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.
80254 Commits
32 Branches
759MB
Tree: 218bb8b3f3
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '218bb8b3f3'
${ noResults }
FFmpeg/libswscale/utils.c

2388 lines
84KB
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/libm.h"

  49. #include "libavutil/mathematics.h"

  50. #include "libavutil/opt.h"

  51. #include "libavutil/pixdesc.h"

  52. #include "libavutil/aarch64/cpu.h"

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

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

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

  56. 

  57. // We have to implement deprecated functions until they are removed, this is the

  58. // simplest way to prevent warnings

  59. #undef attribute_deprecated

  60. #define attribute_deprecated

  61. 

  62. #include "rgb2rgb.h"

  63. #include "swscale.h"

  64. #include "swscale_internal.h"

  65. 

  66. #if !FF_API_SWS_VECTOR

  67. static SwsVector *sws_getIdentityVec(void);

  68. static void sws_addVec(SwsVector *a, SwsVector *b);

  69. static void sws_shiftVec(SwsVector *a, int shift);

  70. static void sws_printVec2(SwsVector *a, AVClass *log_ctx, int log_level);

  71. #endif

  72. 

  73. static void handle_formats(SwsContext *c);

  74. 

  75. unsigned swscale_version(void)

  76. {

  77.     av_assert0(LIBSWSCALE_VERSION_MICRO >= 100);

  78.     return LIBSWSCALE_VERSION_INT;

  79. }

  80. 

  81. const char *swscale_configuration(void)

  82. {

  83.     return FFMPEG_CONFIGURATION;

  84. }

  85. 

  86. const char *swscale_license(void)

  87. {

  88. #define LICENSE_PREFIX "libswscale license: "

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

  90. }

  91. 

  92. typedef struct FormatEntry {

  93.     uint8_t is_supported_in         :1;

  94.     uint8_t is_supported_out        :1;

  95.     uint8_t is_supported_endianness :1;

  96. } FormatEntry;

  97. 

  98. static const FormatEntry format_entries[AV_PIX_FMT_NB] = {

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

  223.     [AV_PIX_FMT_GBRAP12LE]   = { 1, 0 },

  224.     [AV_PIX_FMT_GBRAP12BE]   = { 1, 0 },

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

  247.     [AV_PIX_FMT_P010LE]      = { 1, 0 },

  248.     [AV_PIX_FMT_P010BE]      = { 1, 0 },

  249. };

  250. 

  251. int sws_isSupportedInput(enum AVPixelFormat pix_fmt)

  252. {

  253.     return (unsigned)pix_fmt < AV_PIX_FMT_NB ?

  254.            format_entries[pix_fmt].is_supported_in : 0;

  255. }

  256. 

  257. int sws_isSupportedOutput(enum AVPixelFormat pix_fmt)

  258. {

  259.     return (unsigned)pix_fmt < AV_PIX_FMT_NB ?

  260.            format_entries[pix_fmt].is_supported_out : 0;

  261. }

  262. 

  263. int sws_isSupportedEndiannessConversion(enum AVPixelFormat pix_fmt)

  264. {

  265.     return (unsigned)pix_fmt < AV_PIX_FMT_NB ?

  266.            format_entries[pix_fmt].is_supported_endianness : 0;

  267. }

  268. 

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

  270.                              double dist)

  271. {

  272.     if (dist <= 1.0)

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

  274.     else

  275.         return getSplineCoeff(0.0,

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

  277.                                c + 3.0 * d,

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

  279.                               dist - 1.0);

  280. }

  281. 

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

  283. {

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

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

  286.     }

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

  288.     return pos >> chr_subsample;

  289. }

  290. 

  291. typedef struct {

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

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

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

  295. } ScaleAlgorithm;

  296. 

  297. static const ScaleAlgorithm scale_algorithms[] = {

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

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

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

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

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

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

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

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

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

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

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

  309. };

  310. 

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

  312.                               int *outFilterSize, int xInc, int srcW,

  313.                               int dstW, int filterAlign, int one,

  314.                               int flags, int cpu_flags,

  315.                               SwsVector *srcFilter, SwsVector *dstFilter,

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

  317. {

  318.     int i;

  319.     int filterSize;

  320.     int filter2Size;

  321.     int minFilterSize;

  322.     int64_t *filter    = NULL;

  323.     int64_t *filter2   = NULL;

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

  325.     int ret            = -1;

  326. 

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

  328. 

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

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

  331. 

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

  333.         int i;

  334.         filterSize = 1;

  335.         FF_ALLOCZ_ARRAY_OR_GOTO(NULL, filter,

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

  337. 

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

  339.             filter[i * filterSize] = fone;

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

  341.         }

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

  343.         int i;

  344.         int64_t xDstInSrc;

  345.         filterSize = 1;

  346.         FF_ALLOC_ARRAY_OR_GOTO(NULL, filter,

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

  348. 

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

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

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

  352. 

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

  354.             filter[i]       = fone;

  355.             xDstInSrc      += xInc;

  356.         }

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

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

  359.         int i;

  360.         int64_t xDstInSrc;

  361.         filterSize = 2;

  362.         FF_ALLOC_ARRAY_OR_GOTO(NULL, filter,

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

  364. 

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

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

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

  368.             int j;

  369. 

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

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

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

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

  374.                 if (coeff < 0)

  375.                     coeff = 0;

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

  377.                 xx++;

  378.             }

  379.             xDstInSrc += xInc;

  380.         }

  381.     } else {

  382.         int64_t xDstInSrc;

  383.         int sizeFactor = -1;

  384. 

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

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

  387.                 sizeFactor = scale_algorithms[i].size_factor;

  388.                 break;

  389.             }

  390.         }

  391.         if (flags & SWS_LANCZOS)

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

  393.         av_assert0(sizeFactor > 0);

  394. 

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

  396.             filterSize = 1 + sizeFactor;    // upscale

  397.         else

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

  399. 

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

  401.         filterSize = FFMAX(filterSize, 1);

  402. 

  403.         FF_ALLOC_ARRAY_OR_GOTO(NULL, filter,

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

  405. 

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

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

  408.             int xx = (xDstInSrc - (filterSize - 2) * (1LL<<16)) / (1 << 17);

  409.             int j;

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

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

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

  413.                 double floatd;

  414.                 int64_t coeff;

  415. 

  416.                 if (xInc > 1 << 16)

  417.                     d = d * dstW / srcW;

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

  419. 

  420.                 if (flags & SWS_BICUBIC) {

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

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

  423. 

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

  425.                         coeff = 0.0;

  426.                     } else {

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

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

  429. 

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

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

  432.                                     (-18 * (1 << 24) + 12 * B + 6 * C) *  dd +

  433.                                       (6 * (1 << 24) -  2 * B)         * (1 << 30);

  434.                         else

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

  436.                                       (6 * B + 30 * C) * dd  +

  437.                                     (-12 * B - 48 * C) * d   +

  438.                                       (8 * B + 24 * C) * (1 << 30);

  439.                     }

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

  441.                 }

  442. #if 0

  443.                 else if (flags & SWS_X) {

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

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

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

  447.                 }

  448. #endif

  449.                 else if (flags & SWS_X) {

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

  451.                     double c;

  452. 

  453.                     if (floatd < 1.0)

  454.                         c = cos(floatd * M_PI);

  455.                     else

  456.                         c = -1.0;

  457.                     if (c < 0.0)

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

  459.                     else

  460.                         c = pow(c, A);

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

  462.                 } else if (flags & SWS_AREA) {

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

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

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

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

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

  468.                     else

  469.                         coeff = 0.0;

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

  471.                 } else if (flags & SWS_GAUSS) {

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

  473.                     coeff = exp2(-p * floatd * floatd) * fone;

  474.                 } else if (flags & SWS_SINC) {

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

  476.                 } else if (flags & SWS_LANCZOS) {

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

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

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

  480.                     if (floatd > p)

  481.                         coeff = 0;

  482.                 } else if (flags & SWS_BILINEAR) {

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

  484.                     if (coeff < 0)

  485.                         coeff = 0;

  486.                     coeff *= fone >> 30;

  487.                 } else if (flags & SWS_SPLINE) {

  488.                     double p = -2.196152422706632;

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

  490.                 } else {

  491.                     av_assert0(0);

  492.                 }

  493. 

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

  495.                 xx++;

  496.             }

  497.             xDstInSrc += 2 * xInc;

  498.         }

  499.     }

  500. 

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

  502.      * av_free(filter);

  503.      */

  504.     av_assert0(filterSize > 0);

  505.     filter2Size = filterSize;

  506.     if (srcFilter)

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

  508.     if (dstFilter)

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

  510.     av_assert0(filter2Size > 0);

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

  512. 

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

  514.         int j, k;

  515. 

  516.         if (srcFilter) {

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

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

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

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

  521.             }

  522.         } else {

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

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

  525.         }

  526.         // FIXME dstFilter

  527. 

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

  529.     }

  530.     av_freep(&filter);

  531. 

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

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

  534.     minFilterSize = 0;

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

  536.         int min = filter2Size;

  537.         int j;

  538.         int64_t cutOff = 0.0;

  539. 

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

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

  542.             int k;

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

  544. 

  545.             if (cutOff > SWS_MAX_REDUCE_CUTOFF * fone)

  546.                 break;

  547. 

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

  549.              * filter otherwise */

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

  551.                 break;

  552. 

  553.             // move filter coefficients left

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

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

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

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

  558.         }

  559. 

  560.         cutOff = 0;

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

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

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

  564. 

  565.             if (cutOff > SWS_MAX_REDUCE_CUTOFF * fone)

  566.                 break;

  567.             min--;

  568.         }

  569. 

  570.         if (min > minFilterSize)

  571.             minFilterSize = min;

  572.     }

  573. 

  574.     if (PPC_ALTIVEC(cpu_flags)) {

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

  576.         if (minFilterSize < 5)

  577.             filterAlign = 4;

  578. 

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

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

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

  582.         if (minFilterSize < 3)

  583.             filterAlign = 1;

  584.     }

  585. 

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

  587.         // special case for unscaled vertical filtering

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

  589.             filterAlign = 1;

  590.     }

  591. 

  592.     av_assert0(minFilterSize > 0);

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

  594.     av_assert0(filterSize > 0);

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

  596.     if (!filter)

  597.         goto fail;

  598.     if (filterSize >= MAX_FILTER_SIZE * 16 /

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

  600.         ret = RETCODE_USE_CASCADE;

  601.         goto fail;

  602.     }

  603.     *outFilterSize = filterSize;

  604. 

  605.     if (flags & SWS_PRINT_INFO)

  606.         av_log(NULL, AV_LOG_VERBOSE,

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

  608.                filter2Size, filterSize);

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

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

  611.         int j;

  612. 

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

  614.             if (j >= filter2Size)

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

  616.             else

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

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

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

  620.         }

  621.     }

  622. 

  623.     // FIXME try to align filterPos if possible

  624. 

  625.     // fix borders

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

  627.         int j;

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

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

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

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

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

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

  634.             }

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

  636.         }

  637. 

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

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

  640.             int64_t acc = 0;

  641. 

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

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

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

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

  646.                 }

  647.             }

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

  649.                 if (j < shift) {

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

  651.                 } else {

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

  653.                 }

  654.             }

  655. 

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

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

  658.         }

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

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

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

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

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

  664.             }

  665.         }

  666.     }

  667. 

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

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

  670.     FF_ALLOCZ_ARRAY_OR_GOTO(NULL, *outFilter,

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

  672. 

  673.     /* normalize & store in outFilter */

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

  675.         int j;

  676.         int64_t error = 0;

  677.         int64_t sum   = 0;

  678. 

