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.
30264 Commits
32 Branches
1.0GB
Tree: 2b6bfff2b2
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '2b6bfff2b2'
${ noResults }
FFmpeg/libswscale/swscale.c

2118 lines
77KB
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. /*

  22.   supported Input formats: YV12, I420/IYUV, YUY2, UYVY, BGR32, BGR32_1, BGR24, BGR16, BGR15, RGB32, RGB32_1, RGB24, Y8/Y800, YVU9/IF09, PAL8

  23.   supported output formats: YV12, I420/IYUV, YUY2, UYVY, {BGR,RGB}{1,4,8,15,16,24,32}, Y8/Y800, YVU9/IF09

  24.   {BGR,RGB}{1,4,8,15,16} support dithering

  25. 

  26.   unscaled special converters (YV12=I420=IYUV, Y800=Y8)

  27.   YV12 -> {BGR,RGB}{1,4,8,12,15,16,24,32}

  28.   x -> x

  29.   YUV9 -> YV12

  30.   YUV9/YV12 -> Y800

  31.   Y800 -> YUV9/YV12

  32.   BGR24 -> BGR32 & RGB24 -> RGB32

  33.   BGR32 -> BGR24 & RGB32 -> RGB24

  34.   BGR15 -> BGR16

  35. */

  36. 

  37. /*

  38. tested special converters (most are tested actually, but I did not write it down ...)

  39.  YV12 -> BGR12/BGR16

  40.  YV12 -> YV12

  41.  BGR15 -> BGR16

  42.  BGR16 -> BGR16

  43.  YVU9 -> YV12

  44. 

  45. untested special converters

  46.   YV12/I420 -> BGR15/BGR24/BGR32 (it is the yuv2rgb stuff, so it should be OK)

  47.   YV12/I420 -> YV12/I420

  48.   YUY2/BGR15/BGR24/BGR32/RGB24/RGB32 -> same format

  49.   BGR24 -> BGR32 & RGB24 -> RGB32

  50.   BGR32 -> BGR24 & RGB32 -> RGB24

  51.   BGR24 -> YV12

  52. */

  53. 

  54. #include <inttypes.h>

  55. #include <string.h>

  56. #include <math.h>

  57. #include <stdio.h>

  58. #include "config.h"

  59. #include <assert.h>

  60. #include "swscale.h"

  61. #include "swscale_internal.h"

  62. #include "rgb2rgb.h"

  63. #include "libavutil/avassert.h"

  64. #include "libavutil/intreadwrite.h"

  65. #include "libavutil/x86_cpu.h"

  66. #include "libavutil/cpu.h"

  67. #include "libavutil/avutil.h"

  68. #include "libavutil/mathematics.h"

  69. #include "libavutil/bswap.h"

  70. #include "libavutil/pixdesc.h"

  71. 

  72. #undef MOVNTQ

  73. #undef PAVGB

  74. 

  75. #define DITHER1XBPP

  76. 

  77. #define isPacked(x)         (       \

  78.            (x)==PIX_FMT_PAL8        \

  79.         || (x)==PIX_FMT_YUYV422     \

  80.         || (x)==PIX_FMT_UYVY422     \

  81.         || (x)==PIX_FMT_GRAY8A       \

  82.         || isAnyRGB(x)              \

  83.     )

  84. 

  85. #define RGB2YUV_SHIFT 15

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

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

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

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

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

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

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

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

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

  95. 

  96. static const double rgb2yuv_table[8][9]={

  97.     {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5}, //ITU709

  98.     {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5}, //ITU709

  99.     {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M

  100.     {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M

  101.     {0.59  , 0.11  , 0.30  , -0.331, 0.5, -0.169, -0.421, -0.079, 0.5}, //FCC

  102.     {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M

  103.     {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M

  104.     {0.701 , 0.087 , 0.212 , -0.384, 0.5, -0.116, -0.445, -0.055, 0.5}, //SMPTE 240M

  105. };

  106. 

  107. /*

  108. NOTES

  109. Special versions: fast Y 1:1 scaling (no interpolation in y direction)

  110. 

  111. TODO

  112. more intelligent misalignment avoidance for the horizontal scaler

  113. write special vertical cubic upscale version

  114. optimize C code (YV12 / minmax)

  115. add support for packed pixel YUV input & output

  116. add support for Y8 output

  117. optimize BGR24 & BGR32

  118. add BGR4 output support

  119. write special BGR->BGR scaler

  120. */

  121. 

  122. DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_4)[2][8]={

  123. {  1,   3,   1,   3,   1,   3,   1,   3, },

  124. {  2,   0,   2,   0,   2,   0,   2,   0, },

  125. };

  126. 

  127. DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_8)[2][8]={

  128. {  6,   2,   6,   2,   6,   2,   6,   2, },

  129. {  0,   4,   0,   4,   0,   4,   0,   4, },

  130. };

  131. 

  132. DECLARE_ALIGNED(8, const uint8_t, dither_4x4_16)[4][8]={

  133. {  8,   4,  11,   7,   8,   4,  11,   7, },

  134. {  2,  14,   1,  13,   2,  14,   1,  13, },

  135. { 10,   6,   9,   5,  10,   6,   9,   5, },

  136. {  0,  12,   3,  15,   0,  12,   3,  15, },

  137. };

  138. 

  139. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_32)[8][8]={

  140. { 17,   9,  23,  15,  16,   8,  22,  14, },

  141. {  5,  29,   3,  27,   4,  28,   2,  26, },

  142. { 21,  13,  19,  11,  20,  12,  18,  10, },

  143. {  0,  24,   6,  30,   1,  25,   7,  31, },

  144. { 16,   8,  22,  14,  17,   9,  23,  15, },

  145. {  4,  28,   2,  26,   5,  29,   3,  27, },

  146. { 20,  12,  18,  10,  21,  13,  19,  11, },

  147. {  1,  25,   7,  31,   0,  24,   6,  30, },

  148. };

  149. 

  150. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_73)[8][8]={

  151. {  0,  55,  14,  68,   3,  58,  17,  72, },

  152. { 37,  18,  50,  32,  40,  22,  54,  35, },

  153. {  9,  64,   5,  59,  13,  67,   8,  63, },

  154. { 46,  27,  41,  23,  49,  31,  44,  26, },

  155. {  2,  57,  16,  71,   1,  56,  15,  70, },

  156. { 39,  21,  52,  34,  38,  19,  51,  33, },

  157. { 11,  66,   7,  62,  10,  65,   6,  60, },

  158. { 48,  30,  43,  25,  47,  29,  42,  24, },

  159. };

  160. 

  161. #if 1

  162. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={

  163. {117,  62, 158, 103, 113,  58, 155, 100, },

  164. { 34, 199,  21, 186,  31, 196,  17, 182, },

  165. {144,  89, 131,  76, 141,  86, 127,  72, },

  166. {  0, 165,  41, 206,  10, 175,  52, 217, },

  167. {110,  55, 151,  96, 120,  65, 162, 107, },

  168. { 28, 193,  14, 179,  38, 203,  24, 189, },

  169. {138,  83, 124,  69, 148,  93, 134,  79, },

  170. {  7, 172,  48, 213,   3, 168,  45, 210, },

  171. };

  172. #elif 1

  173. // tries to correct a gamma of 1.5

  174. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={

  175. {  0, 143,  18, 200,   2, 156,  25, 215, },

  176. { 78,  28, 125,  64,  89,  36, 138,  74, },

  177. { 10, 180,   3, 161,  16, 195,   8, 175, },

  178. {109,  51,  93,  38, 121,  60, 105,  47, },

  179. {  1, 152,  23, 210,   0, 147,  20, 205, },

  180. { 85,  33, 134,  71,  81,  30, 130,  67, },

  181. { 14, 190,   6, 171,  12, 185,   5, 166, },

  182. {117,  57, 101,  44, 113,  54,  97,  41, },

  183. };

  184. #elif 1

  185. // tries to correct a gamma of 2.0

  186. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={

  187. {  0, 124,   8, 193,   0, 140,  12, 213, },

  188. { 55,  14, 104,  42,  66,  19, 119,  52, },

  189. {  3, 168,   1, 145,   6, 187,   3, 162, },

  190. { 86,  31,  70,  21,  99,  39,  82,  28, },

  191. {  0, 134,  11, 206,   0, 129,   9, 200, },

  192. { 62,  17, 114,  48,  58,  16, 109,  45, },

  193. {  5, 181,   2, 157,   4, 175,   1, 151, },

  194. { 95,  36,  78,  26,  90,  34,  74,  24, },

  195. };

  196. #else

  197. // tries to correct a gamma of 2.5

  198. DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={

  199. {  0, 107,   3, 187,   0, 125,   6, 212, },

  200. { 39,   7,  86,  28,  49,  11, 102,  36, },

  201. {  1, 158,   0, 131,   3, 180,   1, 151, },

  202. { 68,  19,  52,  12,  81,  25,  64,  17, },

  203. {  0, 119,   5, 203,   0, 113,   4, 195, },

  204. { 45,   9,  96,  33,  42,   8,  91,  30, },

  205. {  2, 172,   1, 144,   2, 165,   0, 137, },

  206. { 77,  23,  60,  15,  72,  21,  56,  14, },

  207. };

  208. #endif

  209. 

