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.
31701 Commits
32 Branches
789MB
Tree: 4556ebfb7d
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '4556ebfb7d'
${ noResults }
FFmpeg/libswscale/swscale_unscaled.c

1058 lines
42KB
Raw Blame History 

  1. /*

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

  3.  *

  4.  * This file is part of Libav.

  5.  *

  6.  * Libav is free software; you can redistribute it and/or

  7.  * modify it under the terms of the GNU Lesser General Public

  8.  * License as published by the Free Software Foundation; either

  9.  * version 2.1 of the License, or (at your option) any later version.

  10.  *

  11.  * Libav is distributed in the hope that it will be useful,

  12.  * but WITHOUT ANY WARRANTY; without even the implied warranty of

  13.  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU

  14.  * Lesser General Public License for more details.

  15.  *

  16.  * You should have received a copy of the GNU Lesser General Public

  17.  * License along with Libav; if not, write to the Free Software

  18.  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA

  19.  */

  20. 

  21. #include <inttypes.h>

  22. #include <string.h>

  23. #include <math.h>

  24. #include <stdio.h>

  25. #include "config.h"

  26. #include <assert.h>

  27. #include "swscale.h"

  28. #include "swscale_internal.h"

  29. #include "rgb2rgb.h"

  30. #include "libavutil/intreadwrite.h"

  31. #include "libavutil/cpu.h"

  32. #include "libavutil/avutil.h"

  33. #include "libavutil/mathematics.h"

  34. #include "libavutil/bswap.h"

  35. #include "libavutil/pixdesc.h"

  36. 

  37. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_1)[8][8] = {

  38.     {   0,  1,  0,  1,  0,  1,  0,  1,},

  39.     {   1,  0,  1,  0,  1,  0,  1,  0,},

  40.     {   0,  1,  0,  1,  0,  1,  0,  1,},

  41.     {   1,  0,  1,  0,  1,  0,  1,  0,},

  42.     {   0,  1,  0,  1,  0,  1,  0,  1,},

  43.     {   1,  0,  1,  0,  1,  0,  1,  0,},

  44.     {   0,  1,  0,  1,  0,  1,  0,  1,},

  45.     {   1,  0,  1,  0,  1,  0,  1,  0,},

  46. };

  47. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_3)[8][8] = {

  48.     {   1,  2,  1,  2,  1,  2,  1,  2,},

  49.     {   3,  0,  3,  0,  3,  0,  3,  0,},

  50.     {   1,  2,  1,  2,  1,  2,  1,  2,},

  51.     {   3,  0,  3,  0,  3,  0,  3,  0,},

  52.     {   1,  2,  1,  2,  1,  2,  1,  2,},

  53.     {   3,  0,  3,  0,  3,  0,  3,  0,},

  54.     {   1,  2,  1,  2,  1,  2,  1,  2,},

  55.     {   3,  0,  3,  0,  3,  0,  3,  0,},

  56. };

  57. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_64)[8][8] = {

  58.     {  18, 34, 30, 46, 17, 33, 29, 45,},

  59.     {  50,  2, 62, 14, 49,  1, 61, 13,},

  60.     {  26, 42, 22, 38, 25, 41, 21, 37,},

  61.     {  58, 10, 54,  6, 57,  9, 53,  5,},

  62.     {  16, 32, 28, 44, 19, 35, 31, 47,},

  63.     {  48,  0, 60, 12, 51,  3, 63, 15,},

  64.     {  24, 40, 20, 36, 27, 43, 23, 39,},

  65.     {  56,  8, 52,  4, 59, 11, 55,  7,},

  66. };

  67. extern const uint8_t dither_8x8_128[8][8];

  68. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_256)[8][8] = {

  69.     {  72, 136, 120, 184,  68, 132, 116, 180,},

  70.     { 200,   8, 248,  56, 196,   4, 244,  52,},

  71.     { 104, 168,  88, 152, 100, 164,  84, 148,},

  72.     { 232,  40, 216,  24, 228,  36, 212,  20,},

  73.     {  64, 128, 102, 176,  76, 140, 124, 188,},

  74.     { 192,   0, 240,  48, 204,  12, 252,  60,},

  75.     {  96, 160,  80, 144, 108, 172,  92, 156,},

  76.     { 224,  32, 208,  16, 236,  44, 220,  28,},

  77. };

  78. 

  79. #define RGB2YUV_SHIFT 15

  80. #define BY ( (int) (0.114 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  81. #define BV (-(int) (0.081 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  82. #define BU ( (int) (0.500 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  83. #define GY ( (int) (0.587 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  84. #define GV (-(int) (0.419 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  85. #define GU (-(int) (0.331 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  86. #define RY ( (int) (0.299 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  87. #define RV ( (int) (0.500 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  88. #define RU (-(int) (0.169 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))

  89. 

  90. static void fillPlane(uint8_t *plane, int stride, int width, int height, int y,

  91.                       uint8_t val)

  92. {

  93.     int i;

  94.     uint8_t *ptr = plane + stride * y;

  95.     for (i = 0; i < height; i++) {

  96.         memset(ptr, val, width);

  97.         ptr += stride;

  98.     }

  99. }

  100. 

  101. static void copyPlane(const uint8_t *src, int srcStride,

  102.                       int srcSliceY, int srcSliceH, int width,

  103.                       uint8_t *dst, int dstStride)

  104. {

  105.     dst += dstStride * srcSliceY;

  106.     if (dstStride == srcStride && srcStride > 0) {

  107.         memcpy(dst, src, srcSliceH * dstStride);

  108.     } else {

  109.         int i;

  110.         for (i = 0; i < srcSliceH; i++) {

  111.             memcpy(dst, src, width);

  112.             src += srcStride;

  113.             dst += dstStride;

  114.         }

  115.     }

  116. }

  117. 

  118. static int planarToNv12Wrapper(SwsContext *c, const uint8_t *src[],

  119.                                int srcStride[], int srcSliceY,

  120.                                int srcSliceH, uint8_t *dstParam[],

  121.                                int dstStride[])

  122. {

  123.     uint8_t *dst = dstParam[1] + dstStride[1] * srcSliceY / 2;

  124. 

  125.     copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,

  126.               dstParam[0], dstStride[0]);

  127. 

