULL;
1569 
1570  for (i = 0; i < a->length; i++) {
1571  vec->coeff[i + (length - 1) / 2 -
1572  (a->length - 1) / 2 - shift] = a->coeff[i];
1573  }
1574 
1575  return vec;
1576 }
1577 
1578 void sws_shiftVec(SwsVector *a, int shift)
1579 {
1580  SwsVector *shifted = sws_getShiftedVec(a, shift);
1581  av_free(a->coeff);
1582  a->coeff = shifted->coeff;
1583  a->length = shifted->length;
1584  av_free(shifted);
1585 }
1586 
1588 {
1589  SwsVector *sum = sws_sumVec(a, b);
1590  av_free(a->coeff);
1591  a->coeff = sum->coeff;
1592  a->length = sum->length;
1593  av_free(sum);
1594 }
1595 
1597 {
1598  SwsVector *diff = sws_diffVec(a, b);
1599  av_free(a->coeff);
1600  a->coeff = diff->coeff;
1601  a->length = diff->length;
1602  av_free(diff);
1603 }
1604 
1606 {
1607  SwsVector *conv = sws_getConvVec(a, b);
1608  av_free(a->coeff);
1609  a->coeff = conv->coeff;
1610  a->length = conv->length;
1611  av_free(conv);
1612 }
1613 
1615 {
1616  int i;
1617  SwsVector *vec = sws_allocVec(a->length);
1618 
1619  if (!vec)
1620  return NULL;
1621 
1622  for (i = 0; i < a->length; i++)
1623  vec->coeff[i] = a->coeff[i];
1624 
1625  return vec;
1626 }
1627 
1628 void sws_printVec2(SwsVector *a, AVClass *log_ctx, int log_level)
1629 {
1630  int i;
1631  double max = 0;
1632  double min = 0;
1633  double range;
1634 
1635  for (i = 0; i < a->length; i++)
1636  if (a->coeff[i] > max)
1637  max = a->coeff[i];
1638 
1639  for (i = 0; i < a->length; i++)
1640  if (a->coeff[i] < min)
1641  min = a->coeff[i];
1642 
1643  range = max - min;
1644 
1645  for (i = 0; i < a->length; i++) {
1646  int x = (int)((a->coeff[i] - min) * 60.0 / range + 0.5);
1647  av_log(log_ctx, log_level, "%1.3f ", a->coeff[i]);
1648  for (; x > 0; x--)
1649  av_log(log_ctx, log_level, " ");
1650  av_log(log_ctx, log_level, "|\n");
1651  }
1652 }
1653 
1655 {
1656  if (!a)
1657  return;
1658  av_freep(&a->coeff);
1659  a->length = 0;
1660  av_free(a);
1661 }
1662 
1664 {
1665  if (!filter)
1666  return;
1667 
1668  if (filter->lumH)
1669  sws_freeVec(filter->lumH);
1670  if (filter->lumV)
1671  sws_freeVec(filter->lumV);
1672  if (filter->chrH)
1673  sws_freeVec(filter->chrH);
1674  if (filter->chrV)
1675  sws_freeVec(filter->chrV);
1676  av_free(filter);
1677 }
1678 
1680 {
1681  int i;
1682  if (!c)
1683  return;
1684 
1685  if (c->lumPixBuf) {
1686  for (i = 0; i < c->vLumBufSize; i++)
1687  av_freep(&c->lumPixBuf[i]);
1688  av_freep(&c->lumPixBuf);
1689  }
1690 
1691  if (c->chrUPixBuf) {
1692  for (i = 0; i < c->vChrBufSize; i++)
1693  av_freep(&c->chrUPixBuf[i]);
1694  av_freep(&c->chrUPixBuf);
1695  av_freep(&c->chrVPixBuf);
1696  }
1697 
1698  if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1699  for (i = 0; i < c->vLumBufSize; i++)
1700  av_freep(&c->alpPixBuf[i]);
1701  av_freep(&c->alpPixBuf);
1702  }
1703 
1704  av_freep(&c->vLumFilter);
1705  av_freep(&c->vChrFilter);
1706  av_freep(&c->hLumFilter);
1707  av_freep(&c->hChrFilter);
1708 #if HAVE_ALTIVEC
1709  av_freep(&c->vYCoeffsBank);
1710  av_freep(&c->vCCoeffsBank);
1711 #endif
1712 
1713  av_freep(&c->vLumFilterPos);
1714  av_freep(&c->vChrFilterPos);
1715  av_freep(&c->hLumFilterPos);
1716  av_freep(&c->hChrFilterPos);
1717 
1718 #if HAVE_MMX_INLINE
1719 #if USE_MMAP
1720  if (c->lumMmxextFilterCode)
1722  if (c->chrMmxextFilterCode)
1724 #elif HAVE_VIRTUALALLOC
1725  if (c->lumMmxextFilterCode)
1726  VirtualFree(c->lumMmxextFilterCode, 0, MEM_RELEASE);
1727  if (c->chrMmxextFilterCode)
1728  VirtualFree(c->chrMmxextFilterCode, 0, MEM_RELEASE);
1729 #else
1732 #endif
1735 #endif /* HAVE_MMX_INLINE */
1736 
1737  av_freep(&c->yuvTable);
1739 
1740  av_free(c);
1741 }
1742 
1743 struct SwsContext *sws_getCachedContext(struct SwsContext *context, int srcW,
