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.
30472 Commits
32 Branches
303MB
Tree: b932eb1be6
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'b932eb1be6'
${ noResults }
FFmpeg/libswscale/swscale.c

2113 lines
78KB
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. static const uint8_t flat64[8]={64,64,64,64,64,64,64,64};

  286. 

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

  288. {    2,    3,    3,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,},

  289. {    2,    3,    7,    7,   13,   13,   25,   25,   25,   25,   25,   25,   25,   25,   25,   25,},

  290. {    3,    3,    4,   15,   15,   29,   57,   57,   57,  113,  113,  113,  113,  113,  113,  113,},

  291. {    3,    4,    4,    5,   31,   31,   61,  121,  241,  241,  241,  241,  481,  481,  481,  481,},

  292. {    3,    4,    5,    5,    6,   63,   63,  125,  249,  497,  993,  993,  993,  993,  993, 1985,},

  293. {    3,    5,    6,    6,    6,    7,  127,  127,  253,  505, 1009, 2017, 4033, 4033, 4033, 4033,},

  294. {    3,    5,    6,    7,    7,    7,    8,  255,  255,  509, 1017, 2033, 4065, 8129,16257,16257,},

  295. {    3,    5,    6,    8,    8,    8,    8,    9,  511,  511, 1021, 2041, 4081, 8161,16321,32641,},

  296. {    3,    5,    7,    8,    9,    9,    9,    9,   10, 1023, 1023, 2045, 4089, 8177,16353,32705,},

  297. {    3,    5,    7,    8,   10,   10,   10,   10,   10,   11, 2047, 2047, 4093, 8185,16369,32737,},

  298. {    3,    5,    7,    8,   10,   11,   11,   11,   11,   11,   12, 4095, 4095, 8189,16377,32753,},

  299. {    3,    5,    7,    9,   10,   12,   12,   12,   12,   12,   12,   13, 8191, 8191,16381,32761,},

  300. {    3,    5,    7,    9,   10,   12,   13,   13,   13,   13,   13,   13,   14,16383,16383,32765,},

  301. {    3,    5,    7,    9,   10,   12,   14,   14,   14,   14,   14,   14,   14,   15,32767,32767,},

  302. {    3,    5,    7,    9,   11,   12,   14,   15,   15,   15,   15,   15,   15,   15,   16,65535,},

  303. };

  304. 

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

  306.                                                     const int16_t *chrFilter, const int16_t **chrUSrc,

  307.                                                     const int16_t **chrVSrc, int chrFilterSize,

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

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

  310. {

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

  312.     int i;

  313.     int shift = 11 + 16 - output_bits;

  314. 

  315. #define output_pixel(pos, val) \

  316.     if (big_endian) { \

  317.         if (output_bits == 16) { \

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

  319.         } else { \

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

  321.         } \

  322.     } else { \

  323.         if (output_bits == 16) { \

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

  325.         } else { \

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

  327.         } \

  328.     }

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

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

  331.         int j;

  332. 

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

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

  335. 

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

  337.     }

  338. 

  339.     if (uDest) {

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

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

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

  343.             int j;

  344. 

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

  346.                 u += chrUSrc[j][i] * chrFilter[j];

  347.                 v += chrVSrc[j][i] * chrFilter[j];

  348.             }

  349. 

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

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

  352.         }

  353.     }

  354. 

  355.     if (CONFIG_SWSCALE_ALPHA && aDest) {

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

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

  358.             int j;

  359. 

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

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

  362. 

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

  364.         }

  365.     }

  366. }

  367. 

  368. #define yuv2NBPS(bits, BE_LE, is_be) \

  369. static void yuv2yuvX ## bits ## BE_LE ## _c(const int16_t *lumFilter, \

  370.                               const int16_t **lumSrc, int lumFilterSize, \

  371.                               const int16_t *chrFilter, const int16_t **chrUSrc, \

  372.                               const int16_t **chrVSrc, \

  373.                               int chrFilterSize, const int16_t **alpSrc, \

  374.                               uint16_t *dest, uint16_t *uDest, uint16_t *vDest, \

  375.                               uint16_t *aDest, int dstW, int chrDstW) \

  376. { \

  377.     yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize, \

  378.                            chrFilter, chrUSrc, chrVSrc, chrFilterSize, \

  379.                            alpSrc, \

  380.                            dest, uDest, vDest, aDest, \

  381.                            dstW, chrDstW, is_be, bits); \

  382. }

  383. yuv2NBPS( 9, BE, 1);

  384. yuv2NBPS( 9, LE, 0);

  385. yuv2NBPS(10, BE, 1);

  386. yuv2NBPS(10, LE, 0);

  387. yuv2NBPS(16, BE, 1);

  388. yuv2NBPS(16, LE, 0);

  389. 

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

  391.                                  const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize,

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

  393.                                  enum PixelFormat dstFormat)

  394. {

  395.     if (isNBPS(dstFormat)) {

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

  397.         yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,

  398.                               chrFilter, chrUSrc, chrVSrc, chrFilterSize,

  399.                               alpSrc,

  400.                               dest, uDest, vDest, aDest,

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

  402.     } else {

  403.         if (isBE(dstFormat)) {

  404.             yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,

  405.                                    chrFilter, chrUSrc, chrVSrc, chrFilterSize,

  406.                                    alpSrc,

  407.                                    dest, uDest, vDest, aDest,

  408.                                    dstW, chrDstW, 1, 16);

  409.         } else {

  410.             yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,

  411.                                    chrFilter, chrUSrc, chrVSrc, chrFilterSize,

  412.                                    alpSrc,

  413.                                    dest, uDest, vDest, aDest,

  414.                                    dstW, chrDstW, 0, 16);

  415.         }

  416.     }

  417. }

  418. 

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

  420.                                const int16_t *chrFilter, const int16_t **chrUSrc,

  421.                                const int16_t **chrVSrc, int chrFilterSize,

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

  423.                                const uint8_t *lumDither, const uint8_t *chrDither)

  424. {

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

  426.     int i;

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

  428.         int val = lumDither[i&7] << 12;

  429.         int j;

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

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

  432. 

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

  434.     }

  435. 

  436.     if (uDest)

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

  438.             int u = chrDither[i&7] << 12;

  439.             int v = chrDither[(i+3)&7] << 12;

  440.             int j;

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

  442.                 u += chrUSrc[j][i] * chrFilter[j];

  443.                 v += chrVSrc[j][i] * chrFilter[j];

  444.             }

  445. 

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

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

  448.         }

  449. 

  450.     if (CONFIG_SWSCALE_ALPHA && aDest)

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

  452.             int val = lumDither[i&7] << 12;

  453.             int j;

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

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

  456. 

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

  458.         }

  459. 

  460. }

  461. 

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

  463.                                 const int16_t *chrFilter, const int16_t **chrUSrc,

  464.                                 const int16_t **chrVSrc, int chrFilterSize,

  465.                                 uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat,

  466.                                 const uint8_t *lumDither, const uint8_t *chrDither)

  467. {

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

  469.     int i;

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

  471.         int val = lumDither[i&7]<<12;

  472.         int j;

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

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

  475. 

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

  477.     }

  478. 

  479.     if (!uDest)

  480.         return;

  481. 