  128.     if (c->dstFormat == PIX_FMT_NV12)

  129.         interleaveBytes(src[1], src[2], dst, c->srcW / 2, srcSliceH / 2,

  130.                         srcStride[1], srcStride[2], dstStride[0]);

  131.     else

  132.         interleaveBytes(src[2], src[1], dst, c->srcW / 2, srcSliceH / 2,

  133.                         srcStride[2], srcStride[1], dstStride[0]);

  134. 

  135.     return srcSliceH;

  136. }

  137. 

  138. static int planarToYuy2Wrapper(SwsContext *c, const uint8_t *src[],

  139.                                int srcStride[], int srcSliceY, int srcSliceH,

  140.                                uint8_t *dstParam[], int dstStride[])

  141. {

  142.     uint8_t *dst = dstParam[0] + dstStride[0] * srcSliceY;

  143. 

  144.     yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0],

  145.                srcStride[1], dstStride[0]);

  146. 

  147.     return srcSliceH;

  148. }

  149. 

  150. static int planarToUyvyWrapper(SwsContext *c, const uint8_t *src[],

  151.                                int srcStride[], int srcSliceY, int srcSliceH,

  152.                                uint8_t *dstParam[], int dstStride[])

  153. {

  154.     uint8_t *dst = dstParam[0] + dstStride[0] * srcSliceY;

  155. 

  156.     yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0],

  157.                srcStride[1], dstStride[0]);

  158. 

  159.     return srcSliceH;

  160. }

  161. 

  162. static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t *src[],

  163.                                 int srcStride[], int srcSliceY, int srcSliceH,

  164.                                 uint8_t *dstParam[], int dstStride[])

  165. {

  166.     uint8_t *dst = dstParam[0] + dstStride[0] * srcSliceY;

  167. 

  168.     yuv422ptoyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0],

  169.                   srcStride[1], dstStride[0]);

  170. 

  171.     return srcSliceH;

  172. }

  173. 

  174. static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t *src[],

  175.                                 int srcStride[], int srcSliceY, int srcSliceH,

  176.                                 uint8_t *dstParam[], int dstStride[])

  177. {

  178.     uint8_t *dst = dstParam[0] + dstStride[0] * srcSliceY;

  179. 

  180.     yuv422ptouyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0],

  181.                   srcStride[1], dstStride[0]);

  182. 

  183.     return srcSliceH;

  184. }

  185. 

  186. static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t *src[],

  187.                                int srcStride[], int srcSliceY, int srcSliceH,

  188.                                uint8_t *dstParam[], int dstStride[])

  189. {

  190.     uint8_t *ydst = dstParam[0] + dstStride[0] * srcSliceY;

  191.     uint8_t *udst = dstParam[1] + dstStride[1] * srcSliceY / 2;

  192.     uint8_t *vdst = dstParam[2] + dstStride[2] * srcSliceY / 2;

  193. 

  194.     yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0],

  195.                  dstStride[1], srcStride[0]);

  196. 

  197.     if (dstParam[3])

  198.         fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  199. 

  200.     return srcSliceH;

  201. }

  202. 

  203. static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t *src[],

  204.                                int srcStride[], int srcSliceY, int srcSliceH,

  205.                                uint8_t *dstParam[], int dstStride[])

  206. {

  207.     uint8_t *ydst = dstParam[0] + dstStride[0] * srcSliceY;

  208.     uint8_t *udst = dstParam[1] + dstStride[1] * srcSliceY;

  209.     uint8_t *vdst = dstParam[2] + dstStride[2] * srcSliceY;

  210. 

  211.     yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0],

  212.                  dstStride[1], srcStride[0]);

  213. 

  214.     return srcSliceH;

  215. }

  216. 

  217. static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t *src[],

  218.                                int srcStride[], int srcSliceY, int srcSliceH,

  219.                                uint8_t *dstParam[], int dstStride[])

  220. {

  221.     uint8_t *ydst = dstParam[0] + dstStride[0] * srcSliceY;

  222.     uint8_t *udst = dstParam[1] + dstStride[1] * srcSliceY / 2;

  223.     uint8_t *vdst = dstParam[2] + dstStride[2] * srcSliceY / 2;

  224. 

  225.     uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0],

  226.                  dstStride[1], srcStride[0]);

  227. 

  228.     if (dstParam[3])

  229.         fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  230. 

  231.     return srcSliceH;

  232. }

  233. 

  234. static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t *src[],

  235.                                int srcStride[], int srcSliceY, int srcSliceH,

  236.                                uint8_t *dstParam[], int dstStride[])

  237. {

  238.     uint8_t *ydst = dstParam[0] + dstStride[0] * srcSliceY;

  239.     uint8_t *udst = dstParam[1] + dstStride[1] * srcSliceY;

  240.     uint8_t *vdst = dstParam[2] + dstStride[2] * srcSliceY;

  241. 

  242.     uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0],

  243.                  dstStride[1], srcStride[0]);

  244. 

  245.     return srcSliceH;

  246. }

  247. 

  248. static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, int num_pixels,

  249.                              const uint8_t *palette)

  250. {

  251.     int i;

  252.     for (i = 0; i < num_pixels; i++)

  253.         ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i << 1]] | (src[(i << 1) + 1] << 24);

  254. }

  255. 

  256. static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, int num_pixels,

  257.                                const uint8_t *palette)

  258. {

  259.     int i;

  260. 

  261.     for (i = 0; i < num_pixels; i++)

  262.         ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i << 1]] | src[(i << 1) + 1];

  263. }

  264. 

  265. static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, int num_pixels,

  266.                              const uint8_t *palette)

  267. {

  268.     int i;

  269. 

  270.     for (i = 0; i < num_pixels; i++) {

  271.         //FIXME slow?

  272.         dst[0] = palette[src[i << 1] * 4 + 0];

  273.         dst[1] = palette[src[i << 1] * 4 + 1];

  274.         dst[2] = palette[src[i << 1] * 4 + 2];

  275.         dst += 3;

  276.     }

  277. }

  278. 

  279. static int packed_16bpc_bswap(SwsContext *c, const uint8_t *src[],

  280.                               int srcStride[], int srcSliceY, int srcSliceH,

  281.                               uint8_t *dst[], int dstStride[])

  282. {

  283.     int i, j;

  284.     int srcstr = srcStride[0] >> 1;

  285.     int dststr = dstStride[0] >> 1;

  286.     uint16_t       *dstPtr =       (uint16_t *) dst[0];

  287.     const uint16_t *srcPtr = (const uint16_t *) src[0];

  288.     int min_stride         = FFMIN(srcstr, dststr);

  289. 