  210. DECLARE_ALIGNED(8, const uint8_t, dithers)[8][8][8]={

  211. {

  212.   {   0,  1,  0,  1,  0,  1,  0,  1,},

  213.   {   1,  0,  1,  0,  1,  0,  1,  0,},

  214.   {   0,  1,  0,  1,  0,  1,  0,  1,},

  215.   {   1,  0,  1,  0,  1,  0,  1,  0,},

  216.   {   0,  1,  0,  1,  0,  1,  0,  1,},

  217.   {   1,  0,  1,  0,  1,  0,  1,  0,},

  218.   {   0,  1,  0,  1,  0,  1,  0,  1,},

  219.   {   1,  0,  1,  0,  1,  0,  1,  0,},

  220. },{

  221.   {   1,  2,  1,  2,  1,  2,  1,  2,},

  222.   {   3,  0,  3,  0,  3,  0,  3,  0,},

  223.   {   1,  2,  1,  2,  1,  2,  1,  2,},

  224.   {   3,  0,  3,  0,  3,  0,  3,  0,},

  225.   {   1,  2,  1,  2,  1,  2,  1,  2,},

  226.   {   3,  0,  3,  0,  3,  0,  3,  0,},

  227.   {   1,  2,  1,  2,  1,  2,  1,  2,},

  228.   {   3,  0,  3,  0,  3,  0,  3,  0,},

  229. },{

  230.   {   2,  4,  3,  5,  2,  4,  3,  5,},

  231.   {   6,  0,  7,  1,  6,  0,  7,  1,},

  232.   {   3,  5,  2,  4,  3,  5,  2,  4,},

  233.   {   7,  1,  6,  0,  7,  1,  6,  0,},

  234.   {   2,  4,  3,  5,  2,  4,  3,  5,},

  235.   {   6,  0,  7,  1,  6,  0,  7,  1,},

  236.   {   3,  5,  2,  4,  3,  5,  2,  4,},

  237.   {   7,  1,  6,  0,  7,  1,  6,  0,},

  238. },{

  239.   {   4,  8,  7, 11,  4,  8,  7, 11,},

  240.   {  12,  0, 15,  3, 12,  0, 15,  3,},

  241.   {   6, 10,  5,  9,  6, 10,  5,  9,},

  242.   {  14,  2, 13,  1, 14,  2, 13,  1,},

  243.   {   4,  8,  7, 11,  4,  8,  7, 11,},

  244.   {  12,  0, 15,  3, 12,  0, 15,  3,},

  245.   {   6, 10,  5,  9,  6, 10,  5,  9,},

  246.   {  14,  2, 13,  1, 14,  2, 13,  1,},

  247. },{

  248.   {   9, 17, 15, 23,  8, 16, 14, 22,},

  249.   {  25,  1, 31,  7, 24,  0, 30,  6,},

  250.   {  13, 21, 11, 19, 12, 20, 10, 18,},

  251.   {  29,  5, 27,  3, 28,  4, 26,  2,},

  252.   {   8, 16, 14, 22,  9, 17, 15, 23,},

  253.   {  24,  0, 30,  6, 25,  1, 31,  7,},

  254.   {  12, 20, 10, 18, 13, 21, 11, 19,},

  255.   {  28,  4, 26,  2, 29,  5, 27,  3,},

  256. },{

  257.   {  18, 34, 30, 46, 17, 33, 29, 45,},

  258.   {  50,  2, 62, 14, 49,  1, 61, 13,},

  259.   {  26, 42, 22, 38, 25, 41, 21, 37,},

  260.   {  58, 10, 54,  6, 57,  9, 53,  5,},

  261.   {  16, 32, 28, 44, 19, 35, 31, 47,},

  262.   {  48,  0, 60, 12, 51,  3, 63, 15,},

  263.   {  24, 40, 20, 36, 27, 43, 23, 39,},

  264.   {  56,  8, 52,  4, 59, 11, 55,  7,},

  265. },{

  266.   {  18, 34, 30, 46, 17, 33, 29, 45,},

  267.   {  50,  2, 62, 14, 49,  1, 61, 13,},

  268.   {  26, 42, 22, 38, 25, 41, 21, 37,},

  269.   {  58, 10, 54,  6, 57,  9, 53,  5,},

  270.   {  16, 32, 28, 44, 19, 35, 31, 47,},

  271.   {  48,  0, 60, 12, 51,  3, 63, 15,},

  272.   {  24, 40, 20, 36, 27, 43, 23, 39,},

  273.   {  56,  8, 52,  4, 59, 11, 55,  7,},

  274. },{

  275.   {  36, 68, 60, 92, 34, 66, 58, 90,},

  276.   { 100,  4,124, 28, 98,  2,122, 26,},

  277.   {  52, 84, 44, 76, 50, 82, 42, 74,},

  278.   { 116, 20,108, 12,114, 18,106, 10,},

  279.   {  32, 64, 56, 88, 38, 70, 62, 94,},

  280.   {  96,  0,120, 24,102,  6,126, 30,},

  281.   {  48, 80, 40, 72, 54, 86, 46, 78,},

  282.   { 112, 16,104,  8,118, 22,110, 14,},

  283. }};

  284. 

  285. uint16_t dither_scale[15][16]={

  286. {    2,    3,    3,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,},

  287. {    2,    3,    7,    7,   13,   13,   25,   25,   25,   25,   25,   25,   25,   25,   25,   25,},

  288. {    3,    3,    4,   15,   15,   29,   57,   57,   57,  113,  113,  113,  113,  113,  113,  113,},

  289. {    3,    4,    4,    5,   31,   31,   61,  121,  241,  241,  241,  241,  481,  481,  481,  481,},

  290. {    3,    4,    5,    5,    6,   63,   63,  125,  249,  497,  993,  993,  993,  993,  993, 1985,},

  291. {    3,    5,    6,    6,    6,    7,  127,  127,  253,  505, 1009, 2017, 4033, 4033, 4033, 4033,},

  292. {    3,    5,    6,    7,    7,    7,    8,  255,  255,  509, 1017, 2033, 4065, 8129,16257,16257,},

  293. {    3,    5,    6,    8,    8,    8,    8,    9,  511,  511, 1021, 2041, 4081, 8161,16321,32641,},

  294. {    3,    5,    7,    8,    9,    9,    9,    9,   10, 1023, 1023, 2045, 4089, 8177,16353,32705,},

  295. {    3,    5,    7,    8,   10,   10,   10,   10,   10,   11, 2047, 2047, 4093, 8185,16369,32737,},

  296. {    3,    5,    7,    8,   10,   11,   11,   11,   11,   11,   12, 4095, 4095, 8189,16377,32753,},

  297. {    3,    5,    7,    9,   10,   12,   12,   12,   12,   12,   12,   13, 8191, 8191,16381,32761,},

  298. {    3,    5,    7,    9,   10,   12,   13,   13,   13,   13,   13,   13,   14,16383,16383,32765,},

  299. {    3,    5,    7,    9,   10,   12,   14,   14,   14,   14,   14,   14,   14,   15,32767,32767,},

  300. {    3,    5,    7,    9,   11,   12,   14,   15,   15,   15,   15,   15,   15,   15,   16,65535,},

  301. };

  302. 

  303. static av_always_inline void yuv2yuvX16inC_template(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

  304.                                                     const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,

  305.                                                     const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest,

  306.                                                     int dstW, int chrDstW, int big_endian, int output_bits)

  307. {

  308.     //FIXME Optimize (just quickly written not optimized..)

  309.     int i;

  310.     int shift = 11 + 16 - output_bits;

  311. 

  312. #define output_pixel(pos, val) \

  313.     if (big_endian) { \

  314.         if (output_bits == 16) { \

  315.             AV_WB16(pos, av_clip_uint16(val >> shift)); \

  316.         } else { \

  317.             AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \

  318.         } \

  319.     } else { \

  320.         if (output_bits == 16) { \

  321.             AV_WL16(pos, av_clip_uint16(val >> shift)); \

  322.         } else { \

  323.             AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \

  324.         } \

  325.     }

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

  327.         int val = 1 << (26-output_bits);

  328.         int j;

  329. 

  330.         for (j = 0; j < lumFilterSize; j++)

  331.             val += lumSrc[j][i] * lumFilter[j];

  332. 

  333.         output_pixel(&dest[i], val);

  334.     }

  335. 

  336.     if (uDest) {

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

  338.             int u = 1 << (26-output_bits);

  339.             int v = 1 << (26-output_bits);

  340.             int j;

  341. 

  342.             for (j = 0; j < chrFilterSize; j++) {

  343.                 u += chrSrc[j][i       ] * chrFilter[j];

  344.                 v += chrSrc[j][i + VOFW] * chrFilter[j];

  345.             }

  346. 

  347.             output_pixel(&uDest[i], u);

  348.             output_pixel(&vDest[i], v);

  349.         }

  350.     }

  351. 

  352.     if (CONFIG_SWSCALE_ALPHA && aDest) {

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

  354.             int val = 1 << (26-output_bits);

  355.             int j;

  356. 

  357.             for (j = 0; j < lumFilterSize; j++)

  358.                 val += alpSrc[j][i] * lumFilter[j];

  359. 

  360.             output_pixel(&aDest[i], val);

  361.         }

  362.     }

  363. }

  364. 

  365. static inline void yuv2yuvX16inC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

  366.                                  const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,

  367.                                  const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest, int dstW, int chrDstW,

  368.                                  enum PixelFormat dstFormat)

  369. {

  370.     if (isNBPS(dstFormat)) {

  371.         const int depth = av_pix_fmt_descriptors[dstFormat].comp[0].depth_minus1+1;

  372.         yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,

  373.                               chrFilter, chrSrc, chrFilterSize,

  374.                               alpSrc,

  375.                               dest, uDest, vDest, aDest,

  376.                               dstW, chrDstW, isBE(dstFormat), depth);

  377.     } else {

  378.         if (isBE(dstFormat)) {

  379.             yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,

  380.                                    chrFilter, chrSrc, chrFilterSize,

  381.                                    alpSrc,

  382.                                    dest, uDest, vDest, aDest,

  383.                                    dstW, chrDstW, 1, 16);

  384.         } else {

  385.             yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,

  386.                                    chrFilter, chrSrc, chrFilterSize,

  387.                                    alpSrc,

  388.                                    dest, uDest, vDest, aDest,

  389.                                    dstW, chrDstW, 0, 16);

  390.         }

  391.     }

  392. }

  393. 

  394. static inline void yuv2yuvXinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

  395.                                const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,

  396.                                const int16_t **alpSrc, uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, int dstW, int chrDstW)

  397. {

  398.     //FIXME Optimize (just quickly written not optimized..)

  399.     int i;

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

  401.         int val=1<<18;

  402.         int j;

  403.         for (j=0; j<lumFilterSize; j++)

  404.             val += lumSrc[j][i] * lumFilter[j];

  405. 

  406.         dest[i]= av_clip_uint8(val>>19);

  407.     }

  408. 

  409.     if (uDest)

  410.         for (i=0; i<chrDstW; i++) {

  411.             int u=1<<18;

  412.             int v=1<<18;

  413.             int j;

  414.             for (j=0; j<chrFilterSize; j++) {

  415.                 u += chrSrc[j][i] * chrFilter[j];

  416.                 v += chrSrc[j][i + VOFW] * chrFilter[j];

  417.             }

  418. 

  419.             uDest[i]= av_clip_uint8(u>>19);

  420.             vDest[i]= av_clip_uint8(v>>19);

  421.         }

  422. 

  423.     if (CONFIG_SWSCALE_ALPHA && aDest)

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

  425.             int val=1<<18;

  426.             int j;

  427.             for (j=0; j<lumFilterSize; j++)

  428.                 val += alpSrc[j][i] * lumFilter[j];

  429. 

  430.             aDest[i]= av_clip_uint8(val>>19);

  431.         }

  432. 

  433. }

  434. 

  435. static inline void yuv2nv12XinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

  436.                                 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,

  437.                                 uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat)

  438. {

  439.     //FIXME Optimize (just quickly written not optimized..)

  440.     int i;

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

  442.         int val=1<<18;

  443.         int j;

  444.         for (j=0; j<lumFilterSize; j++)

  445.             val += lumSrc[j][i] * lumFilter[j];

  446. 

  447.         dest[i]= av_clip_uint8(val>>19);

  448.     }

  449. 

  450.     if (!uDest)

  451.         return;

  452. 

  453.     if (dstFormat == PIX_FMT_NV12)

  454.         for (i=0; i<chrDstW; i++) {

  455.             int u=1<<18;

  456.             int v=1<<18;

  457.             int j;

  458.             for (j=0; j<chrFilterSize; j++) {

  459.                 u += chrSrc[j][i] * chrFilter[j];

  460.                 v += chrSrc[j][i + VOFW] * chrFilter[j];

  461.             }

  462. 

  463.             uDest[2*i]= av_clip_uint8(u>>19);

  464.             uDest[2*i+1]= av_clip_uint8(v>>19);

  465.         }

  466.     else

  467.         for (i=0; i<chrDstW; i++) {

  468.             int u=1<<18;

  469.             int v=1<<18;

  470.             int j;

  471.             for (j=0; j<chrFilterSize; j++) {

  472.                 u += chrSrc[j][i] * chrFilter[j];

  473.                 v += chrSrc[j][i + VOFW] * chrFilter[j];

  474.             }

  475. 

  476.             uDest[2*i]= av_clip_uint8(v>>19);

  477.             uDest[2*i+1]= av_clip_uint8(u>>19);

  478.         }

  479. }

  480. 

  481. #define YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha) \

  482.     for (i=0; i<(dstW>>1); i++) {\

  483.         int j;\

  484.         int Y1 = 1<<18;\

  485.         int Y2 = 1<<18;\

  486.         int U  = 1<<18;\

  487.         int V  = 1<<18;\

  488.         int av_unused A1, A2;\

  489.         type av_unused *r, *b, *g;\

  490.         const int i2= 2*i;\

  491.         \

  492.         for (j=0; j<lumFilterSize; j++) {\

  493.             Y1 += lumSrc[j][i2] * lumFilter[j];\

  494.             Y2 += lumSrc[j][i2+1] * lumFilter[j];\

  495.         }\

  496.         for (j=0; j<chrFilterSize; j++) {\

  497.             U += chrSrc[j][i] * chrFilter[j];\

  498.             V += chrSrc[j][i+VOFW] * chrFilter[j];\

  499.         }\

  500.         Y1>>=19;\

  501.         Y2>>=19;\

  502.         U >>=19;\

  503.         V >>=19;\

  504.         if (alpha) {\

  505.             A1 = 1<<18;\

  506.             A2 = 1<<18;\

  507.             for (j=0; j<lumFilterSize; j++) {\

  508.                 A1 += alpSrc[j][i2  ] * lumFilter[j];\

  509.                 A2 += alpSrc[j][i2+1] * lumFilter[j];\

  510.             }\

  511.             A1>>=19;\

  512.             A2>>=19;\

  513.         }

  514. 

  515. #define YSCALE_YUV_2_PACKEDX_C(type,alpha) \

  516.         YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha)\

  517.         if ((Y1|Y2|U|V)&256) {\

  518.             if (Y1>255)   Y1=255; \

  519.             else if (Y1<0)Y1=0;   \

  520.             if (Y2>255)   Y2=255; \

  521.             else if (Y2<0)Y2=0;   \

  522.             if (U>255)    U=255;  \

  523.             else if (U<0) U=0;    \

  524.             if (V>255)    V=255;  \

  525.             else if (V<0) V=0;    \

  526.         }\

  527.         if (alpha && ((A1|A2)&256)) {\

  528.             A1=av_clip_uint8(A1);\

  529.             A2=av_clip_uint8(A2);\

  530.         }

  531. 

