falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
76381 Commits
32 Branches
844MB
Tree: 4d0d85c94a
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '4d0d85c94a'
${ noResults }
FFmpeg/libswscale/utils.c

2380 lines
86KB
Raw Blame History 

  1. /*

  2.  * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>

  3.  *

  4.  * This file is part of FFmpeg.

  5.  *

  6.  * FFmpeg 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.  * FFmpeg 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 FFmpeg; 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 _DEFAULT_SOURCE

  24. #define _SVID_SOURCE // needed for MAP_ANONYMOUS

  25. #define _DARWIN_C_SOURCE // needed for MAP_ANON

  26. #include <inttypes.h>

  27. #include <math.h>

  28. #include <stdio.h>

  29. #include <string.h>

  30. #if HAVE_SYS_MMAN_H

  31. #include <sys/mman.h>

  32. #if defined(MAP_ANON) && !defined(MAP_ANONYMOUS)

  33. #define MAP_ANONYMOUS MAP_ANON

  34. #endif

  35. #endif

  36. #if HAVE_VIRTUALALLOC

  37. #define WIN32_LEAN_AND_MEAN

  38. #include <windows.h>

  39. #endif

  40. 

  41. #include "libavutil/attributes.h"

  42. #include "libavutil/avassert.h"

  43. #include "libavutil/avutil.h"

  44. #include "libavutil/bswap.h"

  45. #include "libavutil/cpu.h"

  46. #include "libavutil/imgutils.h"

  47. #include "libavutil/intreadwrite.h"

  48. #include "libavutil/mathematics.h"

  49. #include "libavutil/opt.h"

  50. #include "libavutil/pixdesc.h"

  51. #include "libavutil/ppc/cpu.h"

  52. #include "libavutil/x86/asm.h"

  53. #include "libavutil/x86/cpu.h"

  54. #include "rgb2rgb.h"

  55. #include "swscale.h"

  56. #include "swscale_internal.h"

  57. 

  58. static void handle_formats(SwsContext *c);

  59. 

  60. unsigned swscale_version(void)

  61. {

  62.     av_assert0(LIBSWSCALE_VERSION_MICRO >= 100);

  63.     return LIBSWSCALE_VERSION_INT;

  64. }

  65. 

  66. const char *swscale_configuration(void)

  67. {

  68.     return FFMPEG_CONFIGURATION;

  69. }

  70. 

  71. const char *swscale_license(void)

  72. {

  73. #define LICENSE_PREFIX "libswscale license: "

  74.     return LICENSE_PREFIX FFMPEG_LICENSE + sizeof(LICENSE_PREFIX) - 1;

  75. }

  76. 

  77. typedef struct FormatEntry {

  78.     uint8_t is_supported_in         :1;

  79.     uint8_t is_supported_out        :1;

  80.     uint8_t is_supported_endianness :1;

  81. } FormatEntry;

  82. 

  83. static const FormatEntry format_entries[AV_PIX_FMT_NB] = {

  84.     [AV_PIX_FMT_YUV420P]     = { 1, 1 },

  85.     [AV_PIX_FMT_YUYV422]     = { 1, 1 },

  86.     [AV_PIX_FMT_RGB24]       = { 1, 1 },

  87.     [AV_PIX_FMT_BGR24]       = { 1, 1 },

  88.     [AV_PIX_FMT_YUV422P]     = { 1, 1 },

  89.     [AV_PIX_FMT_YUV444P]     = { 1, 1 },

  90.     [AV_PIX_FMT_YUV410P]     = { 1, 1 },

  91.     [AV_PIX_FMT_YUV411P]     = { 1, 1 },

  92.     [AV_PIX_FMT_GRAY8]       = { 1, 1 },

  93.     [AV_PIX_FMT_MONOWHITE]   = { 1, 1 },

  94.     [AV_PIX_FMT_MONOBLACK]   = { 1, 1 },

  95.     [AV_PIX_FMT_PAL8]        = { 1, 0 },

  96.     [AV_PIX_FMT_YUVJ420P]    = { 1, 1 },

  97.     [AV_PIX_FMT_YUVJ411P]    = { 1, 1 },

  98.     [AV_PIX_FMT_YUVJ422P]    = { 1, 1 },

  99.     [AV_PIX_FMT_YUVJ444P]    = { 1, 1 },

  100.     [AV_PIX_FMT_YVYU422]     = { 1, 1 },

  101.     [AV_PIX_FMT_UYVY422]     = { 1, 1 },

  102.     [AV_PIX_FMT_UYYVYY411]   = { 0, 0 },

  103.     [AV_PIX_FMT_BGR8]        = { 1, 1 },

  104.     [AV_PIX_FMT_BGR4]        = { 0, 1 },

  105.     [AV_PIX_FMT_BGR4_BYTE]   = { 1, 1 },

  106.     [AV_PIX_FMT_RGB8]        = { 1, 1 },

  107.     [AV_PIX_FMT_RGB4]        = { 0, 1 },

  108.     [AV_PIX_FMT_RGB4_BYTE]   = { 1, 1 },

  109.     [AV_PIX_FMT_NV12]        = { 1, 1 },

  110.     [AV_PIX_FMT_NV21]        = { 1, 1 },

  111.     [AV_PIX_FMT_ARGB]        = { 1, 1 },

  112.     [AV_PIX_FMT_RGBA]        = { 1, 1 },

  113.     [AV_PIX_FMT_ABGR]        = { 1, 1 },

  114.     [AV_PIX_FMT_BGRA]        = { 1, 1 },

  115.     [AV_PIX_FMT_0RGB]        = { 1, 1 },

  116.     [AV_PIX_FMT_RGB0]        = { 1, 1 },

  117.     [AV_PIX_FMT_0BGR]        = { 1, 1 },

  118.     [AV_PIX_FMT_BGR0]        = { 1, 1 },

  119.     [AV_PIX_FMT_GRAY16BE]    = { 1, 1 },

  120.     [AV_PIX_FMT_GRAY16LE]    = { 1, 1 },

  121.     [AV_PIX_FMT_YUV440P]     = { 1, 1 },

  122.     [AV_PIX_FMT_YUVJ440P]    = { 1, 1 },

  123.     [AV_PIX_FMT_YUV440P10LE] = { 1, 1 },

  124.     [AV_PIX_FMT_YUV440P10BE] = { 1, 1 },

  125.     [AV_PIX_FMT_YUV440P12LE] = { 1, 1 },

  126.     [AV_PIX_FMT_YUV440P12BE] = { 1, 1 },

  127.     [AV_PIX_FMT_YUVA420P]    = { 1, 1 },

  128.     [AV_PIX_FMT_YUVA422P]    = { 1, 1 },

  129.     [AV_PIX_FMT_YUVA444P]    = { 1, 1 },

  130.     [AV_PIX_FMT_YUVA420P9BE] = { 1, 1 },

  131.     [AV_PIX_FMT_YUVA420P9LE] = { 1, 1 },

  132.     [AV_PIX_FMT_YUVA422P9BE] = { 1, 1 },

  133.     [AV_PIX_FMT_YUVA422P9LE] = { 1, 1 },

  134.     [AV_PIX_FMT_YUVA444P9BE] = { 1, 1 },

  135.     [AV_PIX_FMT_YUVA444P9LE] = { 1, 1 },

  136.     [AV_PIX_FMT_YUVA420P10BE]= { 1, 1 },

  137.     [AV_PIX_FMT_YUVA420P10LE]= { 1, 1 },

  138.     [AV_PIX_FMT_YUVA422P10BE]= { 1, 1 },

  139.     [AV_PIX_FMT_YUVA422P10LE]= { 1, 1 },

  140.     [AV_PIX_FMT_YUVA444P10BE]= { 1, 1 },

  141.     [AV_PIX_FMT_YUVA444P10LE]= { 1, 1 },

  142.     [AV_PIX_FMT_YUVA420P16BE]= { 1, 1 },

  143.     [AV_PIX_FMT_YUVA420P16LE]= { 1, 1 },

  144.     [AV_PIX_FMT_YUVA422P16BE]= { 1, 1 },

  145.     [AV_PIX_FMT_YUVA422P16LE]= { 1, 1 },

  146.     [AV_PIX_FMT_YUVA444P16BE]= { 1, 1 },

  147.     [AV_PIX_FMT_YUVA444P16LE]= { 1, 1 },

  148.     [AV_PIX_FMT_RGB48BE]     = { 1, 1 },

  149.     [AV_PIX_FMT_RGB48LE]     = { 1, 1 },

  150.     [AV_PIX_FMT_RGBA64BE]    = { 1, 1, 1 },

  151.     [AV_PIX_FMT_RGBA64LE]    = { 1, 1, 1 },

  152.     [AV_PIX_FMT_RGB565BE]    = { 1, 1 },

  153.     [AV_PIX_FMT_RGB565LE]    = { 1, 1 },

  154.     [AV_PIX_FMT_RGB555BE]    = { 1, 1 },

  155.     [AV_PIX_FMT_RGB555LE]    = { 1, 1 },

  156.     [AV_PIX_FMT_BGR565BE]    = { 1, 1 },

  157.     [AV_PIX_FMT_BGR565LE]    = { 1, 1 },

  158.     [AV_PIX_FMT_BGR555BE]    = { 1, 1 },

  159.     [AV_PIX_FMT_BGR555LE]    = { 1, 1 },

  160.     [AV_PIX_FMT_YUV420P16LE] = { 1, 1 },

  161.     [AV_PIX_FMT_YUV420P16BE] = { 1, 1 },

  162.     [AV_PIX_FMT_YUV422P16LE] = { 1, 1 },

  163.     [AV_PIX_FMT_YUV422P16BE] = { 1, 1 },

  164.     [AV_PIX_FMT_YUV444P16LE] = { 1, 1 },

  165.     [AV_PIX_FMT_YUV444P16BE] = { 1, 1 },

  166.     [AV_PIX_FMT_RGB444LE]    = { 1, 1 },

  167.     [AV_PIX_FMT_RGB444BE]    = { 1, 1 },

  168.     [AV_PIX_FMT_BGR444LE]    = { 1, 1 },

  169.     [AV_PIX_FMT_BGR444BE]    = { 1, 1 },

  170.     [AV_PIX_FMT_YA8]         = { 1, 1 },

  171.     [AV_PIX_FMT_YA16BE]      = { 1, 0 },

  172.     [AV_PIX_FMT_YA16LE]      = { 1, 0 },

  173.     [AV_PIX_FMT_BGR48BE]     = { 1, 1 },

  174.     [AV_PIX_FMT_BGR48LE]     = { 1, 1 },

  175.     [AV_PIX_FMT_BGRA64BE]    = { 1, 1, 1 },

  176.     [AV_PIX_FMT_BGRA64LE]    = { 1, 1, 1 },

  177.     [AV_PIX_FMT_YUV420P9BE]  = { 1, 1 },

  178.     [AV_PIX_FMT_YUV420P9LE]  = { 1, 1 },

  179.     [AV_PIX_FMT_YUV420P10BE] = { 1, 1 },

  180.     [AV_PIX_FMT_YUV420P10LE] = { 1, 1 },

  181.     [AV_PIX_FMT_YUV420P12BE] = { 1, 1 },

  182.     [AV_PIX_FMT_YUV420P12LE] = { 1, 1 },

  183.     [AV_PIX_FMT_YUV420P14BE] = { 1, 1 },

  184.     [AV_PIX_FMT_YUV420P14LE] = { 1, 1 },

  185.     [AV_PIX_FMT_YUV422P9BE]  = { 1, 1 },

  186.     [AV_PIX_FMT_YUV422P9LE]  = { 1, 1 },

  187.     [AV_PIX_FMT_YUV422P10BE] = { 1, 1 },

  188.     [AV_PIX_FMT_YUV422P10LE] = { 1, 1 },

  189.     [AV_PIX_FMT_YUV422P12BE] = { 1, 1 },

  190.     [AV_PIX_FMT_YUV422P12LE] = { 1, 1 },

  191.     [AV_PIX_FMT_YUV422P14BE] = { 1, 1 },

  192.     [AV_PIX_FMT_YUV422P14LE] = { 1, 1 },

  193.     [AV_PIX_FMT_YUV444P9BE]  = { 1, 1 },

  194.     [AV_PIX_FMT_YUV444P9LE]  = { 1, 1 },

  195.     [AV_PIX_FMT_YUV444P10BE] = { 1, 1 },

  196.     [AV_PIX_FMT_YUV444P10LE] = { 1, 1 },

  197.     [AV_PIX_FMT_YUV444P12BE] = { 1, 1 },

  198.     [AV_PIX_FMT_YUV444P12LE] = { 1, 1 },

  199.     [AV_PIX_FMT_YUV444P14BE] = { 1, 1 },

  200.     [AV_PIX_FMT_YUV444P14LE] = { 1, 1 },

  201.     [AV_PIX_FMT_GBRP]        = { 1, 1 },

  202.     [AV_PIX_FMT_GBRP9LE]     = { 1, 1 },

  203.     [AV_PIX_FMT_GBRP9BE]     = { 1, 1 },

  204.     [AV_PIX_FMT_GBRP10LE]    = { 1, 1 },

  205.     [AV_PIX_FMT_GBRP10BE]    = { 1, 1 },

  206.     [AV_PIX_FMT_GBRP12LE]    = { 1, 1 },

  207.     [AV_PIX_FMT_GBRP12BE]    = { 1, 1 },

  208.     [AV_PIX_FMT_GBRP14LE]    = { 1, 1 },

  209.     [AV_PIX_FMT_GBRP14BE]    = { 1, 1 },

  210.     [AV_PIX_FMT_GBRP16LE]    = { 1, 0 },

  211.     [AV_PIX_FMT_GBRP16BE]    = { 1, 0 },

  212.     [AV_PIX_FMT_GBRAP]       = { 1, 1 },

  213.     [AV_PIX_FMT_GBRAP16LE]   = { 1, 0 },

  214.     [AV_PIX_FMT_GBRAP16BE]   = { 1, 0 },

  215.     [AV_PIX_FMT_BAYER_BGGR8] = { 1, 0 },

  216.     [AV_PIX_FMT_BAYER_RGGB8] = { 1, 0 },

  217.     [AV_PIX_FMT_BAYER_GBRG8] = { 1, 0 },

  218.     [AV_PIX_FMT_BAYER_GRBG8] = { 1, 0 },

  219.     [AV_PIX_FMT_BAYER_BGGR16LE] = { 1, 0 },

  220.     [AV_PIX_FMT_BAYER_BGGR16BE] = { 1, 0 },

  221.     [AV_PIX_FMT_BAYER_RGGB16LE] = { 1, 0 },

  222.     [AV_PIX_FMT_BAYER_RGGB16BE] = { 1, 0 },

  223.     [AV_PIX_FMT_BAYER_GBRG16LE] = { 1, 0 },

  224.     [AV_PIX_FMT_BAYER_GBRG16BE] = { 1, 0 },

  225.     [AV_PIX_FMT_BAYER_GRBG16LE] = { 1, 0 },

  226.     [AV_PIX_FMT_BAYER_GRBG16BE] = { 1, 0 },

  227.     [AV_PIX_FMT_XYZ12BE]     = { 1, 1, 1 },

  228.     [AV_PIX_FMT_XYZ12LE]     = { 1, 1, 1 },

  229.     [AV_PIX_FMT_AYUV64LE]    = { 1, 1},

  230. };

  231. 

  232. int sws_isSupportedInput(enum AVPixelFormat pix_fmt)

  233. {

  234.     return (unsigned)pix_fmt < AV_PIX_FMT_NB ?

  235.            format_entries[pix_fmt].is_supported_in : 0;

  236. }

  237. 

  238. int sws_isSupportedOutput(enum AVPixelFormat pix_fmt)

  239. {

  240.     return (unsigned)pix_fmt < AV_PIX_FMT_NB ?

  241.            format_entries[pix_fmt].is_supported_out : 0;

  242. }

  243. 

  244. int sws_isSupportedEndiannessConversion(enum AVPixelFormat pix_fmt)

  245. {

  246.     return (unsigned)pix_fmt < AV_PIX_FMT_NB ?