  290.     for (i = 0; i < srcSliceH; i++) {

  291.         for (j = 0; j < min_stride; j++) {

  292.             dstPtr[j] = av_bswap16(srcPtr[j]);

  293.         }

  294.         srcPtr += srcstr;

  295.         dstPtr += dststr;

  296.     }

  297. 

  298.     return srcSliceH;

  299. }

  300. 

  301. static int palToRgbWrapper(SwsContext *c, const uint8_t *src[], int srcStride[],

  302.                            int srcSliceY, int srcSliceH, uint8_t *dst[],

  303.                            int dstStride[])

  304. {

  305.     const enum PixelFormat srcFormat = c->srcFormat;

  306.     const enum PixelFormat dstFormat = c->dstFormat;

  307.     void (*conv)(const uint8_t *src, uint8_t *dst, int num_pixels,

  308.                  const uint8_t *palette) = NULL;

  309.     int i;

  310.     uint8_t *dstPtr = dst[0] + dstStride[0] * srcSliceY;

  311.     const uint8_t *srcPtr = src[0];

  312. 

  313.     if (srcFormat == PIX_FMT_Y400A) {

  314.         switch (dstFormat) {

  315.         case PIX_FMT_RGB32  : conv = gray8aToPacked32; break;

  316.         case PIX_FMT_BGR32  : conv = gray8aToPacked32; break;

  317.         case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;

  318.         case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;

  319.         case PIX_FMT_RGB24  : conv = gray8aToPacked24; break;

  320.         case PIX_FMT_BGR24  : conv = gray8aToPacked24; break;

  321.         }

  322.     } else if (usePal(srcFormat)) {

  323.         switch (dstFormat) {

  324.         case PIX_FMT_RGB32  : conv = sws_convertPalette8ToPacked32; break;

  325.         case PIX_FMT_BGR32  : conv = sws_convertPalette8ToPacked32; break;

  326.         case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;

  327.         case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;

  328.         case PIX_FMT_RGB24  : conv = sws_convertPalette8ToPacked24; break;

  329.         case PIX_FMT_BGR24  : conv = sws_convertPalette8ToPacked24; break;

  330.         }

  331.     }

  332. 

  333.     if (!conv)

  334.         av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",

  335.                sws_format_name(srcFormat), sws_format_name(dstFormat));

  336.     else {

  337.         for (i = 0; i < srcSliceH; i++) {

  338.             conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);

  339.             srcPtr += srcStride[0];

  340.             dstPtr += dstStride[0];

  341.         }

  342.     }

  343. 

  344.     return srcSliceH;

  345. }

  346. 

  347. #define isRGBA32(x) (            \

  348.            (x) == PIX_FMT_ARGB   \

  349.         || (x) == PIX_FMT_RGBA   \

  350.         || (x) == PIX_FMT_BGRA   \

  351.         || (x) == PIX_FMT_ABGR   \

  352.         )

  353. 

  354. /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */

  355. static int rgbToRgbWrapper(SwsContext *c, const uint8_t *src[], int srcStride[],

  356.                            int srcSliceY, int srcSliceH, uint8_t *dst[],

  357.                            int dstStride[])

  358. {

  359.     const enum PixelFormat srcFormat = c->srcFormat;

  360.     const enum PixelFormat dstFormat = c->dstFormat;

  361.     const int srcBpp = (c->srcFormatBpp + 7) >> 3;

  362.     const int dstBpp = (c->dstFormatBpp + 7) >> 3;

  363.     const int srcId = c->srcFormatBpp;

  364.     const int dstId = c->dstFormatBpp;

  365.     void (*conv)(const uint8_t *src, uint8_t *dst, int src_size) = NULL;

  366. 

  367. #define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)

  368. 

  369.     if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {

  370.         if (     CONV_IS(ABGR, RGBA)

  371.               || CONV_IS(ARGB, BGRA)

  372.               || CONV_IS(BGRA, ARGB)

  373.               || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;

  374.         else if (CONV_IS(ABGR, ARGB)

  375.               || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;

  376.         else if (CONV_IS(ABGR, BGRA)

  377.               || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;

  378.         else if (CONV_IS(BGRA, RGBA)

  379.               || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;

  380.         else if (CONV_IS(BGRA, ABGR)

  381.               || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;

  382.     } else

  383.     /* BGR -> BGR */

  384.     if ((isBGRinInt(srcFormat) && isBGRinInt(dstFormat)) ||

  385.         (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {

  386.         switch (srcId | (dstId << 16)) {

  387.         case 0x000F0010: conv = rgb16to15; break;

  388.         case 0x000F0018: conv = rgb24to15; break;

  389.         case 0x000F0020: conv = rgb32to15; break;

  390.         case 0x0010000F: conv = rgb15to16; break;

  391.         case 0x00100018: conv = rgb24to16; break;

  392.         case 0x00100020: conv = rgb32to16; break;

  393.         case 0x0018000F: conv = rgb15to24; break;

  394.         case 0x00180010: conv = rgb16to24; break;

  395.         case 0x00180020: conv = rgb32to24; break;

  396.         case 0x0020000F: conv = rgb15to32; break;

  397.         case 0x00200010: conv = rgb16to32; break;

  398.         case 0x00200018: conv = rgb24to32; break;

  399.         }

  400.     } else if ((isBGRinInt(srcFormat) && isRGBinInt(dstFormat)) ||

  401.                (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {

  402.         switch (srcId | (dstId << 16)) {

  403.         case 0x000F000F: conv = rgb15tobgr15; break;

  404.         case 0x000F0010: conv = rgb16tobgr15; break;

  405.         case 0x000F0018: conv = rgb24tobgr15; break;

  406.         case 0x000F0020: conv = rgb32tobgr15; break;

  407.         case 0x0010000F: conv = rgb15tobgr16; break;

  408.         case 0x00100010: conv = rgb16tobgr16; break;

  409.         case 0x00100018: conv = rgb24tobgr16; break;

  410.         case 0x00100020: conv = rgb32tobgr16; break;

  411.         case 0x0018000F: conv = rgb15tobgr24; break;

  412.         case 0x00180010: conv = rgb16tobgr24; break;

  413.         case 0x00180018: conv = rgb24tobgr24; break;

  414.         case 0x00180020: conv = rgb32tobgr24; break;

  415.         case 0x0020000F: conv = rgb15tobgr32; break;

  416.         case 0x00200010: conv = rgb16tobgr32; break;

  417.         case 0x00200018: conv = rgb24tobgr32; break;

  418.         }

  419.     }

  420. 