  532. #define YSCALE_YUV_2_PACKEDX_FULL_C(rnd,alpha) \

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

  534.         int j;\

  535.         int Y = 0;\

  536.         int U = -128<<19;\

  537.         int V = -128<<19;\

  538.         int av_unused A;\

  539.         int R,G,B;\

  540.         \

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

  542.             Y += lumSrc[j][i     ] * lumFilter[j];\

  543.         }\

  544.         for (j=0; j<chrFilterSize; j++) {\

  545.             U += chrSrc[j][i     ] * chrFilter[j];\

  546.             V += chrSrc[j][i+VOFW] * chrFilter[j];\

  547.         }\

  548.         Y >>=10;\

  549.         U >>=10;\

  550.         V >>=10;\

  551.         if (alpha) {\

  552.             A = rnd;\

  553.             for (j=0; j<lumFilterSize; j++)\

  554.                 A += alpSrc[j][i     ] * lumFilter[j];\

  555.             A >>=19;\

  556.             if (A&256)\

  557.                 A = av_clip_uint8(A);\

  558.         }

  559. 

  560. #define YSCALE_YUV_2_RGBX_FULL_C(rnd,alpha) \

  561.     YSCALE_YUV_2_PACKEDX_FULL_C(rnd>>3,alpha)\

  562.         Y-= c->yuv2rgb_y_offset;\

  563.         Y*= c->yuv2rgb_y_coeff;\

  564.         Y+= rnd;\

  565.         R= Y + V*c->yuv2rgb_v2r_coeff;\

  566.         G= Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;\

  567.         B= Y +                          U*c->yuv2rgb_u2b_coeff;\

  568.         if ((R|G|B)&(0xC0000000)) {\

  569.             if (R>=(256<<22))   R=(256<<22)-1; \

  570.             else if (R<0)R=0;   \

  571.             if (G>=(256<<22))   G=(256<<22)-1; \

  572.             else if (G<0)G=0;   \

  573.             if (B>=(256<<22))   B=(256<<22)-1; \

  574.             else if (B<0)B=0;   \

  575.         }

  576. 

  577. #define YSCALE_YUV_2_GRAY16_C \

  578.     for (i=0; i<(dstW>>1); i++) {\

  579.         int j;\

  580.         int Y1 = 1<<18;\

  581.         int Y2 = 1<<18;\

  582.         int U  = 1<<18;\

  583.         int V  = 1<<18;\

  584.         \

  585.         const int i2= 2*i;\

  586.         \

  587.         for (j=0; j<lumFilterSize; j++) {\

  588.             Y1 += lumSrc[j][i2] * lumFilter[j];\

  589.             Y2 += lumSrc[j][i2+1] * lumFilter[j];\

  590.         }\

  591.         Y1>>=11;\

  592.         Y2>>=11;\

  593.         if ((Y1|Y2|U|V)&65536) {\

  594.             if (Y1>65535)   Y1=65535; \

  595.             else if (Y1<0)Y1=0;   \

  596.             if (Y2>65535)   Y2=65535; \

  597.             else if (Y2<0)Y2=0;   \

  598.         }

  599. 

  600. #define YSCALE_YUV_2_RGBX_C(type,alpha) \

  601.     YSCALE_YUV_2_PACKEDX_C(type,alpha)  /* FIXME fix tables so that clipping is not needed and then use _NOCLIP*/\

  602.     r = (type *)c->table_rV[V];   \

  603.     g = (type *)(c->table_gU[U] + c->table_gV[V]); \

  604.     b = (type *)c->table_bU[U];

  605. 

  606. #define YSCALE_YUV_2_PACKED2_C(type,alpha)   \

  607.     for (i=0; i<(dstW>>1); i++) { \

  608.         const int i2= 2*i;       \

  609.         int Y1= (buf0[i2  ]*yalpha1+buf1[i2  ]*yalpha)>>19;           \

  610.         int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19;           \

  611.         int U= (uvbuf0[i     ]*uvalpha1+uvbuf1[i     ]*uvalpha)>>19;  \

  612.         int V= (uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19;  \

  613.         type av_unused *r, *b, *g;                                    \

  614.         int av_unused A1, A2;                                         \

  615.         if (alpha) {\

  616.             A1= (abuf0[i2  ]*yalpha1+abuf1[i2  ]*yalpha)>>19;         \

  617.             A2= (abuf0[i2+1]*yalpha1+abuf1[i2+1]*yalpha)>>19;         \

  618.         }

  619. 

  620. #define YSCALE_YUV_2_GRAY16_2_C   \

  621.     for (i=0; i<(dstW>>1); i++) { \

  622.         const int i2= 2*i;       \

  623.         int Y1= (buf0[i2  ]*yalpha1+buf1[i2  ]*yalpha)>>11;           \

  624.         int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>11;

  625. 

  626. #define YSCALE_YUV_2_RGB2_C(type,alpha) \

  627.     YSCALE_YUV_2_PACKED2_C(type,alpha)\

  628.     r = (type *)c->table_rV[V];\

  629.     g = (type *)(c->table_gU[U] + c->table_gV[V]);\

  630.     b = (type *)c->table_bU[U];

  631. 

  632. #define YSCALE_YUV_2_PACKED1_C(type,alpha) \

  633.     for (i=0; i<(dstW>>1); i++) {\

  634.         const int i2= 2*i;\

  635.         int Y1= buf0[i2  ]>>7;\

  636.         int Y2= buf0[i2+1]>>7;\

  637.         int U= (uvbuf1[i     ])>>7;\

  638.         int V= (uvbuf1[i+VOFW])>>7;\

  639.         type av_unused *r, *b, *g;\

  640.         int av_unused A1, A2;\

  641.         if (alpha) {\

  642.             A1= abuf0[i2  ]>>7;\

  643.             A2= abuf0[i2+1]>>7;\

  644.         }

  645. 

  646. #define YSCALE_YUV_2_GRAY16_1_C \

  647.     for (i=0; i<(dstW>>1); i++) {\

  648.         const int i2= 2*i;\

  649.         int Y1= buf0[i2  ]<<1;\

  650.         int Y2= buf0[i2+1]<<1;

  651. 

  652. #define YSCALE_YUV_2_RGB1_C(type,alpha) \

  653.     YSCALE_YUV_2_PACKED1_C(type,alpha)\

  654.     r = (type *)c->table_rV[V];\

  655.     g = (type *)(c->table_gU[U] + c->table_gV[V]);\

  656.     b = (type *)c->table_bU[U];

  657. 

  658. #define YSCALE_YUV_2_PACKED1B_C(type,alpha) \

  659.     for (i=0; i<(dstW>>1); i++) {\

  660.         const int i2= 2*i;\

  661.         int Y1= buf0[i2  ]>>7;\

  662.         int Y2= buf0[i2+1]>>7;\

  663.         int U= (uvbuf0[i     ] + uvbuf1[i     ])>>8;\

  664.         int V= (uvbuf0[i+VOFW] + uvbuf1[i+VOFW])>>8;\

  665.         type av_unused *r, *b, *g;\

  666.         int av_unused A1, A2;\

  667.         if (alpha) {\

  668.             A1= abuf0[i2  ]>>7;\

  669.             A2= abuf0[i2+1]>>7;\

  670.         }

  671. 

  672. #define YSCALE_YUV_2_RGB1B_C(type,alpha) \

  673.     YSCALE_YUV_2_PACKED1B_C(type,alpha)\

  674.     r = (type *)c->table_rV[V];\

  675.     g = (type *)(c->table_gU[U] + c->table_gV[V]);\

  676.     b = (type *)c->table_bU[U];

  677. 

  678. #define YSCALE_YUV_2_MONO2_C \

  679.     const uint8_t * const d128=dither_8x8_220[y&7];\

  680.     uint8_t *g= c->table_gU[128] + c->table_gV[128];\

  681.     for (i=0; i<dstW-7; i+=8) {\

  682.         int acc;\

  683.         acc =       g[((buf0[i  ]*yalpha1+buf1[i  ]*yalpha)>>19) + d128[0]];\

  684.         acc+= acc + g[((buf0[i+1]*yalpha1+buf1[i+1]*yalpha)>>19) + d128[1]];\

  685.         acc+= acc + g[((buf0[i+2]*yalpha1+buf1[i+2]*yalpha)>>19) + d128[2]];\

  686.         acc+= acc + g[((buf0[i+3]*yalpha1+buf1[i+3]*yalpha)>>19) + d128[3]];\

  687.         acc+= acc + g[((buf0[i+4]*yalpha1+buf1[i+4]*yalpha)>>19) + d128[4]];\

  688.         acc+= acc + g[((buf0[i+5]*yalpha1+buf1[i+5]*yalpha)>>19) + d128[5]];\

  689.         acc+= acc + g[((buf0[i+6]*yalpha1+buf1[i+6]*yalpha)>>19) + d128[6]];\

  690.         acc+= acc + g[((buf0[i+7]*yalpha1+buf1[i+7]*yalpha)>>19) + d128[7]];\

  691.         ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\

  692.         dest++;\

  693.     }

  694. 

  695. #define YSCALE_YUV_2_MONOX_C \

  696.     const uint8_t * const d128=dither_8x8_220[y&7];\

  697.     uint8_t *g= c->table_gU[128] + c->table_gV[128];\

  698.     int acc=0;\

  699.     for (i=0; i<dstW-1; i+=2) {\

  700.         int j;\

  701.         int Y1=1<<18;\

  702.         int Y2=1<<18;\

  703. \

  704.         for (j=0; j<lumFilterSize; j++) {\

  705.             Y1 += lumSrc[j][i] * lumFilter[j];\

  706.             Y2 += lumSrc[j][i+1] * lumFilter[j];\

  707.         }\

  708.         Y1>>=19;\

  709.         Y2>>=19;\

  710.         if ((Y1|Y2)&256) {\

  711.             if (Y1>255)   Y1=255;\

  712.             else if (Y1<0)Y1=0;\

  713.             if (Y2>255)   Y2=255;\

  714.             else if (Y2<0)Y2=0;\

  715.         }\

  716.         acc+= acc + g[Y1+d128[(i+0)&7]];\

  717.         acc+= acc + g[Y2+d128[(i+1)&7]];\

  718.         if ((i&7)==6) {\

  719.             ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\

  720.             dest++;\

  721.         }\

  722.     }

  723. 