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

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

  681.         }

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

  683.         if (!sum) {

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

  685.             sum = 1;

  686.         }

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

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

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

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

  691.             error                                  = v - intV * sum;

  692.         }

  693.     }

  694. 

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

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

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

  698.                                                       * read over the end */

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

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

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

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

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

  704.     }

  705. 

  706.     ret = 0;

  707. 

  708. fail:

  709.     if(ret < 0)

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

  711.     av_free(filter);

  712.     av_free(filter2);

  713.     return ret;

  714. }

  715. 

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

  717. {

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

  719.     int64_t vr =  table[0];

  720.     int64_t ub =  table[1];

  721.     int64_t ug = -table[2];

  722.     int64_t vg = -table[3];

  723.     int64_t ONE = 65536;

  724.     int64_t cy = ONE;

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

  726.     int i;

  727.     static const int8_t map[] = {

  728.     BY_IDX, GY_IDX, -1    , BY_IDX, BY_IDX, GY_IDX, -1    , BY_IDX,

  729.     RY_IDX, -1    , GY_IDX, RY_IDX, RY_IDX, -1    , GY_IDX, RY_IDX,

  730.     RY_IDX, GY_IDX, -1    , RY_IDX, RY_IDX, GY_IDX, -1    , RY_IDX,

  731.     BY_IDX, -1    , GY_IDX, BY_IDX, BY_IDX, -1    , GY_IDX, BY_IDX,

  732.     BU_IDX, GU_IDX, -1    , BU_IDX, BU_IDX, GU_IDX, -1    , BU_IDX,

  733.     RU_IDX, -1    , GU_IDX, RU_IDX, RU_IDX, -1    , GU_IDX, RU_IDX,

  734.     RU_IDX, GU_IDX, -1    , RU_IDX, RU_IDX, GU_IDX, -1    , RU_IDX,

  735.     BU_IDX, -1    , GU_IDX, BU_IDX, BU_IDX, -1    , GU_IDX, BU_IDX,

  736.     BV_IDX, GV_IDX, -1    , BV_IDX, BV_IDX, GV_IDX, -1    , BV_IDX,

  737.     RV_IDX, -1    , GV_IDX, RV_IDX, RV_IDX, -1    , GV_IDX, RV_IDX,

  738.     RV_IDX, GV_IDX, -1    , RV_IDX, RV_IDX, GV_IDX, -1    , RV_IDX,

  739.     BV_IDX, -1    , GV_IDX, BV_IDX, BV_IDX, -1    , GV_IDX, BV_IDX,

  740.     RY_IDX, BY_IDX, RY_IDX, BY_IDX, RY_IDX, BY_IDX, RY_IDX, BY_IDX,

  741.     BY_IDX, RY_IDX, BY_IDX, RY_IDX, BY_IDX, RY_IDX, BY_IDX, RY_IDX,

  742.     GY_IDX, -1    , GY_IDX, -1    , GY_IDX, -1    , GY_IDX, -1    ,

  743.     -1    , GY_IDX, -1    , GY_IDX, -1    , GY_IDX, -1    , GY_IDX,

  744.     RU_IDX, BU_IDX, RU_IDX, BU_IDX, RU_IDX, BU_IDX, RU_IDX, BU_IDX,

  745.     BU_IDX, RU_IDX, BU_IDX, RU_IDX, BU_IDX, RU_IDX, BU_IDX, RU_IDX,

  746.     GU_IDX, -1    , GU_IDX, -1    , GU_IDX, -1    , GU_IDX, -1    ,

  747.     -1    , GU_IDX, -1    , GU_IDX, -1    , GU_IDX, -1    , GU_IDX,

  748.     RV_IDX, BV_IDX, RV_IDX, BV_IDX, RV_IDX, BV_IDX, RV_IDX, BV_IDX,

  749.     BV_IDX, RV_IDX, BV_IDX, RV_IDX, BV_IDX, RV_IDX, BV_IDX, RV_IDX,

  750.     GV_IDX, -1    , GV_IDX, -1    , GV_IDX, -1    , GV_IDX, -1    ,

  751.     -1    , GV_IDX, -1    , GV_IDX, -1    , GV_IDX, -1    , GV_IDX, //23

  752.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //24

  753.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //25

  754.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //26

  755.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //27

  756.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //28

  757.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //29

  758.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //30

  759.     -1    , -1    , -1    , -1    , -1    , -1    , -1    , -1    , //31

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

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

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

  763.     };

  764. 

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

  766. 

  767.     if (!dstRange) {

  768.         cy = cy * 255 / 219;

  769.     } else {

  770.         vr = vr * 224 / 255;

  771.         ub = ub * 224 / 255;

  772.         ug = ug * 224 / 255;

  773.         vg = vg * 224 / 255;

  774.     }

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

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

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

  778. 

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

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

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

  782. 

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

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

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

  786. 

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

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

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

  790. 

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

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

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

  794. 

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

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

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

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

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

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

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

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

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

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

  805.     }

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

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

  808. }

  809. 

  810. static void fill_xyztables(struct SwsContext *c)

  811. {

  812.     int i;

  813.     double xyzgamma = XYZ_GAMMA;

  814.     double rgbgamma = 1.0 / RGB_GAMMA;

  815.     double xyzgammainv = 1.0 / XYZ_GAMMA;

  816.     double rgbgammainv = RGB_GAMMA;

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

  818.         {13270, -6295, -2041},

  819.         {-3969,  7682,   170},

  820.         {  228,  -835,  4329} };

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

  822.         {1689, 1464,  739},

  823.         { 871, 2929,  296},

  824.         {  79,  488, 3891} };

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

  826. 

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

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

  829.     c->xyzgamma = xyzgamma_tab;

  830.     c->rgbgamma = rgbgamma_tab;

  831.     c->xyzgammainv = xyzgammainv_tab;

  832.     c->rgbgammainv = rgbgammainv_tab;

  833. 

  834.     if (rgbgamma_tab[4095])

  835.         return;

  836. 

  837.     /* set gamma vectors */

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

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

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

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

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

  843.     }

  844. }

  845. 

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

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

  848.                              int brightness, int contrast, int saturation)

  849. {

  850.     const AVPixFmtDescriptor *desc_dst;

  851.     const AVPixFmtDescriptor *desc_src;

  852.     int need_reinit = 0;

  853. 

  854.     handle_formats(c);

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

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

  857. 

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

  859.         dstRange = 0;

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

  861.         srcRange = 0;

  862. 

  863.     if (c->srcRange != srcRange ||

  864.         c->dstRange != dstRange ||

  865.         c->brightness != brightness ||

  866.         c->contrast   != contrast ||

  867.         c->saturation != saturation ||

  868.         memcmp(c->srcColorspaceTable, inv_table, sizeof(int) * 4) ||

  869.         memcmp(c->dstColorspaceTable,     table, sizeof(int) * 4)

  870.     )

  871.         need_reinit = 1;

  872. 

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

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

  875. 

  876. 

  877. 

  878.     c->brightness = brightness;

  879.     c->contrast   = contrast;

  880.     c->saturation = saturation;

  881.     c->srcRange   = srcRange;

  882.     c->dstRange   = dstRange;

  883. 

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

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

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

  887.         ff_sws_init_range_convert(c);

  888. 

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

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

  891. 

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

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

  894. 

  895.     if (!need_reinit)

  896.         return 0;

  897. 

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

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

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

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

  902.             enum AVPixelFormat tmp_format;

  903.             int tmp_width, tmp_height;

  904.             int srcW = c->srcW;

  905.             int srcH = c->srcH;

  906.             int dstW = c->dstW;

  907.             int dstH = c->dstH;

  908.             int ret;

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

  910. 

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

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

  913.                     tmp_format = AV_PIX_FMT_BGRA64;

  914.                 } else {

  915.                     tmp_format = AV_PIX_FMT_BGR48;

  916.                 }

  917.             } else {

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

  919.                     tmp_format = AV_PIX_FMT_BGRA;

  920.                 } else {

  921.                     tmp_format = AV_PIX_FMT_BGR24;

  922.                 }

  923.             }

  924. 

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

  926.                 tmp_width  = dstW;

  927.                 tmp_height = dstH;

  928.             } else {

  929.                 tmp_width  = srcW;

  930.                 tmp_height = srcH;

  931.             }

  932. 

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

  934.                                 tmp_width, tmp_height, tmp_format, 64);

  935.             if (ret < 0)

  936.                 return ret;

  937. 

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

  939.                                                         tmp_width, tmp_height, tmp_format,

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

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

  942.                 return -1;

  943. 

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

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

  946.             if (ret < 0)

  947.                 return ret;

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

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

  950.                                      srcRange, table, dstRange,

  951.                                      brightness, contrast, saturation);

  952. 

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

  954.                                                     dstW, dstH, c->dstFormat,

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

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

  957.                 return -1;

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

  959.                                      srcRange, table, dstRange,

  960.                                      0, 1 << 16, 1 << 16);

  961.             return 0;

  962.         }

  963.         return -1;

  964.     }

  965. 

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

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

  968.                                  contrast, saturation);

  969.         // FIXME factorize

  970. 

  971.         if (ARCH_PPC)

  972.             ff_yuv2rgb_init_tables_ppc(c, inv_table, brightness,

  973.                                        contrast, saturation);

  974.     }

  975. 

  976.     fill_rgb2yuv_table(c, table, dstRange);

  977. 

  978.     return 0;

  979. }

  980. 

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

  982.                              int *srcRange, int **table, int *dstRange,

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

  984. {

  985.     if (!c )

  986.         return -1;

  987. 

  988.     *inv_table  = c->srcColorspaceTable;

  989.     *table      = c->dstColorspaceTable;

  990.     *srcRange   = c->srcRange;

  991.     *dstRange   = c->dstRange;

  992.     *brightness = c->brightness;

  993.     *contrast   = c->contrast;

  994.     *saturation = c->saturation;

  995. 

  996.     return 0;

  997. }

  998. 

  999. static int handle_jpeg(enum AVPixelFormat *format)

  1000. {

  1001.     switch (*format) {

  1002.     case AV_PIX_FMT_YUVJ420P:

  1003.         *format = AV_PIX_FMT_YUV420P;

  1004.         return 1;

  1005.     case AV_PIX_FMT_YUVJ411P:

  1006.         *format = AV_PIX_FMT_YUV411P;

  1007.         return 1;

  1008.     case AV_PIX_FMT_YUVJ422P:

  1009.         *format = AV_PIX_FMT_YUV422P;

  1010.         return 1;

  1011.     case AV_PIX_FMT_YUVJ444P:

  1012.         *format = AV_PIX_FMT_YUV444P;

  1013.         return 1;

  1014.     case AV_PIX_FMT_YUVJ440P:

  1015.         *format = AV_PIX_FMT_YUV440P;

  1016.         return 1;

  1017.     case AV_PIX_FMT_GRAY8:

  1018.     case AV_PIX_FMT_YA8:

  1019.     case AV_PIX_FMT_GRAY16LE:

  1020.     case AV_PIX_FMT_GRAY16BE:

  1021.     case AV_PIX_FMT_YA16BE:

  1022.     case AV_PIX_FMT_YA16LE:

  1023.         return 1;

  1024.     default:

  1025.         return 0;

  1026.     }

  1027. }

  1028. 

  1029. static int handle_0alpha(enum AVPixelFormat *format)

  1030. {

  1031.     switch (*format) {

  1032.     case AV_PIX_FMT_0BGR    : *format = AV_PIX_FMT_ABGR   ; return 1;

  1033.     case AV_PIX_FMT_BGR0    : *format = AV_PIX_FMT_BGRA   ; return 4;

  1034.     case AV_PIX_FMT_0RGB    : *format = AV_PIX_FMT_ARGB   ; return 1;

  1035.     case AV_PIX_FMT_RGB0    : *format = AV_PIX_FMT_RGBA   ; return 4;

  1036.     default:                                          return 0;

  1037.     }

  1038. }

  1039. 