  421.     if (!conv) {

  422.         av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",

  423.                sws_format_name(srcFormat), sws_format_name(dstFormat));

  424.     } else {

  425.         const uint8_t *srcPtr = src[0];

  426.               uint8_t *dstPtr = dst[0];

  427.         if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) &&

  428.             !isRGBA32(dstFormat))

  429.             srcPtr += ALT32_CORR;

  430. 

  431.         if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) &&

  432.             !isRGBA32(srcFormat))

  433.             dstPtr += ALT32_CORR;

  434. 

  435.         if (dstStride[0] * srcBpp == srcStride[0] * dstBpp && srcStride[0] > 0 &&

  436.             !(srcStride[0] % srcBpp))

  437.             conv(srcPtr, dstPtr + dstStride[0] * srcSliceY,

  438.                  srcSliceH * srcStride[0]);

  439.         else {

  440.             int i;

  441.             dstPtr += dstStride[0] * srcSliceY;

  442. 

  443.             for (i = 0; i < srcSliceH; i++) {

  444.                 conv(srcPtr, dstPtr, c->srcW * srcBpp);

  445.                 srcPtr += srcStride[0];

  446.                 dstPtr += dstStride[0];

  447.             }

  448.         }

  449.     }

  450.     return srcSliceH;

  451. }

  452. 

  453. static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t *src[],

  454.                               int srcStride[], int srcSliceY, int srcSliceH,

  455.                               uint8_t *dst[], int dstStride[])

  456. {

  457.     rgb24toyv12(

  458.         src[0],

  459.         dst[0] +  srcSliceY       * dstStride[0],

  460.         dst[1] + (srcSliceY >> 1) * dstStride[1],

  461.         dst[2] + (srcSliceY >> 1) * dstStride[2],

  462.         c->srcW, srcSliceH,

  463.         dstStride[0], dstStride[1], srcStride[0]);

  464.     if (dst[3])

  465.         fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  466.     return srcSliceH;

  467. }

  468. 

  469. static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t *src[],

  470.                              int srcStride[], int srcSliceY, int srcSliceH,

  471.                              uint8_t *dst[], int dstStride[])

  472. {

  473.     copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,

  474.               dst[0], dstStride[0]);

  475. 

  476.     planar2x(src[1], dst[1] + dstStride[1] * (srcSliceY >> 1), c->chrSrcW,

  477.              srcSliceH >> 2, srcStride[1], dstStride[1]);

  478.     planar2x(src[2], dst[2] + dstStride[2] * (srcSliceY >> 1), c->chrSrcW,

  479.              srcSliceH >> 2, srcStride[2], dstStride[2]);

  480.     if (dst[3])

  481.         fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  482.     return srcSliceH;

  483. }

  484. 

  485. /* unscaled copy like stuff (assumes nearly identical formats) */

  486. static int packedCopyWrapper(SwsContext *c, const uint8_t *src[],

  487.                              int srcStride[], int srcSliceY, int srcSliceH,

  488.                              uint8_t *dst[], int dstStride[])

  489. {

  490.     if (dstStride[0] == srcStride[0] && srcStride[0] > 0)

  491.         memcpy(dst[0] + dstStride[0] * srcSliceY, src[0], srcSliceH * dstStride[0]);

  492.     else {

  493.         int i;

  494.         const uint8_t *srcPtr = src[0];

  495.         uint8_t *dstPtr = dst[0] + dstStride[0] * srcSliceY;

  496.         int length = 0;

  497. 

  498.         /* universal length finder */

  499.         while (length + c->srcW <= FFABS(dstStride[0]) &&

  500.                length + c->srcW <= FFABS(srcStride[0]))

  501.             length += c->srcW;

  502.         assert(length != 0);

  503. 

  504.         for (i = 0; i < srcSliceH; i++) {

  505.             memcpy(dstPtr, srcPtr, length);

  506.             srcPtr += srcStride[0];

  507.             dstPtr += dstStride[0];

  508.         }

  509.     }

  510.     return srcSliceH;

  511. }

  512. 

  513. #define clip9(x)  av_clip_uintp2(x,  9)

  514. #define clip10(x) av_clip_uintp2(x, 10)

  515. #define DITHER_COPY(dst, dstStride, wfunc, src, srcStride, rfunc, dithers, shift, clip) \

  516.     for (i = 0; i < height; i++) { \

  517.         const uint8_t *dither = dithers[i & 7]; \

  518.         for (j = 0; j < length - 7; j += 8) { \

  519.             wfunc(&dst[j + 0], clip((rfunc(&src[j + 0]) + dither[0]) >> shift)); \

  520.             wfunc(&dst[j + 1], clip((rfunc(&src[j + 1]) + dither[1]) >> shift)); \

  521.             wfunc(&dst[j + 2], clip((rfunc(&src[j + 2]) + dither[2]) >> shift)); \

  522.             wfunc(&dst[j + 3], clip((rfunc(&src[j + 3]) + dither[3]) >> shift)); \

  523.             wfunc(&dst[j + 4], clip((rfunc(&src[j + 4]) + dither[4]) >> shift)); \

  524.             wfunc(&dst[j + 5], clip((rfunc(&src[j + 5]) + dither[5]) >> shift)); \

  525.             wfunc(&dst[j + 6], clip((rfunc(&src[j + 6]) + dither[6]) >> shift)); \

  526.             wfunc(&dst[j + 7], clip((rfunc(&src[j + 7]) + dither[7]) >> shift)); \

  527.         } \

  528.         for (; j < length; j++) \

  529.             wfunc(&dst[j],     (rfunc(&src[j]) + dither[j & 7]) >> shift); \

  530.         dst += dstStride; \

  531.         src += srcStride; \

  532.     }

  533. 

  534. static int planarCopyWrapper(SwsContext *c, const uint8_t *src[],

  535.                              int srcStride[], int srcSliceY, int srcSliceH,

  536.                              uint8_t *dst[], int dstStride[])

  537. {

  538.     int plane, i, j;

  539.     for (plane = 0; plane < 4; plane++) {

  540.         int length = (plane == 0 || plane == 3) ? c->srcW  : -((-c->srcW  ) >> c->chrDstHSubSample);

  541.         int y =      (plane == 0 || plane == 3) ? srcSliceY: -((-srcSliceY) >> c->chrDstVSubSample);

  542.         int height = (plane == 0 || plane == 3) ? srcSliceH: -((-srcSliceH) >> c->chrDstVSubSample);

  543.         const uint8_t *srcPtr = src[plane];

  544.         uint8_t *dstPtr = dst[plane] + dstStride[plane] * y;

  545. 