  724. #define YSCALE_YUV_2_ANYRGB_C(func, func2, func_g16, func_monoblack)\

  725.     switch(c->dstFormat) {\

  726.     case PIX_FMT_RGB48BE:\

  727.     case PIX_FMT_RGB48LE:\

  728.         func(uint8_t,0)\

  729.             ((uint8_t*)dest)[ 0]= r[Y1];\

  730.             ((uint8_t*)dest)[ 1]= r[Y1];\

  731.             ((uint8_t*)dest)[ 2]= g[Y1];\

  732.             ((uint8_t*)dest)[ 3]= g[Y1];\

  733.             ((uint8_t*)dest)[ 4]= b[Y1];\

  734.             ((uint8_t*)dest)[ 5]= b[Y1];\

  735.             ((uint8_t*)dest)[ 6]= r[Y2];\

  736.             ((uint8_t*)dest)[ 7]= r[Y2];\

  737.             ((uint8_t*)dest)[ 8]= g[Y2];\

  738.             ((uint8_t*)dest)[ 9]= g[Y2];\

  739.             ((uint8_t*)dest)[10]= b[Y2];\

  740.             ((uint8_t*)dest)[11]= b[Y2];\

  741.             dest+=12;\

  742.         }\

  743.         break;\

  744.     case PIX_FMT_BGR48BE:\

  745.     case PIX_FMT_BGR48LE:\

  746.         func(uint8_t,0)\

  747.             ((uint8_t*)dest)[ 0] = ((uint8_t*)dest)[ 1] = b[Y1];\

  748.             ((uint8_t*)dest)[ 2] = ((uint8_t*)dest)[ 3] = g[Y1];\

  749.             ((uint8_t*)dest)[ 4] = ((uint8_t*)dest)[ 5] = r[Y1];\

  750.             ((uint8_t*)dest)[ 6] = ((uint8_t*)dest)[ 7] = b[Y2];\

  751.             ((uint8_t*)dest)[ 8] = ((uint8_t*)dest)[ 9] = g[Y2];\

  752.             ((uint8_t*)dest)[10] = ((uint8_t*)dest)[11] = r[Y2];\

  753.             dest+=12;\

  754.         }\

  755.         break;\

  756.     case PIX_FMT_RGBA:\

  757.     case PIX_FMT_BGRA:\

  758.         if (CONFIG_SMALL) {\

  759.             int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\

  760.             func(uint32_t,needAlpha)\

  761.                 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? (A1<<24) : 0);\

  762.                 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? (A2<<24) : 0);\

  763.             }\

  764.         } else {\

  765.             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\

  766.                 func(uint32_t,1)\

  767.                     ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (A1<<24);\

  768.                     ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (A2<<24);\

  769.                 }\

  770.             } else {\

  771.                 func(uint32_t,0)\

  772.                     ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\

  773.                     ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\

  774.                 }\

  775.             }\

  776.         }\

  777.         break;\

  778.     case PIX_FMT_ARGB:\

  779.     case PIX_FMT_ABGR:\

  780.         if (CONFIG_SMALL) {\

  781.             int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\

  782.             func(uint32_t,needAlpha)\

  783.                 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? A1 : 0);\

  784.                 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? A2 : 0);\

  785.             }\

  786.         } else {\

  787.             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\

  788.                 func(uint32_t,1)\

  789.                     ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + A1;\

  790.                     ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + A2;\

  791.                 }\

  792.             } else {\

  793.                 func(uint32_t,0)\

  794.                     ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\

  795.                     ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\

  796.                 }\

  797.             }\

  798.         }                \

  799.         break;\

  800.     case PIX_FMT_RGB24:\

  801.         func(uint8_t,0)\

  802.             ((uint8_t*)dest)[0]= r[Y1];\

  803.             ((uint8_t*)dest)[1]= g[Y1];\

  804.             ((uint8_t*)dest)[2]= b[Y1];\

  805.             ((uint8_t*)dest)[3]= r[Y2];\

  806.             ((uint8_t*)dest)[4]= g[Y2];\

  807.             ((uint8_t*)dest)[5]= b[Y2];\

  808.             dest+=6;\

  809.         }\

  810.         break;\

  811.     case PIX_FMT_BGR24:\

  812.         func(uint8_t,0)\

  813.             ((uint8_t*)dest)[0]= b[Y1];\

  814.             ((uint8_t*)dest)[1]= g[Y1];\

  815.             ((uint8_t*)dest)[2]= r[Y1];\

  816.             ((uint8_t*)dest)[3]= b[Y2];\

  817.             ((uint8_t*)dest)[4]= g[Y2];\

  818.             ((uint8_t*)dest)[5]= r[Y2];\

  819.             dest+=6;\

  820.         }\

  821.         break;\

  822.     case PIX_FMT_RGB565BE:\

  823.     case PIX_FMT_RGB565LE:\

  824.     case PIX_FMT_BGR565BE:\

  825.     case PIX_FMT_BGR565LE:\

  826.         {\

  827.             const int dr1= dither_2x2_8[y&1    ][0];\

  828.             const int dg1= dither_2x2_4[y&1    ][0];\

  829.             const int db1= dither_2x2_8[(y&1)^1][0];\

  830.             const int dr2= dither_2x2_8[y&1    ][1];\

  831.             const int dg2= dither_2x2_4[y&1    ][1];\

  832.             const int db2= dither_2x2_8[(y&1)^1][1];\

  833.             func(uint16_t,0)\

  834.                 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\

  835.                 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\

  836.             }\

  837.         }\

  838.         break;\

  839.     case PIX_FMT_RGB555BE:\

  840.     case PIX_FMT_RGB555LE:\

  841.     case PIX_FMT_BGR555BE:\

  842.     case PIX_FMT_BGR555LE:\

  843.         {\

  844.             const int dr1= dither_2x2_8[y&1    ][0];\

  845.             const int dg1= dither_2x2_8[y&1    ][1];\

  846.             const int db1= dither_2x2_8[(y&1)^1][0];\

  847.             const int dr2= dither_2x2_8[y&1    ][1];\

  848.             const int dg2= dither_2x2_8[y&1    ][0];\

  849.             const int db2= dither_2x2_8[(y&1)^1][1];\

  850.             func(uint16_t,0)\

  851.                 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\

  852.                 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\

  853.             }\

  854.         }\

  855.         break;\

  856.     case PIX_FMT_RGB444BE:\

  857.     case PIX_FMT_RGB444LE:\

  858.     case PIX_FMT_BGR444BE:\

  859.     case PIX_FMT_BGR444LE:\

  860.         {\

  861.             const int dr1= dither_4x4_16[y&3    ][0];\

  862.             const int dg1= dither_4x4_16[y&3    ][1];\

  863.             const int db1= dither_4x4_16[(y&3)^3][0];\

  864.             const int dr2= dither_4x4_16[y&3    ][1];\

  865.             const int dg2= dither_4x4_16[y&3    ][0];\

  866.             const int db2= dither_4x4_16[(y&3)^3][1];\

  867.             func(uint16_t,0)\

  868.                 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\

  869.                 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\

  870.             }\

  871.         }\

  872.         break;\

  873.     case PIX_FMT_RGB8:\

  874.     case PIX_FMT_BGR8:\

  875.         {\

  876.             const uint8_t * const d64= dither_8x8_73[y&7];\

  877.             const uint8_t * const d32= dither_8x8_32[y&7];\

  878.             func(uint8_t,0)\

  879.                 ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]];\

  880.                 ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]];\

  881.             }\

  882.         }\

  883.         break;\

  884.     case PIX_FMT_RGB4:\

  885.     case PIX_FMT_BGR4:\

  886.         {\

  887.             const uint8_t * const d64= dither_8x8_73 [y&7];\

  888.             const uint8_t * const d128=dither_8x8_220[y&7];\

  889.             func(uint8_t,0)\

  890.                 ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]\

  891.                                  + ((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4);\

  892.             }\

  893.         }\

  894.         break;\

  895.     case PIX_FMT_RGB4_BYTE:\

  896.     case PIX_FMT_BGR4_BYTE:\

  897.         {\

  898.             const uint8_t * const d64= dither_8x8_73 [y&7];\

  899.             const uint8_t * const d128=dither_8x8_220[y&7];\

  900.             func(uint8_t,0)\

  901.                 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\

  902.                 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\

  903.             }\

  904.         }\

  905.         break;\

  906.     case PIX_FMT_MONOBLACK:\

  907.     case PIX_FMT_MONOWHITE:\

  908.         {\

  909.             func_monoblack\

  910.         }\

  911.         break;\

  912.     case PIX_FMT_YUYV422:\

  913.         func2\

  914.             ((uint8_t*)dest)[2*i2+0]= Y1;\

  915.             ((uint8_t*)dest)[2*i2+1]= U;\

  916.             ((uint8_t*)dest)[2*i2+2]= Y2;\

  917.             ((uint8_t*)dest)[2*i2+3]= V;\

  918.         }                \

  919.         break;\

  920.     case PIX_FMT_UYVY422:\

  921.         func2\

  922.             ((uint8_t*)dest)[2*i2+0]= U;\

  923.             ((uint8_t*)dest)[2*i2+1]= Y1;\

  924.             ((uint8_t*)dest)[2*i2+2]= V;\

  925.             ((uint8_t*)dest)[2*i2+3]= Y2;\

  926.         }                \

  927.         break;\

  928.     case PIX_FMT_GRAY16BE:\

  929.         func_g16\

  930.             ((uint8_t*)dest)[2*i2+0]= Y1>>8;\

  931.             ((uint8_t*)dest)[2*i2+1]= Y1;\

  932.             ((uint8_t*)dest)[2*i2+2]= Y2>>8;\

  933.             ((uint8_t*)dest)[2*i2+3]= Y2;\

  934.         }                \

  935.         break;\

  936.     case PIX_FMT_GRAY16LE:\

  937.         func_g16\

  938.             ((uint8_t*)dest)[2*i2+0]= Y1;\

  939.             ((uint8_t*)dest)[2*i2+1]= Y1>>8;\

  940.             ((uint8_t*)dest)[2*i2+2]= Y2;\

  941.             ((uint8_t*)dest)[2*i2+3]= Y2>>8;\

  942.         }                \

  943.         break;\

  944.     }

  945. 

  946. static inline void yuv2packedXinC(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

  947.                                   const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,

  948.                                   const int16_t **alpSrc, uint8_t *dest, int dstW, int y)

  949. {

  950.     int i;

  951.     YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGBX_C, YSCALE_YUV_2_PACKEDX_C(void,0), YSCALE_YUV_2_GRAY16_C, YSCALE_YUV_2_MONOX_C)

  952. }

  953. 

  954. static inline void yuv2rgbXinC_full(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

  955.                                     const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,

  956.                                     const int16_t **alpSrc, uint8_t *dest, int dstW, int y)

  957. {

  958.     int i;

  959.     int step= c->dstFormatBpp/8;

  960.     int aidx= 3;

  961. 

  962.     switch(c->dstFormat) {

  963.     case PIX_FMT_ARGB:

  964.         dest++;

  965.         aidx= 0;

  966.     case PIX_FMT_RGB24:

  967.         aidx--;

  968.     case PIX_FMT_RGBA:

  969.         if (CONFIG_SMALL) {

  970.             int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;

  971.             YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)

  972.                 dest[aidx]= needAlpha ? A : 255;

  973.                 dest[0]= R>>22;

  974.                 dest[1]= G>>22;

  975.                 dest[2]= B>>22;

  976.                 dest+= step;

  977.             }

  978.         } else {

  979.             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  980.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)

  981.                     dest[aidx]= A;

  982.                     dest[0]= R>>22;

  983.                     dest[1]= G>>22;

  984.                     dest[2]= B>>22;

  985.                     dest+= step;

  986.                 }

  987.             } else {

  988.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)

  989.                     dest[aidx]= 255;

  990.                     dest[0]= R>>22;

  991.                     dest[1]= G>>22;

  992.                     dest[2]= B>>22;

  993.                     dest+= step;

  994.                 }

  995.             }

  996.         }

  997.         break;

  998.     case PIX_FMT_ABGR:

  999.         dest++;

  1000.         aidx= 0;

  1001.     case PIX_FMT_BGR24:

  1002.         aidx--;

  1003.     case PIX_FMT_BGRA:

  1004.         if (CONFIG_SMALL) {

  1005.             int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;

  1006.             YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)

  1007.                 dest[aidx]= needAlpha ? A : 255;

  1008.                 dest[0]= B>>22;

  1009.                 dest[1]= G>>22;

  1010.                 dest[2]= R>>22;

  1011.                 dest+= step;

  1012.             }

  1013.         } else {

  1014.             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  1015.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)

  1016.                     dest[aidx]= A;

  1017.                     dest[0]= B>>22;

  1018.                     dest[1]= G>>22;

  1019.                     dest[2]= R>>22;

  1020.                     dest+= step;

  1021.                 }

  1022.             } else {

  1023.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)

  1024.                     dest[aidx]= 255;

  1025.                     dest[0]= B>>22;

  1026.                     dest[1]= G>>22;

  1027.                     dest[2]= R>>22;

  1028.                     dest+= step;

  1029.                 }

  1030.             }

  1031.         }

  1032.         break;

  1033.     default:

  1034.         assert(0);

  1035.     }

  1036. }

  1037. 

  1038. static void fillPlane(uint8_t* plane, int stride, int width, int height, int y, uint8_t val)

  1039. {

  1040.     int i;

  1041.     uint8_t *ptr = plane + stride*y;

  1042.     for (i=0; i<height; i++) {

  1043.         memset(ptr, val, width);

  1044.         ptr += stride;

  1045.     }

  1046. }

  1047. 

  1048. static inline void rgb48ToY(int16_t *dst, const uint8_t *src, long width,

  1049.                             uint32_t *unused)

  1050. {

  1051.     int i;

  1052.     for (i = 0; i < width; i++) {

  1053.         int r = src[i*6+0];

  1054.         int g = src[i*6+2];

  1055.         int b = src[i*6+4];

  1056. 