  482.     if (dstFormat == PIX_FMT_NV12)

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

  484.             int u = chrDither[i&7]<<12;

  485.             int v = chrDither[(i+3)&7]<<12;

  486.             int j;

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

  488.                 u += chrUSrc[j][i] * chrFilter[j];

  489.                 v += chrVSrc[j][i] * chrFilter[j];

  490.             }

  491. 

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

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

  494.         }

  495.     else

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

  497.             int u = chrDither[i&7]<<12;

  498.             int v = chrDither[(i+3)&7]<<12;

  499.             int j;

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

  501.                 u += chrUSrc[j][i] * chrFilter[j];

  502.                 v += chrVSrc[j][i] * chrFilter[j];

  503.             }

  504. 

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

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

  507.         }

  508. }

  509. 

  510. #define YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha) \

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

  512.         int j;\

  513.         int Y1 = 1<<18;\

  514.         int Y2 = 1<<18;\

  515.         int U  = 1<<18;\

  516.         int V  = 1<<18;\

  517.         int av_unused A1, A2;\

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

  519.         const int i2= 2*i;\

  520.         \

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

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

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

  524.         }\

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

  526.             U += chrUSrc[j][i] * chrFilter[j];\

  527.             V += chrVSrc[j][i] * chrFilter[j];\

  528.         }\

  529.         Y1>>=19;\

  530.         Y2>>=19;\

  531.         U >>=19;\

  532.         V >>=19;\

  533.         if (alpha) {\

  534.             A1 = 1<<18;\

  535.             A2 = 1<<18;\

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

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

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

  539.             }\

  540.             A1>>=19;\

  541.             A2>>=19;\

  542.         }

  543. 

  544. #define YSCALE_YUV_2_PACKEDX_C(type,alpha) \

  545.         YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha)\

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

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

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

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

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

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

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

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

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

  555.         }\

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

  557.             A1=av_clip_uint8(A1);\

  558.             A2=av_clip_uint8(A2);\

  559.         }

  560. 

  561. #define YSCALE_YUV_2_PACKEDX_FULL_C(rnd,alpha) \

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

  563.         int j;\

  564.         int Y = 1<<9;\

  565.         int U = (1<<9)-(128<<19);\

  566.         int V = (1<<9)-(128<<19);\

  567.         int av_unused A;\

  568.         int R,G,B;\

  569.         \

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

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

  572.         }\

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

  574.             U += chrUSrc[j][i] * chrFilter[j];\

  575.             V += chrVSrc[j][i] * chrFilter[j];\

  576.         }\

  577.         Y >>=10;\

  578.         U >>=10;\

  579.         V >>=10;\

  580.         if (alpha) {\

  581.             A = rnd;\

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

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

  584.             A >>=19;\

  585.             if (A&256)\

  586.                 A = av_clip_uint8(A);\

  587.         }

  588. 

  589. #define YSCALE_YUV_2_RGBX_FULL_C(rnd,alpha) \

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

  591.         Y-= c->yuv2rgb_y_offset;\

  592.         Y*= c->yuv2rgb_y_coeff;\

  593.         Y+= rnd;\

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

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

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

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

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

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

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

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

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

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

  604.         }

  605. 

  606. #define YSCALE_YUV_2_GRAY16_C \

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

  608.         int j;\

  609.         int Y1 = 1<<18;\

  610.         int Y2 = 1<<18;\

  611.         int U  = 1<<18;\

  612.         int V  = 1<<18;\

  613.         \

  614.         const int i2= 2*i;\

  615.         \

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

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

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

  619.         }\

  620.         Y1>>=11;\

  621.         Y2>>=11;\

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

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

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

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

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

  627.         }

  628. 

  629. #define YSCALE_YUV_2_RGBX_C(type,alpha) \

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

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

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

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

  634. 

  635. #define YSCALE_YUV_2_PACKED2_C(type,alpha)   \

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

  637.         const int i2= 2*i;       \

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

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

  640.         int U= (ubuf0[i]*uvalpha1+ubuf1[i]*uvalpha)>>19;              \

  641.         int V= (vbuf0[i]*uvalpha1+vbuf1[i]*uvalpha)>>19;              \

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

  643.         int av_unused A1, A2;                                         \

  644.         if (alpha) {\

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

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

  647.         }

  648. 

  649. #define YSCALE_YUV_2_GRAY16_2_C   \

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

  651.         const int i2= 2*i;       \

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

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

  654. 

  655. #define YSCALE_YUV_2_RGB2_C(type,alpha) \

  656.     YSCALE_YUV_2_PACKED2_C(type,alpha)\

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

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

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

  660. 

  661. #define YSCALE_YUV_2_PACKED1_C(type,alpha) \

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

  663.         const int i2= 2*i;\

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

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

  666.         int U= (ubuf1[i])>>7;\

  667.         int V= (vbuf1[i])>>7;\

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

  669.         int av_unused A1, A2;\

  670.         if (alpha) {\

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

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

  673.         }

  674. 

  675. #define YSCALE_YUV_2_GRAY16_1_C \

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

  677.         const int i2= 2*i;\

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

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

  680. 

  681. #define YSCALE_YUV_2_RGB1_C(type,alpha) \

  682.     YSCALE_YUV_2_PACKED1_C(type,alpha)\

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

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

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

  686. 

  687. #define YSCALE_YUV_2_PACKED1B_C(type,alpha) \

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

  689.         const int i2= 2*i;\

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

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

  692.         int U= (ubuf0[i] + ubuf1[i])>>8;\

  693.         int V= (vbuf0[i] + vbuf1[i])>>8;\

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

  695.         int av_unused A1, A2;\

  696.         if (alpha) {\

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

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

  699.         }

  700. 

  701. #define YSCALE_YUV_2_RGB1B_C(type,alpha) \

  702.     YSCALE_YUV_2_PACKED1B_C(type,alpha)\

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

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

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

  706. 

  707. #define YSCALE_YUV_2_MONO2_C \

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

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

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

  711.         int acc;\

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

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

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

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

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

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

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

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

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

  721.         dest++;\

  722.     }

  723. 

  724. #define YSCALE_YUV_2_MONOX_C \

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

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

  727.     int acc=0;\

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

  729.         int j;\

  730.         int Y1=1<<18;\

  731.         int Y2=1<<18;\

  732. \

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

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

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

  736.         }\

  737.         Y1>>=19;\

  738.         Y2>>=19;\

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

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

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

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

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

  744.         }\

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

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

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

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

  749.             dest++;\

  750.         }\

  751.     }

  752. 