1744  int srcH, enum AVPixelFormat srcFormat,
1745  int dstW, int dstH,
1746  enum AVPixelFormat dstFormat, int flags,
1747  SwsFilter *srcFilter,
1748  SwsFilter *dstFilter,
1749  const double *param)
1750 {
1751  static const double default_param[2] = { SWS_PARAM_DEFAULT,
1753 
1754  if (!param)
1755  param = default_param;
1756 
1757  if (context &&
1758  (context->srcW != srcW ||
1759  context->srcH != srcH ||
1760  context->srcFormat != srcFormat ||
1761  context->dstW != dstW ||
1762  context->dstH != dstH ||
1763  context->dstFormat != dstFormat ||
1764  context->flags != flags ||
1765  context->param[0] != param[0] ||
1766  context->param[1] != param[1])) {
1767  sws_freeContext(context);
1768  context = NULL;
1769  }
1770 
1771  if (!context) {
1772  if (!(context = sws_alloc_context()))
1773  return NULL;
1774  context->srcW = srcW;
1775  context->srcH = srcH;
1776  context->srcRange = handle_jpeg(&srcFormat);
1777  context->srcFormat = srcFormat;
1778  context->dstW = dstW;
1779  context->dstH = dstH;
1780  context->dstRange = handle_jpeg(&dstFormat);
1781  context->dstFormat = dstFormat;
1782  context->flags = flags;
1783  context->param[0] = param[0];
1784  context->param[1] = param[1];
1786  context->srcRange,
1787  ff_yuv2rgb_coeffs[SWS_CS_DEFAULT] /* FIXME*/,
1788  context->dstRange, 0, 1 << 16, 1 << 16);
1789  if (sws_init_context(context, srcFilter, dstFilter) < 0) {
1790  sws_freeContext(context);
1791  return NULL;
1792  }
1793  }
1794  return context;
1795 }
packed YUV 4:2:2, 16bpp, Cb Y0 Cr Y1
Definition: pixfmt.h:78
SwsVector * chrV
Definition: swscale.h:116
uint8_t is_supported_out
Definition: utils.c:75
int16_t ** alpPixBuf
Ring buffer for scaled horizontal alpha plane lines to be fed to the vertical scaler.
int sws_isSupportedOutput(enum AVPixelFormat pix_fmt)
Return a positive value if pix_fmt is a supported output format, 0 otherwise.
Definition: utils.c:198
static const FormatEntry format_entries[AV_PIX_FMT_NB]
Definition: utils.c:79
planar YUV 4:2:2, 18bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian
Definition: pixfmt.h:159
ptrdiff_t uv_off_px
offset (in pixels) between u and v planes
void * av_malloc(size_t size)
Allocate a block of size bytes with alignment suitable for all memory accesses (including vectors if ...
Definition: mem.c:62
static SwsVector * sws_sumVec(SwsVector *a, SwsVector *b)
Definition: utils.c:1526
av_cold void ff_yuv2rgb_init_tables_ppc(SwsContext *c, const int inv_table[4], int brightness, int contrast, int saturation)
#define SWS_SRC_V_CHR_DROP_MASK
Definition: swscale.h:69
const AVPixFmtDescriptor * av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt)
Definition: pixdesc.c:1768
int chrSrcH
Height of source chroma planes.
#define SWS_X
Definition: swscale.h:60
static av_always_inline int isPlanarRGB(enum AVPixelFormat pix_fmt)
#define SWS_BICUBIC
Definition: swscale.h:59
#define LIBAV_CONFIGURATION
Definition: config.h:4
uint8_t * chrMmxextFilterCode
Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes.
planar YUV 4:2:0, 15bpp, (1 Cr & Cb sample per 2x2 Y samples), little-endian
Definition: pixfmt.h:152
8 bits gray, 8 bits alpha
Definition: pixfmt.h:143
av_cold void ff_rgb2rgb_init(void)
Definition: rgb2rgb.c:132
packed RGBA 16:16:16:16, 64bpp, 16B, 16G, 16R, 16A, the 2-byte value for each R/G/B/A component is st...
Definition: pixfmt.h:199
#define ARCH_PPC
Definition: config.h:24
const char * sws_format_name(enum AVPixelFormat format)
Definition: utils.c:210
uint8_t * lumMmxextFilterCode
Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes.
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
Definition: pixfmt.h:64
SwsVector * lumV
Definition: swscale.h:114
packed RGB 8:8:8, 24bpp, RGBRGB...