  1057.         dst[i] = (RY*r + GY*g + BY*b + (32<<(RGB2YUV_SHIFT-1)) + (1<<(RGB2YUV_SHIFT-7))) >> (RGB2YUV_SHIFT-6);

  1058.     }

  1059. }

  1060. 

  1061. static inline void rgb48ToUV(int16_t *dstU, int16_t *dstV,

  1062.                              const uint8_t *src1, const uint8_t *src2,

  1063.                              long width, uint32_t *unused)

  1064. {

  1065.     int i;

  1066.     assert(src1==src2);

  1067.     for (i = 0; i < width; i++) {

  1068.         int r = src1[6*i + 0];

  1069.         int g = src1[6*i + 2];

  1070.         int b = src1[6*i + 4];

  1071. 

  1072.         dstU[i] = (RU*r + GU*g + BU*b + (256<<(RGB2YUV_SHIFT-1)) + (1<<(RGB2YUV_SHIFT-7))) >> (RGB2YUV_SHIFT-6);

  1073.         dstV[i] = (RV*r + GV*g + BV*b + (256<<(RGB2YUV_SHIFT-1)) + (1<<(RGB2YUV_SHIFT-7))) >> (RGB2YUV_SHIFT-6);

  1074.     }

  1075. }

  1076. 

  1077. static inline void rgb48ToUV_half(int16_t *dstU, int16_t *dstV,

  1078.                                   const uint8_t *src1, const uint8_t *src2,

  1079.                                   long width, uint32_t *unused)

  1080. {

  1081.     int i;

  1082.     assert(src1==src2);

  1083.     for (i = 0; i < width; i++) {

  1084.         int r= src1[12*i + 0] + src1[12*i + 6];

  1085.         int g= src1[12*i + 2] + src1[12*i + 8];

  1086.         int b= src1[12*i + 4] + src1[12*i + 10];

  1087. 

  1088.         dstU[i]= (RU*r + GU*g + BU*b + (256<<(RGB2YUV_SHIFT)) + (1<<(RGB2YUV_SHIFT-6))) >> (RGB2YUV_SHIFT-5);

  1089.         dstV[i]= (RV*r + GV*g + BV*b + (256<<(RGB2YUV_SHIFT)) + (1<<(RGB2YUV_SHIFT-6))) >> (RGB2YUV_SHIFT-5);

  1090.     }

  1091. }

  1092. 

  1093. static inline void bgr48ToY(int16_t *dst, const uint8_t *src, long width,

  1094.                             uint32_t *unused)

  1095. {

  1096.     int i;

  1097.     for (i = 0; i < width; i++) {

  1098.         int b = src[i*6+0];

  1099.         int g = src[i*6+2];

  1100.         int r = src[i*6+4];

  1101. 

  1102.         dst[i] = (RY*r + GY*g + BY*b + (32<<(RGB2YUV_SHIFT-1)) + (1<<(RGB2YUV_SHIFT-7))) >> (RGB2YUV_SHIFT-6);

  1103.     }

  1104. }

  1105. 

  1106. static inline void bgr48ToUV(int16_t *dstU, int16_t *dstV,

  1107.                              const uint8_t *src1, const uint8_t *src2,

  1108.                              long width, uint32_t *unused)

  1109. {

  1110.     int i;

  1111.     for (i = 0; i < width; i++) {

  1112.         int b = src1[6*i + 0];

  1113.         int g = src1[6*i + 2];

  1114.         int r = src1[6*i + 4];

  1115. 

  1116.         dstU[i] = (RU*r + GU*g + BU*b + (256<<(RGB2YUV_SHIFT-1)) + (1<<(RGB2YUV_SHIFT-7))) >> (RGB2YUV_SHIFT-6);

  1117.         dstV[i] = (RV*r + GV*g + BV*b + (256<<(RGB2YUV_SHIFT-1)) + (1<<(RGB2YUV_SHIFT-7))) >> (RGB2YUV_SHIFT-6);

  1118.     }

  1119. }

  1120. 

  1121. static inline void bgr48ToUV_half(int16_t *dstU, int16_t *dstV,

  1122.                                   const uint8_t *src1, const uint8_t *src2,

  1123.                                   long width, uint32_t *unused)

  1124. {

  1125.     int i;

  1126.     for (i = 0; i < width; i++) {

  1127.         int b= src1[12*i + 0] + src1[12*i + 6];

  1128.         int g= src1[12*i + 2] + src1[12*i + 8];

  1129.         int r= src1[12*i + 4] + src1[12*i + 10];

  1130. 

  1131.         dstU[i]= (RU*r + GU*g + BU*b + (256<<(RGB2YUV_SHIFT)) + (1<<(RGB2YUV_SHIFT-6))) >> (RGB2YUV_SHIFT-5);

  1132.         dstV[i]= (RV*r + GV*g + BV*b + (256<<(RGB2YUV_SHIFT)) + (1<<(RGB2YUV_SHIFT-6))) >> (RGB2YUV_SHIFT-5);

  1133.     }

  1134. }

  1135. 

  1136. #define BGR2Y(type, name, shr, shg, shb, maskr, maskg, maskb, RY, GY, BY, S)\

  1137. static inline void name(int16_t *dst, const uint8_t *src, long width, uint32_t *unused)\

  1138. {\

  1139.     int i;\

  1140.     for (i=0; i<width; i++) {\

  1141.         int b= (((const type*)src)[i]>>shb)&maskb;\

  1142.         int g= (((const type*)src)[i]>>shg)&maskg;\

  1143.         int r= (((const type*)src)[i]>>shr)&maskr;\

  1144. \

  1145.         dst[i]= (((RY)*r + (GY)*g + (BY)*b + (32<<((S)-1)) + (1<<(S-7)))>>((S)-6));\

  1146.     }\

  1147. }

  1148. 

  1149. BGR2Y(uint32_t, bgr32ToY,16, 0, 0, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY   , BY<< 8, RGB2YUV_SHIFT+8)

  1150. BGR2Y(uint32_t,bgr321ToY,16,16, 0, 0xFF00, 0x00FF, 0xFF00, RY    , GY<<8, BY    , RGB2YUV_SHIFT+8)

  1151. BGR2Y(uint32_t, rgb32ToY, 0, 0,16, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY   , BY<< 8, RGB2YUV_SHIFT+8)

  1152. BGR2Y(uint32_t,rgb321ToY, 0,16,16, 0xFF00, 0x00FF, 0xFF00, RY    , GY<<8, BY    , RGB2YUV_SHIFT+8)

  1153. BGR2Y(uint16_t, bgr16ToY, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RY<<11, GY<<5, BY    , RGB2YUV_SHIFT+8)

  1154. BGR2Y(uint16_t, bgr15ToY, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RY<<10, GY<<5, BY    , RGB2YUV_SHIFT+7)

  1155. BGR2Y(uint16_t, rgb16ToY, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RY    , GY<<5, BY<<11, RGB2YUV_SHIFT+8)

  1156. BGR2Y(uint16_t, rgb15ToY, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RY    , GY<<5, BY<<10, RGB2YUV_SHIFT+7)

  1157. 

  1158. static inline void abgrToA(int16_t *dst, const uint8_t *src, long width, uint32_t *unused)

  1159. {

  1160.     int i;

  1161.     for (i=0; i<width; i++) {

  1162.         dst[i]= src[4*i]<<6;

  1163.     }

  1164. }

  1165. 

  1166. #define BGR2UV(type, name, shr, shg, shb, shp, maskr, maskg, maskb, RU, GU, BU, RV, GV, BV, S) \

  1167. static inline void name(int16_t *dstU, int16_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\

  1168. {\

  1169.     int i;\

  1170.     for (i=0; i<width; i++) {\

  1171.         int b= ((((const type*)src)[i]>>shp)&maskb)>>shb;\

  1172.         int g= ((((const type*)src)[i]>>shp)&maskg)>>shg;\

  1173.         int r= ((((const type*)src)[i]>>shp)&maskr)>>shr;\

  1174. \

  1175.         dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (256<<((S)-1)) + (1<<(S-7)))>>((S)-6);\

  1176.         dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (256<<((S)-1)) + (1<<(S-7)))>>((S)-6);\

  1177.     }\

  1178. }\

  1179. static inline void name ## _half(int16_t *dstU, int16_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\

  1180. {\

  1181.     int i;\

  1182.     for (i=0; i<width; i++) {\

  1183.         int pix0= ((const type*)src)[2*i+0]>>shp;\

  1184.         int pix1= ((const type*)src)[2*i+1]>>shp;\

  1185.         int g= (pix0&~(maskr|maskb))+(pix1&~(maskr|maskb));\

  1186.         int b= ((pix0+pix1-g)&(maskb|(2*maskb)))>>shb;\

  1187.         int r= ((pix0+pix1-g)&(maskr|(2*maskr)))>>shr;\

  1188.         g&= maskg|(2*maskg);\

  1189. \

  1190.         g>>=shg;\

  1191. \

  1192.         dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (256U<<(S)) + (1<<(S-6)))>>((S)-6+1);\

  1193.         dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (256U<<(S)) + (1<<(S-6)))>>((S)-6+1);\

  1194.     }\

  1195. }

  1196. 

  1197. BGR2UV(uint32_t, bgr32ToUV,16, 0, 0, 0, 0xFF0000, 0xFF00,   0x00FF, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1198. BGR2UV(uint32_t,bgr321ToUV,16, 0, 0, 8, 0xFF0000, 0xFF00,   0x00FF, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1199. BGR2UV(uint32_t, rgb32ToUV, 0, 0,16, 0,   0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1200. BGR2UV(uint32_t,rgb321ToUV, 0, 0,16, 8,   0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1201. BGR2UV(uint16_t, bgr16ToUV, 0, 0, 0, 0,   0x001F, 0x07E0,   0xF800, RU<<11, GU<<5, BU    , RV<<11, GV<<5, BV    , RGB2YUV_SHIFT+8)

  1202. BGR2UV(uint16_t, bgr15ToUV, 0, 0, 0, 0,   0x001F, 0x03E0,   0x7C00, RU<<10, GU<<5, BU    , RV<<10, GV<<5, BV    , RGB2YUV_SHIFT+7)

  1203. BGR2UV(uint16_t, rgb16ToUV, 0, 0, 0, 0,   0xF800, 0x07E0,   0x001F, RU    , GU<<5, BU<<11, RV    , GV<<5, BV<<11, RGB2YUV_SHIFT+8)

  1204. BGR2UV(uint16_t, rgb15ToUV, 0, 0, 0, 0,   0x7C00, 0x03E0,   0x001F, RU    , GU<<5, BU<<10, RV    , GV<<5, BV<<10, RGB2YUV_SHIFT+7)

  1205. 

  1206. static inline void palToA(int16_t *dst, const uint8_t *src, long width, uint32_t *pal)

  1207. {

  1208.     int i;

  1209.     for (i=0; i<width; i++) {

  1210.         int d= src[i];

  1211. 

  1212.         dst[i]= (pal[d] >> 24)<<6;

  1213.     }

  1214. }

  1215. 

  1216. static inline void palToY(int16_t *dst, const uint8_t *src, long width, uint32_t *pal)

  1217. {

  1218.     int i;

  1219.     for (i=0; i<width; i++) {

  1220.         int d= src[i];

  1221. 

  1222.         dst[i]= (pal[d] & 0xFF)<<6;

  1223.     }

  1224. }

  1225. 

  1226. static inline void palToUV(uint16_t *dstU, int16_t *dstV,

  1227.                            const uint8_t *src1, const uint8_t *src2,

  1228.                            long width, uint32_t *pal)

  1229. {

  1230.     int i;

  1231.     assert(src1 == src2);

  1232.     for (i=0; i<width; i++) {

  1233.         int p= pal[src1[i]];

  1234. 

  1235.         dstU[i]= (uint8_t)(p>> 8)<<6;

  1236.         dstV[i]= (uint8_t)(p>>16)<<6;

  1237.     }

  1238. }

  1239. 

  1240. static inline void monowhite2Y(int16_t *dst, const uint8_t *src, long width, uint32_t *unused)

  1241. {

  1242.     int i, j;

  1243.     for (i=0; i<width/8; i++) {

  1244.         int d= ~src[i];

  1245.         for(j=0; j<8; j++)

  1246.             dst[8*i+j]= ((d>>(7-j))&1)*16383;

  1247.     }

  1248. }

  1249. 