  546.         if (!dst[plane])

  547.             continue;

  548.         // ignore palette for GRAY8

  549.         if (plane == 1 && !dst[2]) continue;

  550.         if (!src[plane] || (plane == 1 && !src[2])) {

  551.             if (is16BPS(c->dstFormat))

  552.                 length *= 2;

  553.             fillPlane(dst[plane], dstStride[plane], length, height, y,

  554.                       (plane == 3) ? 255 : 128);

  555.         } else {

  556.             if (is9_OR_10BPS(c->srcFormat)) {

  557.                 const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1 + 1;

  558.                 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1 + 1;

  559.                 const uint16_t *srcPtr2 = (const uint16_t *) srcPtr;

  560. 

  561.                 if (is16BPS(c->dstFormat)) {

  562.                     uint16_t *dstPtr2 = (uint16_t *) dstPtr;

  563. #define COPY9_OR_10TO16(rfunc, wfunc) \

  564.                     for (i = 0; i < height; i++) { \

  565.                         for (j = 0; j < length; j++) { \

  566.                             int srcpx = rfunc(&srcPtr2[j]); \

  567.                             wfunc(&dstPtr2[j], (srcpx << (16 - src_depth)) | (srcpx >> (2 * src_depth - 16))); \

  568.                         } \

  569.                         dstPtr2 += dstStride[plane] / 2; \

  570.                         srcPtr2 += srcStride[plane] / 2; \

  571.                     }

  572.                     if (isBE(c->dstFormat)) {

  573.                         if (isBE(c->srcFormat)) {

  574.                             COPY9_OR_10TO16(AV_RB16, AV_WB16);

  575.                         } else {

  576.                             COPY9_OR_10TO16(AV_RL16, AV_WB16);

  577.                         }

  578.                     } else {

  579.                         if (isBE(c->srcFormat)) {

  580.                             COPY9_OR_10TO16(AV_RB16, AV_WL16);

  581.                         } else {

  582.                             COPY9_OR_10TO16(AV_RL16, AV_WL16);

  583.                         }

  584.                     }

  585.                 } else if (is9_OR_10BPS(c->dstFormat)) {

  586.                     uint16_t *dstPtr2 = (uint16_t *) dstPtr;

  587. #define COPY9_OR_10TO9_OR_10(loop) \

  588.                     for (i = 0; i < height; i++) { \

  589.                         for (j = 0; j < length; j++) { \

  590.                             loop; \

  591.                         } \

  592.                         dstPtr2 += dstStride[plane] / 2; \

  593.                         srcPtr2 += srcStride[plane] / 2; \

  594.                     }

  595. #define COPY9_OR_10TO9_OR_10_2(rfunc, wfunc) \

  596.                     if (dst_depth > src_depth) { \

  597.                         COPY9_OR_10TO9_OR_10(int srcpx = rfunc(&srcPtr2[j]); \

  598.                             wfunc(&dstPtr2[j], (srcpx << 1) | (srcpx >> 9))); \

  599.                     } else if (dst_depth < src_depth) { \

  600.                         DITHER_COPY(dstPtr2, dstStride[plane] / 2, wfunc, \

  601.                                     srcPtr2, srcStride[plane] / 2, rfunc, \

  602.                                     dither_8x8_1, 1, clip9); \

  603.                     } else { \

  604.                         COPY9_OR_10TO9_OR_10(wfunc(&dstPtr2[j], rfunc(&srcPtr2[j]))); \

  605.                     }

  606.                     if (isBE(c->dstFormat)) {

  607.                         if (isBE(c->srcFormat)) {

  608.                             COPY9_OR_10TO9_OR_10_2(AV_RB16, AV_WB16);

  609.                         } else {

  610.                             COPY9_OR_10TO9_OR_10_2(AV_RL16, AV_WB16);

  611.                         }

  612.                     } else {

  613.                         if (isBE(c->srcFormat)) {

  614.                             COPY9_OR_10TO9_OR_10_2(AV_RB16, AV_WL16);

  615.                         } else {

  616.                             COPY9_OR_10TO9_OR_10_2(AV_RL16, AV_WL16);

  617.                         }

  618.                     }

  619.                 } else {

  620. #define W8(a, b) { *(a) = (b); }

  621. #define COPY9_OR_10TO8(rfunc) \

  622.                     if (src_depth == 9) { \

  623.                         DITHER_COPY(dstPtr,  dstStride[plane],   W8, \

  624.                                     srcPtr2, srcStride[plane] / 2, rfunc, \

  625.                                     dither_8x8_1, 1, av_clip_uint8); \

  626.                     } else { \

  627.                         DITHER_COPY(dstPtr,  dstStride[plane],   W8, \

  628.                                     srcPtr2, srcStride[plane] / 2, rfunc, \

  629.                                     dither_8x8_3, 2, av_clip_uint8); \

  630.                     }

  631.                     if (isBE(c->srcFormat)) {

  632.                         COPY9_OR_10TO8(AV_RB16);

  633.                     } else {

  634.                         COPY9_OR_10TO8(AV_RL16);

  635.                     }

  636.                 }

  637.             } else if (is9_OR_10BPS(c->dstFormat)) {

  638.                 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1 + 1;

  639.                 uint16_t *dstPtr2 = (uint16_t *) dstPtr;

  640. 