Definition: pixfmt.h:61
packed RGB 1:2:1 bitstream, 4bpp, (msb)1B 2G 1R(lsb), a byte contains two pixels, the first pixel in ...
Definition: pixfmt.h:81
const char * desc
Definition: nvenc.c:101
int sws_getColorspaceDetails(struct SwsContext *c, int **inv_table, int *srcRange, int **table, int *dstRange, int *brightness, int *contrast, int *saturation)
Definition: utils.c:820
planar YUV 4:4:4, 27bpp, (1 Cr & Cb sample per 1x1 Y samples), big-endian
Definition: pixfmt.h:155
int av_get_bits_per_pixel(const AVPixFmtDescriptor *pixdesc)
Return the number of bits per pixel used by the pixel format described by pixdesc.
Definition: pixdesc.c:1740
SwsVector * sws_getGaussianVec(double variance, double quality)
Return a normalized Gaussian curve used to filter stuff quality = 3 is high quality, lower is lower quality.
Definition: utils.c:1444
void av_opt_set_defaults(void *s)
Set the values of all AVOption fields to their default values.
Definition: opt.c:599
int vChrDrop
Binary logarithm of extra vertical subsampling factor in source image chroma planes specified by user...
int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4], int fullRange, int brightness, int contrast, int saturation)
Definition: yuv2rgb.c:671
planar GBR 4:4:4 24bpp
Definition: pixfmt.h:162
packed RGBA 16:16:16:16, 64bpp, 16B, 16G, 16R, 16A, the 2-byte value for each R/G/B/A component is st...
Definition: pixfmt.h:198
packed RGB 5:5:5, 16bpp, (msb)1A 5R 5G 5B(lsb), little-endian, most significant bit to 0 ...
Definition: pixfmt.h:112
#define SWS_BICUBLIN
Definition: swscale.h:63
av_log(ac->avr, AV_LOG_TRACE, "%d samples - audio_convert: %s to %s (%s)\, len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt), use_generic ? ac->func_descr_generic :ac->func_descr)
static double getSplineCoeff(double a, double b, double c, double d, double dist)
Definition: utils.c:219
int dstFormatBpp
Number of bits per pixel of the destination pixel format.
planar YUV 4:4:4 40bpp, (1 Cr & Cb sample per 1x1 Y & A samples, little-endian)
Definition: pixfmt.h:182
external API header
static int handle_jpeg(enum AVPixelFormat *format)
Definition: utils.c:838
int sws_isSupportedEndiannessConversion(enum AVPixelFormat pix_fmt)
Definition: utils.c:204
packed BGR 5:6:5, 16bpp, (msb) 5B 6G 5R(lsb), little-endian
Definition: pixfmt.h:115
#define SWS_SRC_V_CHR_DROP_SHIFT
Definition: swscale.h:70
const char * swscale_configuration(void)
Return the libswscale build-time configuration.
Definition: utils.c:60
planar YUV 4:2:0, 13.5bpp, (1 Cr & Cb sample per 2x2 Y samples), big-endian
Definition: pixfmt.h:149
packed RGB 4:4:4, 16bpp, (msb)4A 4R 4G 4B(lsb), big-endian, most significant bits to 0 ...
Definition: pixfmt.h:140
void sws_subVec(SwsVector *a, SwsVector *b)
Definition: utils.c:1596
int srcRange
0 = MPG YUV range, 1 = JPG YUV range (source image).
#define SWS_PRINT_INFO
Definition: swscale.h:74
enum AVPixelFormat av_pix_fmt_swap_endianness(enum AVPixelFormat pix_fmt)
Utility function to swap the endianness of a pixel format.
Definition: pixdesc.c:1821
planar YUV 4:2:0, 24bpp, (1 Cr & Cb sample per 2x2 Y samples), little-endian
Definition: pixfmt.h:128
uint8_t log2_chroma_w
Amount to shift the luma width right to find the chroma width.
Definition: pixdesc.h:91
packed RGB 1:2:1 bitstream, 4bpp, (msb)1R 2G 1B(lsb), a byte contains two pixels, the first pixel in ...
Definition: pixfmt.h:84
Macro definitions for various function/variable attributes.
void av_freep(void *arg)
Free a memory block which has been allocated with av_malloc(z)() or av_realloc() and set the pointer ...
Definition: mem.c:202
packed RGB 5:6:5, 16bpp, (msb) 5R 6G 5B(lsb), little-endian
Definition: pixfmt.h:110
int srcH
Height of source luma/alpha planes.
packed RGB 1:2:1, 8bpp, (msb)1B 2G 1R(lsb)
Definition: pixfmt.h:82
#define SWS_BILINEAR
Definition: swscale.h:58
planar YUV 4:2:0 22.5bpp, (1 Cr & Cb sample per 2x2 Y & A samples), little-endian ...