  1250. static inline void monoblack2Y(int16_t *dst, const uint8_t *src, long width, uint32_t *unused)

  1251. {

  1252.     int i, j;

  1253.     for (i=0; i<width/8; i++) {

  1254.         int d= src[i];

  1255.         for(j=0; j<8; j++)

  1256.             dst[8*i+j]= ((d>>(7-j))&1)*16383;

  1257.     }

  1258. }

  1259. 

  1260. //Note: we have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW+MMX2 one

  1261. //Plain C versions

  1262. 

  1263. #define COMPILE_TEMPLATE_MMX2 0

  1264. #define COMPILE_TEMPLATE_ALTIVEC 0

  1265. 

  1266. #include "swscale_template.c"

  1267. 

  1268. #if HAVE_ALTIVEC

  1269. #undef RENAME

  1270. #undef COMPILE_TEMPLATE_ALTIVEC

  1271. #define COMPILE_TEMPLATE_ALTIVEC 1

  1272. #define RENAME(a) a ## _altivec

  1273. #include "ppc/swscale_template.c"

  1274. #endif

  1275. 

  1276. //MMX versions

  1277. #if HAVE_MMX

  1278. #undef RENAME

  1279. #undef COMPILE_TEMPLATE_MMX2

  1280. #define COMPILE_TEMPLATE_MMX2 0

  1281. #define RENAME(a) a ## _MMX

  1282. #include "x86/swscale_template.c"

  1283. #endif

  1284. 

  1285. //MMX2 versions

  1286. #if HAVE_MMX2

  1287. #undef RENAME

  1288. #undef COMPILE_TEMPLATE_MMX2

  1289. #define COMPILE_TEMPLATE_MMX2 1

  1290. #define RENAME(a) a ## _MMX2

  1291. #include "x86/swscale_template.c"

  1292. #endif

  1293. 

  1294. SwsFunc ff_getSwsFunc(SwsContext *c)

  1295. {

  1296.     int cpu_flags = av_get_cpu_flags();

  1297. 

  1298.     sws_init_swScale_c(c);

  1299. 

  1300. #if HAVE_MMX

  1301.     if (cpu_flags & AV_CPU_FLAG_MMX)

  1302.         sws_init_swScale_MMX(c);

  1303. #endif

  1304. #if HAVE_MMX2

  1305.     if (cpu_flags & AV_CPU_FLAG_MMX2)

  1306.         sws_init_swScale_MMX2(c);

  1307. #endif

  1308. #if HAVE_ALTIVEC

  1309.     if (cpu_flags & AV_CPU_FLAG_ALTIVEC)

  1310.         sws_init_swScale_altivec(c);

  1311. #endif

  1312. 

  1313.     return swScale_c;

  1314. }

  1315. 

  1316. static void copyPlane(const uint8_t *src, int srcStride,

  1317.                       int srcSliceY, int srcSliceH, int width,

  1318.                       uint8_t *dst, int dstStride)

  1319. {

  1320.     dst += dstStride * srcSliceY;

  1321.     if (dstStride == srcStride && srcStride > 0) {

  1322.         memcpy(dst, src, srcSliceH * dstStride);

  1323.     } else {

  1324.         int i;

  1325.         for (i=0; i<srcSliceH; i++) {

  1326.             memcpy(dst, src, width);

  1327.             src += srcStride;

  1328.             dst += dstStride;

  1329.         }

  1330.     }

  1331. }

  1332. 

  1333. static int planarToNv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1334.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1335. {

  1336.     uint8_t *dst = dstParam[1] + dstStride[1]*srcSliceY/2;

  1337. 

  1338.     copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,

  1339.               dstParam[0], dstStride[0]);

  1340. 

  1341.     if (c->dstFormat == PIX_FMT_NV12)

  1342.         interleaveBytes(src[1], src[2], dst, c->srcW/2, srcSliceH/2, srcStride[1], srcStride[2], dstStride[0]);

  1343.     else

  1344.         interleaveBytes(src[2], src[1], dst, c->srcW/2, srcSliceH/2, srcStride[2], srcStride[1], dstStride[0]);

  1345. 

  1346.     return srcSliceH;

  1347. }

  1348. 

  1349. static int planarToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1350.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1351. {

  1352.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1353. 

  1354.     yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);

  1355. 

  1356.     return srcSliceH;

  1357. }

  1358. 

  1359. static int planarToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1360.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1361. {

  1362.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1363. 

  1364.     yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);

  1365. 

  1366.     return srcSliceH;

  1367. }

  1368. 

  1369. static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1370.                                 int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1371. {

  1372.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1373. 

  1374.     yuv422ptoyuy2(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);

  1375. 

  1376.     return srcSliceH;

  1377. }

  1378. 

  1379. static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1380.                                 int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1381. {

  1382.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1383. 

  1384.     yuv422ptouyvy(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);

  1385. 

  1386.     return srcSliceH;

  1387. }

  1388. 

  1389. static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1390.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1391. {

  1392.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1393.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;

  1394.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;

  1395. 

  1396.     yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1397. 

  1398.     if (dstParam[3])

  1399.         fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1400. 

  1401.     return srcSliceH;

  1402. }

  1403. 

  1404. static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1405.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1406. {

  1407.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1408.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;

  1409.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;

  1410. 

  1411.     yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1412. 

  1413.     return srcSliceH;

  1414. }

  1415. 

  1416. static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1417.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1418. {

  1419.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1420.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;

  1421.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;

  1422. 

  1423.     uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1424. 

  1425.     if (dstParam[3])

  1426.         fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1427. 

  1428.     return srcSliceH;

  1429. }

  1430. 

  1431. static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1432.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1433. {

  1434.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1435.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;

  1436.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;

  1437. 

  1438.     uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1439. 

  1440.     return srcSliceH;

  1441. }

  1442. 

  1443. static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  1444. {

  1445.     long i;

  1446.     for (i=0; i<num_pixels; i++)

  1447.         ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | (src[(i<<1)+1] << 24);

  1448. }

  1449. 

  1450. static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  1451. {

  1452.     long i;

  1453. 

  1454.     for (i=0; i<num_pixels; i++)

  1455.         ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | src[(i<<1)+1];

  1456. }

  1457. 

  1458. static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  1459. {

  1460.     long i;

  1461. 

  1462.     for (i=0; i<num_pixels; i++) {

  1463.         //FIXME slow?

  1464.         dst[0]= palette[src[i<<1]*4+0];

  1465.         dst[1]= palette[src[i<<1]*4+1];

  1466.         dst[2]= palette[src[i<<1]*4+2];

  1467.         dst+= 3;

  1468.     }

  1469. }

  1470. 

  1471. static int palToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1472.                            int srcSliceH, uint8_t* dst[], int dstStride[])

  1473. {

  1474.     const enum PixelFormat srcFormat= c->srcFormat;

  1475.     const enum PixelFormat dstFormat= c->dstFormat;

  1476.     void (*conv)(const uint8_t *src, uint8_t *dst, long num_pixels,

  1477.                  const uint8_t *palette)=NULL;

  1478.     int i;

  1479.     uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;

  1480.     const uint8_t *srcPtr= src[0];

  1481. 

  1482.     if (srcFormat == PIX_FMT_GRAY8A) {

  1483.         switch (dstFormat) {

  1484.         case PIX_FMT_RGB32  : conv = gray8aToPacked32; break;

  1485.         case PIX_FMT_BGR32  : conv = gray8aToPacked32; break;

  1486.         case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;

  1487.         case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;

  1488.         case PIX_FMT_RGB24  : conv = gray8aToPacked24; break;

  1489.         case PIX_FMT_BGR24  : conv = gray8aToPacked24; break;

  1490.         }

  1491.     } else if (usePal(srcFormat)) {

  1492.         switch (dstFormat) {

  1493.         case PIX_FMT_RGB32  : conv = sws_convertPalette8ToPacked32; break;

  1494.         case PIX_FMT_BGR32  : conv = sws_convertPalette8ToPacked32; break;

  1495.         case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;

  1496.         case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;

  1497.         case PIX_FMT_RGB24  : conv = sws_convertPalette8ToPacked24; break;

  1498.         case PIX_FMT_BGR24  : conv = sws_convertPalette8ToPacked24; break;

  1499.         }

  1500.     }

  1501. 

  1502.     if (!conv)

  1503.         av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",

  1504.                sws_format_name(srcFormat), sws_format_name(dstFormat));

  1505.     else {

  1506.         for (i=0; i<srcSliceH; i++) {

  1507.             conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);

  1508.             srcPtr+= srcStride[0];

  1509.             dstPtr+= dstStride[0];

  1510.         }

  1511.     }

  1512. 

  1513.     return srcSliceH;

  1514. }

  1515. 

  1516. #define isRGBA32(x) (            \

  1517.            (x) == PIX_FMT_ARGB   \

  1518.         || (x) == PIX_FMT_RGBA   \

  1519.         || (x) == PIX_FMT_BGRA   \

  1520.         || (x) == PIX_FMT_ABGR   \

  1521.         )

  1522. 

  1523. /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */

  1524. static int rgbToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1525.                            int srcSliceH, uint8_t* dst[], int dstStride[])

  1526. {

  1527.     const enum PixelFormat srcFormat= c->srcFormat;

  1528.     const enum PixelFormat dstFormat= c->dstFormat;

  1529.     const int srcBpp= (c->srcFormatBpp + 7) >> 3;

  1530.     const int dstBpp= (c->dstFormatBpp + 7) >> 3;

  1531.     const int srcId= c->srcFormatBpp >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */

  1532.     const int dstId= c->dstFormatBpp >> 2;

  1533.     void (*conv)(const uint8_t *src, uint8_t *dst, long src_size)=NULL;

  1534. 

  1535. #define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)

  1536. 

  1537.     if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {

  1538.         if (     CONV_IS(ABGR, RGBA)

  1539.               || CONV_IS(ARGB, BGRA)

  1540.               || CONV_IS(BGRA, ARGB)

  1541.               || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;

  1542.         else if (CONV_IS(ABGR, ARGB)

  1543.               || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;

  1544.         else if (CONV_IS(ABGR, BGRA)

  1545.               || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;

  1546.         else if (CONV_IS(BGRA, RGBA)

  1547.               || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;

  1548.         else if (CONV_IS(BGRA, ABGR)

  1549.               || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;

  1550.     } else

  1551.     /* BGR -> BGR */

  1552.     if (  (isBGRinInt(srcFormat) && isBGRinInt(dstFormat))

  1553.        || (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {

  1554.         switch(srcId | (dstId<<4)) {

  1555.         case 0x34: conv= rgb16to15; break;

  1556.         case 0x36: conv= rgb24to15; break;

  1557.         case 0x38: conv= rgb32to15; break;

  1558.         case 0x43: conv= rgb15to16; break;

  1559.         case 0x46: conv= rgb24to16; break;

  1560.         case 0x48: conv= rgb32to16; break;

  1561.         case 0x63: conv= rgb15to24; break;

  1562.         case 0x64: conv= rgb16to24; break;

  1563.         case 0x68: conv= rgb32to24; break;

  1564.         case 0x83: conv= rgb15to32; break;

  1565.         case 0x84: conv= rgb16to32; break;

  1566.         case 0x86: conv= rgb24to32; break;

  1567.         }

  1568.     } else if (  (isBGRinInt(srcFormat) && isRGBinInt(dstFormat))

  1569.              || (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {

  1570.         switch(srcId | (dstId<<4)) {

  1571.         case 0x33: conv= rgb15tobgr15; break;

  1572.         case 0x34: conv= rgb16tobgr15; break;

  1573.         case 0x36: conv= rgb24tobgr15; break;

  1574.         case 0x38: conv= rgb32tobgr15; break;

  1575.         case 0x43: conv= rgb15tobgr16; break;

  1576.         case 0x44: conv= rgb16tobgr16; break;

  1577.         case 0x46: conv= rgb24tobgr16; break;

  1578.         case 0x48: conv= rgb32tobgr16; break;

  1579.         case 0x63: conv= rgb15tobgr24; break;

  1580.         case 0x64: conv= rgb16tobgr24; break;

  1581.         case 0x66: conv= rgb24tobgr24; break;

  1582.         case 0x68: conv= rgb32tobgr24; break;

  1583.         case 0x83: conv= rgb15tobgr32; break;

  1584.         case 0x84: conv= rgb16tobgr32; break;

  1585.         case 0x86: conv= rgb24tobgr32; break;

  1586.         }

  1587.     }

  1588. 