  1040. static int handle_xyz(enum AVPixelFormat *format)

  1041. {

  1042.     switch (*format) {

  1043.     case AV_PIX_FMT_XYZ12BE : *format = AV_PIX_FMT_RGB48BE; return 1;

  1044.     case AV_PIX_FMT_XYZ12LE : *format = AV_PIX_FMT_RGB48LE; return 1;

  1045.     default:                                                return 0;

  1046.     }

  1047. }

  1048. 

  1049. static void handle_formats(SwsContext *c)

  1050. {

  1051.     c->src0Alpha |= handle_0alpha(&c->srcFormat);

  1052.     c->dst0Alpha |= handle_0alpha(&c->dstFormat);

  1053.     c->srcXYZ    |= handle_xyz(&c->srcFormat);

  1054.     c->dstXYZ    |= handle_xyz(&c->dstFormat);

  1055.     if (c->srcXYZ || c->dstXYZ)

  1056.         fill_xyztables(c);

  1057. }

  1058. 

  1059. SwsContext *sws_alloc_context(void)

  1060. {

  1061.     SwsContext *c = av_mallocz(sizeof(SwsContext));

  1062. 

  1063.     av_assert0(offsetof(SwsContext, redDither) + DITHER32_INT == offsetof(SwsContext, dither32));

  1064. 

  1065.     if (c) {

  1066.         c->av_class = &ff_sws_context_class;

  1067.         av_opt_set_defaults(c);

  1068.     }

  1069. 

  1070.     return c;

  1071. }

  1072. 

  1073. static uint16_t * alloc_gamma_tbl(double e)

  1074. {

  1075.     int i = 0;

  1076.     uint16_t * tbl;

  1077.     tbl = (uint16_t*)av_malloc(sizeof(uint16_t) * 1 << 16);

  1078.     if (!tbl)

  1079.         return NULL;

  1080. 

  1081.     for (i = 0; i < 65536; ++i) {

  1082.         tbl[i] = pow(i / 65535.0, e) * 65535.0;

  1083.     }

  1084.     return tbl;

  1085. }

  1086. 

  1087. static enum AVPixelFormat alphaless_fmt(enum AVPixelFormat fmt)

  1088. {

  1089.     switch(fmt) {

  1090.     case AV_PIX_FMT_ARGB:       return AV_PIX_FMT_RGB24;

  1091.     case AV_PIX_FMT_RGBA:       return AV_PIX_FMT_RGB24;

  1092.     case AV_PIX_FMT_ABGR:       return AV_PIX_FMT_BGR24;

  1093.     case AV_PIX_FMT_BGRA:       return AV_PIX_FMT_BGR24;

  1094.     case AV_PIX_FMT_YA8:        return AV_PIX_FMT_GRAY8;

  1095. 

  1096.     case AV_PIX_FMT_YUVA420P:   return AV_PIX_FMT_YUV420P;

  1097.     case AV_PIX_FMT_YUVA422P:   return AV_PIX_FMT_YUV422P;

  1098.     case AV_PIX_FMT_YUVA444P:           return AV_PIX_FMT_YUV444P;

  1099. 

  1100.     case AV_PIX_FMT_GBRAP:              return AV_PIX_FMT_GBRP;

  1101. 

  1102.     case AV_PIX_FMT_GBRAP12LE:          return AV_PIX_FMT_GBRP12;

  1103.     case AV_PIX_FMT_GBRAP12BE:          return AV_PIX_FMT_GBRP12;

  1104. 

  1105.     case AV_PIX_FMT_GBRAP16LE:          return AV_PIX_FMT_GBRP16;

  1106.     case AV_PIX_FMT_GBRAP16BE:          return AV_PIX_FMT_GBRP16;

  1107. 

  1108.     case AV_PIX_FMT_RGBA64LE:   return AV_PIX_FMT_RGB48;

  1109.     case AV_PIX_FMT_RGBA64BE:   return AV_PIX_FMT_RGB48;

  1110.     case AV_PIX_FMT_BGRA64LE:   return AV_PIX_FMT_BGR48;

  1111.     case AV_PIX_FMT_BGRA64BE:   return AV_PIX_FMT_BGR48;

  1112. 

  1113.     case AV_PIX_FMT_YA16BE:             return AV_PIX_FMT_GRAY16;

  1114.     case AV_PIX_FMT_YA16LE:             return AV_PIX_FMT_GRAY16;

  1115. 

  1116.     case AV_PIX_FMT_YUVA420P9BE:        return AV_PIX_FMT_YUV420P9;

  1117.     case AV_PIX_FMT_YUVA422P9BE:        return AV_PIX_FMT_YUV422P9;

  1118.     case AV_PIX_FMT_YUVA444P9BE:        return AV_PIX_FMT_YUV444P9;

  1119.     case AV_PIX_FMT_YUVA420P9LE:        return AV_PIX_FMT_YUV420P9;

  1120.     case AV_PIX_FMT_YUVA422P9LE:        return AV_PIX_FMT_YUV422P9;

  1121.     case AV_PIX_FMT_YUVA444P9LE:        return AV_PIX_FMT_YUV444P9;

  1122.     case AV_PIX_FMT_YUVA420P10BE:       return AV_PIX_FMT_YUV420P10;

  1123.     case AV_PIX_FMT_YUVA422P10BE:       return AV_PIX_FMT_YUV422P10;

  1124.     case AV_PIX_FMT_YUVA444P10BE:       return AV_PIX_FMT_YUV444P10;

  1125.     case AV_PIX_FMT_YUVA420P10LE:       return AV_PIX_FMT_YUV420P10;

  1126.     case AV_PIX_FMT_YUVA422P10LE:       return AV_PIX_FMT_YUV422P10;

  1127.     case AV_PIX_FMT_YUVA444P10LE:       return AV_PIX_FMT_YUV444P10;

  1128.     case AV_PIX_FMT_YUVA420P16BE:       return AV_PIX_FMT_YUV420P16;

  1129.     case AV_PIX_FMT_YUVA422P16BE:       return AV_PIX_FMT_YUV422P16;

  1130.     case AV_PIX_FMT_YUVA444P16BE:       return AV_PIX_FMT_YUV444P16;

  1131.     case AV_PIX_FMT_YUVA420P16LE:       return AV_PIX_FMT_YUV420P16;

  1132.     case AV_PIX_FMT_YUVA422P16LE:       return AV_PIX_FMT_YUV422P16;

  1133.     case AV_PIX_FMT_YUVA444P16LE:       return AV_PIX_FMT_YUV444P16;

  1134. 

  1135. //     case AV_PIX_FMT_AYUV64LE:

  1136. //     case AV_PIX_FMT_AYUV64BE:

  1137. //     case AV_PIX_FMT_PAL8:

  1138.     default: return AV_PIX_FMT_NONE;

  1139.     }

  1140. }

  1141. 

  1142. av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter,

  1143.                              SwsFilter *dstFilter)

  1144. {

  1145.     int i;

  1146.     int usesVFilter, usesHFilter;

  1147.     int unscaled;

  1148.     SwsFilter dummyFilter = { NULL, NULL, NULL, NULL };

  1149.     int srcW              = c->srcW;

  1150.     int srcH              = c->srcH;

  1151.     int dstW              = c->dstW;

  1152.     int dstH              = c->dstH;

  1153.     int dst_stride        = FFALIGN(dstW * sizeof(int16_t) + 66, 16);

  1154.     int flags, cpu_flags;

  1155.     enum AVPixelFormat srcFormat = c->srcFormat;

  1156.     enum AVPixelFormat dstFormat = c->dstFormat;

  1157.     const AVPixFmtDescriptor *desc_src;

  1158.     const AVPixFmtDescriptor *desc_dst;

  1159.     int ret = 0;

  1160.     enum AVPixelFormat tmpFmt;

  1161. 

  1162.     cpu_flags = av_get_cpu_flags();

  1163.     flags     = c->flags;

  1164.     emms_c();

  1165.     if (!rgb15to16)

  1166.         ff_sws_rgb2rgb_init();

  1167. 

  1168.     unscaled = (srcW == dstW && srcH == dstH);

  1169. 

  1170.     c->srcRange |= handle_jpeg(&c->srcFormat);

  1171.     c->dstRange |= handle_jpeg(&c->dstFormat);

  1172. 

  1173.     if(srcFormat!=c->srcFormat || dstFormat!=c->dstFormat)

  1174.         av_log(c, AV_LOG_WARNING, "deprecated pixel format used, make sure you did set range correctly\n");

  1175. 

  1176.     if (!c->contrast && !c->saturation && !c->dstFormatBpp)

  1177.         sws_setColorspaceDetails(c, ff_yuv2rgb_coeffs[SWS_CS_DEFAULT], c->srcRange,

  1178.                                  ff_yuv2rgb_coeffs[SWS_CS_DEFAULT],

  1179.                                  c->dstRange, 0, 1 << 16, 1 << 16);

  1180. 

  1181.     handle_formats(c);

  1182.     srcFormat = c->srcFormat;

  1183.     dstFormat = c->dstFormat;

  1184.     desc_src = av_pix_fmt_desc_get(srcFormat);

  1185.     desc_dst = av_pix_fmt_desc_get(dstFormat);

  1186. 

  1187.     // If the source has no alpha then disable alpha blendaway

  1188.     if (c->src0Alpha)

  1189.         c->alphablend = SWS_ALPHA_BLEND_NONE;

  1190. 

  1191.     if (!(unscaled && sws_isSupportedEndiannessConversion(srcFormat) &&

  1192.           av_pix_fmt_swap_endianness(srcFormat) == dstFormat)) {

  1193.     if (!sws_isSupportedInput(srcFormat)) {

  1194.         av_log(c, AV_LOG_ERROR, "%s is not supported as input pixel format\n",

  1195.                av_get_pix_fmt_name(srcFormat));

  1196.         return AVERROR(EINVAL);

  1197.     }

  1198.     if (!sws_isSupportedOutput(dstFormat)) {

  1199.         av_log(c, AV_LOG_ERROR, "%s is not supported as output pixel format\n",

  1200.                av_get_pix_fmt_name(dstFormat));

  1201.         return AVERROR(EINVAL);

  1202.     }

  1203.     }

  1204.     av_assert2(desc_src && desc_dst);

  1205. 

  1206.     i = flags & (SWS_POINT         |

  1207.                  SWS_AREA          |

  1208.                  SWS_BILINEAR      |

  1209.                  SWS_FAST_BILINEAR |

  1210.                  SWS_BICUBIC       |

  1211.                  SWS_X             |

  1212.                  SWS_GAUSS         |

  1213.                  SWS_LANCZOS       |

  1214.                  SWS_SINC          |

  1215.                  SWS_SPLINE        |

  1216.                  SWS_BICUBLIN);

  1217. 