  641.                 if (is16BPS(c->srcFormat)) {

  642.                     const uint16_t *srcPtr2 = (const uint16_t *) srcPtr;

  643. #define COPY16TO9_OR_10(rfunc, wfunc) \

  644.                     if (dst_depth == 9) { \

  645.                         DITHER_COPY(dstPtr2, dstStride[plane] / 2, wfunc, \

  646.                                     srcPtr2, srcStride[plane] / 2, rfunc, \

  647.                                     dither_8x8_128, 7, clip9); \

  648.                     } else { \

  649.                         DITHER_COPY(dstPtr2, dstStride[plane] / 2, wfunc, \

  650.                                     srcPtr2, srcStride[plane] / 2, rfunc, \

  651.                                     dither_8x8_64, 6, clip10); \

  652.                     }

  653.                     if (isBE(c->dstFormat)) {

  654.                         if (isBE(c->srcFormat)) {

  655.                             COPY16TO9_OR_10(AV_RB16, AV_WB16);

  656.                         } else {

  657.                             COPY16TO9_OR_10(AV_RL16, AV_WB16);

  658.                         }

  659.                     } else {

  660.                         if (isBE(c->srcFormat)) {

  661.                             COPY16TO9_OR_10(AV_RB16, AV_WL16);

  662.                         } else {

  663.                             COPY16TO9_OR_10(AV_RL16, AV_WL16);

  664.                         }

  665.                     }

  666.                 } else /* 8bit */ {

  667. #define COPY8TO9_OR_10(wfunc) \

  668.                     for (i = 0; i < height; i++) { \

  669.                         for (j = 0; j < length; j++) { \

  670.                             const int srcpx = srcPtr[j]; \

  671.                             wfunc(&dstPtr2[j], (srcpx << (dst_depth - 8)) | (srcpx >> (16 - dst_depth))); \

  672.                         } \

  673.                         dstPtr2 += dstStride[plane] / 2; \

  674.                         srcPtr  += srcStride[plane]; \

  675.                     }

  676.                     if (isBE(c->dstFormat)) {

  677.                         COPY8TO9_OR_10(AV_WB16);

  678.                     } else {

  679.                         COPY8TO9_OR_10(AV_WL16);

  680.                     }

  681.                 }

  682.             } else if (is16BPS(c->srcFormat) && !is16BPS(c->dstFormat)) {

  683.                 const uint16_t *srcPtr2 = (const uint16_t *) srcPtr;

  684. #define COPY16TO8(rfunc) \

  685.                     DITHER_COPY(dstPtr,  dstStride[plane],   W8, \

  686.                                 srcPtr2, srcStride[plane] / 2, rfunc, \

  687.                                 dither_8x8_256, 8, av_clip_uint8);

  688.                 if (isBE(c->srcFormat)) {

  689.                     COPY16TO8(AV_RB16);

  690.                 } else {

  691.                     COPY16TO8(AV_RL16);

  692.                 }

  693.             } else if (!is16BPS(c->srcFormat) && is16BPS(c->dstFormat)) {

  694.                 for (i = 0; i < height; i++) {

  695.                     for (j = 0; j < length; j++) {

  696.                         dstPtr[ j << 1     ] = srcPtr[j];

  697.                         dstPtr[(j << 1) + 1] = srcPtr[j];

  698.                     }

  699.                     srcPtr += srcStride[plane];

  700.                     dstPtr += dstStride[plane];

  701.                 }

  702.             } else if (is16BPS(c->srcFormat) && is16BPS(c->dstFormat) &&

  703.                       isBE(c->srcFormat) != isBE(c->dstFormat)) {

  704. 

  705.                 for (i = 0; i < height; i++) {

  706.                     for (j = 0; j < length; j++)

  707.                         ((uint16_t *) dstPtr)[j] = av_bswap16(((const uint16_t *) srcPtr)[j]);

  708.                     srcPtr += srcStride[plane];

  709.                     dstPtr += dstStride[plane];

  710.                 }

  711.             } else if (dstStride[plane] == srcStride[plane] &&

  712.                        srcStride[plane] > 0 && srcStride[plane] == length) {

  713.                 memcpy(dst[plane] + dstStride[plane] * y, src[plane],

  714.                        height * dstStride[plane]);

  715.             } else {

  716.                 if (is16BPS(c->srcFormat) && is16BPS(c->dstFormat))

  717.                     length *= 2;

  718.                 for (i = 0; i < height; i++) {

  719.                     memcpy(dstPtr, srcPtr, length);

  720.                     srcPtr += srcStride[plane];

  721.                     dstPtr += dstStride[plane];

  722.                 }

  723.             }

  724.         }

  725.     }

  726.     return srcSliceH;

  727. }

  728. 

  729. 

  730. #define IS_DIFFERENT_ENDIANESS(src_fmt, dst_fmt, pix_fmt)          \

  731.     ((src_fmt == pix_fmt ## BE && dst_fmt == pix_fmt ## LE) ||     \

  732.      (src_fmt == pix_fmt ## LE && dst_fmt == pix_fmt ## BE))

  733. 

  734. 

  735. void ff_get_unscaled_swscale(SwsContext *c)

  736. {

  737.     const enum PixelFormat srcFormat = c->srcFormat;

  738.     const enum PixelFormat dstFormat = c->dstFormat;

  739.     const int flags = c->flags;

  740.     const int dstH = c->dstH;

  741.     int needsDither;

  742. 

  743.     needsDither = isAnyRGB(dstFormat) &&

  744.             c->dstFormatBpp < 24 &&

  745.            (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));

  746. 

  747.     /* yv12_to_nv12 */

  748.     if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) &&

  749.         (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {

  750.         c->swScale = planarToNv12Wrapper;

  751.     }

  752.     /* yuv2bgr */

  753.     if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUV422P ||

  754.          srcFormat == PIX_FMT_YUVA420P) && isAnyRGB(dstFormat) &&

  755.         !(flags & SWS_ACCURATE_RND) && !(dstH & 1)) {

  756.         c->swScale = ff_yuv2rgb_get_func_ptr(c);

  757.     }

  758. 

  759.     if (srcFormat == PIX_FMT_YUV410P &&

  760.         (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P) &&

  761.         !(flags & SWS_BITEXACT)) {

  762.         c->swScale = yvu9ToYv12Wrapper;

  763.     }

  764. 

  765.     /* bgr24toYV12 */

  766.     if (srcFormat == PIX_FMT_BGR24 &&

  767.         (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P) &&

  768.         !(flags & SWS_ACCURATE_RND))

  769.         c->swScale = bgr24ToYv12Wrapper;

  770. 

  771.     /* RGB/BGR -> RGB/BGR (no dither needed forms) */

  772.     if (   isAnyRGB(srcFormat)

  773.         && isAnyRGB(dstFormat)

  774.         && srcFormat != PIX_FMT_BGR8      && dstFormat != PIX_FMT_BGR8

  775.         && srcFormat != PIX_FMT_RGB8      && dstFormat != PIX_FMT_RGB8

  776.         && srcFormat != PIX_FMT_BGR4      && dstFormat != PIX_FMT_BGR4

  777.         && srcFormat != PIX_FMT_RGB4      && dstFormat != PIX_FMT_RGB4

  778.         && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE

  779.         && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE

  780.         && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK

  781.         && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE

  782.         && srcFormat != PIX_FMT_RGB48LE   && dstFormat != PIX_FMT_RGB48LE

  783.         && srcFormat != PIX_FMT_RGB48BE   && dstFormat != PIX_FMT_RGB48BE

  784.         && srcFormat != PIX_FMT_BGR48LE   && dstFormat != PIX_FMT_BGR48LE

  785.         && srcFormat != PIX_FMT_BGR48BE   && dstFormat != PIX_FMT_BGR48BE

  786.         && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))

  787.         c->swScale= rgbToRgbWrapper;

  788. 