  1589.     if (!conv) {

  1590.         av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",

  1591.                sws_format_name(srcFormat), sws_format_name(dstFormat));

  1592.     } else {

  1593.         const uint8_t *srcPtr= src[0];

  1594.               uint8_t *dstPtr= dst[0];

  1595.         if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) && !isRGBA32(dstFormat))

  1596.             srcPtr += ALT32_CORR;

  1597. 

  1598.         if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) && !isRGBA32(srcFormat))

  1599.             dstPtr += ALT32_CORR;

  1600. 

  1601.         if (dstStride[0]*srcBpp == srcStride[0]*dstBpp && srcStride[0] > 0 && !(srcStride[0]%srcBpp))

  1602.             conv(srcPtr, dstPtr + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]);

  1603.         else {

  1604.             int i;

  1605.             dstPtr += dstStride[0]*srcSliceY;

  1606. 

  1607.             for (i=0; i<srcSliceH; i++) {

  1608.                 conv(srcPtr, dstPtr, c->srcW*srcBpp);

  1609.                 srcPtr+= srcStride[0];

  1610.                 dstPtr+= dstStride[0];

  1611.             }

  1612.         }

  1613.     }

  1614.     return srcSliceH;

  1615. }

  1616. 

  1617. static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1618.                               int srcSliceH, uint8_t* dst[], int dstStride[])

  1619. {

  1620.     rgb24toyv12(

  1621.         src[0],

  1622.         dst[0]+ srcSliceY    *dstStride[0],

  1623.         dst[1]+(srcSliceY>>1)*dstStride[1],

  1624.         dst[2]+(srcSliceY>>1)*dstStride[2],

  1625.         c->srcW, srcSliceH,

  1626.         dstStride[0], dstStride[1], srcStride[0]);

  1627.     if (dst[3])

  1628.         fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1629.     return srcSliceH;

  1630. }

  1631. 

  1632. static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1633.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1634. {

  1635.     copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,

  1636.               dst[0], dstStride[0]);

  1637. 

  1638.     planar2x(src[1], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,

  1639.              srcSliceH >> 2, srcStride[1], dstStride[1]);

  1640.     planar2x(src[2], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,

  1641.              srcSliceH >> 2, srcStride[2], dstStride[2]);

  1642.     if (dst[3])

  1643.         fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1644.     return srcSliceH;

  1645. }

  1646. 

  1647. /* unscaled copy like stuff (assumes nearly identical formats) */

  1648. static int packedCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1649.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1650. {

  1651.     if (dstStride[0]==srcStride[0] && srcStride[0] > 0)

  1652.         memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]);

  1653.     else {

  1654.         int i;

  1655.         const uint8_t *srcPtr= src[0];

  1656.         uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;

  1657.         int length=0;

  1658. 

  1659.         /* universal length finder */

  1660.         while(length+c->srcW <= FFABS(dstStride[0])

  1661.            && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW;

  1662.         assert(length!=0);

  1663. 

  1664.         for (i=0; i<srcSliceH; i++) {

  1665.             memcpy(dstPtr, srcPtr, length);

  1666.             srcPtr+= srcStride[0];

  1667.             dstPtr+= dstStride[0];

  1668.         }

  1669.     }

  1670.     return srcSliceH;

  1671. }

  1672. 

  1673. #define DITHER_COPY(dst, dstStride, src, srcStride, bswap, dbswap)\

  1674.     uint16_t scale= dither_scale[dst_depth-1][src_depth-1];\

  1675.     int shift= src_depth-dst_depth + dither_scale[src_depth-2][dst_depth-1];\

  1676.     for (i = 0; i < height; i++) {\

  1677.         uint8_t *dither= dithers[src_depth-9][i&7];\

  1678.         for (j = 0; j < length-7; j+=8){\

  1679.             dst[j+0] = dbswap((bswap(src[j+0]) + dither[0])*scale>>shift);\

  1680.             dst[j+1] = dbswap((bswap(src[j+1]) + dither[1])*scale>>shift);\

  1681.             dst[j+2] = dbswap((bswap(src[j+2]) + dither[2])*scale>>shift);\

  1682.             dst[j+3] = dbswap((bswap(src[j+3]) + dither[3])*scale>>shift);\

  1683.             dst[j+4] = dbswap((bswap(src[j+4]) + dither[4])*scale>>shift);\

  1684.             dst[j+5] = dbswap((bswap(src[j+5]) + dither[5])*scale>>shift);\

  1685.             dst[j+6] = dbswap((bswap(src[j+6]) + dither[6])*scale>>shift);\

  1686.             dst[j+7] = dbswap((bswap(src[j+7]) + dither[7])*scale>>shift);\

  1687.         }\

  1688.         for (; j < length; j++)\

  1689.             dst[j] = dbswap((bswap(src[j]) + dither[j&7])*scale>>shift);\

  1690.         dst += dstStride;\

  1691.         src += srcStride;\

  1692.     }

  1693. 

  1694. 

  1695. static int planarCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1696.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1697. {

  1698.     int plane, i, j;

  1699.     for (plane=0; plane<4; plane++) {

  1700.         int length= (plane==0 || plane==3) ? c->srcW  : -((-c->srcW  )>>c->chrDstHSubSample);

  1701.         int y=      (plane==0 || plane==3) ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample);

  1702.         int height= (plane==0 || plane==3) ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample);

  1703.         const uint8_t *srcPtr= src[plane];

  1704.         uint8_t *dstPtr= dst[plane] + dstStride[plane]*y;

  1705. 

  1706.         if (!dst[plane]) continue;

  1707.         // ignore palette for GRAY8

  1708.         if (plane == 1 && !dst[2]) continue;

  1709.         if (!src[plane] || (plane == 1 && !src[2])) {

  1710.             if(is16BPS(c->dstFormat))

  1711.                 length*=2;

  1712.             fillPlane(dst[plane], dstStride[plane], length, height, y, (plane==3) ? 255 : 128);

  1713.         } else {

  1714.             if(isNBPS(c->srcFormat) || isNBPS(c->dstFormat)

  1715.                || (is16BPS(c->srcFormat) != is16BPS(c->dstFormat))

  1716.             ) {

  1717.                 const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1+1;

  1718.                 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1+1;

  1719.                 const uint16_t *srcPtr2 = (const uint16_t*)srcPtr;

  1720.                 uint16_t *dstPtr2 = (uint16_t*)dstPtr;

  1721. 

  1722.                 if (dst_depth == 8) {

  1723.                     if(isBE(c->srcFormat) == HAVE_BIGENDIAN){

  1724.                         DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, , )

  1725.                     } else {

  1726.                         DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, av_bswap16, )

  1727.                     }

  1728.                 } else if (src_depth == 8) {

  1729.                     for (i = 0; i < height; i++) {

  1730.                         if(isBE(c->dstFormat)){

  1731.                             for (j = 0; j < length; j++)

  1732.                                 AV_WB16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |

  1733.                                                      (srcPtr[j]>>(2*8-dst_depth)));

  1734.                         } else {

  1735.                             for (j = 0; j < length; j++)

  1736.                                 AV_WL16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |

  1737.                                                      (srcPtr[j]>>(2*8-dst_depth)));

  1738.                         }

  1739.                         dstPtr2 += dstStride[plane]/2;

  1740.                         srcPtr  += srcStride[plane];

  1741.                     }

  1742.                 } else if (src_depth <= dst_depth) {

  1743.                     for (i = 0; i < height; i++) {

  1744. #define COPY_UP(r,w) \

  1745.     for (j = 0; j < length; j++){ \

  1746.         unsigned int v= r(&srcPtr2[j]);\

  1747.         w(&dstPtr2[j], (v<<(dst_depth-src_depth)) | \

  1748.                        (v>>(2*src_depth-dst_depth)));\

  1749.     }

  1750.                         if(isBE(c->srcFormat)){

  1751.                             if(isBE(c->dstFormat)){

  1752.                                 COPY_UP(AV_RB16, AV_WB16)

  1753.                             } else {

  1754.                                 COPY_UP(AV_RB16, AV_WL16)

  1755.                             }

  1756.                         } else {

  1757.                             if(isBE(c->dstFormat)){

  1758.                                 COPY_UP(AV_RL16, AV_WB16)

  1759.                             } else {

  1760.                                 COPY_UP(AV_RL16, AV_WL16)

  1761.                             }

  1762.                         }

  1763.                         dstPtr2 += dstStride[plane]/2;

  1764.                         srcPtr2 += srcStride[plane]/2;

  1765.                     }

  1766.                 } else {

  1767.                     if(isBE(c->srcFormat) == HAVE_BIGENDIAN){

  1768.                         if(isBE(c->dstFormat) == HAVE_BIGENDIAN){

  1769.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , )

  1770.                         } else {

  1771.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , av_bswap16)

  1772.                         }

  1773.                     }else{

  1774.                         if(isBE(c->dstFormat) == HAVE_BIGENDIAN){

  1775.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, )

  1776.                         } else {

  1777.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, av_bswap16)

  1778.                         }

  1779.                     }

  1780.                 }

  1781.             } else if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat)

  1782.                   && isBE(c->srcFormat) != isBE(c->dstFormat)) {

  1783. 

  1784.                 for (i=0; i<height; i++) {

  1785.                     for (j=0; j<length; j++)

  1786.                         ((uint16_t*)dstPtr)[j] = av_bswap16(((const uint16_t*)srcPtr)[j]);

  1787.                     srcPtr+= srcStride[plane];

  1788.                     dstPtr+= dstStride[plane];

  1789.                 }

  1790.             } else if (dstStride[plane] == srcStride[plane] &&

  1791.                        srcStride[plane] > 0 && srcStride[plane] == length) {

  1792.                 memcpy(dst[plane] + dstStride[plane]*y, src[plane],

  1793.                        height*dstStride[plane]);

  1794.             } else {

  1795.                 if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat))

  1796.                     length*=2;

  1797.                 for (i=0; i<height; i++) {

  1798.                     memcpy(dstPtr, srcPtr, length);

  1799.                     srcPtr+= srcStride[plane];

  1800.                     dstPtr+= dstStride[plane];

  1801.                 }

  1802.             }

  1803.         }

  1804.     }

  1805.     return srcSliceH;

  1806. }

  1807. 

  1808. void ff_get_unscaled_swscale(SwsContext *c)

  1809. {

  1810.     const enum PixelFormat srcFormat = c->srcFormat;

  1811.     const enum PixelFormat dstFormat = c->dstFormat;

  1812.     const int flags = c->flags;

  1813.     const int dstH = c->dstH;

  1814.     int needsDither;

  1815. 

  1816.     needsDither= isAnyRGB(dstFormat)

  1817.         &&  c->dstFormatBpp < 24

  1818.         && (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));

  1819. 

  1820.     /* yv12_to_nv12 */

  1821.     if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {

  1822.         c->swScale= planarToNv12Wrapper;

  1823.     }

  1824.     /* yuv2bgr */

  1825.     if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P || srcFormat==PIX_FMT_YUVA420P) && isAnyRGB(dstFormat)

  1826.         && !(flags & SWS_ACCURATE_RND) && !(dstH&1)) {

  1827.         c->swScale= ff_yuv2rgb_get_func_ptr(c);

  1828.     }

  1829. 

  1830.     if (srcFormat==PIX_FMT_YUV410P && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_BITEXACT)) {

  1831.         c->swScale= yvu9ToYv12Wrapper;

  1832.     }

  1833. 

  1834.     /* bgr24toYV12 */

  1835.     if (srcFormat==PIX_FMT_BGR24 && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_ACCURATE_RND))

  1836.         c->swScale= bgr24ToYv12Wrapper;

  1837. 

  1838.     /* RGB/BGR -> RGB/BGR (no dither needed forms) */

  1839.     if (   isAnyRGB(srcFormat)

  1840.         && isAnyRGB(dstFormat)

  1841.         && srcFormat != PIX_FMT_BGR8      && dstFormat != PIX_FMT_BGR8

  1842.         && srcFormat != PIX_FMT_RGB8      && dstFormat != PIX_FMT_RGB8

  1843.         && srcFormat != PIX_FMT_BGR4      && dstFormat != PIX_FMT_BGR4

  1844.         && srcFormat != PIX_FMT_RGB4      && dstFormat != PIX_FMT_RGB4

  1845.         && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE

  1846.         && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE

  1847.         && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK

  1848.         && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE

  1849.         && srcFormat != PIX_FMT_RGB48LE   && dstFormat != PIX_FMT_RGB48LE

  1850.         && srcFormat != PIX_FMT_RGB48BE   && dstFormat != PIX_FMT_RGB48BE

  1851.         && srcFormat != PIX_FMT_BGR48LE   && dstFormat != PIX_FMT_BGR48LE

  1852.         && srcFormat != PIX_FMT_BGR48BE   && dstFormat != PIX_FMT_BGR48BE

  1853.         && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))

  1854.         c->swScale= rgbToRgbWrapper;

  1855. 