  247.            format_entries[pix_fmt].is_supported_endianness : 0;

  248. }

  249. 

  250. static double getSplineCoeff(double a, double b, double c, double d,

  251.                              double dist)

  252. {

  253.     if (dist <= 1.0)

  254.         return ((d * dist + c) * dist + b) * dist + a;

  255.     else

  256.         return getSplineCoeff(0.0,

  257.                                b + 2.0 * c + 3.0 * d,

  258.                                c + 3.0 * d,

  259.                               -b - 3.0 * c - 6.0 * d,

  260.                               dist - 1.0);

  261. }

  262. 

  263. static av_cold int get_local_pos(SwsContext *s, int chr_subsample, int pos, int dir)

  264. {

  265.     if (pos == -1 || pos <= -513) {

  266.         pos = (128 << chr_subsample) - 128;

  267.     }

  268.     pos += 128; // relative to ideal left edge

  269.     return pos >> chr_subsample;

  270. }

  271. 

  272. typedef struct {

  273.     int flag;                   ///< flag associated to the algorithm

  274.     const char *description;    ///< human-readable description

  275.     int size_factor;            ///< size factor used when initing the filters

  276. } ScaleAlgorithm;

  277. 

  278. static const ScaleAlgorithm scale_algorithms[] = {

  279.     { SWS_AREA,          "area averaging",                  1 /* downscale only, for upscale it is bilinear */ },

  280.     { SWS_BICUBIC,       "bicubic",                         4 },

  281.     { SWS_BICUBLIN,      "luma bicubic / chroma bilinear", -1 },

  282.     { SWS_BILINEAR,      "bilinear",                        2 },

  283.     { SWS_FAST_BILINEAR, "fast bilinear",                  -1 },

  284.     { SWS_GAUSS,         "Gaussian",                        8 /* infinite ;) */ },

  285.     { SWS_LANCZOS,       "Lanczos",                        -1 /* custom */ },

  286.     { SWS_POINT,         "nearest neighbor / point",       -1 },

  287.     { SWS_SINC,          "sinc",                           20 /* infinite ;) */ },

  288.     { SWS_SPLINE,        "bicubic spline",                 20 /* infinite :)*/ },

  289.     { SWS_X,             "experimental",                    8 },

  290. };

  291. 

  292. static av_cold int initFilter(int16_t **outFilter, int32_t **filterPos,

  293.                               int *outFilterSize, int xInc, int srcW,

  294.                               int dstW, int filterAlign, int one,

  295.                               int flags, int cpu_flags,

  296.                               SwsVector *srcFilter, SwsVector *dstFilter,

  297.                               double param[2], int srcPos, int dstPos)

  298. {

  299.     int i;

  300.     int filterSize;

  301.     int filter2Size;

  302.     int minFilterSize;

  303.     int64_t *filter    = NULL;

  304.     int64_t *filter2   = NULL;

  305.     const int64_t fone = 1LL << (54 - FFMIN(av_log2(srcW/dstW), 8));

  306.     int ret            = -1;

  307. 

  308.     emms_c(); // FIXME should not be required but IS (even for non-MMX versions)

  309. 

  310.     // NOTE: the +3 is for the MMX(+1) / SSE(+3) scaler which reads over the end

  311.     FF_ALLOC_ARRAY_OR_GOTO(NULL, *filterPos, (dstW + 3), sizeof(**filterPos), fail);

  312. 

  313.     if (FFABS(xInc - 0x10000) < 10 && srcPos == dstPos) { // unscaled

  314.         int i;

  315.         filterSize = 1;

  316.         FF_ALLOCZ_ARRAY_OR_GOTO(NULL, filter,

  317.                                 dstW, sizeof(*filter) * filterSize, fail);

  318. 

  319.         for (i = 0; i < dstW; i++) {

  320.             filter[i * filterSize] = fone;

  321.             (*filterPos)[i]        = i;

  322.         }

  323.     } else if (flags & SWS_POINT) { // lame looking point sampling mode

  324.         int i;

  325.         int64_t xDstInSrc;

  326.         filterSize = 1;

  327.         FF_ALLOC_ARRAY_OR_GOTO(NULL, filter,

  328.                                dstW, sizeof(*filter) * filterSize, fail);

  329. 

  330.         xDstInSrc = ((dstPos*(int64_t)xInc)>>8) - ((srcPos*0x8000LL)>>7);

  331.         for (i = 0; i < dstW; i++) {

  332.             int xx = (xDstInSrc - ((filterSize - 1) << 15) + (1 << 15)) >> 16;

  333. 

  334.             (*filterPos)[i] = xx;

  335.             filter[i]       = fone;

  336.             xDstInSrc      += xInc;

  337.         }

  338.     } else if ((xInc <= (1 << 16) && (flags & SWS_AREA)) ||

  339.                (flags & SWS_FAST_BILINEAR)) { // bilinear upscale

  340.         int i;

  341.         int64_t xDstInSrc;

  342.         filterSize = 2;

  343.         FF_ALLOC_ARRAY_OR_GOTO(NULL, filter,

  344.                                dstW, sizeof(*filter) * filterSize, fail);

  345. 

  346.         xDstInSrc = ((dstPos*(int64_t)xInc)>>8) - ((srcPos*0x8000LL)>>7);

  347.         for (i = 0; i < dstW; i++) {

  348.             int xx = (xDstInSrc - ((filterSize - 1) << 15) + (1 << 15)) >> 16;

  349.             int j;

  350. 

  351.             (*filterPos)[i] = xx;

  352.             // bilinear upscale / linear interpolate / area averaging

  353.             for (j = 0; j < filterSize; j++) {

  354.                 int64_t coeff= fone - FFABS(((int64_t)xx<<16) - xDstInSrc)*(fone>>16);

  355.                 if (coeff < 0)

  356.                     coeff = 0;

  357.                 filter[i * filterSize + j] = coeff;

  358.                 xx++;

  359.             }

  360.             xDstInSrc += xInc;

  361.         }

  362.     } else {

  363.         int64_t xDstInSrc;

  364.         int sizeFactor = -1;

  365. 

  366.         for (i = 0; i < FF_ARRAY_ELEMS(scale_algorithms); i++) {

  367.             if (flags & scale_algorithms[i].flag && scale_algorithms[i].size_factor > 0) {

  368.                 sizeFactor = scale_algorithms[i].size_factor;

  369.                 break;

  370.             }

  371.         }

  372.         if (flags & SWS_LANCZOS)

  373.             sizeFactor = param[0] != SWS_PARAM_DEFAULT ? ceil(2 * param[0]) : 6;

  374.         av_assert0(sizeFactor > 0);

  375. 

  376.         if (xInc <= 1 << 16)

  377.             filterSize = 1 + sizeFactor;    // upscale

  378.         else

  379.             filterSize = 1 + (sizeFactor * srcW + dstW - 1) / dstW;

  380. 

  381.         filterSize = FFMIN(filterSize, srcW - 2);

  382.         filterSize = FFMAX(filterSize, 1);

  383. 

  384.         FF_ALLOC_ARRAY_OR_GOTO(NULL, filter,

  385.                                dstW, sizeof(*filter) * filterSize, fail);

  386. 

  387.         xDstInSrc = ((dstPos*(int64_t)xInc)>>7) - ((srcPos*0x10000LL)>>7);

  388.         for (i = 0; i < dstW; i++) {

  389.             int xx = (xDstInSrc - ((int64_t)(filterSize - 2) << 16)) / (1 << 17);

  390.             int j;

  391.             (*filterPos)[i] = xx;

  392.             for (j = 0; j < filterSize; j++) {

  393.                 int64_t d = (FFABS(((int64_t)xx * (1 << 17)) - xDstInSrc)) << 13;

  394.                 double floatd;

  395.                 int64_t coeff;

  396. 

  397.                 if (xInc > 1 << 16)

  398.                     d = d * dstW / srcW;

  399.                 floatd = d * (1.0 / (1 << 30));

  400. 

  401.                 if (flags & SWS_BICUBIC) {

  402.                     int64_t B = (param[0] != SWS_PARAM_DEFAULT ? param[0] :   0) * (1 << 24);

  403.                     int64_t C = (param[1] != SWS_PARAM_DEFAULT ? param[1] : 0.6) * (1 << 24);

  404. 

  405.                     if (d >= 1LL << 31) {

  406.                         coeff = 0.0;

  407.                     } else {

  408.                         int64_t dd  = (d  * d) >> 30;

  409.                         int64_t ddd = (dd * d) >> 30;

  410. 

  411.                         if (d < 1LL << 30)

  412.                             coeff =  (12 * (1 << 24) -  9 * B - 6 * C) * ddd +

  413.                                     (-18 * (1 << 24) + 12 * B + 6 * C) *  dd +

  414.                                       (6 * (1 << 24) -  2 * B)         * (1 << 30);

  415.                         else

  416.                             coeff =      (-B -  6 * C) * ddd +

  417.                                       (6 * B + 30 * C) * dd  +

  418.                                     (-12 * B - 48 * C) * d   +

  419.                                       (8 * B + 24 * C) * (1 << 30);

  420.                     }

  421.                     coeff /= (1LL<<54)/fone;

  422.                 }

  423. #if 0

  424.                 else if (flags & SWS_X) {

  425.                     double p  = param ? param * 0.01 : 0.3;

  426.                     coeff     = d ? sin(d * M_PI) / (d * M_PI) : 1.0;

  427.                     coeff    *= pow(2.0, -p * d * d);

  428.                 }

  429. #endif

  430.                 else if (flags & SWS_X) {

  431.                     double A = param[0] != SWS_PARAM_DEFAULT ? param[0] : 1.0;

  432.                     double c;

  433. 

  434.                     if (floatd < 1.0)

  435.                         c = cos(floatd * M_PI);

  436.                     else

  437.                         c = -1.0;

  438.                     if (c < 0.0)

  439.                         c = -pow(-c, A);

  440.                     else

  441.                         c = pow(c, A);

  442.                     coeff = (c * 0.5 + 0.5) * fone;

  443.                 } else if (flags & SWS_AREA) {

  444.                     int64_t d2 = d - (1 << 29);

  445.                     if (d2 * xInc < -(1LL << (29 + 16)))

  446.                         coeff = 1.0 * (1LL << (30 + 16));

  447.                     else if (d2 * xInc < (1LL << (29 + 16)))

  448.                         coeff = -d2 * xInc + (1LL << (29 + 16));

  449.                     else

  450.                         coeff = 0.0;

  451.                     coeff *= fone >> (30 + 16);

  452.                 } else if (flags & SWS_GAUSS) {

  453.                     double p = param[0] != SWS_PARAM_DEFAULT ? param[0] : 3.0;

  454.                     coeff = (pow(2.0, -p * floatd * floatd)) * fone;

  455.                 } else if (flags & SWS_SINC) {

  456.                     coeff = (d ? sin(floatd * M_PI) / (floatd * M_PI) : 1.0) * fone;

  457.                 } else if (flags & SWS_LANCZOS) {

  458.                     double p = param[0] != SWS_PARAM_DEFAULT ? param[0] : 3.0;

  459.                     coeff = (d ? sin(floatd * M_PI) * sin(floatd * M_PI / p) /

  460.                              (floatd * floatd * M_PI * M_PI / p) : 1.0) * fone;

  461.                     if (floatd > p)

  462.                         coeff = 0;

  463.                 } else if (flags & SWS_BILINEAR) {

  464.                     coeff = (1 << 30) - d;

  465.                     if (coeff < 0)

  466.                         coeff = 0;

  467.                     coeff *= fone >> 30;

  468.                 } else if (flags & SWS_SPLINE) {

  469.                     double p = -2.196152422706632;

  470.                     coeff = getSplineCoeff(1.0, 0.0, p, -p - 1.0, floatd) * fone;

  471.                 } else {

  472.                     av_assert0(0);

  473.                 }

  474. 

  475.                 filter[i * filterSize + j] = coeff;

  476.                 xx++;

  477.             }

  478.             xDstInSrc += 2 * xInc;

  479.         }

  480.     }

  481. 

  482.     /* apply src & dst Filter to filter -> filter2

  483.      * av_free(filter);

  484.      */

  485.     av_assert0(filterSize > 0);

  486.     filter2Size = filterSize;

  487.     if (srcFilter)

  488.         filter2Size += srcFilter->length - 1;

  489.     if (dstFilter)

  490.         filter2Size += dstFilter->length - 1;

  491.     av_assert0(filter2Size > 0);

  492.     FF_ALLOCZ_ARRAY_OR_GOTO(NULL, filter2, dstW, filter2Size * sizeof(*filter2), fail);

  493. 

  494.     for (i = 0; i < dstW; i++) {

  495.         int j, k;

  496. 

  497.         if (srcFilter) {

  498.             for (k = 0; k < srcFilter->length; k++) {

  499.                 for (j = 0; j < filterSize; j++)

  500.                     filter2[i * filter2Size + k + j] +=

  501.                         srcFilter->coeff[k] * filter[i * filterSize + j];

  502.             }

  503.         } else {

  504.             for (j = 0; j < filterSize; j++)

  505.                 filter2[i * filter2Size + j] = filter[i * filterSize + j];

  506.         }

  507.         // FIXME dstFilter

  508. 

  509.         (*filterPos)[i] += (filterSize - 1) / 2 - (filter2Size - 1) / 2;

  510.     }

  511.     av_freep(&filter);

  512. 

  513.     /* try to reduce the filter-size (step1 find size and shift left) */

  514.     // Assume it is near normalized (*0.5 or *2.0 is OK but * 0.001 is not).

  515.     minFilterSize = 0;

  516.     for (i = dstW - 1; i >= 0; i--) {

  517.         int min = filter2Size;

  518.         int j;

  519.         int64_t cutOff = 0.0;

  520. 

  521.         /* get rid of near zero elements on the left by shifting left */

  522.         for (j = 0; j < filter2Size; j++) {

  523.             int k;

  524.             cutOff += FFABS(filter2[i * filter2Size]);

  525. 

  526.             if (cutOff > SWS_MAX_REDUCE_CUTOFF * fone)

  527.                 break;

  528. 

  529.             /* preserve monotonicity because the core can't handle the

  530.              * filter otherwise */

  531.             if (i < dstW - 1 && (*filterPos)[i] >= (*filterPos)[i + 1])

  532.                 break;

  533. 

  534.             // move filter coefficients left

  535.             for (k = 1; k < filter2Size; k++)

  536.                 filter2[i * filter2Size + k - 1] = filter2[i * filter2Size + k];

  537.             filter2[i * filter2Size + k - 1] = 0;

  538.             (*filterPos)[i]++;

  539.         }

  540. 

  541.         cutOff = 0;

  542.         /* count near zeros on the right */

  543.         for (j = filter2Size - 1; j > 0; j--) {

  544.             cutOff += FFABS(filter2[i * filter2Size + j]);

  545. 

  546.             if (cutOff > SWS_MAX_REDUCE_CUTOFF * fone)

  547.                 break;

  548.             min--;

  549.         }

  550. 

  551.         if (min > minFilterSize)

  552.             minFilterSize = min;

  553.     }

  554. 

  555.     if (PPC_ALTIVEC(cpu_flags)) {

  556.         // we can handle the special case 4, so we don't want to go the full 8

  557.         if (minFilterSize < 5)

  558.             filterAlign = 4;

  559. 

  560.         /* We really don't want to waste our time doing useless computation, so

  561.          * fall back on the scalar C code for very small filters.

  562.          * Vectorizing is worth it only if you have a decent-sized vector. */

  563.         if (minFilterSize < 3)

  564.             filterAlign = 1;

  565.     }

  566. 

  567.     if (HAVE_MMX && cpu_flags & AV_CPU_FLAG_MMX) {

  568.         // special case for unscaled vertical filtering

  569.         if (minFilterSize == 1 && filterAlign == 2)

  570.             filterAlign = 1;

  571.     }

  572. 