Definition: pixfmt.h:172
planar GBRA 4:4:4:4 64bpp, big-endian
Definition: pixfmt.h:209
planar YUV 4:2:0 40bpp, (1 Cr & Cb sample per 2x2 Y & A samples, big-endian)
Definition: pixfmt.h:183
const int32_t ff_yuv2rgb_coeffs[8][4]
Definition: yuv2rgb.c:49
planar YUV 4:2:0, 20bpp, (1 Cr & Cb sample per 2x2 Y & A samples)
Definition: pixfmt.h:98
int chrDstVSubSample
Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in destination i...
AVComponentDescriptor comp[4]
Parameters that describe how pixels are packed.
Definition: pixdesc.h:119
uint8_t
#define av_cold
Definition: attributes.h:66
int length
number of coefficients in the vector
Definition: swscale.h:108
#define SWS_LANCZOS
Definition: swscale.h:66
8 bits with AV_PIX_FMT_RGB32 palette
Definition: pixfmt.h:70
AVOptions.
int x86_reg
Definition: asm.h:71
packed RGB 16:16:16, 48bpp, 16R, 16G, 16B, the 2-byte value for each R/G/B component is stored as lit...
Definition: pixfmt.h:107
Definition: vf_drawbox.c:37
int vChrFilterSize
Vertical filter size for chroma pixels.
#define b
Definition: input.c:52
packed RGBA 16:16:16:16, 64bpp, 16R, 16G, 16B, 16A, the 2-byte value for each R/G/B/A component is st...
Definition: pixfmt.h:196
int16_t ** lumPixBuf
Ring buffer for scaled horizontal luma plane lines to be fed to the vertical scaler.
packed RGB 4:4:4, 16bpp, (msb)4A 4R 4G 4B(lsb), little-endian, most significant bits to 0 ...
Definition: pixfmt.h:139
void sws_addVec(SwsVector *a, SwsVector *b)
Definition: utils.c:1587
#define emms_c()
Definition: internal.h:48
#define SWS_FULL_CHR_H_INT
Definition: swscale.h:78
packed RGB 5:6:5, 16bpp, (msb) 5R 6G 5B(lsb), big-endian
Definition: pixfmt.h:109
packed ABGR 8:8:8:8, 32bpp, ABGRABGR...
Definition: pixfmt.h:91
planar YUV 4:2:0 40bpp, (1 Cr & Cb sample per 2x2 Y & A samples, little-endian)
Definition: pixfmt.h:184
#define SWS_FAST_BILINEAR
Definition: swscale.h:57
planar GBR 4:4:4 48bpp, big-endian
Definition: pixfmt.h:167
planar YUV 4:4:0 full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV440P and setting color_range...
Definition: pixfmt.h:97
static int flags
Definition: log.c:50
SwsContext * sws_getContext(int srcW, int srcH, enum AVPixelFormat srcFormat, int dstW, int dstH, enum AVPixelFormat dstFormat, int flags, SwsFilter *srcFilter, SwsFilter *dstFilter, const double *param)
Allocate and return an SwsContext.
Definition: utils.c:1319
planar YUV 4:2:2, 16bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV422P and setting col...
Definition: pixfmt.h:72
planar YUV 4:4:4 64bpp, (1 Cr & Cb sample per 1x1 Y & A samples, big-endian)
Definition: pixfmt.h:187
av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter, SwsFilter *dstFilter)
Initialize the swscaler context sws_context.
Definition: utils.c:870
#define AV_LOG_VERBOSE
Detailed information.
Definition: log.h:140
#define isAnyRGB(x)
external api for the swscale stuff
enum AVPixelFormat dstFormat
Destination pixel format.
#define isALPHA(x)
Definition: swscale.c:50
#define B
Definition: huffyuv.h:49
#define FFALIGN(x, a)
Definition: macros.h:48
int chrSrcHSubSample
Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in source imag...
static av_always_inline int isYUV(enum AVPixelFormat pix_fmt)
static SwsVector * sws_getShiftedVec(SwsVector *a, int shift)
Definition: utils.c:1561
uint64_t vRounder
const char * name
Definition: pixdesc.h:81
#define ROUNDED_DIV(a, b)
Definition: common.h:52
int32_t * vChrFilterPos
Array of vertical filter starting positions for each dst[i] for chroma planes.
int dstH
Height of destination luma/alpha planes.
planar YUV 4:2:0, 13.5bpp, (1 Cr & Cb sample per 2x2 Y samples), little-endian
Definition: pixfmt.h:150
SwsFilter * sws_getDefaultFilter(float lumaGBlur, float chromaGBlur, float lumaSharpen, float chromaSharpen, float chromaHShift, float chromaVShift, int verbose)
Definition: utils.c:1355
planar GBR 4:4:4 27bpp, big-endian
Definition: pixfmt.h:163
#define INLINE_MMX(flags)
Definition: cpu.h:79
planar YUV 4:4:4, 30bpp, (1 Cr & Cb sample per 1x1 Y samples), little-endian
Definition: pixfmt.h:158
planar YUV 4:2:2 24bpp, (1 Cr & Cb sample per 2x1 Y & A samples)
Definition: pixfmt.h:169
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:124
void av_free(void *ptr)
Free a memory block which has been allocated with av_malloc(z)() or av_realloc(). ...