  1856.     if ((usePal(srcFormat) && (

  1857.         dstFormat == PIX_FMT_RGB32   ||

  1858.         dstFormat == PIX_FMT_RGB32_1 ||

  1859.         dstFormat == PIX_FMT_RGB24   ||

  1860.         dstFormat == PIX_FMT_BGR32   ||

  1861.         dstFormat == PIX_FMT_BGR32_1 ||

  1862.         dstFormat == PIX_FMT_BGR24)))

  1863.         c->swScale= palToRgbWrapper;

  1864. 

  1865.     if (srcFormat == PIX_FMT_YUV422P) {

  1866.         if (dstFormat == PIX_FMT_YUYV422)

  1867.             c->swScale= yuv422pToYuy2Wrapper;

  1868.         else if (dstFormat == PIX_FMT_UYVY422)

  1869.             c->swScale= yuv422pToUyvyWrapper;

  1870.     }

  1871. 

  1872.     /* LQ converters if -sws 0 or -sws 4*/

  1873.     if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {

  1874.         /* yv12_to_yuy2 */

  1875.         if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {

  1876.             if (dstFormat == PIX_FMT_YUYV422)

  1877.                 c->swScale= planarToYuy2Wrapper;

  1878.             else if (dstFormat == PIX_FMT_UYVY422)

  1879.                 c->swScale= planarToUyvyWrapper;

  1880.         }

  1881.     }

  1882.     if(srcFormat == PIX_FMT_YUYV422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))

  1883.         c->swScale= yuyvToYuv420Wrapper;

  1884.     if(srcFormat == PIX_FMT_UYVY422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))

  1885.         c->swScale= uyvyToYuv420Wrapper;

  1886.     if(srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)

  1887.         c->swScale= yuyvToYuv422Wrapper;

  1888.     if(srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)

  1889.         c->swScale= uyvyToYuv422Wrapper;

  1890. 

  1891. #if HAVE_ALTIVEC

  1892.     if ((av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC) &&

  1893.         !(c->flags & SWS_BITEXACT) &&

  1894.         srcFormat == PIX_FMT_YUV420P) {

  1895.         // unscaled YV12 -> packed YUV, we want speed

  1896.         if (dstFormat == PIX_FMT_YUYV422)

  1897.             c->swScale= yv12toyuy2_unscaled_altivec;

  1898.         else if (dstFormat == PIX_FMT_UYVY422)

  1899.             c->swScale= yv12touyvy_unscaled_altivec;

  1900.     }

  1901. #endif

  1902. 

  1903.     /* simple copy */

  1904.     if (  srcFormat == dstFormat

  1905.         || (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P)

  1906.         || (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P)

  1907.         || (isPlanarYUV(srcFormat) && isGray(dstFormat))

  1908.         || (isPlanarYUV(dstFormat) && isGray(srcFormat))

  1909.         || (isGray(dstFormat) && isGray(srcFormat))

  1910.         || (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat)

  1911.             && c->chrDstHSubSample == c->chrSrcHSubSample

  1912.             && c->chrDstVSubSample == c->chrSrcVSubSample

  1913.             && dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21

  1914.             && srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))

  1915.     {

  1916.         if (isPacked(c->srcFormat))

  1917.             c->swScale= packedCopyWrapper;

  1918.         else /* Planar YUV or gray */

  1919.             c->swScale= planarCopyWrapper;

  1920.     }

  1921. #if ARCH_BFIN

  1922.     ff_bfin_get_unscaled_swscale (c);

  1923. #endif

  1924. }

  1925. 

  1926. static void reset_ptr(const uint8_t* src[], int format)

  1927. {

  1928.     if(!isALPHA(format))

  1929.         src[3]=NULL;

  1930.     if(!isPlanarYUV(format)) {

  1931.         src[3]=src[2]=NULL;

  1932. 

  1933.         if (!usePal(format))

  1934.             src[1]= NULL;

  1935.     }

  1936. }

  1937. 

  1938. static int check_image_pointers(uint8_t *data[4], enum PixelFormat pix_fmt,

  1939.                                 const int linesizes[4])

  1940. {

  1941.     const AVPixFmtDescriptor *desc = &av_pix_fmt_descriptors[pix_fmt];

  1942.     int i;

  1943. 

  1944.     for (i = 0; i < 4; i++) {

  1945.         int plane = desc->comp[i].plane;

  1946.         if (!data[plane] || !linesizes[plane])

  1947.             return 0;

  1948.     }

  1949. 

  1950.     return 1;

  1951. }

  1952. 

  1953. /**

  1954.  * swscale wrapper, so we don't need to export the SwsContext.

  1955.  * Assumes planar YUV to be in YUV order instead of YVU.

  1956.  */

  1957. int sws_scale(SwsContext *c, const uint8_t* const src[], const int srcStride[], int srcSliceY,

  1958.               int srcSliceH, uint8_t* const dst[], const int dstStride[])

  1959. {

  1960.     int i;

  1961.     const uint8_t* src2[4]= {src[0], src[1], src[2], src[3]};

  1962.     uint8_t* dst2[4]= {dst[0], dst[1], dst[2], dst[3]};

  1963. 

  1964.     // do not mess up sliceDir if we have a "trailing" 0-size slice

  1965.     if (srcSliceH == 0)

  1966.         return 0;

  1967. 

  1968.     if (!check_image_pointers(src, c->srcFormat, srcStride)) {

  1969.         av_log(c, AV_LOG_ERROR, "bad src image pointers\n");

  1970.         return 0;

  1971.     }

  1972.     if (!check_image_pointers(dst, c->dstFormat, dstStride)) {

  1973.         av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");

  1974.         return 0;

  1975.     }

  1976. 

  1977.     if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {

  1978.         av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");

  1979.         return 0;

  1980.     }

  1981.     if (c->sliceDir == 0) {

  1982.         if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;

  1983.     }

  1984. 

  1985.     if (usePal(c->srcFormat)) {

  1986.         for (i=0; i<256; i++) {

  1987.             int p, r, g, b, y, u, v, a = 0xff;

  1988.             if(c->srcFormat == PIX_FMT_PAL8) {

  1989.                 p=((const uint32_t*)(src[1]))[i];

  1990.                 a= (p>>24)&0xFF;

  1991.                 r= (p>>16)&0xFF;

  1992.                 g= (p>> 8)&0xFF;

  1993.                 b=  p     &0xFF;

  1994.             } else if(c->srcFormat == PIX_FMT_RGB8) {

  1995.                 r= (i>>5    )*36;

  1996.                 g= ((i>>2)&7)*36;

  1997.                 b= (i&3     )*85;

  1998.             } else if(c->srcFormat == PIX_FMT_BGR8) {

  1999.                 b= (i>>6    )*85;

  2000.                 g= ((i>>3)&7)*36;

  2001.                 r= (i&7     )*36;

  2002.             } else if(c->srcFormat == PIX_FMT_RGB4_BYTE) {

  2003.                 r= (i>>3    )*255;

  2004.                 g= ((i>>1)&3)*85;

  2005.                 b= (i&1     )*255;

  2006.             } else if(c->srcFormat == PIX_FMT_GRAY8 || c->srcFormat == PIX_FMT_GRAY8A) {

  2007.                 r = g = b = i;

  2008.             } else {

  2009.                 assert(c->srcFormat == PIX_FMT_BGR4_BYTE);

  2010.                 b= (i>>3    )*255;

  2011.                 g= ((i>>1)&3)*85;

  2012.                 r= (i&1     )*255;

  2013.             }

  2014.             y= av_clip_uint8((RY*r + GY*g + BY*b + ( 33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2015.             u= av_clip_uint8((RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2016.             v= av_clip_uint8((RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2017.             c->pal_yuv[i]= y + (u<<8) + (v<<16) + (a<<24);

  2018. 

  2019.             switch(c->dstFormat) {

  2020.             case PIX_FMT_BGR32:

  2021. #if !HAVE_BIGENDIAN

  2022.             case PIX_FMT_RGB24:

  2023. #endif

  2024.                 c->pal_rgb[i]=  r + (g<<8) + (b<<16) + (a<<24);

  2025.                 break;

  2026.             case PIX_FMT_BGR32_1:

  2027. #if HAVE_BIGENDIAN

  2028.             case PIX_FMT_BGR24:

  2029. #endif

  2030.                 c->pal_rgb[i]= a + (r<<8) + (g<<16) + (b<<24);

  2031.                 break;

  2032.             case PIX_FMT_RGB32_1:

  2033. #if HAVE_BIGENDIAN

  2034.             case PIX_FMT_RGB24:

  2035. #endif

  2036.                 c->pal_rgb[i]= a + (b<<8) + (g<<16) + (r<<24);

  2037.                 break;

  2038.             case PIX_FMT_RGB32:

  2039. #if !HAVE_BIGENDIAN

  2040.             case PIX_FMT_BGR24:

  2041. #endif

  2042.             default:

  2043.                 c->pal_rgb[i]=  b + (g<<8) + (r<<16) + (a<<24);

  2044.             }

  2045.         }

  2046.     }

  2047. 

  2048.     // copy strides, so they can safely be modified

  2049.     if (c->sliceDir == 1) {

  2050.         // slices go from top to bottom

  2051.         int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2], srcStride[3]};

  2052.         int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2], dstStride[3]};

  2053. 

  2054.         reset_ptr(src2, c->srcFormat);

  2055.         reset_ptr((const uint8_t**)dst2, c->dstFormat);

  2056. 

  2057.         /* reset slice direction at end of frame */

  2058.         if (srcSliceY + srcSliceH == c->srcH)

  2059.             c->sliceDir = 0;

  2060. 

  2061.         return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2, dstStride2);

  2062.     } else {

  2063.         // slices go from bottom to top => we flip the image internally

  2064.         int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2], -srcStride[3]};

  2065.         int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2], -dstStride[3]};

  2066. 

  2067.         src2[0] += (srcSliceH-1)*srcStride[0];

  2068.         if (!usePal(c->srcFormat))

  2069.             src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1];

  2070.         src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2];

  2071.         src2[3] += (srcSliceH-1)*srcStride[3];

  2072.         dst2[0] += ( c->dstH                      -1)*dstStride[0];

  2073.         dst2[1] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1];

  2074.         dst2[2] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2];

  2075.         dst2[3] += ( c->dstH                      -1)*dstStride[3];

  2076. 

  2077.         reset_ptr(src2, c->srcFormat);

  2078.         reset_ptr((const uint8_t**)dst2, c->dstFormat);

  2079. 

  2080.         /* reset slice direction at end of frame */

  2081.         if (!srcSliceY)

  2082.             c->sliceDir = 0;

  2083. 

  2084.         return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2);

  2085.     }

  2086. }

  2087. 

  2088. #if LIBSWSCALE_VERSION_MAJOR < 1

  2089. int sws_scale_ordered(SwsContext *c, const uint8_t* const src[], int srcStride[], int srcSliceY,

  2090.                       int srcSliceH, uint8_t* dst[], int dstStride[])

  2091. {

  2092.     return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride);

  2093. }

  2094. #endif

  2095. 

  2096. /* Convert the palette to the same packed 32-bit format as the palette */

  2097. void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  2098. {

  2099.     long i;

  2100. 

  2101.     for (i=0; i<num_pixels; i++)

  2102.         ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];

  2103. }

  2104. 

  2105. /* Palette format: ABCD -> dst format: ABC */

  2106. void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  2107. {

  2108.     long i;

  2109. 

  2110.     for (i=0; i<num_pixels; i++) {

  2111.         //FIXME slow?

  2112.         dst[0]= palette[src[i]*4+0];

  2113.         dst[1]= palette[src[i]*4+1];

  2114.         dst[2]= palette[src[i]*4+2];

  2115.         dst+= 3;

  2116.     }

  2117. }