  1218.     /* provide a default scaler if not set by caller */

  1219.     if (!i) {

  1220.         if (dstW < srcW && dstH < srcH)

  1221.             flags |= SWS_BICUBIC;

  1222.         else if (dstW > srcW && dstH > srcH)

  1223.             flags |= SWS_BICUBIC;

  1224.         else

  1225.             flags |= SWS_BICUBIC;

  1226.         c->flags = flags;

  1227.     } else if (i & (i - 1)) {

  1228.         av_log(c, AV_LOG_ERROR,

  1229.                "Exactly one scaler algorithm must be chosen, got %X\n", i);

  1230.         return AVERROR(EINVAL);

  1231.     }

  1232.     /* sanity check */

  1233.     if (srcW < 1 || srcH < 1 || dstW < 1 || dstH < 1) {

  1234.         /* FIXME check if these are enough and try to lower them after

  1235.          * fixing the relevant parts of the code */

  1236.         av_log(c, AV_LOG_ERROR, "%dx%d -> %dx%d is invalid scaling dimension\n",

  1237.                srcW, srcH, dstW, dstH);

  1238.         return AVERROR(EINVAL);

  1239.     }

  1240.     if (flags & SWS_FAST_BILINEAR) {

  1241.         if (srcW < 8 || dstW < 8) {

  1242.             flags ^= SWS_FAST_BILINEAR | SWS_BILINEAR;

  1243.             c->flags = flags;

  1244.         }

  1245.     }

  1246. 

  1247.     if (!dstFilter)

  1248.         dstFilter = &dummyFilter;

  1249.     if (!srcFilter)

  1250.         srcFilter = &dummyFilter;

  1251. 

  1252.     c->lumXInc      = (((int64_t)srcW << 16) + (dstW >> 1)) / dstW;

  1253.     c->lumYInc      = (((int64_t)srcH << 16) + (dstH >> 1)) / dstH;

  1254.     c->dstFormatBpp = av_get_bits_per_pixel(desc_dst);

  1255.     c->srcFormatBpp = av_get_bits_per_pixel(desc_src);

  1256.     c->vRounder     = 4 * 0x0001000100010001ULL;

  1257. 

  1258.     usesVFilter = (srcFilter->lumV && srcFilter->lumV->length > 1) ||

  1259.                   (srcFilter->chrV && srcFilter->chrV->length > 1) ||

  1260.                   (dstFilter->lumV && dstFilter->lumV->length > 1) ||

  1261.                   (dstFilter->chrV && dstFilter->chrV->length > 1);

  1262.     usesHFilter = (srcFilter->lumH && srcFilter->lumH->length > 1) ||

  1263.                   (srcFilter->chrH && srcFilter->chrH->length > 1) ||

  1264.                   (dstFilter->lumH && dstFilter->lumH->length > 1) ||

  1265.                   (dstFilter->chrH && dstFilter->chrH->length > 1);

  1266. 

  1267.     av_pix_fmt_get_chroma_sub_sample(srcFormat, &c->chrSrcHSubSample, &c->chrSrcVSubSample);

  1268.     av_pix_fmt_get_chroma_sub_sample(dstFormat, &c->chrDstHSubSample, &c->chrDstVSubSample);

  1269. 

  1270.     if (isAnyRGB(dstFormat) && !(flags&SWS_FULL_CHR_H_INT)) {

  1271.         if (dstW&1) {

  1272.             av_log(c, AV_LOG_DEBUG, "Forcing full internal H chroma due to odd output size\n");

  1273.             flags |= SWS_FULL_CHR_H_INT;

  1274.             c->flags = flags;

  1275.         }

  1276. 

  1277.         if (   c->chrSrcHSubSample == 0

  1278.             && c->chrSrcVSubSample == 0

  1279.             && c->dither != SWS_DITHER_BAYER //SWS_FULL_CHR_H_INT is currently not supported with SWS_DITHER_BAYER

  1280.             && !(c->flags & SWS_FAST_BILINEAR)

  1281.         ) {

  1282.             av_log(c, AV_LOG_DEBUG, "Forcing full internal H chroma due to input having non subsampled chroma\n");

  1283.             flags |= SWS_FULL_CHR_H_INT;

  1284.             c->flags = flags;

  1285.         }

  1286.     }

  1287. 

  1288.     if (c->dither == SWS_DITHER_AUTO) {

  1289.         if (flags & SWS_ERROR_DIFFUSION)

  1290.             c->dither = SWS_DITHER_ED;

  1291.     }

  1292. 

  1293.     if(dstFormat == AV_PIX_FMT_BGR4_BYTE ||

  1294.        dstFormat == AV_PIX_FMT_RGB4_BYTE ||

  1295.        dstFormat == AV_PIX_FMT_BGR8 ||

  1296.        dstFormat == AV_PIX_FMT_RGB8) {

  1297.         if (c->dither == SWS_DITHER_AUTO)

  1298.             c->dither = (flags & SWS_FULL_CHR_H_INT) ? SWS_DITHER_ED : SWS_DITHER_BAYER;

  1299.         if (!(flags & SWS_FULL_CHR_H_INT)) {

  1300.             if (c->dither == SWS_DITHER_ED || c->dither == SWS_DITHER_A_DITHER || c->dither == SWS_DITHER_X_DITHER) {

  1301.                 av_log(c, AV_LOG_DEBUG,

  1302.                     "Desired dithering only supported in full chroma interpolation for destination format '%s'\n",

  1303.                     av_get_pix_fmt_name(dstFormat));

  1304.                 flags   |= SWS_FULL_CHR_H_INT;

  1305.                 c->flags = flags;

  1306.             }

  1307.         }

  1308.         if (flags & SWS_FULL_CHR_H_INT) {

  1309.             if (c->dither == SWS_DITHER_BAYER) {

  1310.                 av_log(c, AV_LOG_DEBUG,

  1311.                     "Ordered dither is not supported in full chroma interpolation for destination format '%s'\n",

  1312.                     av_get_pix_fmt_name(dstFormat));

  1313.                 c->dither = SWS_DITHER_ED;

  1314.             }

  1315.         }

  1316.     }

  1317.     if (isPlanarRGB(dstFormat)) {

  1318.         if (!(flags & SWS_FULL_CHR_H_INT)) {

  1319.             av_log(c, AV_LOG_DEBUG,

  1320.                    "%s output is not supported with half chroma resolution, switching to full\n",

  1321.                    av_get_pix_fmt_name(dstFormat));

  1322.             flags   |= SWS_FULL_CHR_H_INT;

  1323.             c->flags = flags;

  1324.         }

  1325.     }

  1326. 

  1327.     /* reuse chroma for 2 pixels RGB/BGR unless user wants full

  1328.      * chroma interpolation */

  1329.     if (flags & SWS_FULL_CHR_H_INT &&

  1330.         isAnyRGB(dstFormat)        &&

  1331.         !isPlanarRGB(dstFormat)    &&

  1332.         dstFormat != AV_PIX_FMT_RGBA64LE &&

  1333.         dstFormat != AV_PIX_FMT_RGBA64BE &&

  1334.         dstFormat != AV_PIX_FMT_BGRA64LE &&

  1335.         dstFormat != AV_PIX_FMT_BGRA64BE &&

  1336.         dstFormat != AV_PIX_FMT_RGB48LE &&

  1337.         dstFormat != AV_PIX_FMT_RGB48BE &&

  1338.         dstFormat != AV_PIX_FMT_BGR48LE &&

  1339.         dstFormat != AV_PIX_FMT_BGR48BE &&

  1340.         dstFormat != AV_PIX_FMT_RGBA  &&

  1341.         dstFormat != AV_PIX_FMT_ARGB  &&

  1342.         dstFormat != AV_PIX_FMT_BGRA  &&

  1343.         dstFormat != AV_PIX_FMT_ABGR  &&

  1344.         dstFormat != AV_PIX_FMT_RGB24 &&

  1345.         dstFormat != AV_PIX_FMT_BGR24 &&

  1346.         dstFormat != AV_PIX_FMT_BGR4_BYTE &&

  1347.         dstFormat != AV_PIX_FMT_RGB4_BYTE &&

  1348.         dstFormat != AV_PIX_FMT_BGR8 &&

  1349.         dstFormat != AV_PIX_FMT_RGB8

  1350.     ) {

  1351.         av_log(c, AV_LOG_WARNING,

  1352.                "full chroma interpolation for destination format '%s' not yet implemented\n",

  1353.                av_get_pix_fmt_name(dstFormat));

  1354.         flags   &= ~SWS_FULL_CHR_H_INT;

  1355.         c->flags = flags;

  1356.     }

  1357.     if (isAnyRGB(dstFormat) && !(flags & SWS_FULL_CHR_H_INT))

  1358.         c->chrDstHSubSample = 1;

  1359. 

  1360.     // drop some chroma lines if the user wants it

  1361.     c->vChrDrop          = (flags & SWS_SRC_V_CHR_DROP_MASK) >>

  1362.                            SWS_SRC_V_CHR_DROP_SHIFT;

  1363.     c->chrSrcVSubSample += c->vChrDrop;

  1364. 

  1365.     /* drop every other pixel for chroma calculation unless user

  1366.      * wants full chroma */

  1367.     if (isAnyRGB(srcFormat) && !(flags & SWS_FULL_CHR_H_INP)   &&

  1368.         srcFormat != AV_PIX_FMT_RGB8 && srcFormat != AV_PIX_FMT_BGR8 &&

  1369.         srcFormat != AV_PIX_FMT_RGB4 && srcFormat != AV_PIX_FMT_BGR4 &&

  1370.         srcFormat != AV_PIX_FMT_RGB4_BYTE && srcFormat != AV_PIX_FMT_BGR4_BYTE &&

  1371.         srcFormat != AV_PIX_FMT_GBRP9BE   && srcFormat != AV_PIX_FMT_GBRP9LE  &&

  1372.         srcFormat != AV_PIX_FMT_GBRP10BE  && srcFormat != AV_PIX_FMT_GBRP10LE &&

  1373.         srcFormat != AV_PIX_FMT_GBRP12BE  && srcFormat != AV_PIX_FMT_GBRP12LE &&

  1374.         srcFormat != AV_PIX_FMT_GBRAP12BE && srcFormat != AV_PIX_FMT_GBRAP12LE &&

  1375.         srcFormat != AV_PIX_FMT_GBRP14BE  && srcFormat != AV_PIX_FMT_GBRP14LE &&

  1376.         srcFormat != AV_PIX_FMT_GBRP16BE  && srcFormat != AV_PIX_FMT_GBRP16LE &&

  1377.         srcFormat != AV_PIX_FMT_GBRAP16BE  && srcFormat != AV_PIX_FMT_GBRAP16LE &&

  1378.         ((dstW >> c->chrDstHSubSample) <= (srcW >> 1) ||

  1379.          (flags & SWS_FAST_BILINEAR)))

  1380.         c->chrSrcHSubSample = 1;

  1381. 

  1382.     // Note the AV_CEIL_RSHIFT is so that we always round toward +inf.

  1383.     c->chrSrcW = AV_CEIL_RSHIFT(srcW, c->chrSrcHSubSample);

  1384.     c->chrSrcH = AV_CEIL_RSHIFT(srcH, c->chrSrcVSubSample);

  1385.     c->chrDstW = AV_CEIL_RSHIFT(dstW, c->chrDstHSubSample);

  1386.     c->chrDstH = AV_CEIL_RSHIFT(dstH, c->chrDstVSubSample);

  1387. 

  1388.     FF_ALLOCZ_OR_GOTO(c, c->formatConvBuffer, FFALIGN(srcW*2+78, 16) * 2, fail);

  1389. 

  1390.     c->srcBpc = desc_src->comp[0].depth;

  1391.     if (c->srcBpc < 8)

  1392.         c->srcBpc = 8;

  1393.     c->dstBpc = desc_dst->comp[0].depth;

  1394.     if (c->dstBpc < 8)

  1395.         c->dstBpc = 8;

  1396.     if (isAnyRGB(srcFormat) || srcFormat == AV_PIX_FMT_PAL8)

  1397.         c->srcBpc = 16;

  1398.     if (c->dstBpc == 16)

  1399.         dst_stride <<= 1;

  1400. 