Definition: mem.c:190
16 bits gray, 16 bits alpha (big-endian)
Definition: pixfmt.h:205
uint8_t log2_chroma_h
Amount to shift the luma height right to find the chroma height.
Definition: pixdesc.h:100
int16_t ** chrVPixBuf
Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
int32_t * hChrFilterPos
Array of horizontal filter starting positions for each dst[i] for chroma planes.
planar YUV 4:4:4, 48bpp, (1 Cr & Cb sample per 1x1 Y samples), big-endian
Definition: pixfmt.h:133
#define AVERROR(e)
Definition: error.h:43
int hLumFilterSize
Horizontal filter size for luma/alpha pixels.
SwsFunc ff_getSwsFunc(SwsContext *c)
Return function pointer to fastest main scaler path function depending on architecture and available ...
Definition: swscale.c:773
#define PPC_ALTIVEC(flags)
Definition: cpu.h:26
#define SWS_MAX_REDUCE_CUTOFF
Definition: swscale.h:85
packed BGRA 8:8:8:8, 32bpp, BGRABGRA...
Definition: pixfmt.h:92
#define AV_LOG_DEBUG
Stuff which is only useful for libav* developers.
Definition: log.h:145
planar YUV 4:4:4 36bpp, (1 Cr & Cb sample per 1x1 Y & A samples), big-endian
Definition: pixfmt.h:175
planar YUV 4:2:2, 20bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian
Definition: pixfmt.h:154
planar YUV 4:2:0, 12bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (firs...
Definition: pixfmt.h:86
planar YUV 4:2:2, 32bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian
Definition: pixfmt.h:131
like NV12, with 10bpp per component, data in the high bits, zeros in the low bits, big-endian
Definition: pixfmt.h:231
void sws_printVec2(SwsVector *a, AVClass *log_ctx, int log_level)
Print with av_log() a textual representation of the vector a if log_level <= av_log_level.
Definition: utils.c:1628
#define SWS_CS_DEFAULT
Definition: swscale.h:93
int vChrBufSize
Number of vertical chroma lines allocated in the ring buffer.
#define X86_MMX(flags)
Definition: cpu.h:31
planar YUV 4:4:4 36bpp, (1 Cr & Cb sample per 1x1 Y & A samples), little-endian
Definition: pixfmt.h:176
#define FFMAX(a, b)
Definition: common.h:64
packed ARGB 8:8:8:8, 32bpp, ARGBARGB...
Definition: pixfmt.h:89
void sws_scaleVec(SwsVector *a, double scalar)
Scale all the coefficients of a by the scalar value.
Definition: utils.c:1495
int chrDstW
Width of destination chroma planes.
SwsVector * lumH
Definition: swscale.h:113
#define fail()
Definition: checkasm.h:80
packed RGB 16:16:16, 48bpp, 16B, 16G, 16R, the 2-byte value for each R/G/B component is stored as lit...
Definition: pixfmt.h:148
packed RGBA 8:8:8:8, 32bpp, RGBARGBA...
Definition: pixfmt.h:90
void sws_normalizeVec(SwsVector *a, double height)
Scale all the coefficients of a so that their sum equals height.
Definition: utils.c:1503
planar YUV 4:2:0 25bpp, (1 Cr & Cb sample per 2x2 Y & A samples, big-endian)
Definition: pixfmt.h:177
struct SwsContext * sws_getCachedContext(struct SwsContext *context, int srcW, int srcH, enum AVPixelFormat srcFormat, int dstW, int dstH, enum AVPixelFormat dstFormat, int flags, SwsFilter *srcFilter, SwsFilter *dstFilter, const double *param)
Check if context can be reused, otherwise reallocate a new one.
Definition: utils.c:1743
#define LICENSE_PREFIX
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
Definition: pixfmt.h:63
int32_t * hLumFilterPos
Array of horizontal filter starting positions for each dst[i] for luma/alpha planes.
void sws_freeFilter(SwsFilter *filter)
Definition: utils.c:1663
int hChrFilterSize
Horizontal filter size for chroma pixels.
static void filter(MpegAudioContext *s, int ch, const short *samples, int incr)
Definition: mpegaudioenc.c:307
SwsVector * sws_allocVec(int length)
Allocate and return an uninitialized vector with length coefficients.
Definition: utils.c:1432
ptrdiff_t uv_off_byte
offset (in bytes) between u and v planes
as above, but U and V bytes are swapped
Definition: pixfmt.h:87
int dstRange
0 = MPG YUV range, 1 = JPG YUV range (destination image).
#define APCK_SIZE
#define FFMIN(a, b)
Definition: common.h:66
packed RGB 1:2:1, 8bpp, (msb)1R 2G 1B(lsb)
Definition: pixfmt.h:85
planar YUV 4:2:0, 12bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV420P and setting col...