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

  754.     switch(c->dstFormat) {\

  755.     case PIX_FMT_RGB48BE:\

  756.     case PIX_FMT_RGB48LE:\

  757.         func(uint8_t,0)\

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

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

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

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

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

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

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

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

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

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

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

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

  770.             dest+=12;\

  771.         }\

  772.         break;\

  773.     case PIX_FMT_BGR48BE:\

  774.     case PIX_FMT_BGR48LE:\

  775.         func(uint8_t,0)\

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

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

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

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

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

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

  782.             dest+=12;\

  783.         }\

  784.         break;\

  785.     case PIX_FMT_RGBA:\

  786.     case PIX_FMT_BGRA:\

  787.         if (CONFIG_SMALL) {\

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

  789.             func(uint32_t,needAlpha)\

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

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

  792.             }\

  793.         } else {\

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

  795.                 func(uint32_t,1)\

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

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

  798.                 }\

  799.             } else {\

  800.                 func(uint32_t,0)\

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

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

  803.                 }\

  804.             }\

  805.         }\

  806.         break;\

  807.     case PIX_FMT_ARGB:\

  808.     case PIX_FMT_ABGR:\

  809.         if (CONFIG_SMALL) {\

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

  811.             func(uint32_t,needAlpha)\

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

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

  814.             }\

  815.         } else {\

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

  817.                 func(uint32_t,1)\

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

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

  820.                 }\

  821.             } else {\

  822.                 func(uint32_t,0)\

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

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

  825.                 }\

  826.             }\

  827.         }                \

  828.         break;\

  829.     case PIX_FMT_RGB24:\

  830.         func(uint8_t,0)\

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

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

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

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

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

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

  837.             dest+=6;\

  838.         }\

  839.         break;\

  840.     case PIX_FMT_BGR24:\

  841.         func(uint8_t,0)\

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

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

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

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

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

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

  848.             dest+=6;\

  849.         }\

  850.         break;\

  851.     case PIX_FMT_RGB565BE:\

  852.     case PIX_FMT_RGB565LE:\

  853.     case PIX_FMT_BGR565BE:\

  854.     case PIX_FMT_BGR565LE:\

  855.         {\

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

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

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

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

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

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

  862.             func(uint16_t,0)\

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

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

  865.             }\

  866.         }\

  867.         break;\

  868.     case PIX_FMT_RGB555BE:\

  869.     case PIX_FMT_RGB555LE:\

  870.     case PIX_FMT_BGR555BE:\

  871.     case PIX_FMT_BGR555LE:\

  872.         {\

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

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

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

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

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

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

  879.             func(uint16_t,0)\

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

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

  882.             }\

  883.         }\

  884.         break;\

  885.     case PIX_FMT_RGB444BE:\

  886.     case PIX_FMT_RGB444LE:\

  887.     case PIX_FMT_BGR444BE:\

  888.     case PIX_FMT_BGR444LE:\

  889.         {\

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

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

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

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

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

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

  896.             func(uint16_t,0)\

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

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

  899.             }\

  900.         }\

  901.         break;\

  902.     case PIX_FMT_RGB8:\

  903.     case PIX_FMT_BGR8:\

  904.         {\

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

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

  907.             func(uint8_t,0)\

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

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

  910.             }\

  911.         }\

  912.         break;\

  913.     case PIX_FMT_RGB4:\

  914.     case PIX_FMT_BGR4:\

  915.         {\

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

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

  918.             func(uint8_t,0)\

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

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

  921.             }\

  922.         }\

  923.         break;\

  924.     case PIX_FMT_RGB4_BYTE:\

  925.     case PIX_FMT_BGR4_BYTE:\

  926.         {\

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

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

  929.             func(uint8_t,0)\

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

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

  932.             }\

  933.         }\

  934.         break;\

  935.     case PIX_FMT_MONOBLACK:\

  936.     case PIX_FMT_MONOWHITE:\

  937.         {\

  938.             func_monoblack\

  939.         }\

  940.         break;\

  941.     case PIX_FMT_YUYV422:\

  942.         func2\

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

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

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

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

  947.         }                \

  948.         break;\

  949.     case PIX_FMT_UYVY422:\

  950.         func2\

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

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

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

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

  955.         }                \

  956.         break;\

  957.     case PIX_FMT_GRAY16BE:\

  958.         func_g16\

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

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

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

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

  963.         }                \

  964.         break;\

  965.     case PIX_FMT_GRAY16LE:\

  966.         func_g16\

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

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

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

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

  971.         }                \

  972.         break;\

  973.     }

  974. 

  975. static inline void yuv2packedXinC(SwsContext *c, const int16_t *lumFilter,

  976.                                   const int16_t **lumSrc, int lumFilterSize,

  977.                                   const int16_t *chrFilter, const int16_t **chrUSrc,

  978.                                   const int16_t **chrVSrc, int chrFilterSize,

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

  980. {

  981.     int i;

  982.     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)

  983. }

  984. 

  985. static inline void yuv2rgbXinC_full(SwsContext *c, const int16_t *lumFilter,

  986.                                     const int16_t **lumSrc, int lumFilterSize,

  987.                                     const int16_t *chrFilter, const int16_t **chrUSrc,

  988.                                     const int16_t **chrVSrc, int chrFilterSize,

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

  990. {

  991.     int i;

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

  993.     int aidx= 3;

  994. 

  995.     switch(c->dstFormat) {

  996.     case PIX_FMT_ARGB:

  997.         dest++;

  998.         aidx= 0;

  999.     case PIX_FMT_RGB24:

  1000.         aidx--;

  1001.     case PIX_FMT_RGBA:

  1002.         if (CONFIG_SMALL) {

  1003.             int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;

  1004.             YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)

  1005.                 dest[aidx]= needAlpha ? A : 255;

  1006.                 dest[0]= R>>22;

  1007.                 dest[1]= G>>22;

  1008.                 dest[2]= B>>22;

  1009.                 dest+= step;

  1010.             }

  1011.         } else {

  1012.             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  1013.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)

  1014.                     dest[aidx]= A;

  1015.                     dest[0]= R>>22;

  1016.                     dest[1]= G>>22;

  1017.                     dest[2]= B>>22;

  1018.                     dest+= step;

  1019.                 }

  1020.             } else {

  1021.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)

  1022.                     dest[aidx]= 255;

  1023.                     dest[0]= R>>22;

  1024.                     dest[1]= G>>22;

  1025.                     dest[2]= B>>22;

  1026.                     dest+= step;

  1027.                 }

  1028.             }

  1029.         }

  1030.         break;

  1031.     case PIX_FMT_ABGR:

  1032.         dest++;

  1033.         aidx= 0;

  1034.     case PIX_FMT_BGR24:

  1035.         aidx--;

  1036.     case PIX_FMT_BGRA:

  1037.         if (CONFIG_SMALL) {

  1038.             int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;

  1039.             YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)

  1040.                 dest[aidx]= needAlpha ? A : 255;

  1041.                 dest[0]= B>>22;

  1042.                 dest[1]= G>>22;

  1043.                 dest[2]= R>>22;

  1044.                 dest+= step;

  1045.             }

  1046.         } else {

  1047.             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  1048.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)

  1049.                     dest[aidx]= A;

  1050.                     dest[0]= B>>22;

  1051.                     dest[1]= G>>22;

  1052.                     dest[2]= R>>22;

  1053.                     dest+= step;

  1054.                 }

  1055.             } else {

  1056.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)

  1057.                     dest[aidx]= 255;

  1058.                     dest[0]= B>>22;

  1059.                     dest[1]= G>>22;

  1060.                     dest[2]= R>>22;

  1061.                     dest+= step;

  1062.                 }

  1063.             }

  1064.         }

  1065.         break;

  1066.     default:

  1067.         assert(0);

  1068.     }

  1069. }

  1070. 