  1401.     if (INLINE_MMXEXT(cpu_flags) && c->srcBpc == 8 && c->dstBpc <= 14) {

  1402.         c->canMMXEXTBeUsed = dstW >= srcW && (dstW & 31) == 0 &&

  1403.                              c->chrDstW >= c->chrSrcW &&

  1404.                              (srcW & 15) == 0;

  1405.         if (!c->canMMXEXTBeUsed && dstW >= srcW && c->chrDstW >= c->chrSrcW && (srcW & 15) == 0

  1406. 

  1407.             && (flags & SWS_FAST_BILINEAR)) {

  1408.             if (flags & SWS_PRINT_INFO)

  1409.                 av_log(c, AV_LOG_INFO,

  1410.                        "output width is not a multiple of 32 -> no MMXEXT scaler\n");

  1411.         }

  1412.         if (usesHFilter || isNBPS(c->srcFormat) || is16BPS(c->srcFormat) || isAnyRGB(c->srcFormat))

  1413.             c->canMMXEXTBeUsed = 0;

  1414.     } else

  1415.         c->canMMXEXTBeUsed = 0;

  1416. 

  1417.     c->chrXInc = (((int64_t)c->chrSrcW << 16) + (c->chrDstW >> 1)) / c->chrDstW;

  1418.     c->chrYInc = (((int64_t)c->chrSrcH << 16) + (c->chrDstH >> 1)) / c->chrDstH;

  1419. 

  1420.     /* Match pixel 0 of the src to pixel 0 of dst and match pixel n-2 of src

  1421.      * to pixel n-2 of dst, but only for the FAST_BILINEAR mode otherwise do

  1422.      * correct scaling.

  1423.      * n-2 is the last chrominance sample available.

  1424.      * This is not perfect, but no one should notice the difference, the more

  1425.      * correct variant would be like the vertical one, but that would require

  1426.      * some special code for the first and last pixel */

  1427.     if (flags & SWS_FAST_BILINEAR) {

  1428.         if (c->canMMXEXTBeUsed) {

  1429.             c->lumXInc += 20;

  1430.             c->chrXInc += 20;

  1431.         }

  1432.         // we don't use the x86 asm scaler if MMX is available

  1433.         else if (INLINE_MMX(cpu_flags) && c->dstBpc <= 14) {

  1434.             c->lumXInc = ((int64_t)(srcW       - 2) << 16) / (dstW       - 2) - 20;

  1435.             c->chrXInc = ((int64_t)(c->chrSrcW - 2) << 16) / (c->chrDstW - 2) - 20;

  1436.         }

  1437.     }

  1438. 

  1439.     // hardcoded for now

  1440.     c->gamma_value = 2.2;

  1441.     tmpFmt = AV_PIX_FMT_RGBA64LE;

  1442. 

  1443. 

  1444.     if (!unscaled && c->gamma_flag && (srcFormat != tmpFmt || dstFormat != tmpFmt)) {

  1445.         SwsContext *c2;

  1446.         c->cascaded_context[0] = NULL;

  1447. 

  1448.         ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  1449.                             srcW, srcH, tmpFmt, 64);

  1450.         if (ret < 0)

  1451.             return ret;

  1452. 

  1453.         c->cascaded_context[0] = sws_getContext(srcW, srcH, srcFormat,

  1454.                                                 srcW, srcH, tmpFmt,

  1455.                                                 flags, NULL, NULL, c->param);

  1456.         if (!c->cascaded_context[0]) {

  1457.             return -1;

  1458.         }

  1459. 

  1460.         c->cascaded_context[1] = sws_getContext(srcW, srcH, tmpFmt,

  1461.                                                 dstW, dstH, tmpFmt,

  1462.                                                 flags, srcFilter, dstFilter, c->param);

  1463. 

  1464.         if (!c->cascaded_context[1])

  1465.             return -1;

  1466. 

  1467.         c2 = c->cascaded_context[1];

  1468.         c2->is_internal_gamma = 1;

  1469.         c2->gamma     = alloc_gamma_tbl(    c->gamma_value);

  1470.         c2->inv_gamma = alloc_gamma_tbl(1.f/c->gamma_value);

  1471.         if (!c2->gamma || !c2->inv_gamma)

  1472.             return AVERROR(ENOMEM);

  1473. 

  1474.         // is_internal_flag is set after creating the context

  1475.         // to properly create the gamma convert FilterDescriptor

  1476.         // we have to re-initialize it

  1477.         ff_free_filters(c2);

  1478.         if (ff_init_filters(c2) < 0) {

  1479.             sws_freeContext(c2);

  1480.             return -1;

  1481.         }

  1482. 

  1483.         c->cascaded_context[2] = NULL;

  1484.         if (dstFormat != tmpFmt) {

  1485.             ret = av_image_alloc(c->cascaded1_tmp, c->cascaded1_tmpStride,

  1486.                                 dstW, dstH, tmpFmt, 64);

  1487.             if (ret < 0)

  1488.                 return ret;

  1489. 

  1490.             c->cascaded_context[2] = sws_getContext(dstW, dstH, tmpFmt,

  1491.                                                 dstW, dstH, dstFormat,

  1492.                                                 flags, NULL, NULL, c->param);

  1493.             if (!c->cascaded_context[2])

  1494.                 return -1;

  1495.         }

  1496.         return 0;

  1497.     }

  1498. 

  1499.     if (isBayer(srcFormat)) {

  1500.         if (!unscaled ||

  1501.             (dstFormat != AV_PIX_FMT_RGB24 && dstFormat != AV_PIX_FMT_YUV420P)) {

  1502.             enum AVPixelFormat tmpFormat = AV_PIX_FMT_RGB24;

  1503. 

  1504.             ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  1505.                                 srcW, srcH, tmpFormat, 64);

  1506.             if (ret < 0)

  1507.                 return ret;

  1508. 

  1509.             c->cascaded_context[0] = sws_getContext(srcW, srcH, srcFormat,

  1510.                                                     srcW, srcH, tmpFormat,

  1511.                                                     flags, srcFilter, NULL, c->param);

  1512.             if (!c->cascaded_context[0])

  1513.                 return -1;

  1514. 

  1515.             c->cascaded_context[1] = sws_getContext(srcW, srcH, tmpFormat,

  1516.                                                     dstW, dstH, dstFormat,

  1517.                                                     flags, NULL, dstFilter, c->param);

  1518.             if (!c->cascaded_context[1])

  1519.                 return -1;

  1520.             return 0;

  1521.         }

  1522.     }

  1523. 

  1524.     if (CONFIG_SWSCALE_ALPHA && isALPHA(srcFormat) && !isALPHA(dstFormat)) {

  1525.         enum AVPixelFormat tmpFormat = alphaless_fmt(srcFormat);

  1526. 

  1527.         if (tmpFormat != AV_PIX_FMT_NONE && c->alphablend != SWS_ALPHA_BLEND_NONE)

  1528.         if (!unscaled ||

  1529.             dstFormat != tmpFormat ||

  1530.             usesHFilter || usesVFilter ||

  1531.             c->srcRange != c->dstRange

  1532.         ) {

  1533.             c->cascaded_mainindex = 1;

  1534.             ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  1535.                                 srcW, srcH, tmpFormat, 64);

  1536.             if (ret < 0)

  1537.                 return ret;

  1538. 

  1539.             c->cascaded_context[0] = sws_alloc_set_opts(srcW, srcH, srcFormat,

  1540.                                                         srcW, srcH, tmpFormat,

  1541.                                                         flags, c->param);

  1542.             if (!c->cascaded_context[0])

  1543.                 return -1;

  1544.             c->cascaded_context[0]->alphablend = c->alphablend;

  1545.             ret = sws_init_context(c->cascaded_context[0], NULL , NULL);

  1546.             if (ret < 0)

  1547.                 return ret;

  1548. 

  1549.             c->cascaded_context[1] = sws_alloc_set_opts(srcW, srcH, tmpFormat,

  1550.                                                         dstW, dstH, dstFormat,

  1551.                                                         flags, c->param);

  1552.             if (!c->cascaded_context[1])

  1553.                 return -1;

  1554. 

  1555.             c->cascaded_context[1]->srcRange = c->srcRange;

  1556.             c->cascaded_context[1]->dstRange = c->dstRange;

  1557.             ret = sws_init_context(c->cascaded_context[1], srcFilter , dstFilter);

  1558.             if (ret < 0)

  1559.                 return ret;

  1560. 

  1561.             return 0;

  1562.         }

  1563.     }

  1564. 

  1565. #define USE_MMAP (HAVE_MMAP && HAVE_MPROTECT && defined MAP_ANONYMOUS)

  1566. 

  1567.     /* precalculate horizontal scaler filter coefficients */

  1568.     {

  1569. #if HAVE_MMXEXT_INLINE

  1570. // can't downscale !!!

  1571.         if (c->canMMXEXTBeUsed && (flags & SWS_FAST_BILINEAR)) {

  1572.             c->lumMmxextFilterCodeSize = ff_init_hscaler_mmxext(dstW, c->lumXInc, NULL,

  1573.                                                              NULL, NULL, 8);

  1574.             c->chrMmxextFilterCodeSize = ff_init_hscaler_mmxext(c->chrDstW, c->chrXInc,

  1575.                                                              NULL, NULL, NULL, 4);

  1576. 

  1577. #if USE_MMAP

  1578.             c->lumMmxextFilterCode = mmap(NULL, c->lumMmxextFilterCodeSize,

  1579.                                           PROT_READ | PROT_WRITE,

  1580.                                           MAP_PRIVATE | MAP_ANONYMOUS,

  1581.                                           -1, 0);

  1582.             c->chrMmxextFilterCode = mmap(NULL, c->chrMmxextFilterCodeSize,

  1583.                                           PROT_READ | PROT_WRITE,

  1584.                                           MAP_PRIVATE | MAP_ANONYMOUS,

  1585.                                           -1, 0);

  1586. #elif HAVE_VIRTUALALLOC

  1587.             c->lumMmxextFilterCode = VirtualAlloc(NULL,

  1588.                                                   c->lumMmxextFilterCodeSize,

  1589.                                                   MEM_COMMIT,

  1590.                                                   PAGE_EXECUTE_READWRITE);

  1591.             c->chrMmxextFilterCode = VirtualAlloc(NULL,

  1592.                                                   c->chrMmxextFilterCodeSize,

  1593.                                                   MEM_COMMIT,

  1594.                                                   PAGE_EXECUTE_READWRITE);

  1595. #else

  1596.             c->lumMmxextFilterCode = av_malloc(c->lumMmxextFilterCodeSize);

  1597.             c->chrMmxextFilterCode = av_malloc(c->chrMmxextFilterCodeSize);

  1598. #endif

  1599. 

  1600. #ifdef MAP_ANONYMOUS

  1601.             if (c->lumMmxextFilterCode == MAP_FAILED || c->chrMmxextFilterCode == MAP_FAILED)

  1602. #else

  1603.             if (!c->lumMmxextFilterCode || !c->chrMmxextFilterCode)

  1604. #endif

  1605.             {

  1606.                 av_log(c, AV_LOG_ERROR, "Failed to allocate MMX2FilterCode\n");

  1607.                 return AVERROR(ENOMEM);

  1608.             }

  1609. 

  1610.             FF_ALLOCZ_OR_GOTO(c, c->hLumFilter,    (dstW           / 8 + 8) * sizeof(int16_t), fail);

  1611.             FF_ALLOCZ_OR_GOTO(c, c->hChrFilter,    (c->chrDstW     / 4 + 8) * sizeof(int16_t), fail);

  1612.             FF_ALLOCZ_OR_GOTO(c, c->hLumFilterPos, (dstW       / 2 / 8 + 8) * sizeof(int32_t), fail);

  1613.             FF_ALLOCZ_OR_GOTO(c, c->hChrFilterPos, (c->chrDstW / 2 / 4 + 8) * sizeof(int32_t), fail);

  1614. 