  573.     av_assert0(minFilterSize > 0);

  574.     filterSize = (minFilterSize + (filterAlign - 1)) & (~(filterAlign - 1));

  575.     av_assert0(filterSize > 0);

  576.     filter = av_malloc_array(dstW, filterSize * sizeof(*filter));

  577.     if (!filter)

  578.         goto fail;

  579.     if (filterSize >= MAX_FILTER_SIZE * 16 /

  580.                       ((flags & SWS_ACCURATE_RND) ? APCK_SIZE : 16)) {

  581.         ret = RETCODE_USE_CASCADE;

  582.         goto fail;

  583.     }

  584.     *outFilterSize = filterSize;

  585. 

  586.     if (flags & SWS_PRINT_INFO)

  587.         av_log(NULL, AV_LOG_VERBOSE,

  588.                "SwScaler: reducing / aligning filtersize %d -> %d\n",

  589.                filter2Size, filterSize);

  590.     /* try to reduce the filter-size (step2 reduce it) */

  591.     for (i = 0; i < dstW; i++) {

  592.         int j;

  593. 

  594.         for (j = 0; j < filterSize; j++) {

  595.             if (j >= filter2Size)

  596.                 filter[i * filterSize + j] = 0;

  597.             else

  598.                 filter[i * filterSize + j] = filter2[i * filter2Size + j];

  599.             if ((flags & SWS_BITEXACT) && j >= minFilterSize)

  600.                 filter[i * filterSize + j] = 0;

  601.         }

  602.     }

  603. 

  604.     // FIXME try to align filterPos if possible

  605. 

  606.     // fix borders

  607.     for (i = 0; i < dstW; i++) {

  608.         int j;

  609.         if ((*filterPos)[i] < 0) {

  610.             // move filter coefficients left to compensate for filterPos

  611.             for (j = 1; j < filterSize; j++) {

  612.                 int left = FFMAX(j + (*filterPos)[i], 0);

  613.                 filter[i * filterSize + left] += filter[i * filterSize + j];

  614.                 filter[i * filterSize + j]     = 0;

  615.             }

  616.             (*filterPos)[i]= 0;

  617.         }

  618. 

  619.         if ((*filterPos)[i] + filterSize > srcW) {

  620.             int shift = (*filterPos)[i] + FFMIN(filterSize - srcW, 0);

  621.             int64_t acc = 0;

  622. 

  623.             for (j = filterSize - 1; j >= 0; j--) {

  624.                 if ((*filterPos)[i] + j >= srcW) {

  625.                     acc += filter[i * filterSize + j];

  626.                     filter[i * filterSize + j] = 0;

  627.                 }

  628.             }

  629.             for (j = filterSize - 1; j >= 0; j--) {

  630.                 if (j < shift) {

  631.                     filter[i * filterSize + j] = 0;

  632.                 } else {

  633.                     filter[i * filterSize + j] = filter[i * filterSize + j - shift];

  634.                 }

  635.             }

  636. 

  637.             (*filterPos)[i]-= shift;

  638.             filter[i * filterSize + srcW - 1 - (*filterPos)[i]] += acc;

  639.         }

  640.         av_assert0((*filterPos)[i] >= 0);

  641.         av_assert0((*filterPos)[i] < srcW);

  642.         if ((*filterPos)[i] + filterSize > srcW) {

  643.             for (j = 0; j < filterSize; j++) {

  644.                 av_assert0((*filterPos)[i] + j < srcW || !filter[i * filterSize + j]);

  645.             }

  646.         }

  647.     }

  648. 

  649.     // Note the +1 is for the MMX scaler which reads over the end

  650.     /* align at 16 for AltiVec (needed by hScale_altivec_real) */

  651.     FF_ALLOCZ_ARRAY_OR_GOTO(NULL, *outFilter,

  652.                             (dstW + 3), *outFilterSize * sizeof(int16_t), fail);

  653. 

  654.     /* normalize & store in outFilter */

  655.     for (i = 0; i < dstW; i++) {

  656.         int j;

  657.         int64_t error = 0;

  658.         int64_t sum   = 0;

  659. 

  660.         for (j = 0; j < filterSize; j++) {

  661.             sum += filter[i * filterSize + j];

  662.         }

  663.         sum = (sum + one / 2) / one;

  664.         if (!sum) {

  665.             av_log(NULL, AV_LOG_WARNING, "SwScaler: zero vector in scaling\n");

  666.             sum = 1;

  667.         }

  668.         for (j = 0; j < *outFilterSize; j++) {

  669.             int64_t v = filter[i * filterSize + j] + error;

  670.             int intV  = ROUNDED_DIV(v, sum);

  671.             (*outFilter)[i * (*outFilterSize) + j] = intV;

  672.             error                                  = v - intV * sum;

  673.         }

  674.     }

  675. 

  676.     (*filterPos)[dstW + 0] =

  677.     (*filterPos)[dstW + 1] =

  678.     (*filterPos)[dstW + 2] = (*filterPos)[dstW - 1]; /* the MMX/SSE scaler will

  679.                                                       * read over the end */

  680.     for (i = 0; i < *outFilterSize; i++) {

  681.         int k = (dstW - 1) * (*outFilterSize) + i;

  682.         (*outFilter)[k + 1 * (*outFilterSize)] =

  683.         (*outFilter)[k + 2 * (*outFilterSize)] =

  684.         (*outFilter)[k + 3 * (*outFilterSize)] = (*outFilter)[k];

  685.     }

  686. 

  687.     ret = 0;

  688. 

  689. fail:

  690.     if(ret < 0)

  691.         av_log(NULL, ret == RETCODE_USE_CASCADE ? AV_LOG_DEBUG : AV_LOG_ERROR, "sws: initFilter failed\n");

  692.     av_free(filter);

  693.     av_free(filter2);

  694.     return ret;

  695. }

  696. 

  697. static void fill_rgb2yuv_table(SwsContext *c, const int table[4], int dstRange)

  698. {

  699.     int64_t W, V, Z, Cy, Cu, Cv;

  700.     int64_t vr =  table[0];

  701.     int64_t ub =  table[1];

  702.     int64_t ug = -table[2];

  703.     int64_t vg = -table[3];

  704.     int64_t ONE = 65536;

  705.     int64_t cy = ONE;

  706.     uint8_t *p = (uint8_t*)c->input_rgb2yuv_table;

  707.     int i;

  708.     static const int8_t map[] = {

  709.     BY_IDX, GY_IDX, -1    , BY_IDX, BY_IDX, GY_IDX, -1    , BY_IDX,

  710.     RY_IDX, -1    , GY_IDX, RY_IDX, RY_IDX, -1    , GY_IDX, RY_IDX,

  711.     RY_IDX, GY_IDX, -1    , RY_IDX, RY_IDX, GY_IDX, -1    , RY_IDX,

  712.     BY_IDX, -1    , GY_IDX, BY_IDX, BY_IDX, -1    , GY_IDX, BY_IDX,

  713.     BU_IDX, GU_IDX, -1    , BU_IDX, BU_IDX, GU_IDX, -1    , BU_IDX,

  714.     RU_IDX, -1    , GU_IDX, RU_IDX, RU_IDX, -1    , GU_IDX, RU_IDX,

  715.     RU_IDX, GU_IDX, -1    , RU_IDX, RU_IDX, GU_IDX, -1    , RU_IDX,

  716.     BU_IDX, -1    , GU_IDX, BU_IDX, BU_IDX, -1    , GU_IDX, BU_IDX,

  717.     BV_IDX, GV_IDX, -1    , BV_IDX, BV_IDX, GV_IDX, -1    , BV_IDX,

  718.     RV_IDX, -1    , GV_IDX, RV_IDX, RV_IDX, -1    , GV_IDX, RV_IDX,

  719.     RV_IDX, GV_IDX, -1    , RV_IDX, RV_IDX, GV_IDX, -1    , RV_IDX,

  720.     BV_IDX, -1    , GV_IDX, BV_IDX, BV_IDX, -1    , GV_IDX, BV_IDX,

  721.     RY_IDX, BY_IDX, RY_IDX, BY_IDX, RY_IDX, BY_IDX, RY_IDX, BY_IDX,

  722.     BY_IDX, RY_IDX, BY_IDX, RY_IDX, BY_IDX, RY_IDX, BY_IDX, RY_IDX,

  723.     GY_IDX, -1    , GY_IDX, -1    , GY_IDX, -1    , GY_IDX, -1    ,

  724.     -1    , GY_IDX, -1    , GY_IDX, -1    , GY_IDX, -1    , GY_IDX,

  725.     RU_IDX, BU_IDX, RU_IDX, BU_IDX, RU_IDX, BU_IDX, RU_IDX, BU_IDX,

  726.     BU_IDX, RU_IDX, BU_IDX, RU_IDX, BU_IDX, RU_IDX, BU_IDX, RU_IDX,

  727.     GU_IDX, -1    , GU_IDX, -1    , GU_IDX, -1    , GU_IDX, -1    ,

  728.     -1    , GU_IDX, -1    , GU_IDX, -1    , GU_IDX, -1    , GU_IDX,

  729.     RV_IDX, BV_IDX, RV_IDX, BV_IDX, RV_IDX, BV_IDX, RV_IDX, BV_IDX,

  730.     BV_IDX, RV_IDX, BV_IDX, RV_IDX, BV_IDX, RV_IDX, BV_IDX, RV_IDX,

  731.     GV_IDX, -1    , GV_IDX, -1    , GV_IDX, -1    , GV_IDX, -1    ,

  732.     -1    , GV_IDX, -1    , GV_IDX, -1    , GV_IDX, -1    , GV_IDX, //23

  733.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //24

  734.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //25

  735.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //26

  736.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //27

  737.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //28

  738.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //29

  739.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //30

  740.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //31

  741.     BY_IDX, GY_IDX, RY_IDX, -1    , -1    , -1    , -1    , -1    , //32

  742.     BU_IDX, GU_IDX, RU_IDX, -1    , -1    , -1    , -1    , -1    , //33

  743.     BV_IDX, GV_IDX, RV_IDX, -1    , -1    , -1    , -1    , -1    , //34

  744.     };

  745. 

  746.     dstRange = 0; //FIXME range = 1 is handled elsewhere

  747. 

  748.     if (!dstRange) {

  749.         cy = cy * 255 / 219;

  750.     } else {

  751.         vr = vr * 224 / 255;

  752.         ub = ub * 224 / 255;

  753.         ug = ug * 224 / 255;

  754.         vg = vg * 224 / 255;

  755.     }

  756.     W = ROUNDED_DIV(ONE*ONE*ug, ub);

  757.     V = ROUNDED_DIV(ONE*ONE*vg, vr);

  758.     Z = ONE*ONE-W-V;

  759. 

  760.     Cy = ROUNDED_DIV(cy*Z, ONE);

  761.     Cu = ROUNDED_DIV(ub*Z, ONE);

  762.     Cv = ROUNDED_DIV(vr*Z, ONE);

  763. 

  764.     c->input_rgb2yuv_table[RY_IDX] = -ROUNDED_DIV((1 << RGB2YUV_SHIFT)*V        , Cy);

  765.     c->input_rgb2yuv_table[GY_IDX] =  ROUNDED_DIV((1 << RGB2YUV_SHIFT)*ONE*ONE  , Cy);

  766.     c->input_rgb2yuv_table[BY_IDX] = -ROUNDED_DIV((1 << RGB2YUV_SHIFT)*W        , Cy);

  767. 

  768.     c->input_rgb2yuv_table[RU_IDX] =  ROUNDED_DIV((1 << RGB2YUV_SHIFT)*V        , Cu);

  769.     c->input_rgb2yuv_table[GU_IDX] = -ROUNDED_DIV((1 << RGB2YUV_SHIFT)*ONE*ONE  , Cu);

  770.     c->input_rgb2yuv_table[BU_IDX] =  ROUNDED_DIV((1 << RGB2YUV_SHIFT)*(Z+W)    , Cu);

  771. 

  772.     c->input_rgb2yuv_table[RV_IDX] =  ROUNDED_DIV((1 << RGB2YUV_SHIFT)*(V+Z)    , Cv);

  773.     c->input_rgb2yuv_table[GV_IDX] = -ROUNDED_DIV((1 << RGB2YUV_SHIFT)*ONE*ONE  , Cv);

  774.     c->input_rgb2yuv_table[BV_IDX] =  ROUNDED_DIV((1 << RGB2YUV_SHIFT)*W        , Cv);

  775. 

  776.     if(/*!dstRange && */!memcmp(table, ff_yuv2rgb_coeffs[SWS_CS_DEFAULT], sizeof(ff_yuv2rgb_coeffs[SWS_CS_DEFAULT]))) {

  777.         c->input_rgb2yuv_table[BY_IDX] =  ((int)(0.114 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  778.         c->input_rgb2yuv_table[BV_IDX] = (-(int)(0.081 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  779.         c->input_rgb2yuv_table[BU_IDX] =  ((int)(0.500 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  780.         c->input_rgb2yuv_table[GY_IDX] =  ((int)(0.587 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  781.         c->input_rgb2yuv_table[GV_IDX] = (-(int)(0.419 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  782.         c->input_rgb2yuv_table[GU_IDX] = (-(int)(0.331 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  783.         c->input_rgb2yuv_table[RY_IDX] =  ((int)(0.299 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  784.         c->input_rgb2yuv_table[RV_IDX] =  ((int)(0.500 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  785.         c->input_rgb2yuv_table[RU_IDX] = (-(int)(0.169 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5));

  786.     }

  787.     for(i=0; i<FF_ARRAY_ELEMS(map); i++)

  788.         AV_WL16(p + 16*4 + 2*i, map[i] >= 0 ? c->input_rgb2yuv_table[map[i]] : 0);

  789. }

  790. 

  791. static void fill_xyztables(struct SwsContext *c)

  792. {

  793.     int i;

  794.     double xyzgamma = XYZ_GAMMA;

  795.     double rgbgamma = 1.0 / RGB_GAMMA;

  796.     double xyzgammainv = 1.0 / XYZ_GAMMA;

  797.     double rgbgammainv = RGB_GAMMA;

  798.     static const int16_t xyz2rgb_matrix[3][4] = {

  799.         {13270, -6295, -2041},

  800.         {-3969,  7682,   170},

  801.         {  228,  -835,  4329} };

  802.     static const int16_t rgb2xyz_matrix[3][4] = {

  803.         {1689, 1464,  739},

  804.         { 871, 2929,  296},

  805.         {  79,  488, 3891} };

  806.     static int16_t xyzgamma_tab[4096], rgbgamma_tab[4096], xyzgammainv_tab[4096], rgbgammainv_tab[4096];

  807. 

  808.     memcpy(c->xyz2rgb_matrix, xyz2rgb_matrix, sizeof(c->xyz2rgb_matrix));

  809.     memcpy(c->rgb2xyz_matrix, rgb2xyz_matrix, sizeof(c->rgb2xyz_matrix));

  810.     c->xyzgamma = xyzgamma_tab;

  811.     c->rgbgamma = rgbgamma_tab;

  812.     c->xyzgammainv = xyzgammainv_tab;

  813.     c->rgbgammainv = rgbgammainv_tab;

  814. 

  815.     if (rgbgamma_tab[4095])

  816.         return;

  817. 

  818.     /* set gamma vectors */

  819.     for (i = 0; i < 4096; i++) {

  820.         xyzgamma_tab[i] = lrint(pow(i / 4095.0, xyzgamma) * 4095.0);

  821.         rgbgamma_tab[i] = lrint(pow(i / 4095.0, rgbgamma) * 4095.0);

  822.         xyzgammainv_tab[i] = lrint(pow(i / 4095.0, xyzgammainv) * 4095.0);

  823.         rgbgammainv_tab[i] = lrint(pow(i / 4095.0, rgbgammainv) * 4095.0);

  824.     }

  825. }

  826. 