  1071. static void fillPlane(uint8_t* plane, int stride, int width, int height, int y, uint8_t val)

  1072. {

  1073.     int i;

  1074.     uint8_t *ptr = plane + stride*y;

  1075.     for (i=0; i<height; i++) {

  1076.         memset(ptr, val, width);

  1077.         ptr += stride;

  1078.     }

  1079. }

  1080. 

  1081. #define RGB48(name, R, B, READ)\

  1082. static inline void name ## ToY(int16_t *dst, const uint16_t *src, int width, uint32_t *unused)\

  1083. {\

  1084.     int i;\

  1085.     for (i = 0; i < width; i++) {\

  1086.         int r = READ(&src[i*3+R]);\

  1087.         int g = READ(&src[i*3+1]);\

  1088.         int b = READ(&src[i*3+B]);\

  1089. \

  1090.         dst[i] = (RY*r + GY*g + BY*b + (32<<(RGB2YUV_SHIFT-1+8)) + (1<<(RGB2YUV_SHIFT-7+8))) >> (RGB2YUV_SHIFT-6+8);\

  1091.     }\

  1092. }\

  1093. \

  1094. static inline void name ## ToUV(int16_t *dstU, int16_t *dstV,\

  1095.                              const uint16_t *src1, const uint16_t *src2,\

  1096.                              int width, uint32_t *unused)\

  1097. {\

  1098.     int i;\

  1099.     assert(src1==src2);\

  1100.     for (i = 0; i < width; i++) {\

  1101.         int r = READ(&src1[3*i + R]);\

  1102.         int g = READ(&src1[3*i + 1]);\

  1103.         int b = READ(&src1[3*i + B]);\

  1104. \

  1105.         dstU[i] = (RU*r + GU*g + BU*b + (256<<(RGB2YUV_SHIFT-1+8)) + (1<<(RGB2YUV_SHIFT-7+8))) >> (RGB2YUV_SHIFT-6+8);\

  1106.         dstV[i] = (RV*r + GV*g + BV*b + (256<<(RGB2YUV_SHIFT-1+8)) + (1<<(RGB2YUV_SHIFT-7+8))) >> (RGB2YUV_SHIFT-6+8);\

  1107.     }\

  1108. }\

  1109. \

  1110. static inline void name ## ToUV_half(int16_t *dstU, int16_t *dstV,\

  1111.                                   const uint16_t *src1, const uint16_t *src2,\

  1112.                                   int width, uint32_t *unused)\

  1113. {\

  1114.     int i;\

  1115.     assert(src1==src2);\

  1116.     for (i = 0; i < width; i++) {\

  1117.         int r= READ(&src1[6*i + R]) + READ(&src1[6*i + 3+R]);\

  1118.         int g= READ(&src1[6*i + 1]) + READ(&src1[6*i + 4]);\

  1119.         int b= READ(&src1[6*i + B]) + READ(&src1[6*i + 3+B]);\

  1120. \

  1121.         dstU[i]= (RU*r + GU*g + BU*b + (256U<<(RGB2YUV_SHIFT+8)) + (1<<(RGB2YUV_SHIFT-6+8))) >> (RGB2YUV_SHIFT-5+8);\

  1122.         dstV[i]= (RV*r + GV*g + BV*b + (256U<<(RGB2YUV_SHIFT+8)) + (1<<(RGB2YUV_SHIFT-6+8))) >> (RGB2YUV_SHIFT-5+8);\

  1123.     }\

  1124. }

  1125. 

  1126. RGB48(rgb48LE, 0, 2, AV_RL16)

  1127. RGB48(rgb48BE, 0, 2, AV_RB16)

  1128. RGB48(bgr48LE, 2, 0, AV_RL16)

  1129. RGB48(bgr48BE, 2, 0, AV_RB16)

  1130. 

  1131. #define BGR2Y(type, name, shr, shg, shb, maskr, maskg, maskb, RY, GY, BY, S)\

  1132. static inline void name(int16_t *dst, const uint8_t *src, int width, uint32_t *unused)\

  1133. {\

  1134.     int i;\

  1135.     for (i=0; i<width; i++) {\

  1136.         int b= (((const type*)src)[i]>>shb)&maskb;\

  1137.         int g= (((const type*)src)[i]>>shg)&maskg;\

  1138.         int r= (((const type*)src)[i]>>shr)&maskr;\

  1139. \

  1140.         dst[i]= (((RY)*r + (GY)*g + (BY)*b + (32<<((S)-1)) + (1<<(S-7)))>>((S)-6));\

  1141.     }\

  1142. }

  1143. 

  1144. BGR2Y(uint32_t, bgr32ToY,16, 0, 0, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY   , BY<< 8, RGB2YUV_SHIFT+8)

  1145. BGR2Y(uint32_t,bgr321ToY,16,16, 0, 0xFF00, 0x00FF, 0xFF00, RY    , GY<<8, BY    , RGB2YUV_SHIFT+8)

  1146. BGR2Y(uint32_t, rgb32ToY, 0, 0,16, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY   , BY<< 8, RGB2YUV_SHIFT+8)

  1147. BGR2Y(uint32_t,rgb321ToY, 0,16,16, 0xFF00, 0x00FF, 0xFF00, RY    , GY<<8, BY    , RGB2YUV_SHIFT+8)

  1148. BGR2Y(uint16_t, bgr16ToY, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RY<<11, GY<<5, BY    , RGB2YUV_SHIFT+8)

  1149. BGR2Y(uint16_t, bgr15ToY, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RY<<10, GY<<5, BY    , RGB2YUV_SHIFT+7)

  1150. BGR2Y(uint16_t, rgb16ToY, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RY    , GY<<5, BY<<11, RGB2YUV_SHIFT+8)

  1151. BGR2Y(uint16_t, rgb15ToY, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RY    , GY<<5, BY<<10, RGB2YUV_SHIFT+7)

  1152. 

  1153. static inline void abgrToA(int16_t *dst, const uint8_t *src, int width, uint32_t *unused)

  1154. {

  1155.     int i;

  1156.     for (i=0; i<width; i++) {

  1157.         dst[i]= src[4*i]<<6;

  1158.     }

  1159. }

  1160. 

  1161. #define BGR2UV(type, name, shr, shg, shb, shp, maskr, maskg, maskb, RU, GU, BU, RV, GV, BV, S) \

  1162. static inline void name(int16_t *dstU, int16_t *dstV, const uint8_t *src, const uint8_t *dummy, int width, uint32_t *unused)\

  1163. {\

  1164.     int i;\

  1165.     for (i=0; i<width; i++) {\

  1166.         int b= ((((const type*)src)[i]>>shp)&maskb)>>shb;\

  1167.         int g= ((((const type*)src)[i]>>shp)&maskg)>>shg;\

  1168.         int r= ((((const type*)src)[i]>>shp)&maskr)>>shr;\

  1169. \

  1170.         dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (256<<((S)-1)) + (1<<(S-7)))>>((S)-6);\

  1171.         dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (256<<((S)-1)) + (1<<(S-7)))>>((S)-6);\

  1172.     }\

  1173. }\

  1174. static inline void name ## _half(int16_t *dstU, int16_t *dstV, const uint8_t *src, const uint8_t *dummy, int width, uint32_t *unused)\

  1175. {\

  1176.     int i;\

  1177.     for (i=0; i<width; i++) {\

  1178.         int pix0= ((const type*)src)[2*i+0]>>shp;\

  1179.         int pix1= ((const type*)src)[2*i+1]>>shp;\

  1180.         int g= (pix0&~(maskr|maskb))+(pix1&~(maskr|maskb));\

  1181.         int b= ((pix0+pix1-g)&(maskb|(2*maskb)))>>shb;\

  1182.         int r= ((pix0+pix1-g)&(maskr|(2*maskr)))>>shr;\

  1183.         g&= maskg|(2*maskg);\

  1184. \

  1185.         g>>=shg;\

  1186. \

  1187.         dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (256U<<(S)) + (1<<(S-6)))>>((S)-6+1);\

  1188.         dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (256U<<(S)) + (1<<(S-6)))>>((S)-6+1);\

  1189.     }\

  1190. }

  1191. 