  1615.             ff_init_hscaler_mmxext(      dstW, c->lumXInc, c->lumMmxextFilterCode,

  1616.                                 c->hLumFilter, (uint32_t*)c->hLumFilterPos, 8);

  1617.             ff_init_hscaler_mmxext(c->chrDstW, c->chrXInc, c->chrMmxextFilterCode,

  1618.                                 c->hChrFilter, (uint32_t*)c->hChrFilterPos, 4);

  1619. 

  1620. #if USE_MMAP

  1621.             if (   mprotect(c->lumMmxextFilterCode, c->lumMmxextFilterCodeSize, PROT_EXEC | PROT_READ) == -1

  1622.                 || mprotect(c->chrMmxextFilterCode, c->chrMmxextFilterCodeSize, PROT_EXEC | PROT_READ) == -1) {

  1623.                 av_log(c, AV_LOG_ERROR, "mprotect failed, cannot use fast bilinear scaler\n");

  1624.                 goto fail;

  1625.             }

  1626. #endif

  1627.         } else

  1628. #endif /* HAVE_MMXEXT_INLINE */

  1629.         {

  1630.             const int filterAlign = X86_MMX(cpu_flags)     ? 4 :

  1631.                                     PPC_ALTIVEC(cpu_flags) ? 8 :

  1632.                                     have_neon(cpu_flags)   ? 8 : 1;

  1633. 

  1634.             if ((ret = initFilter(&c->hLumFilter, &c->hLumFilterPos,

  1635.                            &c->hLumFilterSize, c->lumXInc,

  1636.                            srcW, dstW, filterAlign, 1 << 14,

  1637.                            (flags & SWS_BICUBLIN) ? (flags | SWS_BICUBIC) : flags,

  1638.                            cpu_flags, srcFilter->lumH, dstFilter->lumH,

  1639.                            c->param,

  1640.                            get_local_pos(c, 0, 0, 0),

  1641.                            get_local_pos(c, 0, 0, 0))) < 0)

  1642.                 goto fail;

  1643.             if ((ret = initFilter(&c->hChrFilter, &c->hChrFilterPos,

  1644.                            &c->hChrFilterSize, c->chrXInc,

  1645.                            c->chrSrcW, c->chrDstW, filterAlign, 1 << 14,

  1646.                            (flags & SWS_BICUBLIN) ? (flags | SWS_BILINEAR) : flags,

  1647.                            cpu_flags, srcFilter->chrH, dstFilter->chrH,

  1648.                            c->param,

  1649.                            get_local_pos(c, c->chrSrcHSubSample, c->src_h_chr_pos, 0),

  1650.                            get_local_pos(c, c->chrDstHSubSample, c->dst_h_chr_pos, 0))) < 0)

  1651.                 goto fail;

  1652.         }

  1653.     } // initialize horizontal stuff

  1654. 

  1655.     /* precalculate vertical scaler filter coefficients */

  1656.     {

  1657.         const int filterAlign = X86_MMX(cpu_flags)     ? 2 :

  1658.                                 PPC_ALTIVEC(cpu_flags) ? 8 :

  1659.                                 have_neon(cpu_flags)   ? 2 : 1;

  1660. 

  1661.         if ((ret = initFilter(&c->vLumFilter, &c->vLumFilterPos, &c->vLumFilterSize,

  1662.                        c->lumYInc, srcH, dstH, filterAlign, (1 << 12),

  1663.                        (flags & SWS_BICUBLIN) ? (flags | SWS_BICUBIC) : flags,

  1664.                        cpu_flags, srcFilter->lumV, dstFilter->lumV,

  1665.                        c->param,

  1666.                        get_local_pos(c, 0, 0, 1),

  1667.                        get_local_pos(c, 0, 0, 1))) < 0)

  1668.             goto fail;

  1669.         if ((ret = initFilter(&c->vChrFilter, &c->vChrFilterPos, &c->vChrFilterSize,

  1670.                        c->chrYInc, c->chrSrcH, c->chrDstH,

  1671.                        filterAlign, (1 << 12),

  1672.                        (flags & SWS_BICUBLIN) ? (flags | SWS_BILINEAR) : flags,

  1673.                        cpu_flags, srcFilter->chrV, dstFilter->chrV,

  1674.                        c->param,

  1675.                        get_local_pos(c, c->chrSrcVSubSample, c->src_v_chr_pos, 1),

  1676.                        get_local_pos(c, c->chrDstVSubSample, c->dst_v_chr_pos, 1))) < 0)

  1677. 

  1678.             goto fail;

  1679. 

  1680. #if HAVE_ALTIVEC

  1681.         FF_ALLOC_OR_GOTO(c, c->vYCoeffsBank, sizeof(vector signed short) * c->vLumFilterSize * c->dstH,    fail);

  1682.         FF_ALLOC_OR_GOTO(c, c->vCCoeffsBank, sizeof(vector signed short) * c->vChrFilterSize * c->chrDstH, fail);

  1683. 

  1684.         for (i = 0; i < c->vLumFilterSize * c->dstH; i++) {

  1685.             int j;

  1686.             short *p = (short *)&c->vYCoeffsBank[i];

  1687.             for (j = 0; j < 8; j++)

  1688.                 p[j] = c->vLumFilter[i];

  1689.         }

  1690. 

  1691.         for (i = 0; i < c->vChrFilterSize * c->chrDstH; i++) {

  1692.             int j;

  1693.             short *p = (short *)&c->vCCoeffsBank[i];

  1694.             for (j = 0; j < 8; j++)

  1695.                 p[j] = c->vChrFilter[i];

  1696.         }

  1697. #endif

  1698.     }

  1699. 

  1700.     for (i = 0; i < 4; i++)

  1701.         FF_ALLOCZ_OR_GOTO(c, c->dither_error[i], (c->dstW+2) * sizeof(int), fail);

  1702. 

  1703.     c->needAlpha = (CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat) && isALPHA(c->dstFormat)) ? 1 : 0;

  1704. 

  1705.     // 64 / c->scalingBpp is the same as 16 / sizeof(scaling_intermediate)

  1706.     c->uv_off   = (dst_stride>>1) + 64 / (c->dstBpc &~ 7);

  1707.     c->uv_offx2 = dst_stride + 16;

  1708. 

  1709.     av_assert0(c->chrDstH <= dstH);

  1710. 

  1711.     if (flags & SWS_PRINT_INFO) {

  1712.         const char *scaler = NULL, *cpucaps;

  1713. 

  1714.         for (i = 0; i < FF_ARRAY_ELEMS(scale_algorithms); i++) {

  1715.             if (flags & scale_algorithms[i].flag) {

  1716.                 scaler = scale_algorithms[i].description;

  1717.                 break;

  1718.             }

  1719.         }

  1720.         if (!scaler)

  1721.             scaler =  "ehh flags invalid?!";

  1722.         av_log(c, AV_LOG_INFO, "%s scaler, from %s to %s%s ",

  1723.                scaler,

  1724.                av_get_pix_fmt_name(srcFormat),

  1725. #ifdef DITHER1XBPP

  1726.                dstFormat == AV_PIX_FMT_BGR555   || dstFormat == AV_PIX_FMT_BGR565   ||

  1727.                dstFormat == AV_PIX_FMT_RGB444BE || dstFormat == AV_PIX_FMT_RGB444LE ||

  1728.                dstFormat == AV_PIX_FMT_BGR444BE || dstFormat == AV_PIX_FMT_BGR444LE ?

  1729.                                                              "dithered " : "",

  1730. #else

  1731.                "",

  1732. #endif

  1733.                av_get_pix_fmt_name(dstFormat));

  1734. 

  1735.         if (INLINE_MMXEXT(cpu_flags))

  1736.             cpucaps = "MMXEXT";

  1737.         else if (INLINE_AMD3DNOW(cpu_flags))

  1738.             cpucaps = "3DNOW";

  1739.         else if (INLINE_MMX(cpu_flags))

  1740.             cpucaps = "MMX";

  1741.         else if (PPC_ALTIVEC(cpu_flags))

  1742.             cpucaps = "AltiVec";

  1743.         else

  1744.             cpucaps = "C";

  1745. 

  1746.         av_log(c, AV_LOG_INFO, "using %s\n", cpucaps);

  1747. 

  1748.         av_log(c, AV_LOG_VERBOSE, "%dx%d -> %dx%d\n", srcW, srcH, dstW, dstH);

  1749.         av_log(c, AV_LOG_DEBUG,

  1750.                "lum srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n",

  1751.                c->srcW, c->srcH, c->dstW, c->dstH, c->lumXInc, c->lumYInc);

  1752.         av_log(c, AV_LOG_DEBUG,

  1753.                "chr srcW=%d srcH=%d dstW=%d dstH=%d xInc=%d yInc=%d\n",

  1754.                c->chrSrcW, c->chrSrcH, c->chrDstW, c->chrDstH,

  1755.                c->chrXInc, c->chrYInc);

  1756.     }

  1757. 

  1758.     /* alpha blend special case, note this has been split via cascaded contexts if its scaled */

  1759.     if (unscaled && !usesHFilter && !usesVFilter &&

  1760.         c->alphablend != SWS_ALPHA_BLEND_NONE &&

  1761.         isALPHA(srcFormat) &&

  1762.         (c->srcRange == c->dstRange || isAnyRGB(dstFormat)) &&

  1763.         alphaless_fmt(srcFormat) == dstFormat

  1764.     ) {

  1765.         c->swscale = ff_sws_alphablendaway;

  1766. 

  1767.         if (flags & SWS_PRINT_INFO)

  1768.             av_log(c, AV_LOG_INFO,

  1769.                     "using alpha blendaway %s -> %s special converter\n",

  1770.                     av_get_pix_fmt_name(srcFormat), av_get_pix_fmt_name(dstFormat));

  1771.         return 0;

  1772.     }

  1773. 

  1774.     /* unscaled special cases */

  1775.     if (unscaled && !usesHFilter && !usesVFilter &&

  1776.         (c->srcRange == c->dstRange || isAnyRGB(dstFormat))) {

  1777.         ff_get_unscaled_swscale(c);

  1778. 

  1779.         if (c->swscale) {

  1780.             if (flags & SWS_PRINT_INFO)

  1781.                 av_log(c, AV_LOG_INFO,

  1782.                        "using unscaled %s -> %s special converter\n",

  1783.                        av_get_pix_fmt_name(srcFormat), av_get_pix_fmt_name(dstFormat));

  1784.             return 0;

  1785.         }

  1786.     }

  1787. 

  1788.     c->swscale = ff_getSwsFunc(c);

  1789.     return ff_init_filters(c);

  1790. fail: // FIXME replace things by appropriate error codes

  1791.     if (ret == RETCODE_USE_CASCADE)  {

  1792.         int tmpW = sqrt(srcW * (int64_t)dstW);

  1793.         int tmpH = sqrt(srcH * (int64_t)dstH);

  1794.         enum AVPixelFormat tmpFormat = AV_PIX_FMT_YUV420P;

  1795. 

  1796.         if (isALPHA(srcFormat))

  1797.             tmpFormat = AV_PIX_FMT_YUVA420P;

  1798. 

  1799.         if (srcW*(int64_t)srcH <= 4LL*dstW*dstH)

  1800.             return AVERROR(EINVAL);

  1801. 

  1802.         ret = av_image_alloc(c->cascaded_tmp, c->cascaded_tmpStride,

  1803.                              tmpW, tmpH, tmpFormat, 64);

  1804.         if (ret < 0)

  1805.             return ret;

  1806. 