  827. int sws_setColorspaceDetails(struct SwsContext *c, const int inv_table[4],

  828.                              int srcRange, const int table[4], int dstRange,

  829.                              int brightness, int contrast, int saturation)

  830. {

  831.     const AVPixFmtDescriptor *desc_dst;

  832.     const AVPixFmtDescriptor *desc_src;

  833.     int need_reinit = 0;

  834.     memmove(c->srcColorspaceTable, inv_table, sizeof(int) * 4);

  835.     memmove(c->dstColorspaceTable, table, sizeof(int) * 4);

  836. 

  837.     handle_formats(c);

  838.     desc_dst = av_pix_fmt_desc_get(c->dstFormat);

  839.     desc_src = av_pix_fmt_desc_get(c->srcFormat);

  840. 

  841.     if(!isYUV(c->dstFormat) && !isGray(c->dstFormat))

  842.         dstRange = 0;

  843.     if(!isYUV(c->srcFormat) && !isGray(c->srcFormat))

  844.         srcRange = 0;

  845. 

  846.     c->brightness = brightness;

  847.     c->contrast   = contrast;

  848.     c->saturation = saturation;

  849.     if (c->srcRange != srcRange || c->dstRange != dstRange)

  850.         need_reinit = 1;

  851.     c->srcRange   = srcRange;

  852.     c->dstRange   = dstRange;

  853. 

  854.     //The srcBpc check is possibly wrong but we seem to lack a definitive reference to test this

  855.     //and what we have in ticket 2939 looks better with this check

  856.     if (need_reinit && (c->srcBpc == 8 || !isYUV(c->srcFormat)))

  857.         ff_sws_init_range_convert(c);

  858. 

  859.     c->dstFormatBpp = av_get_bits_per_pixel(desc_dst);

  860.     c->srcFormatBpp = av_get_bits_per_pixel(desc_src);

  861. 

  862.     if (c->cascaded_context[c->cascaded_mainindex])

  863.         return sws_setColorspaceDetails(c->cascaded_context[c->cascaded_mainindex],inv_table, srcRange,table, dstRange, brightness,  contrast, saturation);

  864. 

  865.     if ((isYUV(c->dstFormat) || isGray(c->dstFormat)) && (isYUV(c->srcFormat) || isGray(c->srcFormat))) {

  866.         if (!c->cascaded_context[0] &&

  867.             memcmp(c->dstColorspaceTable, c->srcColorspaceTable, sizeof(int) * 4) &&

  868.             c->srcW && c->srcH && c->dstW && c->dstH) {

  869.             enum AVPixelFormat tmp_format;

  870.             int tmp_width, tmp_height;

  871.             int srcW = c->srcW;

  872.             int srcH = c->srcH;

  873.             int dstW = c->dstW;

  874.             int dstH = c->dstH;

  875.             int ret;

  876.             av_log(c, AV_LOG_VERBOSE, "YUV color matrix differs for YUV->YUV, using intermediate RGB to convert\n");

  877. 

  878.             if (isNBPS(c->dstFormat) || is16BPS(c->dstFormat)) {

  879.                 if (isALPHA(c->srcFormat) && isALPHA(c->dstFormat)) {

  880.                     tmp_format = AV_PIX_FMT_BGRA64;

  881.                 } else {

  882.                     tmp_format = AV_PIX_FMT_BGR48;

  883.                 }

  884.             } else {

  885.                 if (isALPHA(c->srcFormat) && isALPHA(c->dstFormat)) {

  886.                     tmp_format = AV_PIX_FMT_BGRA;

  887.                 } else {

  888.                     tmp_format = AV_PIX_FMT_BGR24;

  889.                 }

  890.             }

  891. 

  892.             if (srcW*srcH > dstW*dstH) {

  893.                 tmp_width  = dstW;

  894.                 tmp_height = dstH;

  895.             } else {

  896.                 tmp_width  = srcW;

  897.                 tmp_height = srcH;

  898.             }

  899. 

  900.             ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  901.                                 tmp_width, tmp_height, tmp_format, 64);

  902.             if (ret < 0)

  903.                 return ret;

  904. 

  905.             c->cascaded_context[0] = sws_alloc_set_opts(srcW, srcH, c->srcFormat,

  906.                                                         tmp_width, tmp_height, tmp_format,

  907.                                                         c->flags, c->param);

  908.             if (!c->cascaded_context[0])

  909.                 return -1;

  910. 

  911.             c->cascaded_context[0]->alphablend = c->alphablend;

  912.             ret = sws_init_context(c->cascaded_context[0], NULL , NULL);

  913.             if (ret < 0)

  914.                 return ret;

  915.             //we set both src and dst depending on that the RGB side will be ignored

  916.             sws_setColorspaceDetails(c->cascaded_context[0], inv_table,

  917.                                      srcRange, table, dstRange,

  918.                                      brightness, contrast, saturation);

  919. 

  920.             c->cascaded_context[1] = sws_getContext(tmp_width, tmp_height, tmp_format,

  921.                                                     dstW, dstH, c->dstFormat,

  922.                                                     c->flags, NULL, NULL, c->param);

  923.             if (!c->cascaded_context[1])

  924.                 return -1;

  925.             sws_setColorspaceDetails(c->cascaded_context[1], inv_table,

  926.                                      srcRange, table, dstRange,

  927.                                      0, 1 << 16, 1 << 16);

  928.             return 0;

  929.         }

  930.         return -1;

  931.     }

  932. 

  933.     if (!isYUV(c->dstFormat) && !isGray(c->dstFormat)) {

  934.         ff_yuv2rgb_c_init_tables(c, inv_table, srcRange, brightness,

  935.                                  contrast, saturation);

  936.         // FIXME factorize

  937. 

  938.         if (ARCH_PPC)

  939.             ff_yuv2rgb_init_tables_ppc(c, inv_table, brightness,

  940.                                        contrast, saturation);

  941.     }

  942. 

  943.     fill_rgb2yuv_table(c, table, dstRange);

  944. 

  945.     return 0;

  946. }

  947. 

  948. int sws_getColorspaceDetails(struct SwsContext *c, int **inv_table,

  949.                              int *srcRange, int **table, int *dstRange,

  950.                              int *brightness, int *contrast, int *saturation)

  951. {

  952.     if (!c )

  953.         return -1;

  954. 

  955.     *inv_table  = c->srcColorspaceTable;

  956.     *table      = c->dstColorspaceTable;

  957.     *srcRange   = c->srcRange;

  958.     *dstRange   = c->dstRange;

  959.     *brightness = c->brightness;

  960.     *contrast   = c->contrast;

  961.     *saturation = c->saturation;

  962. 

  963.     return 0;

  964. }

  965. 

  966. static int handle_jpeg(enum AVPixelFormat *format)

  967. {

  968.     switch (*format) {

  969.     case AV_PIX_FMT_YUVJ420P:

  970.         *format = AV_PIX_FMT_YUV420P;

  971.         return 1;

  972.     case AV_PIX_FMT_YUVJ411P:

  973.         *format = AV_PIX_FMT_YUV411P;

  974.         return 1;

  975.     case AV_PIX_FMT_YUVJ422P:

  976.         *format = AV_PIX_FMT_YUV422P;

  977.         return 1;

  978.     case AV_PIX_FMT_YUVJ444P:

  979.         *format = AV_PIX_FMT_YUV444P;

  980.         return 1;

  981.     case AV_PIX_FMT_YUVJ440P:

  982.         *format = AV_PIX_FMT_YUV440P;

  983.         return 1;

  984.     case AV_PIX_FMT_GRAY8:

  985.     case AV_PIX_FMT_YA8:

  986.     case AV_PIX_FMT_GRAY16LE:

  987.     case AV_PIX_FMT_GRAY16BE:

  988.     case AV_PIX_FMT_YA16BE:

  989.     case AV_PIX_FMT_YA16LE:

  990.         return 1;

  991.     default:

  992.         return 0;

  993.     }

  994. }

  995. 

  996. static int handle_0alpha(enum AVPixelFormat *format)

  997. {

  998.     switch (*format) {

  999.     case AV_PIX_FMT_0BGR    : *format = AV_PIX_FMT_ABGR   ; return 1;

  1000.     case AV_PIX_FMT_BGR0    : *format = AV_PIX_FMT_BGRA   ; return 4;

  1001.     case AV_PIX_FMT_0RGB    : *format = AV_PIX_FMT_ARGB   ; return 1;

  1002.     case AV_PIX_FMT_RGB0    : *format = AV_PIX_FMT_RGBA   ; return 4;

  1003.     default:                                          return 0;

  1004.     }

  1005. }

  1006. 

  1007. static int handle_xyz(enum AVPixelFormat *format)

  1008. {

  1009.     switch (*format) {

  1010.     case AV_PIX_FMT_XYZ12BE : *format = AV_PIX_FMT_RGB48BE; return 1;

  1011.     case AV_PIX_FMT_XYZ12LE : *format = AV_PIX_FMT_RGB48LE; return 1;

  1012.     default:                                                return 0;

  1013.     }

  1014. }

  1015. 

  1016. static void handle_formats(SwsContext *c)

  1017. {

  1018.     c->src0Alpha |= handle_0alpha(&c->srcFormat);

  1019.     c->dst0Alpha |= handle_0alpha(&c->dstFormat);

  1020.     c->srcXYZ    |= handle_xyz(&c->srcFormat);

  1021.     c->dstXYZ    |= handle_xyz(&c->dstFormat);

  1022.     if (c->srcXYZ || c->dstXYZ)

  1023.         fill_xyztables(c);

  1024. }

  1025. 

  1026. SwsContext *sws_alloc_context(void)

  1027. {

  1028.     SwsContext *c = av_mallocz(sizeof(SwsContext));

  1029. 

  1030.     av_assert0(offsetof(SwsContext, redDither) + DITHER32_INT == offsetof(SwsContext, dither32));

  1031. 

  1032.     if (c) {

  1033.         c->av_class = &sws_context_class;

  1034.         av_opt_set_defaults(c);

  1035.     }

  1036. 

  1037.     return c;

  1038. }

  1039. 

  1040. static uint16_t * alloc_gamma_tbl(double e)

  1041. {

  1042.     int i = 0;

  1043.     uint16_t * tbl;

  1044.     tbl = (uint16_t*)av_malloc(sizeof(uint16_t) * 1 << 16);

  1045.     if (!tbl)

  1046.         return NULL;

  1047. 

  1048.     for (i = 0; i < 65536; ++i) {

  1049.         tbl[i] = pow(i / 65535.0, e) * 65535.0;

  1050.     }

  1051.     return tbl;

  1052. }

  1053. 

  1054. static enum AVPixelFormat alphaless_fmt(enum AVPixelFormat fmt)

  1055. {

  1056.     switch(fmt) {

  1057.     case AV_PIX_FMT_ARGB:       return AV_PIX_FMT_RGB24;

  1058.     case AV_PIX_FMT_RGBA:       return AV_PIX_FMT_RGB24;

  1059.     case AV_PIX_FMT_ABGR:       return AV_PIX_FMT_BGR24;

  1060.     case AV_PIX_FMT_BGRA:       return AV_PIX_FMT_BGR24;

  1061.     case AV_PIX_FMT_YA8:        return AV_PIX_FMT_GRAY8;

  1062. 

  1063.     case AV_PIX_FMT_YUVA420P:   return AV_PIX_FMT_YUV420P;

  1064.     case AV_PIX_FMT_YUVA422P:   return AV_PIX_FMT_YUV422P;

  1065.     case AV_PIX_FMT_YUVA444P:           return AV_PIX_FMT_YUV444P;

  1066. 

  1067.     case AV_PIX_FMT_GBRAP:              return AV_PIX_FMT_GBRP;

  1068. 

  1069.     case AV_PIX_FMT_GBRAP16LE:          return AV_PIX_FMT_GBRP16;

  1070.     case AV_PIX_FMT_GBRAP16BE:          return AV_PIX_FMT_GBRP16;

  1071. 

  1072.     case AV_PIX_FMT_RGBA64LE:   return AV_PIX_FMT_RGB48;

  1073.     case AV_PIX_FMT_RGBA64BE:   return AV_PIX_FMT_RGB48;

  1074.     case AV_PIX_FMT_BGRA64LE:   return AV_PIX_FMT_BGR48;

  1075.     case AV_PIX_FMT_BGRA64BE:   return AV_PIX_FMT_BGR48;

  1076. 

  1077.     case AV_PIX_FMT_YA16BE:             return AV_PIX_FMT_GRAY16;

  1078.     case AV_PIX_FMT_YA16LE:             return AV_PIX_FMT_GRAY16;

  1079. 

  1080.     case AV_PIX_FMT_YUVA420P9BE:        return AV_PIX_FMT_YUV420P9;

  1081.     case AV_PIX_FMT_YUVA422P9BE:        return AV_PIX_FMT_YUV422P9;

  1082.     case AV_PIX_FMT_YUVA444P9BE:        return AV_PIX_FMT_YUV444P9;

  1083.     case AV_PIX_FMT_YUVA420P9LE:        return AV_PIX_FMT_YUV420P9;

  1084.     case AV_PIX_FMT_YUVA422P9LE:        return AV_PIX_FMT_YUV422P9;

  1085.     case AV_PIX_FMT_YUVA444P9LE:        return AV_PIX_FMT_YUV444P9;

  1086.     case AV_PIX_FMT_YUVA420P10BE:       return AV_PIX_FMT_YUV420P10;

  1087.     case AV_PIX_FMT_YUVA422P10BE:       return AV_PIX_FMT_YUV422P10;

  1088.     case AV_PIX_FMT_YUVA444P10BE:       return AV_PIX_FMT_YUV444P10;

  1089.     case AV_PIX_FMT_YUVA420P10LE:       return AV_PIX_FMT_YUV420P10;

  1090.     case AV_PIX_FMT_YUVA422P10LE:       return AV_PIX_FMT_YUV422P10;

  1091.     case AV_PIX_FMT_YUVA444P10LE:       return AV_PIX_FMT_YUV444P10;

  1092.     case AV_PIX_FMT_YUVA420P16BE:       return AV_PIX_FMT_YUV420P16;

  1093.     case AV_PIX_FMT_YUVA422P16BE:       return AV_PIX_FMT_YUV422P16;

  1094.     case AV_PIX_FMT_YUVA444P16BE:       return AV_PIX_FMT_YUV444P16;

  1095.     case AV_PIX_FMT_YUVA420P16LE:       return AV_PIX_FMT_YUV420P16;

  1096.     case AV_PIX_FMT_YUVA422P16LE:       return AV_PIX_FMT_YUV422P16;

  1097.     case AV_PIX_FMT_YUVA444P16LE:       return AV_PIX_FMT_YUV444P16;

  1098. 

  1099. //     case AV_PIX_FMT_AYUV64LE:

  1100. //     case AV_PIX_FMT_AYUV64BE:

  1101. //     case AV_PIX_FMT_PAL8:

  1102.     default: return AV_PIX_FMT_NONE;

  1103.     }

  1104. }

  1105. 

  1106. av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter,

  1107.                              SwsFilter *dstFilter)

  1108. {

  1109.     int i, j;

  1110.     int usesVFilter, usesHFilter;

  1111.     int unscaled;

  1112.     SwsFilter dummyFilter = { NULL, NULL, NULL, NULL };

  1113.     int srcW              = c->srcW;

  1114.     int srcH              = c->srcH;

  1115.     int dstW              = c->dstW;

  1116.     int dstH              = c->dstH;

  1117.     int dst_stride        = FFALIGN(dstW * sizeof(int16_t) + 66, 16);

  1118.     int flags, cpu_flags;

  1119.     enum AVPixelFormat srcFormat = c->srcFormat;

  1120.     enum AVPixelFormat dstFormat = c->dstFormat;

  1121.     const AVPixFmtDescriptor *desc_src;

  1122.     const AVPixFmtDescriptor *desc_dst;

  1123.     int ret = 0;

  1124.     enum AVPixelFormat tmpFmt;

  1125. 

  1126.     cpu_flags = av_get_cpu_flags();

  1127.     flags     = c->flags;

  1128.     emms_c();

  1129.     if (!rgb15to16)

  1130.         sws_rgb2rgb_init();

  1131. 

  1132.     unscaled = (srcW == dstW && srcH == dstH);

  1133. 