  1192. BGR2UV(uint32_t, bgr32ToUV,16, 0, 0, 0, 0xFF0000, 0xFF00,   0x00FF, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1193. BGR2UV(uint32_t,bgr321ToUV,16, 0, 0, 8, 0xFF0000, 0xFF00,   0x00FF, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1194. BGR2UV(uint32_t, rgb32ToUV, 0, 0,16, 0,   0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1195. BGR2UV(uint32_t,rgb321ToUV, 0, 0,16, 8,   0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1196. BGR2UV(uint16_t, bgr16ToUV, 0, 0, 0, 0,   0x001F, 0x07E0,   0xF800, RU<<11, GU<<5, BU    , RV<<11, GV<<5, BV    , RGB2YUV_SHIFT+8)

  1197. BGR2UV(uint16_t, bgr15ToUV, 0, 0, 0, 0,   0x001F, 0x03E0,   0x7C00, RU<<10, GU<<5, BU    , RV<<10, GV<<5, BV    , RGB2YUV_SHIFT+7)

  1198. BGR2UV(uint16_t, rgb16ToUV, 0, 0, 0, 0,   0xF800, 0x07E0,   0x001F, RU    , GU<<5, BU<<11, RV    , GV<<5, BV<<11, RGB2YUV_SHIFT+8)

  1199. BGR2UV(uint16_t, rgb15ToUV, 0, 0, 0, 0,   0x7C00, 0x03E0,   0x001F, RU    , GU<<5, BU<<10, RV    , GV<<5, BV<<10, RGB2YUV_SHIFT+7)

  1200. 

  1201. static inline void palToA(int16_t *dst, const uint8_t *src, int width, uint32_t *pal)

  1202. {

  1203.     int i;

  1204.     for (i=0; i<width; i++) {

  1205.         int d= src[i];

  1206. 

  1207.         dst[i]= (pal[d] >> 24)<<6;

  1208.     }

  1209. }

  1210. 

  1211. static inline void palToY(int16_t *dst, const uint8_t *src, long width, uint32_t *pal)

  1212. {

  1213.     int i;

  1214.     for (i=0; i<width; i++) {

  1215.         int d= src[i];

  1216. 

  1217.         dst[i]= (pal[d] & 0xFF)<<6;

  1218.     }

  1219. }

  1220. 

  1221. static inline void palToUV(uint16_t *dstU, int16_t *dstV,

  1222.                            const uint8_t *src1, const uint8_t *src2,

  1223.                            int width, uint32_t *pal)

  1224. {

  1225.     int i;

  1226.     assert(src1 == src2);

  1227.     for (i=0; i<width; i++) {

  1228.         int p= pal[src1[i]];

  1229. 

  1230.         dstU[i]= (uint8_t)(p>> 8)<<6;

  1231.         dstV[i]= (uint8_t)(p>>16)<<6;

  1232.     }

  1233. }

  1234. 

  1235. static inline void monowhite2Y(int16_t *dst, const uint8_t *src, int width, uint32_t *unused)

  1236. {

  1237.     int i, j;

  1238.     for (i=0; i<width/8; i++) {

  1239.         int d= ~src[i];

  1240.         for(j=0; j<8; j++)

  1241.             dst[8*i+j]= ((d>>(7-j))&1)*16383;

  1242.     }

  1243. }

  1244. 

  1245. static inline void monoblack2Y(int16_t *dst, const uint8_t *src, int width, uint32_t *unused)

  1246. {

  1247.     int i, j;

  1248.     for (i=0; i<width/8; i++) {

  1249.         int d= src[i];

  1250.         for(j=0; j<8; j++)

  1251.             dst[8*i+j]= ((d>>(7-j))&1)*16383;

  1252.     }

  1253. }

  1254. 

  1255. //Note: we have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW+MMX2 one

  1256. //Plain C versions

  1257. 

  1258. #define COMPILE_TEMPLATE_MMX2 0

  1259. #define COMPILE_TEMPLATE_ALTIVEC 0

  1260. 

  1261. #include "swscale_template.c"

  1262. 

  1263. #if HAVE_ALTIVEC

  1264. #undef RENAME

  1265. #undef COMPILE_TEMPLATE_ALTIVEC

  1266. #define COMPILE_TEMPLATE_ALTIVEC 1

  1267. #define RENAME(a) a ## _altivec

  1268. #include "ppc/swscale_template.c"

  1269. #endif

  1270. 

  1271. //MMX versions

  1272. #if HAVE_MMX

  1273. #undef RENAME

  1274. #undef COMPILE_TEMPLATE_MMX2

  1275. #define COMPILE_TEMPLATE_MMX2 0

  1276. #define RENAME(a) a ## _MMX

  1277. #include "x86/swscale_template.c"

  1278. #endif

  1279. 

  1280. //MMX2 versions

  1281. #if HAVE_MMX2

  1282. #undef RENAME

  1283. #undef COMPILE_TEMPLATE_MMX2

  1284. #define COMPILE_TEMPLATE_MMX2 1

  1285. #define RENAME(a) a ## _MMX2

  1286. #include "x86/swscale_template.c"

  1287. #endif

  1288. 

  1289. SwsFunc ff_getSwsFunc(SwsContext *c)

  1290. {

  1291.     int cpu_flags = av_get_cpu_flags();

  1292. 

  1293.     sws_init_swScale_c(c);

  1294. 

  1295. #if HAVE_MMX

  1296.     if (cpu_flags & AV_CPU_FLAG_MMX)

  1297.         sws_init_swScale_MMX(c);

  1298. #endif

  1299. #if HAVE_MMX2

  1300.     if (cpu_flags & AV_CPU_FLAG_MMX2)

  1301.         sws_init_swScale_MMX2(c);

  1302. #endif

  1303. #if HAVE_ALTIVEC

  1304.     if (cpu_flags & AV_CPU_FLAG_ALTIVEC)

  1305.         sws_init_swScale_altivec(c);

  1306. #endif

  1307. 

  1308.     return swScale_c;

  1309. }

  1310. 