  789.     /* bswap 16 bits per pixel/component packed formats */

  790.     if (IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGR444) ||

  791.         IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGR48)  ||

  792.         IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGR555) ||

  793.         IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGR565) ||

  794.         IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_GRAY16) ||

  795.         IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGB444) ||

  796.         IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGB48)  ||

  797.         IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGB555) ||

  798.         IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGB565))

  799.         c->swScale = packed_16bpc_bswap;

  800. 

  801.     if ((usePal(srcFormat) && (

  802.         dstFormat == PIX_FMT_RGB32   ||

  803.         dstFormat == PIX_FMT_RGB32_1 ||

  804.         dstFormat == PIX_FMT_RGB24   ||

  805.         dstFormat == PIX_FMT_BGR32   ||

  806.         dstFormat == PIX_FMT_BGR32_1 ||

  807.         dstFormat == PIX_FMT_BGR24)))

  808.         c->swScale = palToRgbWrapper;

  809. 

  810.     if (srcFormat == PIX_FMT_YUV422P) {

  811.         if (dstFormat == PIX_FMT_YUYV422)

  812.             c->swScale = yuv422pToYuy2Wrapper;

  813.         else if (dstFormat == PIX_FMT_UYVY422)

  814.             c->swScale = yuv422pToUyvyWrapper;

  815.     }

  816. 

  817.     /* LQ converters if -sws 0 or -sws 4*/

  818.     if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {

  819.         /* yv12_to_yuy2 */

  820.         if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {

  821.             if (dstFormat == PIX_FMT_YUYV422)

  822.                 c->swScale = planarToYuy2Wrapper;

  823.             else if (dstFormat == PIX_FMT_UYVY422)

  824.                 c->swScale = planarToUyvyWrapper;

  825.         }

  826.     }

  827.     if (srcFormat == PIX_FMT_YUYV422 &&

  828.        (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))

  829.         c->swScale = yuyvToYuv420Wrapper;

  830.     if (srcFormat == PIX_FMT_UYVY422 &&

  831.        (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))

  832.         c->swScale = uyvyToYuv420Wrapper;

  833.     if (srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)

  834.         c->swScale = yuyvToYuv422Wrapper;

  835.     if (srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)

  836.         c->swScale = uyvyToYuv422Wrapper;

  837. 

  838.     /* simple copy */

  839.     if ( srcFormat == dstFormat ||

  840.         (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P) ||

  841.         (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P) ||

  842.         (isPlanarYUV(srcFormat) && isGray(dstFormat)) ||

  843.         (isPlanarYUV(dstFormat) && isGray(srcFormat)) ||

  844.         (isGray(dstFormat) && isGray(srcFormat)) ||

  845.         (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat) &&

  846.          c->chrDstHSubSample == c->chrSrcHSubSample &&

  847.          c->chrDstVSubSample == c->chrSrcVSubSample &&

  848.          dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21 &&

  849.          srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))

  850.     {

  851.         if (isPacked(c->srcFormat))

  852.             c->swScale = packedCopyWrapper;

  853.         else /* Planar YUV or gray */

  854.             c->swScale = planarCopyWrapper;

  855.     }

  856. 

  857.     if (ARCH_BFIN)

  858.         ff_bfin_get_unscaled_swscale(c);

  859.     if (HAVE_ALTIVEC)

  860.         ff_swscale_get_unscaled_altivec(c);

  861. }

  862. 

  863. static void reset_ptr(const uint8_t *src[], int format)

  864. {

  865.     if (!isALPHA(format))

  866.         src[3] = NULL;

  867.     if (!isPlanar(format)) {

  868.         src[3] = src[2] = NULL;

  869. 

  870.         if (!usePal(format))

  871.             src[1] = NULL;

  872.     }

  873. }

  874. 

  875. static int check_image_pointers(uint8_t *data[4], enum PixelFormat pix_fmt,

  876.                                 const int linesizes[4])

  877. {

  878.     const AVPixFmtDescriptor *desc = &av_pix_fmt_descriptors[pix_fmt];

  879.     int i;

  880. 

  881.     for (i = 0; i < 4; i++) {

  882.         int plane = desc->comp[i].plane;

  883.         if (!data[plane] || !linesizes[plane])

  884.             return 0;

  885.     }

  886. 

  887.     return 1;

  888. }

  889. 

  890. /**

  891.  * swscale wrapper, so we don't need to export the SwsContext.

  892.  * Assumes planar YUV to be in YUV order instead of YVU.

  893.  */

  894. int attribute_align_arg sws_scale(struct SwsContext *c,

  895.                                   const uint8_t * const srcSlice[],

  896.                                   const int srcStride[], int srcSliceY,

  897.                                   int srcSliceH, uint8_t *const dst[],

  898.                                   const int dstStride[])

  899. {

  900.     int i;

  901.     const uint8_t *src2[4] = { srcSlice[0], srcSlice[1], srcSlice[2], srcSlice[3] };

  902.     uint8_t *dst2[4] = { dst[0], dst[1], dst[2], dst[3] };

  903. 

  904.     // do not mess up sliceDir if we have a "trailing" 0-size slice

  905.     if (srcSliceH == 0)

  906.         return 0;

  907. 

  908.     if (!check_image_pointers(srcSlice, c->srcFormat, srcStride)) {

  909.         av_log(c, AV_LOG_ERROR, "bad src image pointers\n");

  910.         return 0;

  911.     }

  912.     if (!check_image_pointers(dst, c->dstFormat, dstStride)) {

  913.         av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");

  914.         return 0;

  915.     }

  916. 

  917.     if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {

  918.         av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");

  919.         return 0;

  920.     }

  921.     if (c->sliceDir == 0) {

  922.         if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;

  923.     }

  924. 