  1134.     c->srcRange |= handle_jpeg(&c->srcFormat);

  1135.     c->dstRange |= handle_jpeg(&c->dstFormat);

  1136. 

  1137.     if(srcFormat!=c->srcFormat || dstFormat!=c->dstFormat)

  1138.         av_log(c, AV_LOG_WARNING, "deprecated pixel format used, make sure you did set range correctly\n");

  1139. 

  1140.     if (!c->contrast && !c->saturation && !c->dstFormatBpp)

  1141.         sws_setColorspaceDetails(c, ff_yuv2rgb_coeffs[SWS_CS_DEFAULT], c->srcRange,

  1142.                                  ff_yuv2rgb_coeffs[SWS_CS_DEFAULT],

  1143.                                  c->dstRange, 0, 1 << 16, 1 << 16);

  1144. 

  1145.     handle_formats(c);

  1146.     srcFormat = c->srcFormat;

  1147.     dstFormat = c->dstFormat;

  1148.     desc_src = av_pix_fmt_desc_get(srcFormat);

  1149.     desc_dst = av_pix_fmt_desc_get(dstFormat);

  1150. 

  1151.     // If the source has no alpha then disable alpha blendaway

  1152.     if (c->src0Alpha)

  1153.         c->alphablend = SWS_ALPHA_BLEND_NONE;

  1154. 

  1155.     if (!(unscaled && sws_isSupportedEndiannessConversion(srcFormat) &&

  1156.           av_pix_fmt_swap_endianness(srcFormat) == dstFormat)) {

  1157.     if (!sws_isSupportedInput(srcFormat)) {

  1158.         av_log(c, AV_LOG_ERROR, "%s is not supported as input pixel format\n",

  1159.                av_get_pix_fmt_name(srcFormat));

  1160.         return AVERROR(EINVAL);

  1161.     }

  1162.     if (!sws_isSupportedOutput(dstFormat)) {

  1163.         av_log(c, AV_LOG_ERROR, "%s is not supported as output pixel format\n",

  1164.                av_get_pix_fmt_name(dstFormat));

  1165.         return AVERROR(EINVAL);

  1166.     }

  1167.     }

  1168.     av_assert2(desc_src && desc_dst);

  1169. 

  1170.     i = flags & (SWS_POINT         |

  1171.                  SWS_AREA          |

  1172.                  SWS_BILINEAR      |

  1173.                  SWS_FAST_BILINEAR |

  1174.                  SWS_BICUBIC       |

  1175.                  SWS_X             |

  1176.                  SWS_GAUSS         |

  1177.                  SWS_LANCZOS       |

  1178.                  SWS_SINC          |

  1179.                  SWS_SPLINE        |

  1180.                  SWS_BICUBLIN);

  1181. 

  1182.     /* provide a default scaler if not set by caller */

  1183.     if (!i) {

  1184.         if (dstW < srcW && dstH < srcH)

  1185.             flags |= SWS_BICUBIC;

  1186.         else if (dstW > srcW && dstH > srcH)

  1187.             flags |= SWS_BICUBIC;

  1188.         else

  1189.             flags |= SWS_BICUBIC;

  1190.         c->flags = flags;

  1191.     } else if (i & (i - 1)) {

  1192.         av_log(c, AV_LOG_ERROR,

  1193.                "Exactly one scaler algorithm must be chosen, got %X\n", i);

  1194.         return AVERROR(EINVAL);

  1195.     }

  1196.     /* sanity check */

  1197.     if (srcW < 1 || srcH < 1 || dstW < 1 || dstH < 1) {

  1198.         /* FIXME check if these are enough and try to lower them after

  1199.          * fixing the relevant parts of the code */

  1200.         av_log(c, AV_LOG_ERROR, "%dx%d -> %dx%d is invalid scaling dimension\n",

  1201.                srcW, srcH, dstW, dstH);

  1202.         return AVERROR(EINVAL);

  1203.     }

  1204. 

  1205.     if (!dstFilter)

  1206.         dstFilter = &dummyFilter;

  1207.     if (!srcFilter)

  1208.         srcFilter = &dummyFilter;

  1209. 

  1210.     c->lumXInc      = (((int64_t)srcW << 16) + (dstW >> 1)) / dstW;

  1211.     c->lumYInc      = (((int64_t)srcH << 16) + (dstH >> 1)) / dstH;

  1212.     c->dstFormatBpp = av_get_bits_per_pixel(desc_dst);

  1213.     c->srcFormatBpp = av_get_bits_per_pixel(desc_src);

  1214.     c->vRounder     = 4 * 0x0001000100010001ULL;

  1215. 

  1216.     usesVFilter = (srcFilter->lumV && srcFilter->lumV->length > 1) ||

  1217.                   (srcFilter->chrV && srcFilter->chrV->length > 1) ||

  1218.                   (dstFilter->lumV && dstFilter->lumV->length > 1) ||

  1219.                   (dstFilter->chrV && dstFilter->chrV->length > 1);

  1220.     usesHFilter = (srcFilter->lumH && srcFilter->lumH->length > 1) ||

  1221.                   (srcFilter->chrH && srcFilter->chrH->length > 1) ||

  1222.                   (dstFilter->lumH && dstFilter->lumH->length > 1) ||

  1223.                   (dstFilter->chrH && dstFilter->chrH->length > 1);

  1224. 

  1225.     av_pix_fmt_get_chroma_sub_sample(srcFormat, &c->chrSrcHSubSample, &c->chrSrcVSubSample);

  1226.     av_pix_fmt_get_chroma_sub_sample(dstFormat, &c->chrDstHSubSample, &c->chrDstVSubSample);

  1227. 

  1228.     if (isAnyRGB(dstFormat) && !(flags&SWS_FULL_CHR_H_INT)) {

  1229.         if (dstW&1) {

  1230.             av_log(c, AV_LOG_DEBUG, "Forcing full internal H chroma due to odd output size\n");

  1231.             flags |= SWS_FULL_CHR_H_INT;

  1232.             c->flags = flags;

  1233.         }

  1234. 

  1235.         if (   c->chrSrcHSubSample == 0

  1236.             && c->chrSrcVSubSample == 0

  1237.             && c->dither != SWS_DITHER_BAYER //SWS_FULL_CHR_H_INT is currently not supported with SWS_DITHER_BAYER

  1238.             && !(c->flags & SWS_FAST_BILINEAR)

  1239.         ) {

  1240.             av_log(c, AV_LOG_DEBUG, "Forcing full internal H chroma due to input having non subsampled chroma\n");

  1241.             flags |= SWS_FULL_CHR_H_INT;

  1242.             c->flags = flags;

  1243.         }

  1244.     }

  1245. 

  1246.     if (c->dither == SWS_DITHER_AUTO) {

  1247.         if (flags & SWS_ERROR_DIFFUSION)

  1248.             c->dither = SWS_DITHER_ED;

  1249.     }

  1250. 

  1251.     if(dstFormat == AV_PIX_FMT_BGR4_BYTE ||

  1252.        dstFormat == AV_PIX_FMT_RGB4_BYTE ||

  1253.        dstFormat == AV_PIX_FMT_BGR8 ||

  1254.        dstFormat == AV_PIX_FMT_RGB8) {

  1255.         if (c->dither == SWS_DITHER_AUTO)

  1256.             c->dither = (flags & SWS_FULL_CHR_H_INT) ? SWS_DITHER_ED : SWS_DITHER_BAYER;

  1257.         if (!(flags & SWS_FULL_CHR_H_INT)) {

  1258.             if (c->dither == SWS_DITHER_ED || c->dither == SWS_DITHER_A_DITHER || c->dither == SWS_DITHER_X_DITHER) {

  1259.                 av_log(c, AV_LOG_DEBUG,

  1260.                     "Desired dithering only supported in full chroma interpolation for destination format '%s'\n",

  1261.                     av_get_pix_fmt_name(dstFormat));

  1262.                 flags   |= SWS_FULL_CHR_H_INT;

  1263.                 c->flags = flags;

  1264.             }

  1265.         }

  1266.         if (flags & SWS_FULL_CHR_H_INT) {

  1267.             if (c->dither == SWS_DITHER_BAYER) {

  1268.                 av_log(c, AV_LOG_DEBUG,

  1269.                     "Ordered dither is not supported in full chroma interpolation for destination format '%s'\n",

  1270.                     av_get_pix_fmt_name(dstFormat));

  1271.                 c->dither = SWS_DITHER_ED;

  1272.             }

  1273.         }

  1274.     }

  1275.     if (isPlanarRGB(dstFormat)) {

  1276.         if (!(flags & SWS_FULL_CHR_H_INT)) {

  1277.             av_log(c, AV_LOG_DEBUG,

  1278.                    "%s output is not supported with half chroma resolution, switching to full\n",

  1279.                    av_get_pix_fmt_name(dstFormat));

  1280.             flags   |= SWS_FULL_CHR_H_INT;

  1281.             c->flags = flags;

  1282.         }

  1283.     }

  1284. 

  1285.     /* reuse chroma for 2 pixels RGB/BGR unless user wants full

  1286.      * chroma interpolation */

  1287.     if (flags & SWS_FULL_CHR_H_INT &&

  1288.         isAnyRGB(dstFormat)        &&

  1289.         !isPlanarRGB(dstFormat)    &&

  1290.         dstFormat != AV_PIX_FMT_RGBA64LE &&

  1291.         dstFormat != AV_PIX_FMT_RGBA64BE &&

  1292.         dstFormat != AV_PIX_FMT_BGRA64LE &&

  1293.         dstFormat != AV_PIX_FMT_BGRA64BE &&

  1294.         dstFormat != AV_PIX_FMT_RGB48LE &&

  1295.         dstFormat != AV_PIX_FMT_RGB48BE &&

  1296.         dstFormat != AV_PIX_FMT_BGR48LE &&

  1297.         dstFormat != AV_PIX_FMT_BGR48BE &&

  1298.         dstFormat != AV_PIX_FMT_RGBA  &&

  1299.         dstFormat != AV_PIX_FMT_ARGB  &&

  1300.         dstFormat != AV_PIX_FMT_BGRA  &&

  1301.         dstFormat != AV_PIX_FMT_ABGR  &&

  1302.         dstFormat != AV_PIX_FMT_RGB24 &&

  1303.         dstFormat != AV_PIX_FMT_BGR24 &&

  1304.         dstFormat != AV_PIX_FMT_BGR4_BYTE &&

  1305.         dstFormat != AV_PIX_FMT_RGB4_BYTE &&

  1306.         dstFormat != AV_PIX_FMT_BGR8 &&

  1307.         dstFormat != AV_PIX_FMT_RGB8

  1308.     ) {

  1309.         av_log(c, AV_LOG_WARNING,

  1310.                "full chroma interpolation for destination format '%s' not yet implemented\n",

  1311.                av_get_pix_fmt_name(dstFormat));

  1312.         flags   &= ~SWS_FULL_CHR_H_INT;

  1313.         c->flags = flags;

  1314.     }

  1315.     if (isAnyRGB(dstFormat) && !(flags & SWS_FULL_CHR_H_INT))

  1316.         c->chrDstHSubSample = 1;

  1317. 

  1318.     // drop some chroma lines if the user wants it

  1319.     c->vChrDrop          = (flags & SWS_SRC_V_CHR_DROP_MASK) >>

  1320.                            SWS_SRC_V_CHR_DROP_SHIFT;

  1321.     c->chrSrcVSubSample += c->vChrDrop;

  1322. 

  1323.     /* drop every other pixel for chroma calculation unless user

  1324.      * wants full chroma */

  1325.     if (isAnyRGB(srcFormat) && !(flags & SWS_FULL_CHR_H_INP)   &&

  1326.         srcFormat != AV_PIX_FMT_RGB8 && srcFormat != AV_PIX_FMT_BGR8 &&

  1327.         srcFormat != AV_PIX_FMT_RGB4 && srcFormat != AV_PIX_FMT_BGR4 &&

  1328.         srcFormat != AV_PIX_FMT_RGB4_BYTE && srcFormat != AV_PIX_FMT_BGR4_BYTE &&

  1329.         srcFormat != AV_PIX_FMT_GBRP9BE   && srcFormat != AV_PIX_FMT_GBRP9LE  &&

  1330.         srcFormat != AV_PIX_FMT_GBRP10BE  && srcFormat != AV_PIX_FMT_GBRP10LE &&

  1331.         srcFormat != AV_PIX_FMT_GBRP12BE  && srcFormat != AV_PIX_FMT_GBRP12LE &&

  1332.         srcFormat != AV_PIX_FMT_GBRP14BE  && srcFormat != AV_PIX_FMT_GBRP14LE &&

  1333.         srcFormat != AV_PIX_FMT_GBRP16BE  && srcFormat != AV_PIX_FMT_GBRP16LE &&

  1334.         ((dstW >> c->chrDstHSubSample) <= (srcW >> 1) ||

  1335.          (flags & SWS_FAST_BILINEAR)))

  1336.         c->chrSrcHSubSample = 1;

  1337. 

  1338.     // Note the FF_CEIL_RSHIFT is so that we always round toward +inf.

  1339.     c->chrSrcW = FF_CEIL_RSHIFT(srcW, c->chrSrcHSubSample);

  1340.     c->chrSrcH = FF_CEIL_RSHIFT(srcH, c->chrSrcVSubSample);

  1341.     c->chrDstW = FF_CEIL_RSHIFT(dstW, c->chrDstHSubSample);

  1342.     c->chrDstH = FF_CEIL_RSHIFT(dstH, c->chrDstVSubSample);

  1343. 

  1344.     FF_ALLOCZ_OR_GOTO(c, c->formatConvBuffer, FFALIGN(srcW*2+78, 16) * 2, fail);

  1345. 

  1346.     c->srcBpc = desc_src->comp[0].depth;

  1347.     if (c->srcBpc < 8)

  1348.         c->srcBpc = 8;

  1349.     c->dstBpc = desc_dst->comp[0].depth;

  1350.     if (c->dstBpc < 8)

  1351.         c->dstBpc = 8;

  1352.     if (isAnyRGB(srcFormat) || srcFormat == AV_PIX_FMT_PAL8)

  1353.         c->srcBpc = 16;

  1354.     if (c->dstBpc == 16)

  1355.         dst_stride <<= 1;

  1356. 

  1357.     if (INLINE_MMXEXT(cpu_flags) && c->srcBpc == 8 && c->dstBpc <= 14) {

  1358.         c->canMMXEXTBeUsed = dstW >= srcW && (dstW & 31) == 0 &&

  1359.                              c->chrDstW >= c->chrSrcW &&

  1360.                              (srcW & 15) == 0;

  1361.         if (!c->canMMXEXTBeUsed && dstW >= srcW && c->chrDstW >= c->chrSrcW && (srcW & 15) == 0

  1362. 

  1363.             && (flags & SWS_FAST_BILINEAR)) {

  1364.             if (flags & SWS_PRINT_INFO)

  1365.                 av_log(c, AV_LOG_INFO,

  1366.                        "output width is not a multiple of 32 -> no MMXEXT scaler\n");

  1367.         }

  1368.         if (usesHFilter || isNBPS(c->srcFormat) || is16BPS(c->srcFormat) || isAnyRGB(c->srcFormat))

  1369.             c->canMMXEXTBeUsed = 0;

  1370.     } else

  1371.         c->canMMXEXTBeUsed = 0;

  1372. 

  1373.     c->chrXInc = (((int64_t)c->chrSrcW << 16) + (c->chrDstW >> 1)) / c->chrDstW;

  1374.     c->chrYInc = (((int64_t)c->chrSrcH << 16) + (c->chrDstH >> 1)) / c->chrDstH;

  1375. 

  1376.     /* Match pixel 0 of the src to pixel 0 of dst and match pixel n-2 of src

  1377.      * to pixel n-2 of dst, but only for the FAST_BILINEAR mode otherwise do

  1378.      * correct scaling.

  1379.      * n-2 is the last chrominance sample available.