  1807.         c->cascaded_context[0] = sws_getContext(srcW, srcH, srcFormat,

  1808.                                                 tmpW, tmpH, tmpFormat,

  1809.                                                 flags, srcFilter, NULL, c->param);

  1810.         if (!c->cascaded_context[0])

  1811.             return -1;

  1812. 

  1813.         c->cascaded_context[1] = sws_getContext(tmpW, tmpH, tmpFormat,

  1814.                                                 dstW, dstH, dstFormat,

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

  1816.         if (!c->cascaded_context[1])

  1817.             return -1;

  1818.         return 0;

  1819.     }

  1820.     return -1;

  1821. }

  1822. 

  1823. SwsContext *sws_alloc_set_opts(int srcW, int srcH, enum AVPixelFormat srcFormat,

  1824.                                int dstW, int dstH, enum AVPixelFormat dstFormat,

  1825.                                int flags, const double *param)

  1826. {

  1827.     SwsContext *c;

  1828. 

  1829.     if (!(c = sws_alloc_context()))

  1830.         return NULL;

  1831. 

  1832.     c->flags     = flags;

  1833.     c->srcW      = srcW;

  1834.     c->srcH      = srcH;

  1835.     c->dstW      = dstW;

  1836.     c->dstH      = dstH;

  1837.     c->srcFormat = srcFormat;

  1838.     c->dstFormat = dstFormat;

  1839. 

  1840.     if (param) {

  1841.         c->param[0] = param[0];

  1842.         c->param[1] = param[1];

  1843.     }

  1844. 

  1845.     return c;

  1846. }

  1847. 

  1848. SwsContext *sws_getContext(int srcW, int srcH, enum AVPixelFormat srcFormat,

  1849.                            int dstW, int dstH, enum AVPixelFormat dstFormat,

  1850.                            int flags, SwsFilter *srcFilter,

  1851.                            SwsFilter *dstFilter, const double *param)

  1852. {

  1853.     SwsContext *c;

  1854. 

  1855.     c = sws_alloc_set_opts(srcW, srcH, srcFormat,

  1856.                            dstW, dstH, dstFormat,

  1857.                            flags, param);

  1858.     if (!c)

  1859.         return NULL;

  1860. 

  1861.     if (sws_init_context(c, srcFilter, dstFilter) < 0) {

  1862.         sws_freeContext(c);

  1863.         return NULL;

  1864.     }

  1865. 

  1866.     return c;

  1867. }

  1868. 

  1869. static int isnan_vec(SwsVector *a)

  1870. {

  1871.     int i;

  1872.     for (i=0; i<a->length; i++)

  1873.         if (isnan(a->coeff[i]))

  1874.             return 1;

  1875.     return 0;

  1876. }

  1877. 

  1878. static void makenan_vec(SwsVector *a)

  1879. {

  1880.     int i;

  1881.     for (i=0; i<a->length; i++)

  1882.         a->coeff[i] = NAN;

  1883. }

  1884. 

  1885. SwsFilter *sws_getDefaultFilter(float lumaGBlur, float chromaGBlur,

  1886.                                 float lumaSharpen, float chromaSharpen,

  1887.                                 float chromaHShift, float chromaVShift,

  1888.                                 int verbose)

  1889. {

  1890.     SwsFilter *filter = av_malloc(sizeof(SwsFilter));

  1891.     if (!filter)

  1892.         return NULL;

  1893. 

  1894.     if (lumaGBlur != 0.0) {

  1895.         filter->lumH = sws_getGaussianVec(lumaGBlur, 3.0);

  1896.         filter->lumV = sws_getGaussianVec(lumaGBlur, 3.0);

  1897.     } else {

  1898.         filter->lumH = sws_getIdentityVec();

  1899.         filter->lumV = sws_getIdentityVec();

  1900.     }

  1901. 

  1902.     if (chromaGBlur != 0.0) {

  1903.         filter->chrH = sws_getGaussianVec(chromaGBlur, 3.0);

  1904.         filter->chrV = sws_getGaussianVec(chromaGBlur, 3.0);

  1905.     } else {

  1906.         filter->chrH = sws_getIdentityVec();

  1907.         filter->chrV = sws_getIdentityVec();

  1908.     }

  1909. 

  1910.     if (!filter->lumH || !filter->lumV || !filter->chrH || !filter->chrV)

  1911.         goto fail;

  1912. 

  1913.     if (chromaSharpen != 0.0) {

  1914.         SwsVector *id = sws_getIdentityVec();

  1915.         if (!id)

  1916.             goto fail;

  1917.         sws_scaleVec(filter->chrH, -chromaSharpen);

  1918.         sws_scaleVec(filter->chrV, -chromaSharpen);

  1919.         sws_addVec(filter->chrH, id);

  1920.         sws_addVec(filter->chrV, id);

  1921.         sws_freeVec(id);

  1922.     }

  1923. 

  1924.     if (lumaSharpen != 0.0) {

  1925.         SwsVector *id = sws_getIdentityVec();

  1926.         if (!id)

  1927.             goto fail;

  1928.         sws_scaleVec(filter->lumH, -lumaSharpen);

  1929.         sws_scaleVec(filter->lumV, -lumaSharpen);

  1930.         sws_addVec(filter->lumH, id);

  1931.         sws_addVec(filter->lumV, id);

  1932.         sws_freeVec(id);

  1933.     }

  1934. 

  1935.     if (chromaHShift != 0.0)

  1936.         sws_shiftVec(filter->chrH, (int)(chromaHShift + 0.5));

  1937. 

  1938.     if (chromaVShift != 0.0)

  1939.         sws_shiftVec(filter->chrV, (int)(chromaVShift + 0.5));

  1940. 

  1941.     sws_normalizeVec(filter->chrH, 1.0);

  1942.     sws_normalizeVec(filter->chrV, 1.0);

  1943.     sws_normalizeVec(filter->lumH, 1.0);

  1944.     sws_normalizeVec(filter->lumV, 1.0);

  1945. 

  1946.     if (isnan_vec(filter->chrH) ||

  1947.         isnan_vec(filter->chrV) ||

  1948.         isnan_vec(filter->lumH) ||

  1949.         isnan_vec(filter->lumV))

  1950.         goto fail;

  1951. 

  1952.     if (verbose)

  1953.         sws_printVec2(filter->chrH, NULL, AV_LOG_DEBUG);

  1954.     if (verbose)

  1955.         sws_printVec2(filter->lumH, NULL, AV_LOG_DEBUG);

  1956. 

  1957.     return filter;

  1958. 

  1959. fail:

  1960.     sws_freeVec(filter->lumH);

  1961.     sws_freeVec(filter->lumV);

  1962.     sws_freeVec(filter->chrH);

  1963.     sws_freeVec(filter->chrV);

  1964.     av_freep(&filter);

  1965.     return NULL;

  1966. }

  1967. 

  1968. SwsVector *sws_allocVec(int length)

  1969. {

  1970.     SwsVector *vec;

  1971. 

  1972.     if(length <= 0 || length > INT_MAX/ sizeof(double))

  1973.         return NULL;

  1974. 

  1975.     vec = av_malloc(sizeof(SwsVector));

  1976.     if (!vec)

  1977.         return NULL;

  1978.     vec->length = length;

  1979.     vec->coeff  = av_malloc(sizeof(double) * length);

  1980.     if (!vec->coeff)

  1981.         av_freep(&vec);

  1982.     return vec;

  1983. }

  1984. 

  1985. SwsVector *sws_getGaussianVec(double variance, double quality)

  1986. {

  1987.     const int length = (int)(variance * quality + 0.5) | 1;

  1988.     int i;

  1989.     double middle  = (length - 1) * 0.5;

  1990.     SwsVector *vec;

  1991. 

  1992.     if(variance < 0 || quality < 0)

  1993.         return NULL;

  1994. 

  1995.     vec = sws_allocVec(length);

  1996. 

  1997.     if (!vec)

  1998.         return NULL;

  1999. 

  2000.     for (i = 0; i < length; i++) {

  2001.         double dist = i - middle;

  2002.         vec->coeff[i] = exp(-dist * dist / (2 * variance * variance)) /

  2003.                         sqrt(2 * variance * M_PI);

  2004.     }

  2005. 

  2006.     sws_normalizeVec(vec, 1.0);

  2007. 

  2008.     return vec;

  2009. }

  2010. 

  2011. /**

  2012.  * Allocate and return a vector with length coefficients, all

  2013.  * with the same value c.

  2014.  */

  2015. #if !FF_API_SWS_VECTOR

  2016. static

  2017. #endif

  2018. SwsVector *sws_getConstVec(double c, int length)

  2019. {

  2020.     int i;

  2021.     SwsVector *vec = sws_allocVec(length);

  2022. 

  2023.     if (!vec)

  2024.         return NULL;

  2025. 

  2026.     for (i = 0; i < length; i++)

  2027.         vec->coeff[i] = c;

  2028. 

  2029.     return vec;

  2030. }

  2031. 

  2032. /**

  2033.  * Allocate and return a vector with just one coefficient, with

  2034.  * value 1.0.

  2035.  */

  2036. #if !FF_API_SWS_VECTOR

  2037. static

  2038. #endif

  2039. SwsVector *sws_getIdentityVec(void)

  2040. {

  2041.     return sws_getConstVec(1.0, 1);

  2042. }

  2043. 

  2044. static double sws_dcVec(SwsVector *a)

  2045. {

  2046.     int i;

  2047.     double sum = 0;

  2048. 

  2049.     for (i = 0; i < a->length; i++)

  2050.         sum += a->coeff[i];

  2051. 

  2052.     return sum;

  2053. }

  2054. 

  2055. void sws_scaleVec(SwsVector *a, double scalar)

  2056. {

  2057.     int i;

  2058. 

  2059.     for (i = 0; i < a->length; i++)

  2060.         a->coeff[i] *= scalar;

  2061. }

  2062. 

  2063. void sws_normalizeVec(SwsVector *a, double height)

  2064. {

  2065.     sws_scaleVec(a, height / sws_dcVec(a));

  2066. }

  2067. 

  2068. #if FF_API_SWS_VECTOR

  2069. static SwsVector *sws_getConvVec(SwsVector *a, SwsVector *b)

  2070. {

  2071.     int length = a->length + b->length - 1;

  2072.     int i, j;

  2073.     SwsVector *vec = sws_getConstVec(0.0, length);

  2074. 

  2075.     if (!vec)

  2076.         return NULL;

  2077. 

  2078.     for (i = 0; i < a->length; i++) {

  2079.         for (j = 0; j < b->length; j++) {

  2080.             vec->coeff[i + j] += a->coeff[i] * b->coeff[j];

  2081.         }

  2082.     }

  2083. 

  2084.     return vec;

  2085. }

  2086. #endif

  2087. 

  2088. static SwsVector *sws_sumVec(SwsVector *a, SwsVector *b)

  2089. {

  2090.     int length = FFMAX(a->length, b->length);

  2091.     int i;

  2092.     SwsVector *vec = sws_getConstVec(0.0, length);

  2093. 

  2094.     if (!vec)

  2095.         return NULL;

  2096. 

  2097.     for (i = 0; i < a->length; i++)

  2098.         vec->coeff[i + (length - 1) / 2 - (a->length - 1) / 2] += a->coeff[i];

  2099.     for (i = 0; i < b->length; i++)

  2100.         vec->coeff[i + (length - 1) / 2 - (b->length - 1) / 2] += b->coeff[i];

  2101. 

  2102.     return vec;

  2103. }

  2104. 