  1311. static void copyPlane(const uint8_t *src, int srcStride,

  1312.                       int srcSliceY, int srcSliceH, int width,

  1313.                       uint8_t *dst, int dstStride)

  1314. {

  1315.     dst += dstStride * srcSliceY;

  1316.     if (dstStride == srcStride && srcStride > 0) {

  1317.         memcpy(dst, src, srcSliceH * dstStride);

  1318.     } else {

  1319.         int i;

  1320.         for (i=0; i<srcSliceH; i++) {

  1321.             memcpy(dst, src, width);

  1322.             src += srcStride;

  1323.             dst += dstStride;

  1324.         }

  1325.     }

  1326. }

  1327. 

  1328. static int planarToNv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1329.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1330. {

  1331.     uint8_t *dst = dstParam[1] + dstStride[1]*srcSliceY/2;

  1332. 

  1333.     copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,

  1334.               dstParam[0], dstStride[0]);

  1335. 

  1336.     if (c->dstFormat == PIX_FMT_NV12)

  1337.         interleaveBytes(src[1], src[2], dst, c->srcW/2, srcSliceH/2, srcStride[1], srcStride[2], dstStride[0]);

  1338.     else

  1339.         interleaveBytes(src[2], src[1], dst, c->srcW/2, srcSliceH/2, srcStride[2], srcStride[1], dstStride[0]);

  1340. 

  1341.     return srcSliceH;

  1342. }

  1343. 

  1344. static int planarToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1345.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1346. {

  1347.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1348. 

  1349.     yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);

  1350. 

  1351.     return srcSliceH;

  1352. }

  1353. 

  1354. static int planarToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1355.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1356. {

  1357.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1358. 

  1359.     yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);

  1360. 

  1361.     return srcSliceH;

  1362. }

  1363. 

  1364. static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1365.                                 int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1366. {

  1367.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1368. 

  1369.     yuv422ptoyuy2(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);

  1370. 

  1371.     return srcSliceH;

  1372. }

  1373. 

  1374. static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1375.                                 int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1376. {

  1377.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1378. 

  1379.     yuv422ptouyvy(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);

  1380. 

  1381.     return srcSliceH;

  1382. }

  1383. 

  1384. static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1385.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1386. {

  1387.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1388.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;

  1389.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;

  1390. 

  1391.     yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1392. 

  1393.     if (dstParam[3])

  1394.         fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1395. 

  1396.     return srcSliceH;

  1397. }

  1398. 

  1399. static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1400.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1401. {

  1402.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1403.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;

  1404.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;

  1405. 

  1406.     yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1407. 

  1408.     return srcSliceH;

  1409. }

  1410. 

  1411. static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1412.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1413. {

  1414.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1415.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;

  1416.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;

  1417. 

  1418.     uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1419. 

  1420.     if (dstParam[3])

  1421.         fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1422. 

  1423.     return srcSliceH;

  1424. }

  1425. 

  1426. static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1427.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1428. {

  1429.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1430.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;

  1431.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;

  1432. 

  1433.     uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1434. 

  1435.     return srcSliceH;

  1436. }

  1437. 

  1438. static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)

  1439. {

  1440.     int i;

  1441.     for (i=0; i<num_pixels; i++)

  1442.         ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | (src[(i<<1)+1] << 24);

  1443. }

  1444. 

  1445. static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)

  1446. {

  1447.     int i;

  1448. 

  1449.     for (i=0; i<num_pixels; i++)

  1450.         ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | src[(i<<1)+1];

  1451. }

  1452. 

  1453. static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)

  1454. {

  1455.     int i;

  1456. 

  1457.     for (i=0; i<num_pixels; i++) {

  1458.         //FIXME slow?

  1459.         dst[0]= palette[src[i<<1]*4+0];

  1460.         dst[1]= palette[src[i<<1]*4+1];

  1461.         dst[2]= palette[src[i<<1]*4+2];

  1462.         dst+= 3;

  1463.     }

  1464. }

  1465. 

  1466. static int palToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1467.                            int srcSliceH, uint8_t* dst[], int dstStride[])

  1468. {

  1469.     const enum PixelFormat srcFormat= c->srcFormat;

  1470.     const enum PixelFormat dstFormat= c->dstFormat;

  1471.     void (*conv)(const uint8_t *src, uint8_t *dst, int num_pixels,

  1472.                  const uint8_t *palette)=NULL;

  1473.     int i;

  1474.     uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;

  1475.     const uint8_t *srcPtr= src[0];

  1476. 

  1477.     if (srcFormat == PIX_FMT_GRAY8A) {

  1478.         switch (dstFormat) {

  1479.         case PIX_FMT_RGB32  : conv = gray8aToPacked32; break;

  1480.         case PIX_FMT_BGR32  : conv = gray8aToPacked32; break;

  1481.         case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;

  1482.         case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;

  1483.         case PIX_FMT_RGB24  : conv = gray8aToPacked24; break;

  1484.         case PIX_FMT_BGR24  : conv = gray8aToPacked24; break;

  1485.         }

  1486.     } else if (usePal(srcFormat)) {

  1487.         switch (dstFormat) {

  1488.         case PIX_FMT_RGB32  : conv = sws_convertPalette8ToPacked32; break;

  1489.         case PIX_FMT_BGR32  : conv = sws_convertPalette8ToPacked32; break;

  1490.         case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;

  1491.         case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;

  1492.         case PIX_FMT_RGB24  : conv = sws_convertPalette8ToPacked24; break;

  1493.         case PIX_FMT_BGR24  : conv = sws_convertPalette8ToPacked24; break;

  1494.         }

  1495.     }

  1496. 

  1497.     if (!conv)

  1498.         av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",

  1499.                av_get_pix_fmt_name(srcFormat), av_get_pix_fmt_name(dstFormat));

  1500.     else {

  1501.         for (i=0; i<srcSliceH; i++) {

  1502.             conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);

  1503.             srcPtr+= srcStride[0];

  1504.             dstPtr+= dstStride[0];

  1505.         }

  1506.     }

  1507. 

  1508.     return srcSliceH;

  1509. }

  1510. 

  1511. #define isRGBA32(x) (            \

  1512.            (x) == PIX_FMT_ARGB   \

  1513.         || (x) == PIX_FMT_RGBA   \

  1514.         || (x) == PIX_FMT_BGRA   \

  1515.         || (x) == PIX_FMT_ABGR   \

  1516.         )

  1517. 

  1518. /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */

  1519. static int rgbToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1520.                            int srcSliceH, uint8_t* dst[], int dstStride[])

  1521. {

  1522.     const enum PixelFormat srcFormat= c->srcFormat;

  1523.     const enum PixelFormat dstFormat= c->dstFormat;

  1524.     const int srcBpp= (c->srcFormatBpp + 7) >> 3;

  1525.     const int dstBpp= (c->dstFormatBpp + 7) >> 3;

  1526.     const int srcId= c->srcFormatBpp >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */

  1527.     const int dstId= c->dstFormatBpp >> 2;

  1528.     void (*conv)(const uint8_t *src, uint8_t *dst, int src_size)=NULL;

  1529. 

  1530. #define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)

  1531. 