  925.     if (usePal(c->srcFormat)) {

  926.         for (i = 0; i < 256; i++) {

  927.             int p, r, g, b, y, u, v;

  928.             if (c->srcFormat == PIX_FMT_PAL8) {

  929.                 p = ((const uint32_t *)(srcSlice[1]))[i];

  930.                 r = (p >> 16) & 0xFF;

  931.                 g = (p >>  8) & 0xFF;

  932.                 b =  p        & 0xFF;

  933.             } else if (c->srcFormat == PIX_FMT_RGB8) {

  934.                 r = ( i >> 5     ) * 36;

  935.                 g = ((i >> 2) & 7) * 36;

  936.                 b = ( i       & 3) * 85;

  937.             } else if (c->srcFormat == PIX_FMT_BGR8) {

  938.                 b = ( i >> 6     ) * 85;

  939.                 g = ((i >> 3) & 7) * 36;

  940.                 r = ( i       & 7) * 36;

  941.             } else if (c->srcFormat == PIX_FMT_RGB4_BYTE) {

  942.                 r = ( i >> 3     ) * 255;

  943.                 g = ((i >> 1) & 3) * 85;

  944.                 b = ( i       & 1) * 255;

  945.             } else if (c->srcFormat == PIX_FMT_GRAY8 ||

  946.                       c->srcFormat == PIX_FMT_Y400A) {

  947.                 r = g = b = i;

  948.             } else {

  949.                 assert(c->srcFormat == PIX_FMT_BGR4_BYTE);

  950.                 b = ( i >> 3     ) * 255;

  951.                 g = ((i >> 1) & 3) * 85;

  952.                 r = ( i       & 1) * 255;

  953.             }

  954.             y = av_clip_uint8((RY * r + GY * g + BY * b + ( 33 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);

  955.             u = av_clip_uint8((RU * r + GU * g + BU * b + (257 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);

  956.             v = av_clip_uint8((RV * r + GV * g + BV * b + (257 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);

  957.             c->pal_yuv[i] = y + (u << 8) + (v << 16);

  958. 

  959.             switch (c->dstFormat) {

  960.             case PIX_FMT_BGR32:

  961. #if !HAVE_BIGENDIAN

  962.             case PIX_FMT_RGB24:

  963. #endif

  964.                 c->pal_rgb[i] =  r + (g << 8) + (b << 16);

  965.                 break;

  966.             case PIX_FMT_BGR32_1:

  967. #if HAVE_BIGENDIAN

  968.             case PIX_FMT_BGR24:

  969. #endif

  970.                 c->pal_rgb[i] = (r + (g << 8) + (b << 16)) << 8;

  971.                 break;

  972.             case PIX_FMT_RGB32_1:

  973. #if HAVE_BIGENDIAN

  974.             case PIX_FMT_RGB24:

  975. #endif

  976.                 c->pal_rgb[i] = (b + (g << 8) + (r << 16)) << 8;

  977.                 break;

  978.             case PIX_FMT_RGB32:

  979. #if !HAVE_BIGENDIAN

  980.             case PIX_FMT_BGR24:

  981. #endif

  982.             default:

  983.                 c->pal_rgb[i] =  b + (g << 8) + (r << 16);

  984.             }

  985.         }

  986.     }

  987. 

  988.     // copy strides, so they can safely be modified

  989.     if (c->sliceDir == 1) {

  990.         // slices go from top to bottom

  991.         int srcStride2[4] = { srcStride[0], srcStride[1], srcStride[2],

  992.                               srcStride[3] };

  993.         int dstStride2[4] = { dstStride[0], dstStride[1], dstStride[2],

  994.                               dstStride[3] };

  995. 

  996.         reset_ptr(src2, c->srcFormat);

  997.         reset_ptr((const uint8_t **) dst2, c->dstFormat);

  998. 

  999.         /* reset slice direction at end of frame */

  1000.         if (srcSliceY + srcSliceH == c->srcH)

  1001.             c->sliceDir = 0;

  1002. 

  1003.         return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2,

  1004.                           dstStride2);

  1005.     } else {

  1006.         // slices go from bottom to top => we flip the image internally

  1007.         int srcStride2[4] = { -srcStride[0], -srcStride[1], -srcStride[2],

  1008.                               -srcStride[3] };

  1009.         int dstStride2[4] = { -dstStride[0], -dstStride[1], -dstStride[2],

  1010.                               -dstStride[3] };

  1011. 

  1012.         src2[0] += (srcSliceH - 1) * srcStride[0];

  1013.         if (!usePal(c->srcFormat))

  1014.             src2[1] += ((srcSliceH >> c->chrSrcVSubSample) - 1) * srcStride[1];

  1015.         src2[2] += ((srcSliceH >> c->chrSrcVSubSample) - 1) * srcStride[2];

  1016.         src2[3] += (srcSliceH - 1) * srcStride[3];

  1017.         dst2[0] += ( c->dstH                         - 1) * dstStride[0];

  1018.         dst2[1] += ((c->dstH >> c->chrDstVSubSample) - 1) * dstStride[1];

  1019.         dst2[2] += ((c->dstH >> c->chrDstVSubSample) - 1) * dstStride[2];

  1020.         dst2[3] += ( c->dstH                         - 1) * dstStride[3];

  1021. 

  1022.         reset_ptr(src2, c->srcFormat);

  1023.         reset_ptr((const uint8_t **) dst2, c->dstFormat);

  1024. 

  1025.         /* reset slice direction at end of frame */

  1026.         if (!srcSliceY)

  1027.             c->sliceDir = 0;

  1028. 

  1029.         return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH,

  1030.                           srcSliceH, dst2, dstStride2);

  1031.     }

  1032. }

  1033. 

  1034. /* Convert the palette to the same packed 32-bit format as the palette */

  1035. void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst,

  1036.                                    int num_pixels, const uint8_t *palette)

  1037. {

  1038.     int i;

  1039. 

  1040.     for (i = 0; i < num_pixels; i++)

  1041.         ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];

  1042. }

  1043. 

  1044. /* Palette format: ABCD -> dst format: ABC */

  1045. void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst,

  1046.                                    int num_pixels, const uint8_t *palette)

  1047. {

  1048.     int i;

  1049. 

  1050.     for (i = 0; i < num_pixels; i++) {

  1051.         //FIXME slow?

  1052.         dst[0] = palette[src[i] * 4 + 0];

  1053.         dst[1] = palette[src[i] * 4 + 1];

  1054.         dst[2] = palette[src[i] * 4 + 2];

  1055.         dst += 3;

  1056.     }

  1057. }