  1380.      * This is not perfect, but no one should notice the difference, the more

  1381.      * correct variant would be like the vertical one, but that would require

  1382.      * some special code for the first and last pixel */

  1383.     if (flags & SWS_FAST_BILINEAR) {

  1384.         if (c->canMMXEXTBeUsed) {

  1385.             c->lumXInc += 20;

  1386.             c->chrXInc += 20;

  1387.         }

  1388.         // we don't use the x86 asm scaler if MMX is available

  1389.         else if (INLINE_MMX(cpu_flags) && c->dstBpc <= 14) {

  1390.             c->lumXInc = ((int64_t)(srcW       - 2) << 16) / (dstW       - 2) - 20;

  1391.             c->chrXInc = ((int64_t)(c->chrSrcW - 2) << 16) / (c->chrDstW - 2) - 20;

  1392.         }

  1393.     }

  1394. 

  1395.     // hardcoded for now

  1396.     c->gamma_value = 2.2;

  1397.     tmpFmt = AV_PIX_FMT_RGBA64LE;

  1398. 

  1399. 

  1400.     if (!unscaled && c->gamma_flag && (srcFormat != tmpFmt || dstFormat != tmpFmt)) {

  1401.         SwsContext *c2;

  1402.         c->cascaded_context[0] = NULL;

  1403. 

  1404.         ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  1405.                             srcW, srcH, tmpFmt, 64);

  1406.         if (ret < 0)

  1407.             return ret;

  1408. 

  1409.         c->cascaded_context[0] = sws_getContext(srcW, srcH, srcFormat,

  1410.                                                 srcW, srcH, tmpFmt,

  1411.                                                 flags, NULL, NULL, c->param);

  1412.         if (!c->cascaded_context[0]) {

  1413.             return -1;

  1414.         }

  1415. 

  1416.         c->cascaded_context[1] = sws_getContext(srcW, srcH, tmpFmt,

  1417.                                                 dstW, dstH, tmpFmt,

  1418.                                                 flags, srcFilter, dstFilter, c->param);

  1419. 

  1420.         if (!c->cascaded_context[1])

  1421.             return -1;

  1422. 

  1423.         c2 = c->cascaded_context[1];

  1424.         c2->is_internal_gamma = 1;

  1425.         c2->gamma     = alloc_gamma_tbl(    c->gamma_value);

  1426.         c2->inv_gamma = alloc_gamma_tbl(1.f/c->gamma_value);

  1427.         if (!c2->gamma || !c2->inv_gamma)

  1428.             return AVERROR(ENOMEM);

  1429. 

  1430.         // is_internal_flag is set after creating the context

  1431.         // to properly create the gamma convert FilterDescriptor

  1432.         // we have to re-initialize it

  1433.         ff_free_filters(c2);

  1434.         if (ff_init_filters(c2) < 0) {

  1435.             sws_freeContext(c2);

  1436.             return -1;

  1437.         }

  1438. 

  1439.         c->cascaded_context[2] = NULL;

  1440.         if (dstFormat != tmpFmt) {

  1441.             ret = av_image_alloc(c->cascaded1_tmp, c->cascaded1_tmpStride,

  1442.                                 dstW, dstH, tmpFmt, 64);

  1443.             if (ret < 0)

  1444.                 return ret;

  1445. 

  1446.             c->cascaded_context[2] = sws_getContext(dstW, dstH, tmpFmt,

  1447.                                                 dstW, dstH, dstFormat,

  1448.                                                 flags, NULL, NULL, c->param);

  1449.             if (!c->cascaded_context[2])

  1450.                 return -1;

  1451.         }

  1452.         return 0;

  1453.     }

  1454. 

  1455.     if (isBayer(srcFormat)) {

  1456.         if (!unscaled ||

  1457.             (dstFormat != AV_PIX_FMT_RGB24 && dstFormat != AV_PIX_FMT_YUV420P)) {

  1458.             enum AVPixelFormat tmpFormat = AV_PIX_FMT_RGB24;

  1459. 

  1460.             ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  1461.                                 srcW, srcH, tmpFormat, 64);

  1462.             if (ret < 0)

  1463.                 return ret;

  1464. 

  1465.             c->cascaded_context[0] = sws_getContext(srcW, srcH, srcFormat,

  1466.                                                     srcW, srcH, tmpFormat,

  1467.                                                     flags, srcFilter, NULL, c->param);

  1468.             if (!c->cascaded_context[0])

  1469.                 return -1;

  1470. 

  1471.             c->cascaded_context[1] = sws_getContext(srcW, srcH, tmpFormat,

  1472.                                                     dstW, dstH, dstFormat,

  1473.                                                     flags, NULL, dstFilter, c->param);

  1474.             if (!c->cascaded_context[1])

  1475.                 return -1;

  1476.             return 0;

  1477.         }

  1478.     }

  1479. 

  1480.     if (CONFIG_SWSCALE_ALPHA && isALPHA(srcFormat) && !isALPHA(dstFormat)) {

  1481.         enum AVPixelFormat tmpFormat = alphaless_fmt(srcFormat);

  1482. 

  1483.         if (tmpFormat != AV_PIX_FMT_NONE && c->alphablend != SWS_ALPHA_BLEND_NONE)

  1484.         if (!unscaled ||

  1485.             dstFormat != tmpFormat ||

  1486.             usesHFilter || usesVFilter ||

  1487.             c->srcRange != c->dstRange

  1488.         ) {

  1489.             c->cascaded_mainindex = 1;

  1490.             ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  1491.                                 srcW, srcH, tmpFormat, 64);

  1492.             if (ret < 0)

  1493.                 return ret;

  1494. 

  1495.             c->cascaded_context[0] = sws_alloc_set_opts(srcW, srcH, srcFormat,

  1496.                                                         srcW, srcH, tmpFormat,

  1497.                                                         flags, c->param);

  1498.             if (!c->cascaded_context[0])

  1499.                 return -1;

  1500.             c->cascaded_context[0]->alphablend = c->alphablend;

  1501.             ret = sws_init_context(c->cascaded_context[0], NULL , NULL);

  1502.             if (ret < 0)

  1503.                 return ret;

  1504. 

  1505.             c->cascaded_context[1] = sws_alloc_set_opts(srcW, srcH, tmpFormat,

  1506.                                                         dstW, dstH, dstFormat,

  1507.                                                         flags, c->param);

  1508.             if (!c->cascaded_context[1])

  1509.                 return -1;

  1510. 

  1511.             c->cascaded_context[1]->srcRange = c->srcRange;

  1512.             c->cascaded_context[1]->dstRange = c->dstRange;

  1513.             ret = sws_init_context(c->cascaded_context[1], srcFilter , dstFilter);

  1514.             if (ret < 0)

  1515.                 return ret;

  1516. 

  1517.             return 0;

  1518.         }

  1519.     }

  1520. 

  1521. #define USE_MMAP (HAVE_MMAP && HAVE_MPROTECT && defined MAP_ANONYMOUS)

  1522. 

  1523.     /* precalculate horizontal scaler filter coefficients */

  1524.     {

  1525. #if HAVE_MMXEXT_INLINE

  1526. // can't downscale !!!

  1527.         if (c->canMMXEXTBeUsed && (flags & SWS_FAST_BILINEAR)) {

  1528.             c->lumMmxextFilterCodeSize = ff_init_hscaler_mmxext(dstW, c->lumXInc, NULL,

  1529.                                                              NULL, NULL, 8);

  1530.             c->chrMmxextFilterCodeSize = ff_init_hscaler_mmxext(c->chrDstW, c->chrXInc,

  1531.                                                              NULL, NULL, NULL, 4);

  1532. 

  1533. #if USE_MMAP

  1534.             c->lumMmxextFilterCode = mmap(NULL, c->lumMmxextFilterCodeSize,

  1535.                                           PROT_READ | PROT_WRITE,

  1536.                                           MAP_PRIVATE | MAP_ANONYMOUS,

  1537.                                           -1, 0);

  1538.             c->chrMmxextFilterCode = mmap(NULL, c->chrMmxextFilterCodeSize,

  1539.                                           PROT_READ | PROT_WRITE,

  1540.                                           MAP_PRIVATE | MAP_ANONYMOUS,

  1541.                                           -1, 0);

  1542. #elif HAVE_VIRTUALALLOC

  1543.             c->lumMmxextFilterCode = VirtualAlloc(NULL,

  1544.                                                   c->lumMmxextFilterCodeSize,

  1545.                                                   MEM_COMMIT,

  1546.                                                   PAGE_EXECUTE_READWRITE);

  1547.             c->chrMmxextFilterCode = VirtualAlloc(NULL,

  1548.                                                   c->chrMmxextFilterCodeSize,

  1549.                                                   MEM_COMMIT,

  1550.                                                   PAGE_EXECUTE_READWRITE);

  1551. #else

  1552.             c->lumMmxextFilterCode = av_malloc(c->lumMmxextFilterCodeSize);

  1553.             c->chrMmxextFilterCode = av_malloc(c->chrMmxextFilterCodeSize);

  1554. #endif

  1555. 

  1556. #ifdef MAP_ANONYMOUS

  1557.             if (c->lumMmxextFilterCode == MAP_FAILED || c->chrMmxextFilterCode == MAP_FAILED)

  1558. #else

  1559.             if (!c->lumMmxextFilterCode || !c->chrMmxextFilterCode)

  1560. #endif

  1561.             {

  1562.                 av_log(c, AV_LOG_ERROR, "Failed to allocate MMX2FilterCode\n");

  1563.                 return AVERROR(ENOMEM);

  1564.             }

  1565. 

  1566.             FF_ALLOCZ_OR_GOTO(c, c->hLumFilter,    (dstW           / 8 + 8) * sizeof(int16_t), fail);

  1567.             FF_ALLOCZ_OR_GOTO(c, c->hChrFilter,    (c->chrDstW     / 4 + 8) * sizeof(int16_t), fail);

  1568.             FF_ALLOCZ_OR_GOTO(c, c->hLumFilterPos, (dstW       / 2 / 8 + 8) * sizeof(int32_t), fail);

  1569.             FF_ALLOCZ_OR_GOTO(c, c->hChrFilterPos, (c->chrDstW / 2 / 4 + 8) * sizeof(int32_t), fail);

  1570. 

  1571.             ff_init_hscaler_mmxext(      dstW, c->lumXInc, c->lumMmxextFilterCode,

  1572.                                 c->hLumFilter, (uint32_t*)c->hLumFilterPos, 8);

  1573.             ff_init_hscaler_mmxext(c->chrDstW, c->chrXInc, c->chrMmxextFilterCode,

  1574.                                 c->hChrFilter, (uint32_t*)c->hChrFilterPos, 4);

  1575. 

  1576. #if USE_MMAP

  1577.             if (   mprotect(c->lumMmxextFilterCode, c->lumMmxextFilterCodeSize, PROT_EXEC | PROT_READ) == -1

  1578.                 || mprotect(c->chrMmxextFilterCode, c->chrMmxextFilterCodeSize, PROT_EXEC | PROT_READ) == -1) {

  1579.                 av_log(c, AV_LOG_ERROR, "mprotect failed, cannot use fast bilinear scaler\n");

  1580.                 goto fail;

  1581.             }

  1582. #endif

  1583.         } else

  1584. #endif /* HAVE_MMXEXT_INLINE */

  1585.         {

  1586.             const int filterAlign = X86_MMX(cpu_flags)     ? 4 :

  1587.                                     PPC_ALTIVEC(cpu_flags) ? 8 : 1;

  1588. 

  1589.             if ((ret = initFilter(&c->hLumFilter, &c->hLumFilterPos,

  1590.                            &c->hLumFilterSize, c->lumXInc,

  1591.                            srcW, dstW, filterAlign, 1 << 14,

  1592.                            (flags & SWS_BICUBLIN) ? (flags | SWS_BICUBIC) : flags,

  1593.                            cpu_flags, srcFilter->lumH, dstFilter->lumH,

  1594.                            c->param,

  1595.                            get_local_pos(c, 0, 0, 0),

  1596.                            get_local_pos(c, 0, 0, 0))) < 0)

  1597.                 goto fail;

  1598.             if ((ret = initFilter(&c->hChrFilter, &c->hChrFilterPos,

  1599.                            &c->hChrFilterSize, c->chrXInc,

  1600.                            c->chrSrcW, c->chrDstW, filterAlign, 1 << 14,

  1601.                            (flags & SWS_BICUBLIN) ? (flags | SWS_BILINEAR) : flags,

  1602.                            cpu_flags, srcFilter->chrH, dstFilter->chrH,

  1603.                            c->param,

  1604.                            get_local_pos(c, c->chrSrcHSubSample, c->src_h_chr_pos, 0),

  1605.                            get_local_pos(c, c->chrDstHSubSample, c->dst_h_chr_pos, 0))) < 0)

  1606.                 goto fail;

  1607.         }

  1608.     } // initialize horizontal stuff

  1609. 

  1610.     /* precalculate vertical scaler filter coefficients */

  1611.     {

  1612.         const int filterAlign = X86_MMX(cpu_flags)     ? 2 :

  1613.                                 PPC_ALTIVEC(cpu_flags) ? 8 : 1;

  1614. 

  1615.         if ((ret = initFilter(&c->vLumFilter, &c->vLumFilterPos, &c->vLumFilterSize,

  1616.                        c->lumYInc, srcH, dstH, filterAlign, (1 << 12),

  1617.                        (flags & SWS_BICUBLIN) ? (flags | SWS_BICUBIC) : flags,

  1618.                        cpu_flags, srcFilter->lumV, dstFilter->lumV,

  1619.                        c->param,

  1620.                        get_local_pos(c, 0, 0, 1),

  1621.                        get_local_pos(c, 0, 0, 1))) < 0)

  1622.             goto fail;

  1623.         if ((ret = initFilter(&c->vChrFilter, &c->vChrFilterPos, &c->vChrFilterSize,

  1624.                        c->chrYInc, c->chrSrcH, c->chrDstH,

  1625.                        filterAlign, (1 << 12),

  1626.                        (flags & SWS_BICUBLIN) ? (flags | SWS_BILINEAR) : flags,

  1627.                        cpu_flags, srcFilter->chrV, dstFilter->chrV,

  1628.                        c->param,

  1629.                        get_local_pos(c, c->chrSrcVSubSample, c->src_v_chr_pos, 1),

  1630.                        get_local_pos(c, c->chrDstVSubSample, c->dst_v_chr_pos, 1))) < 0)

  1631. 

  1632.             goto fail;

  1633. 

  1634. #if HAVE_ALTIVEC

  1635.         FF_ALLOC_OR_GOTO(c, c->vYCoeffsBank, sizeof(vector signed short) * c->vLumFilterSize * c->dstH,    fail);

  1636.         FF_ALLOC_OR_GOTO(c, c->vCCoeffsBank, sizeof(vector signed short) * c->vChrFilterSize * c->chrDstH, fail);

  1637. 

  1638.         for (i = 0; i < c->vLumFilterSize * c->dstH; i++) {

  1639.             int j;

  1640.             short *p = (short *)&c->vYCoeffsBank[i];

  1641.             for (j = 0; j < 8; j++)

  1642.                 p[j] = c->vLumFilter[i];

  1643.         }

  1644. 

  1645.         for (i = 0; i < c->vChrFilterSize * c->chrDstH; i++) {

  1646.             int j;

  1647.             short *p = (short *)&c->vCCoeffsBank[i];

  1648.             for (j = 0; j < 8; j++)

  1649.                 p[j] = c->vChrFilter[i];

  1650.         }

  1651. #endif

  1652.     }

  1653. 

  1654.     // calculate buffer sizes so that they won't run out while handling these damn slices

  1655.     c->vLumBufSize = c->vLumFilterSize;

  1656.     c->vChrBufSize = c->vChrFilterSize;

  1657.     for (i = 0; i < dstH; i++) {

  1658.         int chrI      = (int64_t)i * c->chrDstH / dstH;

  1659.         int nextSlice = FFMAX(c->vLumFilterPos[i] + c->vLumFilterSize - 1,

  1660.                               ((c->vChrFilterPos[chrI] + c->vChrFilterSize - 1)

  1661.                                << c->chrSrcVSubSample));

  1662. 