  1532.     if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {

  1533.         if (     CONV_IS(ABGR, RGBA)

  1534.               || CONV_IS(ARGB, BGRA)

  1535.               || CONV_IS(BGRA, ARGB)

  1536.               || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;

  1537.         else if (CONV_IS(ABGR, ARGB)

  1538.               || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;

  1539.         else if (CONV_IS(ABGR, BGRA)

  1540.               || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;

  1541.         else if (CONV_IS(BGRA, RGBA)

  1542.               || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;

  1543.         else if (CONV_IS(BGRA, ABGR)

  1544.               || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;

  1545.     } else

  1546.     /* BGR -> BGR */

  1547.     if (  (isBGRinInt(srcFormat) && isBGRinInt(dstFormat))

  1548.        || (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {

  1549.         switch(srcId | (dstId<<4)) {

  1550.         case 0x34: conv= rgb16to15; break;

  1551.         case 0x36: conv= rgb24to15; break;

  1552.         case 0x38: conv= rgb32to15; break;

  1553.         case 0x43: conv= rgb15to16; break;

  1554.         case 0x46: conv= rgb24to16; break;

  1555.         case 0x48: conv= rgb32to16; break;

  1556.         case 0x63: conv= rgb15to24; break;

  1557.         case 0x64: conv= rgb16to24; break;

  1558.         case 0x68: conv= rgb32to24; break;

  1559.         case 0x83: conv= rgb15to32; break;

  1560.         case 0x84: conv= rgb16to32; break;

  1561.         case 0x86: conv= rgb24to32; break;

  1562.         }

  1563.     } else if (  (isBGRinInt(srcFormat) && isRGBinInt(dstFormat))

  1564.              || (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {

  1565.         switch(srcId | (dstId<<4)) {

  1566.         case 0x33: conv= rgb15tobgr15; break;

  1567.         case 0x34: conv= rgb16tobgr15; break;

  1568.         case 0x36: conv= rgb24tobgr15; break;

  1569.         case 0x38: conv= rgb32tobgr15; break;

  1570.         case 0x43: conv= rgb15tobgr16; break;

  1571.         case 0x44: conv= rgb16tobgr16; break;

  1572.         case 0x46: conv= rgb24tobgr16; break;

  1573.         case 0x48: conv= rgb32tobgr16; break;

  1574.         case 0x63: conv= rgb15tobgr24; break;

  1575.         case 0x64: conv= rgb16tobgr24; break;

  1576.         case 0x66: conv= rgb24tobgr24; break;

  1577.         case 0x68: conv= rgb32tobgr24; break;

  1578.         case 0x83: conv= rgb15tobgr32; break;

  1579.         case 0x84: conv= rgb16tobgr32; break;

  1580.         case 0x86: conv= rgb24tobgr32; break;

  1581.         }

  1582.     }

  1583. 

  1584.     if (!conv) {

  1585.         av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",

  1586.                av_get_pix_fmt_name(srcFormat), av_get_pix_fmt_name(dstFormat));

  1587.     } else {

  1588.         const uint8_t *srcPtr= src[0];

  1589.               uint8_t *dstPtr= dst[0];

  1590.         if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) && !isRGBA32(dstFormat))

  1591.             srcPtr += ALT32_CORR;

  1592. 

  1593.         if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) && !isRGBA32(srcFormat))

  1594.             dstPtr += ALT32_CORR;

  1595. 

  1596.         if (dstStride[0]*srcBpp == srcStride[0]*dstBpp && srcStride[0] > 0 && !(srcStride[0]%srcBpp))

  1597.             conv(srcPtr, dstPtr + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]);

  1598.         else {

  1599.             int i;

  1600.             dstPtr += dstStride[0]*srcSliceY;

  1601. 

  1602.             for (i=0; i<srcSliceH; i++) {

  1603.                 conv(srcPtr, dstPtr, c->srcW*srcBpp);

  1604.                 srcPtr+= srcStride[0];

  1605.                 dstPtr+= dstStride[0];

  1606.             }

  1607.         }

  1608.     }

  1609.     return srcSliceH;

  1610. }

  1611. 

  1612. static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1613.                               int srcSliceH, uint8_t* dst[], int dstStride[])

  1614. {

  1615.     rgb24toyv12(

  1616.         src[0],

  1617.         dst[0]+ srcSliceY    *dstStride[0],

  1618.         dst[1]+(srcSliceY>>1)*dstStride[1],

  1619.         dst[2]+(srcSliceY>>1)*dstStride[2],

  1620.         c->srcW, srcSliceH,

  1621.         dstStride[0], dstStride[1], srcStride[0]);

  1622.     if (dst[3])

  1623.         fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1624.     return srcSliceH;

  1625. }

  1626. 

  1627. static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1628.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1629. {

  1630.     copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,

  1631.               dst[0], dstStride[0]);

  1632. 

  1633.     planar2x(src[1], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,

  1634.              srcSliceH >> 2, srcStride[1], dstStride[1]);

  1635.     planar2x(src[2], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,

  1636.              srcSliceH >> 2, srcStride[2], dstStride[2]);

  1637.     if (dst[3])

  1638.         fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1639.     return srcSliceH;

  1640. }

  1641. 

  1642. /* unscaled copy like stuff (assumes nearly identical formats) */

  1643. static int packedCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1644.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1645. {

  1646.     if (dstStride[0]==srcStride[0] && srcStride[0] > 0)

  1647.         memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]);

  1648.     else {

  1649.         int i;

  1650.         const uint8_t *srcPtr= src[0];

  1651.         uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;

  1652.         int length=0;

  1653. 

  1654.         /* universal length finder */

  1655.         while(length+c->srcW <= FFABS(dstStride[0])

  1656.            && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW;

  1657.         assert(length!=0);

  1658. 

  1659.         for (i=0; i<srcSliceH; i++) {

  1660.             memcpy(dstPtr, srcPtr, length);

  1661.             srcPtr+= srcStride[0];

  1662.             dstPtr+= dstStride[0];

  1663.         }

  1664.     }

  1665.     return srcSliceH;

  1666. }

  1667. 

  1668. #define DITHER_COPY(dst, dstStride, src, srcStride, bswap, dbswap)\

  1669.     uint16_t scale= dither_scale[dst_depth-1][src_depth-1];\

  1670.     int shift= src_depth-dst_depth + dither_scale[src_depth-2][dst_depth-1];\

  1671.     for (i = 0; i < height; i++) {\

  1672.         const uint8_t *dither= dithers[src_depth-9][i&7];\

  1673.         for (j = 0; j < length-7; j+=8){\

  1674.             dst[j+0] = dbswap((bswap(src[j+0]) + dither[0])*scale>>shift);\

  1675.             dst[j+1] = dbswap((bswap(src[j+1]) + dither[1])*scale>>shift);\

  1676.             dst[j+2] = dbswap((bswap(src[j+2]) + dither[2])*scale>>shift);\

  1677.             dst[j+3] = dbswap((bswap(src[j+3]) + dither[3])*scale>>shift);\

  1678.             dst[j+4] = dbswap((bswap(src[j+4]) + dither[4])*scale>>shift);\

  1679.             dst[j+5] = dbswap((bswap(src[j+5]) + dither[5])*scale>>shift);\

  1680.             dst[j+6] = dbswap((bswap(src[j+6]) + dither[6])*scale>>shift);\

  1681.             dst[j+7] = dbswap((bswap(src[j+7]) + dither[7])*scale>>shift);\

  1682.         }\

  1683.         for (; j < length; j++)\

  1684.             dst[j] = dbswap((bswap(src[j]) + dither[j&7])*scale>>shift);\

  1685.         dst += dstStride;\

  1686.         src += srcStride;\

  1687.     }

  1688. 