  2105. #if FF_API_SWS_VECTOR

  2106. static SwsVector *sws_diffVec(SwsVector *a, SwsVector *b)

  2107. {

  2108.     int length = FFMAX(a->length, b->length);

  2109.     int i;

  2110.     SwsVector *vec = sws_getConstVec(0.0, length);

  2111. 

  2112.     if (!vec)

  2113.         return NULL;

  2114. 

  2115.     for (i = 0; i < a->length; i++)

  2116.         vec->coeff[i + (length - 1) / 2 - (a->length - 1) / 2] += a->coeff[i];

  2117.     for (i = 0; i < b->length; i++)

  2118.         vec->coeff[i + (length - 1) / 2 - (b->length - 1) / 2] -= b->coeff[i];

  2119. 

  2120.     return vec;

  2121. }

  2122. #endif

  2123. 

  2124. /* shift left / or right if "shift" is negative */

  2125. static SwsVector *sws_getShiftedVec(SwsVector *a, int shift)

  2126. {

  2127.     int length = a->length + FFABS(shift) * 2;

  2128.     int i;

  2129.     SwsVector *vec = sws_getConstVec(0.0, length);

  2130. 

  2131.     if (!vec)

  2132.         return NULL;

  2133. 

  2134.     for (i = 0; i < a->length; i++) {

  2135.         vec->coeff[i + (length    - 1) / 2 -

  2136.                        (a->length - 1) / 2 - shift] = a->coeff[i];

  2137.     }

  2138. 

  2139.     return vec;

  2140. }

  2141. 

  2142. #if !FF_API_SWS_VECTOR

  2143. static

  2144. #endif

  2145. void sws_shiftVec(SwsVector *a, int shift)

  2146. {

  2147.     SwsVector *shifted = sws_getShiftedVec(a, shift);

  2148.     if (!shifted) {

  2149.         makenan_vec(a);

  2150.         return;

  2151.     }

  2152.     av_free(a->coeff);

  2153.     a->coeff  = shifted->coeff;

  2154.     a->length = shifted->length;

  2155.     av_free(shifted);

  2156. }

  2157. 

  2158. #if !FF_API_SWS_VECTOR

  2159. static

  2160. #endif

  2161. void sws_addVec(SwsVector *a, SwsVector *b)

  2162. {

  2163.     SwsVector *sum = sws_sumVec(a, b);

  2164.     if (!sum) {

  2165.         makenan_vec(a);

  2166.         return;

  2167.     }

  2168.     av_free(a->coeff);

  2169.     a->coeff  = sum->coeff;

  2170.     a->length = sum->length;

  2171.     av_free(sum);

  2172. }

  2173. 

  2174. #if FF_API_SWS_VECTOR

  2175. void sws_subVec(SwsVector *a, SwsVector *b)

  2176. {

  2177.     SwsVector *diff = sws_diffVec(a, b);

  2178.     if (!diff) {

  2179.         makenan_vec(a);

  2180.         return;

  2181.     }

  2182.     av_free(a->coeff);

  2183.     a->coeff  = diff->coeff;

  2184.     a->length = diff->length;

  2185.     av_free(diff);

  2186. }

  2187. 

  2188. void sws_convVec(SwsVector *a, SwsVector *b)

  2189. {

  2190.     SwsVector *conv = sws_getConvVec(a, b);

  2191.     if (!conv) {

  2192.         makenan_vec(a);

  2193.         return;

  2194.     }

  2195.     av_free(a->coeff);

  2196.     a->coeff  = conv->coeff;

  2197.     a->length = conv->length;

  2198.     av_free(conv);

  2199. }

  2200. 

  2201. SwsVector *sws_cloneVec(SwsVector *a)

  2202. {

  2203.     SwsVector *vec = sws_allocVec(a->length);

  2204. 

  2205.     if (!vec)

  2206.         return NULL;

  2207. 

  2208.     memcpy(vec->coeff, a->coeff, a->length * sizeof(*a->coeff));

  2209. 

  2210.     return vec;

  2211. }

  2212. #endif

  2213. 

  2214. /**

  2215.  * Print with av_log() a textual representation of the vector a

  2216.  * if log_level <= av_log_level.

  2217.  */

  2218. #if !FF_API_SWS_VECTOR

  2219. static

  2220. #endif

  2221. void sws_printVec2(SwsVector *a, AVClass *log_ctx, int log_level)

  2222. {

  2223.     int i;

  2224.     double max = 0;

  2225.     double min = 0;

  2226.     double range;

  2227. 

  2228.     for (i = 0; i < a->length; i++)

  2229.         if (a->coeff[i] > max)

  2230.             max = a->coeff[i];

  2231. 

  2232.     for (i = 0; i < a->length; i++)

  2233.         if (a->coeff[i] < min)

  2234.             min = a->coeff[i];

  2235. 

  2236.     range = max - min;

  2237. 

  2238.     for (i = 0; i < a->length; i++) {

  2239.         int x = (int)((a->coeff[i] - min) * 60.0 / range + 0.5);

  2240.         av_log(log_ctx, log_level, "%1.3f ", a->coeff[i]);

  2241.         for (; x > 0; x--)

  2242.             av_log(log_ctx, log_level, " ");

  2243.         av_log(log_ctx, log_level, "|\n");

  2244.     }

  2245. }

  2246. 

  2247. void sws_freeVec(SwsVector *a)

  2248. {

  2249.     if (!a)

  2250.         return;

  2251.     av_freep(&a->coeff);

  2252.     a->length = 0;

  2253.     av_free(a);

  2254. }

  2255. 

  2256. void sws_freeFilter(SwsFilter *filter)

  2257. {

  2258.     if (!filter)

  2259.         return;

  2260. 

  2261.     sws_freeVec(filter->lumH);

  2262.     sws_freeVec(filter->lumV);

  2263.     sws_freeVec(filter->chrH);

  2264.     sws_freeVec(filter->chrV);

  2265.     av_free(filter);

  2266. }

  2267. 

  2268. void sws_freeContext(SwsContext *c)

  2269. {

  2270.     int i;

  2271.     if (!c)

  2272.         return;

  2273. 

  2274.     for (i = 0; i < 4; i++)

  2275.         av_freep(&c->dither_error[i]);

  2276. 

  2277.     av_freep(&c->vLumFilter);

  2278.     av_freep(&c->vChrFilter);

  2279.     av_freep(&c->hLumFilter);

  2280.     av_freep(&c->hChrFilter);

  2281. #if HAVE_ALTIVEC

  2282.     av_freep(&c->vYCoeffsBank);

  2283.     av_freep(&c->vCCoeffsBank);

  2284. #endif

  2285. 

  2286.     av_freep(&c->vLumFilterPos);

  2287.     av_freep(&c->vChrFilterPos);

  2288.     av_freep(&c->hLumFilterPos);

  2289.     av_freep(&c->hChrFilterPos);

  2290. 

  2291. #if HAVE_MMX_INLINE

  2292. #if USE_MMAP

  2293.     if (c->lumMmxextFilterCode)

  2294.         munmap(c->lumMmxextFilterCode, c->lumMmxextFilterCodeSize);

  2295.     if (c->chrMmxextFilterCode)

  2296.         munmap(c->chrMmxextFilterCode, c->chrMmxextFilterCodeSize);

  2297. #elif HAVE_VIRTUALALLOC

  2298.     if (c->lumMmxextFilterCode)

  2299.         VirtualFree(c->lumMmxextFilterCode, 0, MEM_RELEASE);

  2300.     if (c->chrMmxextFilterCode)

  2301.         VirtualFree(c->chrMmxextFilterCode, 0, MEM_RELEASE);

  2302. #else

  2303.     av_free(c->lumMmxextFilterCode);

  2304.     av_free(c->chrMmxextFilterCode);

  2305. #endif

  2306.     c->lumMmxextFilterCode = NULL;

  2307.     c->chrMmxextFilterCode = NULL;

  2308. #endif /* HAVE_MMX_INLINE */

  2309. 

  2310.     av_freep(&c->yuvTable);

  2311.     av_freep(&c->formatConvBuffer);

  2312. 

  2313.     sws_freeContext(c->cascaded_context[0]);

  2314.     sws_freeContext(c->cascaded_context[1]);

  2315.     sws_freeContext(c->cascaded_context[2]);

  2316.     memset(c->cascaded_context, 0, sizeof(c->cascaded_context));

  2317.     av_freep(&c->cascaded_tmp[0]);

  2318.     av_freep(&c->cascaded1_tmp[0]);

  2319. 

  2320.     av_freep(&c->gamma);

  2321.     av_freep(&c->inv_gamma);

  2322. 

  2323.     ff_free_filters(c);

  2324. 

  2325.     av_free(c);

  2326. }

  2327. 

  2328. struct SwsContext *sws_getCachedContext(struct SwsContext *context, int srcW,

  2329.                                         int srcH, enum AVPixelFormat srcFormat,

  2330.                                         int dstW, int dstH,

  2331.                                         enum AVPixelFormat dstFormat, int flags,

  2332.                                         SwsFilter *srcFilter,

  2333.                                         SwsFilter *dstFilter,

  2334.                                         const double *param)

  2335. {

  2336.     static const double default_param[2] = { SWS_PARAM_DEFAULT,

  2337.                                              SWS_PARAM_DEFAULT };

  2338.     int64_t src_h_chr_pos = -513, dst_h_chr_pos = -513,

  2339.             src_v_chr_pos = -513, dst_v_chr_pos = -513;

  2340. 

  2341.     if (!param)

  2342.         param = default_param;

  2343. 

  2344.     if (context &&

  2345.         (context->srcW      != srcW      ||

  2346.          context->srcH      != srcH      ||

  2347.          context->srcFormat != srcFormat ||

  2348.          context->dstW      != dstW      ||

  2349.          context->dstH      != dstH      ||

  2350.          context->dstFormat != dstFormat ||

  2351.          context->flags     != flags     ||

  2352.          context->param[0]  != param[0]  ||

  2353.          context->param[1]  != param[1])) {

  2354. 

  2355.         av_opt_get_int(context, "src_h_chr_pos", 0, &src_h_chr_pos);

  2356.         av_opt_get_int(context, "src_v_chr_pos", 0, &src_v_chr_pos);

  2357.         av_opt_get_int(context, "dst_h_chr_pos", 0, &dst_h_chr_pos);

  2358.         av_opt_get_int(context, "dst_v_chr_pos", 0, &dst_v_chr_pos);

  2359.         sws_freeContext(context);

  2360.         context = NULL;

  2361.     }

  2362. 

  2363.     if (!context) {

  2364.         if (!(context = sws_alloc_context()))

  2365.             return NULL;

  2366.         context->srcW      = srcW;

  2367.         context->srcH      = srcH;

  2368.         context->srcFormat = srcFormat;

  2369.         context->dstW      = dstW;

  2370.         context->dstH      = dstH;

  2371.         context->dstFormat = dstFormat;

  2372.         context->flags     = flags;

  2373.         context->param[0]  = param[0];

  2374.         context->param[1]  = param[1];

  2375. 

  2376.         av_opt_set_int(context, "src_h_chr_pos", src_h_chr_pos, 0);

  2377.         av_opt_set_int(context, "src_v_chr_pos", src_v_chr_pos, 0);

  2378.         av_opt_set_int(context, "dst_h_chr_pos", dst_h_chr_pos, 0);

  2379.         av_opt_set_int(context, "dst_v_chr_pos", dst_v_chr_pos, 0);

  2380. 

  2381.         if (sws_init_context(context, srcFilter, dstFilter) < 0) {

  2382.             sws_freeContext(context);

  2383.             return NULL;

  2384.         }

  2385.     }

  2386.     return context;

  2387. }