  1663.         nextSlice >>= c->chrSrcVSubSample;

  1664.         nextSlice <<= c->chrSrcVSubSample;

  1665.         if (c->vLumFilterPos[i] + c->vLumBufSize < nextSlice)

  1666.             c->vLumBufSize = nextSlice - c->vLumFilterPos[i];

  1667.         if (c->vChrFilterPos[chrI] + c->vChrBufSize <

  1668.             (nextSlice >> c->chrSrcVSubSample))

  1669.             c->vChrBufSize = (nextSlice >> c->chrSrcVSubSample) -

  1670.                              c->vChrFilterPos[chrI];

  1671.     }

  1672. 

  1673.     for (i = 0; i < 4; i++)

  1674.         FF_ALLOCZ_OR_GOTO(c, c->dither_error[i], (c->dstW+2) * sizeof(int), fail);

  1675. 

  1676.     /* Allocate pixbufs (we use dynamic allocation because otherwise we would

  1677.      * need to allocate several megabytes to handle all possible cases) */

  1678.     FF_ALLOCZ_OR_GOTO(c, c->lumPixBuf,  c->vLumBufSize * 3 * sizeof(int16_t *), fail);

  1679.     FF_ALLOCZ_OR_GOTO(c, c->chrUPixBuf, c->vChrBufSize * 3 * sizeof(int16_t *), fail);

  1680.     FF_ALLOCZ_OR_GOTO(c, c->chrVPixBuf, c->vChrBufSize * 3 * sizeof(int16_t *), fail);

  1681.     if (CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat) && isALPHA(c->dstFormat))

  1682.         FF_ALLOCZ_OR_GOTO(c, c->alpPixBuf, c->vLumBufSize * 3 * sizeof(int16_t *), fail);

  1683.     /* Note we need at least one pixel more at the end because of the MMX code

  1684.      * (just in case someone wants to replace the 4000/8000). */

  1685.     /* align at 16 bytes for AltiVec */

  1686.     for (i = 0; i < c->vLumBufSize; i++) {

  1687.         FF_ALLOCZ_OR_GOTO(c, c->lumPixBuf[i + c->vLumBufSize],

  1688.                           dst_stride + 16, fail);

  1689.         c->lumPixBuf[i] = c->lumPixBuf[i + c->vLumBufSize];

  1690.     }

  1691.     // 64 / c->scalingBpp is the same as 16 / sizeof(scaling_intermediate)

  1692.     c->uv_off   = (dst_stride>>1) + 64 / (c->dstBpc &~ 7);

  1693.     c->uv_offx2 = dst_stride + 16;

  1694.     for (i = 0; i < c->vChrBufSize; i++) {

  1695.         FF_ALLOC_OR_GOTO(c, c->chrUPixBuf[i + c->vChrBufSize],

  1696.                          dst_stride * 2 + 32, fail);

  1697.         c->chrUPixBuf[i] = c->chrUPixBuf[i + c->vChrBufSize];

  1698.         c->chrVPixBuf[i] = c->chrVPixBuf[i + c->vChrBufSize]

  1699.                          = c->chrUPixBuf[i] + (dst_stride >> 1) + 8;

  1700.     }

  1701.     if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf)

  1702.         for (i = 0; i < c->vLumBufSize; i++) {

  1703.             FF_ALLOCZ_OR_GOTO(c, c->alpPixBuf[i + c->vLumBufSize],

  1704.                               dst_stride + 16, fail);

  1705.             c->alpPixBuf[i] = c->alpPixBuf[i + c->vLumBufSize];

  1706.         }

  1707. 

  1708.     // try to avoid drawing green stuff between the right end and the stride end

  1709.     for (i = 0; i < c->vChrBufSize; i++)

  1710.         if(desc_dst->comp[0].depth == 16){

  1711.             av_assert0(c->dstBpc > 14);

  1712.             for(j=0; j<dst_stride/2+1; j++)

  1713.                 ((int32_t*)(c->chrUPixBuf[i]))[j] = 1<<18;

  1714.         } else

  1715.             for(j=0; j<dst_stride+1; j++)

  1716.                 ((int16_t*)(c->chrUPixBuf[i]))[j] = 1<<14;

  1717. 

  1718.     av_assert0(c->chrDstH <= dstH);

  1719. 

  1720.     if (flags & SWS_PRINT_INFO) {

  1721.         const char *scaler = NULL, *cpucaps;

  1722. 

  1723.         for (i = 0; i < FF_ARRAY_ELEMS(scale_algorithms); i++) {

  1724.             if (flags & scale_algorithms[i].flag) {

  1725.                 scaler = scale_algorithms[i].description;

  1726.                 break;

  1727.             }

  1728.         }

  1729.         if (!scaler)

  1730.             scaler =  "ehh flags invalid?!";

  1731.         av_log(c, AV_LOG_INFO, "%s scaler, from %s to %s%s ",

  1732.                scaler,

  1733.                av_get_pix_fmt_name(srcFormat),

  1734. #ifdef DITHER1XBPP

  1735.                dstFormat == AV_PIX_FMT_BGR555   || dstFormat == AV_PIX_FMT_BGR565   ||

  1736.                dstFormat == AV_PIX_FMT_RGB444BE || dstFormat == AV_PIX_FMT_RGB444LE ||

  1737.                dstFormat == AV_PIX_FMT_BGR444BE || dstFormat == AV_PIX_FMT_BGR444LE ?

  1738.                                                              "dithered " : "",

  1739. #else

  1740.                "",

  1741. #endif

  1742.                av_get_pix_fmt_name(dstFormat));

  1743. 

  1744.         if (INLINE_MMXEXT(cpu_flags))

  1745.             cpucaps = "MMXEXT";

  1746.         else if (INLINE_AMD3DNOW(cpu_flags))

  1747.             cpucaps = "3DNOW";

  1748.         else if (INLINE_MMX(cpu_flags))

  1749.             cpucaps = "MMX";

  1750.         else if (PPC_ALTIVEC(cpu_flags))

  1751.             cpucaps = "AltiVec";

  1752.         else

  1753.             cpucaps = "C";

  1754. 

  1755.         av_log(c, AV_LOG_INFO, "using %s\n", cpucaps);

  1756. 

  1757.         av_log(c, AV_LOG_VERBOSE, "%dx%d -> %dx%d\n", srcW, srcH, dstW, dstH);

  1758.         av_log(c, AV_LOG_DEBUG,

  1759.                "lum srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n",

  1760.                c->srcW, c->srcH, c->dstW, c->dstH, c->lumXInc, c->lumYInc);

  1761.         av_log(c, AV_LOG_DEBUG,

  1762.                "chr srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n",

  1763.                c->chrSrcW, c->chrSrcH, c->chrDstW, c->chrDstH,

  1764.                c->chrXInc, c->chrYInc);

  1765.     }

  1766. 

  1767.     /* alpha blend special case, note this has been split via cascaded contexts if its scaled */

  1768.     if (unscaled && !usesHFilter && !usesVFilter &&

  1769.         c->alphablend != SWS_ALPHA_BLEND_NONE &&

  1770.         isALPHA(srcFormat) &&

  1771.         (c->srcRange == c->dstRange || isAnyRGB(dstFormat)) &&

  1772.         alphaless_fmt(srcFormat) == dstFormat

  1773.     ) {

  1774.         c->swscale = ff_sws_alphablendaway;

  1775. 

  1776.         if (flags & SWS_PRINT_INFO)

  1777.             av_log(c, AV_LOG_INFO,

  1778.                     "using alpha blendaway %s -> %s special converter\n",

  1779.                     av_get_pix_fmt_name(srcFormat), av_get_pix_fmt_name(dstFormat));

  1780.         return 0;

  1781.     }

  1782. 

  1783.     /* unscaled special cases */

  1784.     if (unscaled && !usesHFilter && !usesVFilter &&

  1785.         (c->srcRange == c->dstRange || isAnyRGB(dstFormat))) {

  1786.         ff_get_unscaled_swscale(c);

  1787. 

  1788.         if (c->swscale) {

  1789.             if (flags & SWS_PRINT_INFO)

  1790.                 av_log(c, AV_LOG_INFO,

  1791.                        "using unscaled %s -> %s special converter\n",

  1792.                        av_get_pix_fmt_name(srcFormat), av_get_pix_fmt_name(dstFormat));

  1793.             return 0;

  1794.         }

  1795.     }

  1796. 

  1797.     c->swscale = ff_getSwsFunc(c);

  1798.     return ff_init_filters(c);

  1799. fail: // FIXME replace things by appropriate error codes

  1800.     if (ret == RETCODE_USE_CASCADE)  {

  1801.         int tmpW = sqrt(srcW * (int64_t)dstW);

  1802.         int tmpH = sqrt(srcH * (int64_t)dstH);

  1803.         enum AVPixelFormat tmpFormat = AV_PIX_FMT_YUV420P;

  1804. 

  1805.         if (srcW*(int64_t)srcH <= 4LL*dstW*dstH)

  1806.             return AVERROR(EINVAL);

  1807. 

  1808.         ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  1809.                              tmpW, tmpH, tmpFormat, 64);

  1810.         if (ret < 0)

  1811.             return ret;

  1812. 

  1813.         c->cascaded_context[0] = sws_getContext(srcW, srcH, srcFormat,

  1814.                                                 tmpW, tmpH, tmpFormat,

  1815.                                                 flags, srcFilter, NULL, c->param);

  1816.         if (!c->cascaded_context[0])

  1817.             return -1;

  1818. 

  1819.         c->cascaded_context[1] = sws_getContext(tmpW, tmpH, tmpFormat,

  1820.                                                 dstW, dstH, dstFormat,

  1821.                                                 flags, NULL, dstFilter, c->param);

  1822.         if (!c->cascaded_context[1])

  1823.             return -1;

  1824.         return 0;

  1825.     }

  1826.     return -1;

  1827. }

  1828. 

  1829. SwsContext *sws_alloc_set_opts(int srcW, int srcH, enum AVPixelFormat srcFormat,

  1830.                                int dstW, int dstH, enum AVPixelFormat dstFormat,

  1831.                                int flags, const double *param)

  1832. {

  1833.     SwsContext *c;

  1834. 

  1835.     if (!(c = sws_alloc_context()))

  1836.         return NULL;

  1837. 

  1838.     c->flags     = flags;

  1839.     c->srcW      = srcW;

  1840.     c->srcH      = srcH;

  1841.     c->dstW      = dstW;

  1842.     c->dstH      = dstH;

  1843.     c->srcFormat = srcFormat;

  1844.     c->dstFormat = dstFormat;

  1845. 

  1846.     if (param) {

  1847.         c->param[0] = param[0];

  1848.         c->param[1] = param[1];

  1849.     }

  1850. 

  1851.     return c;

  1852. }

  1853. 

  1854. SwsContext *sws_getContext(int srcW, int srcH, enum AVPixelFormat srcFormat,

  1855.                            int dstW, int dstH, enum AVPixelFormat dstFormat,

  1856.                            int flags, SwsFilter *srcFilter,

  1857.                            SwsFilter *dstFilter, const double *param)

  1858. {

  1859.     SwsContext *c;

  1860. 

  1861.     c = sws_alloc_set_opts(srcW, srcH, srcFormat,

  1862.                            dstW, dstH, dstFormat,

  1863.                            flags, param);

  1864.     if (!c)

  1865.         return NULL;

  1866. 

  1867.     if (sws_init_context(c, srcFilter, dstFilter) < 0) {

  1868.         sws_freeContext(c);

  1869.         return NULL;

  1870.     }

  1871. 

  1872.     return c;

  1873. }

  1874. 

  1875. static int isnan_vec(SwsVector *a)

  1876. {

  1877.     int i;

  1878.     for (i=0; i<a->length; i++)

  1879.         if (isnan(a->coeff[i]))

  1880.             return 1;

  1881.     return 0;

  1882. }

  1883. 

  1884. static void makenan_vec(SwsVector *a)

  1885. {

  1886.     int i;

  1887.     for (i=0; i<a->length; i++)

  1888.         a->coeff[i] = NAN;

  1889. }

  1890. 

  1891. SwsFilter *sws_getDefaultFilter(float lumaGBlur, float chromaGBlur,

  1892.                                 float lumaSharpen, float chromaSharpen,

  1893.                                 float chromaHShift, float chromaVShift,

  1894.                                 int verbose)

  1895. {

  1896.     SwsFilter *filter = av_malloc(sizeof(SwsFilter));

  1897.     if (!filter)

  1898.         return NULL;

  1899. 

  1900.     if (lumaGBlur != 0.0) {

  1901.         filter->lumH = sws_getGaussianVec(lumaGBlur, 3.0);

  1902.         filter->lumV = sws_getGaussianVec(lumaGBlur, 3.0);

  1903.     } else {

  1904.         filter->lumH = sws_getIdentityVec();

  1905.         filter->lumV = sws_getIdentityVec();

  1906.     }

  1907. 

  1908.     if (chromaGBlur != 0.0) {

  1909.         filter->chrH = sws_getGaussianVec(chromaGBlur, 3.0);

  1910.         filter->chrV = sws_getGaussianVec(chromaGBlur, 3.0);

  1911.     } else {

  1912.         filter->chrH = sws_getIdentityVec();

  1913.         filter->chrV = sws_getIdentityVec();

  1914.     }

  1915. 

  1916.     if (!filter->lumH || !filter->lumV || !filter->chrH || !filter->chrV)

  1917.         goto fail;

  1918. 

  1919.     if (chromaSharpen != 0.0) {

  1920.         SwsVector *id = sws_getIdentityVec();

  1921.         if (!id)

  1922.             goto fail;

  1923.         sws_scaleVec(filter->chrH, -chromaSharpen);

  1924.         sws_scaleVec(filter->chrV, -chromaSharpen);

  1925.         sws_addVec(filter->chrH, id);

  1926.         sws_addVec(filter->chrV, id);

  1927.         sws_freeVec(id);

  1928.     }

  1929. 

  1930.     if (lumaSharpen != 0.0) {

  1931.         SwsVector *id = sws_getIdentityVec();

  1932.         if (!id)

  1933.             goto fail;

  1934.         sws_scaleVec(filter->lumH, -lumaSharpen);

  1935.         sws_scaleVec(filter->lumV, -lumaSharpen);

  1936.         sws_addVec(filter->lumH, id);

  1937.         sws_addVec(filter->lumV, id);

  1938.         sws_freeVec(id);

  1939.     }

  1940. 

  1941.     if (chromaHShift != 0.0)

  1942.         sws_shiftVec(filter->chrH, (int)(chromaHShift + 0.5));

  1943. 

  1944.     if (chromaVShift != 0.0)

  1945.         sws_shiftVec(filter->chrV, (int)(chromaVShift + 0.5));

  1946. 

  1947.     sws_normalizeVec(filter->chrH, 1.0);

  1948.     sws_normalizeVec(filter->chrV, 1.0);

  1949.     sws_normalizeVec(filter->lumH, 1.0);

  1950.     sws_normalizeVec(filter->lumV, 1.0);

  1951. 

  1952.     if (isnan_vec(filter->chrH) ||

  1953.         isnan_vec(filter->chrV) ||

  1954.         isnan_vec(filter->lumH) ||

  1955.         isnan_vec(filter->lumV))

  1956.         goto fail;

  1957. 

  1958.     if (verbose)

  1959.         sws_printVec2(filter->chrH, NULL, AV_LOG_DEBUG);

  1960.     if (verbose)

  1961.         sws_printVec2(filter->lumH, NULL, AV_LOG_DEBUG);

  1962. 

  1963.     return filter;

  1964. 

  1965. fail:

  1966.     sws_freeVec(filter->lumH);

  1967.     sws_freeVec(filter->lumV);

  1968.     sws_freeVec(filter->chrH);

  1969.     sws_freeVec(filter->chrV);

  1970.     av_freep(&filter);

  1971.     return NULL;

  1972. }

  1973. 

  1974. SwsVector *sws_allocVec(int length)

  1975. {

  1976.     SwsVector *vec;

  1977. 