Definition: pixfmt.h:71
#define SWS_GAUSS
Definition: swscale.h:64
SwsVector * chrH
Definition: swscale.h:115
uint8_t * formatConvBuffer
#define INLINE_AMD3DNOW(flags)
Definition: cpu.h:77
int vLumBufSize
Number of vertical luma/alpha lines allocated in the ring buffer.
int16_t ** chrUPixBuf
Ring buffer for scaled horizontal chroma plane lines to be fed to the vertical scaler.
int32_t
SwsVector * sws_getIdentityVec(void)
Allocate and return a vector with just one coefficient, with value 1.0.
Definition: utils.c:1479
void sws_freeContext(SwsContext *c)
Free the swscaler context swsContext.
Definition: utils.c:1679
planar YUV 4:4:4 40bpp, (1 Cr & Cb sample per 1x1 Y & A samples, big-endian)
Definition: pixfmt.h:181
packed YUV 4:2:2, 16bpp, Y0 Cr Y1 Cb
Definition: pixfmt.h:201
#define FFABS(a)
Definition: common.h:61
planar YUV 4:2:2 27bpp, (1 Cr & Cb sample per 2x1 Y & A samples), big-endian
Definition: pixfmt.h:173
packed RGB 8:8:8, 24bpp, BGRBGR...
Definition: pixfmt.h:62
packed XYZ 4:4:4, 36 bpp, (msb) 12X, 12Y, 12Z (lsb), the 2-byte value for each X/Y/Z is stored as big...
Definition: pixfmt.h:191
planar YUV 4:2:0, 15bpp, (1 Cr & Cb sample per 2x2 Y samples), big-endian
Definition: pixfmt.h:151
enum AVPixelFormat pix_fmt
Definition: movenc.c:853
unsigned swscale_version(void)
Definition: utils.c:55
like NV12, with 10bpp per component, data in the high bits, zeros in the low bits, little-endian
Definition: pixfmt.h:230
int sws_setColorspaceDetails(struct SwsContext *c, const int inv_table[4], int srcRange, const int table[4], int dstRange, int brightness, int contrast, int saturation)
Definition: utils.c:790
int srcColorspaceTable[4]
int dstW
Width of destination luma/alpha planes.
planar YUV 4:2:2, 18bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian
Definition: pixfmt.h:160
packed RGB 16:16:16, 48bpp, 16B, 16G, 16R, the 2-byte value for each R/G/B component is stored as big...
Definition: pixfmt.h:147
packed BGR 5:6:5, 16bpp, (msb) 5B 6G 5R(lsb), big-endian
Definition: pixfmt.h:114
int32_t * vLumFilterPos
Array of vertical filter starting positions for each dst[i] for luma/alpha planes.
#define AV_PIX_FMT_BGR555
Definition: pixfmt.h:256
#define LIBAV_LICENSE
Definition: config.h:5
NULL
Definition: eval.c:55
packed RGB 3:3:2, 8bpp, (msb)2B 3G 3R(lsb)
Definition: pixfmt.h:80
double * coeff
pointer to the list of coefficients
Definition: swscale.h:107
#define AV_LOG_INFO
Standard information.
Definition: log.h:135
planar YUV 4:2:0, 24bpp, (1 Cr & Cb sample per 2x2 Y samples), big-endian
Definition: pixfmt.h:129
int dstColorspaceTable[4]
const AVClass * av_class
info on struct for av_log
planar YUV 4:4:4 32bpp, (1 Cr & Cb sample per 1x1 Y & A samples)
Definition: pixfmt.h:170
Descriptor that unambiguously describes how the bits of a pixel are stored in the up to 4 data planes...
Definition: pixdesc.h:80
void sws_freeVec(SwsVector *a)
Definition: utils.c:1654
planar GBR 4:4:4 30bpp, big-endian
Definition: pixfmt.h:165
planar YUV 4:2:2 48bpp, (1 Cr & Cb sample per 2x1 Y & A samples, big-endian)
Definition: pixfmt.h:185
int chrDstH
Height of destination chroma planes.
packed YUV 4:2:2, 16bpp, Y0 Cb Y1 Cr
Definition: pixfmt.h:60
static void getSubSampleFactors(int *h, int *v, enum AVPixelFormat format)
Definition: utils.c:783
#define SWS_AREA
Definition: swscale.h:62
planar YUV 4:1:0, 9bpp, (1 Cr & Cb sample per 4x4 Y samples)
Definition: pixfmt.h:65
planar YUV 4:2:2 48bpp, (1 Cr & Cb sample per 2x1 Y & A samples, little-endian)
Definition: pixfmt.h:186
static SwsVector * sws_diffVec(SwsVector *a, SwsVector *b)
Definition: utils.c:1543
const char * format
Definition: movenc.c:47
void sws_shiftVec(SwsVector *a, int shift)
Definition: utils.c:1578
int lumMmxextFilterCodeSize
Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes...
Describe the class of an AVClass context structure.
Definition: log.h:34
planar YUV 4:4:4 64bpp, (1 Cr & Cb sample per 1x1 Y & A samples, little-endian)
Definition: pixfmt.h:188
#define RET
Definition: utils.c:71
Y , 16bpp, big-endian.
Definition: pixfmt.h:94
int vLumFilterSize
Vertical filter size for luma/alpha pixels.
#define SWS_ACCURATE_RND
Definition: swscale.h:82
byte swapping routines
int chrMmxextFilterCodeSize
Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes.
planar YUV 4:2:0 22.5bpp, (1 Cr & Cb sample per 2x2 Y & A samples), big-endian
Definition: pixfmt.h:171
packed BGR 5:5:5, 16bpp, (msb)1A 5B 5G 5R(lsb), little-endian, most significant bit to 1 ...
Definition: pixfmt.h:117
int16_t * vChrFilter
Array of vertical filter coefficients for chroma planes.
int av_get_cpu_flags(void)
Return the flags which specify extensions supported by the CPU.
Definition: cpu.c:47
SwsVector * sws_cloneVec(SwsVector *a)
Allocate and return a clone of the vector a, that is a vector with the same coefficients as a...
Definition: utils.c:1614
#define SWS_POINT
Definition: swscale.h:61
int16_t * hLumFilter
Array of horizontal filter coefficients for luma/alpha planes.
SwsContext * sws_alloc_context(void)
Allocate an empty SwsContext.
Definition: utils.c:858
#define AV_PIX_FMT_BGR565
Definition: pixfmt.h:255
#define SWS_SPLINE
Definition: swscale.h:67
#define SWS_SINC
Definition: swscale.h:65
planar YUV 4:2:2 30bpp, (1 Cr & Cb sample per 2x1 Y & A samples, little-endian)
Definition: pixfmt.h:180
packed RGB 5:5:5, 16bpp, (msb)1A 5R 5G 5B(lsb), big-endian, most significant bit to 0 ...
Definition: pixfmt.h:111
#define SWS_BITEXACT
Definition: swscale.h:83
packed BGR 4:4:4, 16bpp, (msb)4A 4B 4G 4R(lsb), big-endian, most significant bits to 1 ...
Definition: pixfmt.h:142
planar YUV 4:4:4, 27bpp, (1 Cr & Cb sample per 1x1 Y samples), little-endian
Definition: pixfmt.h:156
planar YUV 4:4:4, 48bpp, (1 Cr & Cb sample per 1x1 Y samples), little-endian
Definition: pixfmt.h:132
int height
Definition: gxfenc.c:72
uint8_t is_supported_in
Definition: utils.c:74
void sws_convVec(SwsVector *a, SwsVector *b)
Definition: utils.c:1605
Y , 1bpp, 0 is black, 1 is white, in each byte pixels are ordered from the msb to the lsb...
Definition: pixfmt.h:69
#define INLINE_MMXEXT(flags)
Definition: cpu.h:80
static double sws_dcVec(SwsVector *a)
Definition: utils.c:1484
#define CONFIG_SWSCALE_ALPHA
Definition: config.h:408
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:59
Y , 8bpp.
Definition: pixfmt.h:67
double param[2]
Input parameters for scaling algorithms that need them.
Y , 1bpp, 0 is white, 1 is black, in each byte pixels are ordered from the msb to the lsb...
Definition: pixfmt.h:68
planar GBRA 4:4:4:4 32bpp
Definition: pixfmt.h:208
#define FF_ALLOC_OR_GOTO(ctx, p, size, label)
Definition: internal.h:120
planar GBR 4:4:4 27bpp, little-endian
Definition: pixfmt.h:164
packed RGB 16:16:16, 48bpp, 16R, 16G, 16B, the 2-byte value for each R/G/B component is stored as big...
Definition: pixfmt.h:106
enum AVPixelFormat srcFormat
Source pixel format.
planar YUV 4:2:2, 32bpp, (1 Cr & Cb sample per 2x1 Y samples), little-endian
Definition: pixfmt.h:130
planar YUV 4:4:4, 24bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV444P and setting col...
Definition: pixfmt.h:73
packed RGB 3:3:2, 8bpp, (msb)2R 3G 3B(lsb)
Definition: pixfmt.h:83
planar YUV 4:1:1, 12bpp, (1 Cr & Cb sample per 4x1 Y samples)
Definition: pixfmt.h:66
planar YUV 4:2:0 25bpp, (1 Cr & Cb sample per 2x2 Y & A samples, little-endian)
Definition: pixfmt.h:178
#define SWS_PARAM_DEFAULT
Definition: swscale.h:72
#define SWS_FULL_CHR_H_INP
Definition: swscale.h:80
SwsFunc swscale
Note that src, dst, srcStride, dstStride will be copied in the sws_scale() wrapper so they can be fre...