  1978.     if(length <= 0 || length > INT_MAX/ sizeof(double))

  1979.         return NULL;

  1980. 

  1981.     vec = av_malloc(sizeof(SwsVector));

  1982.     if (!vec)

  1983.         return NULL;

  1984.     vec->length = length;

  1985.     vec->coeff  = av_malloc(sizeof(double) * length);

  1986.     if (!vec->coeff)

  1987.         av_freep(&vec);

  1988.     return vec;

  1989. }

  1990. 

  1991. SwsVector *sws_getGaussianVec(double variance, double quality)

  1992. {

  1993.     const int length = (int)(variance * quality + 0.5) | 1;

  1994.     int i;

  1995.     double middle  = (length - 1) * 0.5;

  1996.     SwsVector *vec;

  1997. 

  1998.     if(variance < 0 || quality < 0)

  1999.         return NULL;

  2000. 

  2001.     vec = sws_allocVec(length);

  2002. 

  2003.     if (!vec)

  2004.         return NULL;

  2005. 

  2006.     for (i = 0; i < length; i++) {

  2007.         double dist = i - middle;

  2008.         vec->coeff[i] = exp(-dist * dist / (2 * variance * variance)) /

  2009.                         sqrt(2 * variance * M_PI);

  2010.     }

  2011. 

  2012.     sws_normalizeVec(vec, 1.0);

  2013. 

  2014.     return vec;

  2015. }

  2016. 

  2017. SwsVector *sws_getConstVec(double c, int length)

  2018. {

  2019.     int i;

  2020.     SwsVector *vec = sws_allocVec(length);

  2021. 

  2022.     if (!vec)

  2023.         return NULL;

  2024. 

  2025.     for (i = 0; i < length; i++)

  2026.         vec->coeff[i] = c;

  2027. 

  2028.     return vec;

  2029. }

  2030. 

  2031. SwsVector *sws_getIdentityVec(void)

  2032. {

  2033.     return sws_getConstVec(1.0, 1);

  2034. }

  2035. 

  2036. static double sws_dcVec(SwsVector *a)

  2037. {

  2038.     int i;

  2039.     double sum = 0;

  2040. 

  2041.     for (i = 0; i < a->length; i++)

  2042.         sum += a->coeff[i];

  2043. 

  2044.     return sum;

  2045. }

  2046. 

  2047. void sws_scaleVec(SwsVector *a, double scalar)

  2048. {

  2049.     int i;

  2050. 

  2051.     for (i = 0; i < a->length; i++)

  2052.         a->coeff[i] *= scalar;

  2053. }

  2054. 

  2055. void sws_normalizeVec(SwsVector *a, double height)

  2056. {

  2057.     sws_scaleVec(a, height / sws_dcVec(a));

  2058. }

  2059. 

  2060. static SwsVector *sws_getConvVec(SwsVector *a, SwsVector *b)

  2061. {

  2062.     int length = a->length + b->length - 1;

  2063.     int i, j;

  2064.     SwsVector *vec = sws_getConstVec(0.0, length);

  2065. 

  2066.     if (!vec)

  2067.         return NULL;

  2068. 

  2069.     for (i = 0; i < a->length; i++) {

  2070.         for (j = 0; j < b->length; j++) {

  2071.             vec->coeff[i + j] += a->coeff[i] * b->coeff[j];

  2072.         }

  2073.     }

  2074. 

  2075.     return vec;

  2076. }

  2077. 

  2078. static SwsVector *sws_sumVec(SwsVector *a, SwsVector *b)

  2079. {

  2080.     int length = FFMAX(a->length, b->length);

  2081.     int i;

  2082.     SwsVector *vec = sws_getConstVec(0.0, length);

  2083. 

  2084.     if (!vec)

  2085.         return NULL;

  2086. 

  2087.     for (i = 0; i < a->length; i++)

  2088.         vec->coeff[i + (length - 1) / 2 - (a->length - 1) / 2] += a->coeff[i];

  2089.     for (i = 0; i < b->length; i++)

  2090.         vec->coeff[i + (length - 1) / 2 - (b->length - 1) / 2] += b->coeff[i];

  2091. 

  2092.     return vec;

  2093. }

  2094. 

  2095. static SwsVector *sws_diffVec(SwsVector *a, SwsVector *b)

  2096. {

  2097.     int length = FFMAX(a->length, b->length);

  2098.     int i;

  2099.     SwsVector *vec = sws_getConstVec(0.0, length);

  2100. 

  2101.     if (!vec)

  2102.         return NULL;

  2103. 

  2104.     for (i = 0; i < a->length; i++)

  2105.         vec->coeff[i + (length - 1) / 2 - (a->length - 1) / 2] += a->coeff[i];

  2106.     for (i = 0; i < b->length; i++)

  2107.         vec->coeff[i + (length - 1) / 2 - (b->length - 1) / 2] -= b->coeff[i];

  2108. 

  2109.     return vec;

  2110. }

  2111. 

  2112. /* shift left / or right if "shift" is negative */

  2113. static SwsVector *sws_getShiftedVec(SwsVector *a, int shift)

  2114. {

  2115.     int length = a->length + FFABS(shift) * 2;

  2116.     int i;

  2117.     SwsVector *vec = sws_getConstVec(0.0, length);

  2118. 

  2119.     if (!vec)

  2120.         return NULL;

  2121. 

  2122.     for (i = 0; i < a->length; i++) {

  2123.         vec->coeff[i + (length    - 1) / 2 -

  2124.                        (a->length - 1) / 2 - shift] = a->coeff[i];

  2125.     }

  2126. 

  2127.     return vec;

  2128. }

  2129. 

  2130. void sws_shiftVec(SwsVector *a, int shift)

  2131. {

  2132.     SwsVector *shifted = sws_getShiftedVec(a, shift);

  2133.     if (!shifted) {

  2134.         makenan_vec(a);

  2135.         return;

  2136.     }

  2137.     av_free(a->coeff);

  2138.     a->coeff  = shifted->coeff;

  2139.     a->length = shifted->length;

  2140.     av_free(shifted);

  2141. }

  2142. 

  2143. void sws_addVec(SwsVector *a, SwsVector *b)

  2144. {

  2145.     SwsVector *sum = sws_sumVec(a, b);

  2146.     if (!sum) {

  2147.         makenan_vec(a);

  2148.         return;

  2149.     }

  2150.     av_free(a->coeff);

  2151.     a->coeff  = sum->coeff;

  2152.     a->length = sum->length;

  2153.     av_free(sum);

  2154. }

  2155. 

  2156. void sws_subVec(SwsVector *a, SwsVector *b)

  2157. {

  2158.     SwsVector *diff = sws_diffVec(a, b);

  2159.     if (!diff) {

  2160.         makenan_vec(a);

  2161.         return;

  2162.     }

  2163.     av_free(a->coeff);

  2164.     a->coeff  = diff->coeff;

  2165.     a->length = diff->length;

  2166.     av_free(diff);

  2167. }

  2168. 

  2169. void sws_convVec(SwsVector *a, SwsVector *b)

  2170. {

  2171.     SwsVector *conv = sws_getConvVec(a, b);

  2172.     if (!conv) {

  2173.         makenan_vec(a);

  2174.         return;

  2175.     }

  2176.     av_free(a->coeff);

  2177.     a->coeff  = conv->coeff;

  2178.     a->length = conv->length;

  2179.     av_free(conv);

  2180. }

  2181. 

  2182. SwsVector *sws_cloneVec(SwsVector *a)

  2183. {

  2184.     SwsVector *vec = sws_allocVec(a->length);

  2185. 

  2186.     if (!vec)

  2187.         return NULL;

  2188. 

  2189.     memcpy(vec->coeff, a->coeff, a->length * sizeof(*a->coeff));

  2190. 

  2191.     return vec;

  2192. }

  2193. 

  2194. void sws_printVec2(SwsVector *a, AVClass *log_ctx, int log_level)

  2195. {

  2196.     int i;

  2197.     double max = 0;

  2198.     double min = 0;

  2199.     double range;

  2200. 

  2201.     for (i = 0; i < a->length; i++)

  2202.         if (a->coeff[i] > max)

  2203.             max = a->coeff[i];

  2204. 

  2205.     for (i = 0; i < a->length; i++)

  2206.         if (a->coeff[i] < min)

  2207.             min = a->coeff[i];

  2208. 

  2209.     range = max - min;

  2210. 

  2211.     for (i = 0; i < a->length; i++) {

  2212.         int x = (int)((a->coeff[i] - min) * 60.0 / range + 0.5);

  2213.         av_log(log_ctx, log_level, "%1.3f ", a->coeff[i]);

  2214.         for (; x > 0; x--)

  2215.             av_log(log_ctx, log_level, " ");

  2216.         av_log(log_ctx, log_level, "|\n");

  2217.     }

  2218. }

  2219. 

  2220. void sws_freeVec(SwsVector *a)

  2221. {

  2222.     if (!a)

  2223.         return;

  2224.     av_freep(&a->coeff);

  2225.     a->length = 0;

  2226.     av_free(a);

  2227. }

  2228. 

  2229. void sws_freeFilter(SwsFilter *filter)

  2230. {

  2231.     if (!filter)

  2232.         return;

  2233. 

  2234.     sws_freeVec(filter->lumH);

  2235.     sws_freeVec(filter->lumV);

  2236.     sws_freeVec(filter->chrH);

  2237.     sws_freeVec(filter->chrV);

  2238.     av_free(filter);

  2239. }

  2240. 

  2241. void sws_freeContext(SwsContext *c)

  2242. {

  2243.     int i;

  2244.     if (!c)

  2245.         return;

  2246. 

  2247.     if (c->lumPixBuf) {

  2248.         for (i = 0; i < c->vLumBufSize; i++)

  2249.             av_freep(&c->lumPixBuf[i]);

  2250.         av_freep(&c->lumPixBuf);

  2251.     }

  2252. 

  2253.     if (c->chrUPixBuf) {

  2254.         for (i = 0; i < c->vChrBufSize; i++)

  2255.             av_freep(&c->chrUPixBuf[i]);

  2256.         av_freep(&c->chrUPixBuf);

  2257.         av_freep(&c->chrVPixBuf);

  2258.     }

  2259. 

  2260.     if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  2261.         for (i = 0; i < c->vLumBufSize; i++)

  2262.             av_freep(&c->alpPixBuf[i]);

  2263.         av_freep(&c->alpPixBuf);

  2264.     }

  2265. 

  2266.     for (i = 0; i < 4; i++)

  2267.         av_freep(&c->dither_error[i]);

  2268. 

  2269.     av_freep(&c->vLumFilter);

  2270.     av_freep(&c->vChrFilter);

  2271.     av_freep(&c->hLumFilter);

  2272.     av_freep(&c->hChrFilter);

  2273. #if HAVE_ALTIVEC

  2274.     av_freep(&c->vYCoeffsBank);

  2275.     av_freep(&c->vCCoeffsBank);

  2276. #endif

  2277. 

  2278.     av_freep(&c->vLumFilterPos);

  2279.     av_freep(&c->vChrFilterPos);

  2280.     av_freep(&c->hLumFilterPos);

  2281.     av_freep(&c->hChrFilterPos);

  2282. 

  2283. #if HAVE_MMX_INLINE

  2284. #if USE_MMAP

  2285.     if (c->lumMmxextFilterCode)

  2286.         munmap(c->lumMmxextFilterCode, c->lumMmxextFilterCodeSize);

  2287.     if (c->chrMmxextFilterCode)

  2288.         munmap(c->chrMmxextFilterCode, c->chrMmxextFilterCodeSize);

  2289. #elif HAVE_VIRTUALALLOC

  2290.     if (c->lumMmxextFilterCode)

  2291.         VirtualFree(c->lumMmxextFilterCode, 0, MEM_RELEASE);

  2292.     if (c->chrMmxextFilterCode)

  2293.         VirtualFree(c->chrMmxextFilterCode, 0, MEM_RELEASE);

  2294. #else

  2295.     av_free(c->lumMmxextFilterCode);

  2296.     av_free(c->chrMmxextFilterCode);

  2297. #endif

  2298.     c->lumMmxextFilterCode = NULL;

  2299.     c->chrMmxextFilterCode = NULL;

  2300. #endif /* HAVE_MMX_INLINE */

  2301. 

  2302.     av_freep(&c->yuvTable);

  2303.     av_freep(&c->formatConvBuffer);

  2304. 

  2305.     sws_freeContext(c->cascaded_context[0]);

  2306.     sws_freeContext(c->cascaded_context[1]);

  2307.     sws_freeContext(c->cascaded_context[2]);

  2308.     memset(c->cascaded_context, 0, sizeof(c->cascaded_context));

  2309.     av_freep(&c->cascaded_tmp[0]);

  2310.     av_freep(&c->cascaded1_tmp[0]);

  2311. 

  2312.     av_freep(&c->gamma);

  2313.     av_freep(&c->inv_gamma);

  2314. 

  2315.     ff_free_filters(c);

  2316. 

  2317.     av_free(c);

  2318. }

  2319. 

  2320. struct SwsContext *sws_getCachedContext(struct SwsContext *context, int srcW,

  2321.                                         int srcH, enum AVPixelFormat srcFormat,

  2322.                                         int dstW, int dstH,

  2323.                                         enum AVPixelFormat dstFormat, int flags,

  2324.                                         SwsFilter *srcFilter,

  2325.                                         SwsFilter *dstFilter,

  2326.                                         const double *param)

  2327. {

  2328.     static const double default_param[2] = { SWS_PARAM_DEFAULT,

  2329.                                              SWS_PARAM_DEFAULT };

  2330.     int64_t src_h_chr_pos = -513, dst_h_chr_pos = -513,

  2331.             src_v_chr_pos = -513, dst_v_chr_pos = -513;

  2332. 

  2333.     if (!param)

  2334.         param = default_param;

  2335. 

  2336.     if (context &&

  2337.         (context->srcW      != srcW      ||

  2338.          context->srcH      != srcH      ||

  2339.          context->srcFormat != srcFormat ||

  2340.          context->dstW      != dstW      ||

  2341.          context->dstH      != dstH      ||

  2342.          context->dstFormat != dstFormat ||

  2343.          context->flags     != flags     ||

  2344.          context->param[0]  != param[0]  ||

  2345.          context->param[1]  != param[1])) {

  2346. 

  2347.         av_opt_get_int(context, "src_h_chr_pos", 0, &src_h_chr_pos);

  2348.         av_opt_get_int(context, "src_v_chr_pos", 0, &src_v_chr_pos);

  2349.         av_opt_get_int(context, "dst_h_chr_pos", 0, &dst_h_chr_pos);

  2350.         av_opt_get_int(context, "dst_v_chr_pos", 0, &dst_v_chr_pos);

  2351.         sws_freeContext(context);

  2352.         context = NULL;

  2353.     }

  2354. 

  2355.     if (!context) {

  2356.         if (!(context = sws_alloc_context()))

  2357.             return NULL;

  2358.         context->srcW      = srcW;

  2359.         context->srcH      = srcH;

  2360.         context->srcFormat = srcFormat;

  2361.         context->dstW      = dstW;

  2362.         context->dstH      = dstH;

  2363.         context->dstFormat = dstFormat;

  2364.         context->flags     = flags;

  2365.         context->param[0]  = param[0];

  2366.         context->param[1]  = param[1];

  2367. 

  2368.         av_opt_set_int(context, "src_h_chr_pos", src_h_chr_pos, 0);

  2369.         av_opt_set_int(context, "src_v_chr_pos", src_v_chr_pos, 0);

  2370.         av_opt_set_int(context, "dst_h_chr_pos", dst_h_chr_pos, 0);

  2371.         av_opt_set_int(context, "dst_v_chr_pos", dst_v_chr_pos, 0);

  2372. 

  2373.         if (sws_init_context(context, srcFilter, dstFilter) < 0) {

  2374.             sws_freeContext(context);

  2375.             return NULL;

  2376.         }

  2377.     }

  2378.     return context;

  2379. }