#define MAX_FILTER_SIZE
packed XYZ 4:4:4, 36 bpp, (msb) 12X, 12Y, 12Z (lsb), the 2-byte value for each X/Y/Z is stored as lit...
Definition: pixfmt.h:190
int srcFormatBpp
Number of bits per pixel of the source pixel format.
const AVClass ff_sws_context_class
Definition: options.c:69
Y , 16bpp, little-endian.
Definition: pixfmt.h:95
uint8_t is_supported_endianness
Definition: utils.c:76
16 bits gray, 16 bits alpha (little-endian)
Definition: pixfmt.h:206
int sws_isSupportedInput(enum AVPixelFormat pix_fmt)
Return a positive value if pix_fmt is a supported input format, 0 otherwise.
Definition: utils.c:192
planar YUV 4:2:2 30bpp, (1 Cr & Cb sample per 2x1 Y & A samples, big-endian)
Definition: pixfmt.h:179
static av_cold int initFilter(int16_t **outFilter, int32_t **filterPos, int *outFilterSize, int xInc, int srcW, int dstW, int filterAlign, int one, int flags, int cpu_flags, SwsVector *srcFilter, SwsVector *dstFilter, double param[2], int is_horizontal)
Definition: utils.c:232
packed BGR 5:5:5, 16bpp, (msb)1A 5B 5G 5R(lsb), big-endian, most significant bit to 1 ...
Definition: pixfmt.h:116
number of pixel formats, DO NOT USE THIS if you want to link with shared libav* because the number of...
Definition: pixfmt.h:233
int16_t * vLumFilter
Array of vertical filter coefficients for luma/alpha planes.
SwsVector * sws_getConstVec(double c, int length)
Allocate and return a vector with length coefficients, all with the same value c. ...
Definition: utils.c:1465
planar YUV 4:4:0 (1 Cr & Cb sample per 1x2 Y samples)
Definition: pixfmt.h:96
planar GBR 4:4:4 48bpp, little-endian
Definition: pixfmt.h:168
int16_t * hChrFilter
Array of horizontal filter coefficients for chroma planes.
#define LOCAL_MANGLE(a)
Definition: asm.h:108
packed BGR 4:4:4, 16bpp, (msb)4A 4B 4G 4R(lsb), little-endian, most significant bits to 1 ...
Definition: pixfmt.h:141
const char * av_get_pix_fmt_name(enum AVPixelFormat pix_fmt)
Return the short name for a pixel format, NULL in case pix_fmt is unknown.
Definition: pixdesc.c:1704
planar YUV 4:2:2 27bpp, (1 Cr & Cb sample per 2x1 Y & A samples), little-endian
Definition: pixfmt.h:174
int chrDstHSubSample
Binary logarithm of horizontal subsampling factor between luma/alpha and chroma planes in destination...
int chrSrcW
Width of source chroma planes.
#define isGray(x)
int depth
Number of bits in the component.
Definition: pixdesc.h:57
void ff_get_unscaled_swscale(SwsContext *c)
Set c->swscale to an unscaled converter if one exists for the specific source and destination formats...
planar GBRA 4:4:4:4 64bpp, little-endian
Definition: pixfmt.h:210
int srcW
Width of source luma/alpha planes.
packed YUV 4:1:1, 12bpp, Cb Y0 Y1 Cr Y2 Y3
Definition: pixfmt.h:79
float min
int chrSrcVSubSample
Binary logarithm of vertical subsampling factor between luma/alpha and chroma planes in source image...
int flags
Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc...
AVPixelFormat
Pixel format.
Definition: pixfmt.h:57
void * av_mallocz(size_t size)
Allocate a block of size bytes with alignment suitable for all memory accesses (including vectors if ...
Definition: mem.c:211
void(* rgb15to16)(const uint8_t *src, uint8_t *dst, int src_size)
Definition: rgb2rgb.c:51
planar YUV 4:4:4, 30bpp, (1 Cr & Cb sample per 1x1 Y samples), big-endian
Definition: pixfmt.h:157
for(j=16;j >0;--j)
#define FF_ALLOCZ_OR_GOTO(ctx, p, size, label)
Definition: internal.h:129
planar GBR 4:4:4 30bpp, little-endian
Definition: pixfmt.h:166
static SwsVector * sws_getConvVec(SwsVector *a, SwsVector *b)
Definition: utils.c:1508
packed RGBA 16:16:16:16, 64bpp, 16R, 16G, 16B, 16A, the 2-byte value for each R/G/B/A component is st...
Definition: pixfmt.h:197
#define LIBSWSCALE_VERSION_INT
Definition: version.h:33
#define AV_CEIL_RSHIFT(a, b)
Fast a / (1 << b) rounded toward +inf, assuming a >= 0 and b >= 0.
Definition: common.h:57
planar YUV 4:2:2, 20bpp, (1 Cr & Cb sample per 2x1 Y samples), big-endian
Definition: pixfmt.h:153
const char * swscale_license(void)
Return the libswscale license.
Definition: utils.c:65