  1689. 

  1690. static int planarCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1691.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1692. {

  1693.     int plane, i, j;

  1694.     for (plane=0; plane<4; plane++) {

  1695.         int length= (plane==0 || plane==3) ? c->srcW  : -((-c->srcW  )>>c->chrDstHSubSample);

  1696.         int y=      (plane==0 || plane==3) ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample);

  1697.         int height= (plane==0 || plane==3) ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample);

  1698.         const uint8_t *srcPtr= src[plane];

  1699.         uint8_t *dstPtr= dst[plane] + dstStride[plane]*y;

  1700. 

  1701.         if (!dst[plane]) continue;

  1702.         // ignore palette for GRAY8

  1703.         if (plane == 1 && !dst[2]) continue;

  1704.         if (!src[plane] || (plane == 1 && !src[2])) {

  1705.             if(is16BPS(c->dstFormat))

  1706.                 length*=2;

  1707.             fillPlane(dst[plane], dstStride[plane], length, height, y, (plane==3) ? 255 : 128);

  1708.         } else {

  1709.             if(isNBPS(c->srcFormat) || isNBPS(c->dstFormat)

  1710.                || (is16BPS(c->srcFormat) != is16BPS(c->dstFormat))

  1711.             ) {

  1712.                 const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1+1;

  1713.                 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1+1;

  1714.                 const uint16_t *srcPtr2 = (const uint16_t*)srcPtr;

  1715.                 uint16_t *dstPtr2 = (uint16_t*)dstPtr;

  1716. 

  1717.                 if (dst_depth == 8) {

  1718.                     if(isBE(c->srcFormat) == HAVE_BIGENDIAN){

  1719.                         DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, , )

  1720.                     } else {

  1721.                         DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, av_bswap16, )

  1722.                     }

  1723.                 } else if (src_depth == 8) {

  1724.                     for (i = 0; i < height; i++) {

  1725.                         if(isBE(c->dstFormat)){

  1726.                             for (j = 0; j < length; j++)

  1727.                                 AV_WB16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |

  1728.                                                      (srcPtr[j]>>(2*8-dst_depth)));

  1729.                         } else {

  1730.                             for (j = 0; j < length; j++)

  1731.                                 AV_WL16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |

  1732.                                                      (srcPtr[j]>>(2*8-dst_depth)));

  1733.                         }

  1734.                         dstPtr2 += dstStride[plane]/2;

  1735.                         srcPtr  += srcStride[plane];

  1736.                     }

  1737.                 } else if (src_depth <= dst_depth) {

  1738.                     for (i = 0; i < height; i++) {

  1739. #define COPY_UP(r,w) \

  1740.     for (j = 0; j < length; j++){ \

  1741.         unsigned int v= r(&srcPtr2[j]);\

  1742.         w(&dstPtr2[j], (v<<(dst_depth-src_depth)) | \

  1743.                        (v>>(2*src_depth-dst_depth)));\

  1744.     }

  1745.                         if(isBE(c->srcFormat)){

  1746.                             if(isBE(c->dstFormat)){

  1747.                                 COPY_UP(AV_RB16, AV_WB16)

  1748.                             } else {

  1749.                                 COPY_UP(AV_RB16, AV_WL16)

  1750.                             }

  1751.                         } else {

  1752.                             if(isBE(c->dstFormat)){

  1753.                                 COPY_UP(AV_RL16, AV_WB16)

  1754.                             } else {

  1755.                                 COPY_UP(AV_RL16, AV_WL16)

  1756.                             }

  1757.                         }

  1758.                         dstPtr2 += dstStride[plane]/2;

  1759.                         srcPtr2 += srcStride[plane]/2;

  1760.                     }

  1761.                 } else {

  1762.                     if(isBE(c->srcFormat) == HAVE_BIGENDIAN){

  1763.                         if(isBE(c->dstFormat) == HAVE_BIGENDIAN){

  1764.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , )

  1765.                         } else {

  1766.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , av_bswap16)

  1767.                         }

  1768.                     }else{

  1769.                         if(isBE(c->dstFormat) == HAVE_BIGENDIAN){

  1770.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, )

  1771.                         } else {

  1772.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, av_bswap16)

  1773.                         }

  1774.                     }

  1775.                 }

  1776.             } else if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat)

  1777.                   && isBE(c->srcFormat) != isBE(c->dstFormat)) {

  1778. 

  1779.                 for (i=0; i<height; i++) {

  1780.                     for (j=0; j<length; j++)

  1781.                         ((uint16_t*)dstPtr)[j] = av_bswap16(((const uint16_t*)srcPtr)[j]);

  1782.                     srcPtr+= srcStride[plane];

  1783.                     dstPtr+= dstStride[plane];

  1784.                 }

  1785.             } else if (dstStride[plane] == srcStride[plane] &&

  1786.                        srcStride[plane] > 0 && srcStride[plane] == length) {

  1787.                 memcpy(dst[plane] + dstStride[plane]*y, src[plane],

  1788.                        height*dstStride[plane]);

  1789.             } else {

  1790.                 if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat))

  1791.                     length*=2;

  1792.                 for (i=0; i<height; i++) {

  1793.                     memcpy(dstPtr, srcPtr, length);

  1794.                     srcPtr+= srcStride[plane];

  1795.                     dstPtr+= dstStride[plane];

  1796.                 }

  1797.             }

  1798.         }

  1799.     }

  1800.     return srcSliceH;

  1801. }

  1802. 

  1803. void ff_get_unscaled_swscale(SwsContext *c)

  1804. {

  1805.     const enum PixelFormat srcFormat = c->srcFormat;

  1806.     const enum PixelFormat dstFormat = c->dstFormat;

  1807.     const int flags = c->flags;

  1808.     const int dstH = c->dstH;

  1809.     int needsDither;

  1810. 

  1811.     needsDither= isAnyRGB(dstFormat)

  1812.         &&  c->dstFormatBpp < 24

  1813.         && (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));

  1814. 

  1815.     /* yv12_to_nv12 */

  1816.     if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {

  1817.         c->swScale= planarToNv12Wrapper;

  1818.     }

  1819.     /* yuv2bgr */

  1820.     if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P || srcFormat==PIX_FMT_YUVA420P) && isAnyRGB(dstFormat)

  1821.         && !(flags & SWS_ACCURATE_RND) && !(dstH&1)) {

  1822.         c->swScale= ff_yuv2rgb_get_func_ptr(c);

  1823.     }

  1824. 

  1825.     if (srcFormat==PIX_FMT_YUV410P && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_BITEXACT)) {

  1826.         c->swScale= yvu9ToYv12Wrapper;

  1827.     }

  1828. 

  1829.     /* bgr24toYV12 */

  1830.     if (srcFormat==PIX_FMT_BGR24 && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_ACCURATE_RND))

  1831.         c->swScale= bgr24ToYv12Wrapper;

  1832. 

  1833.     /* RGB/BGR -> RGB/BGR (no dither needed forms) */

  1834.     if (   isAnyRGB(srcFormat)

  1835.         && isAnyRGB(dstFormat)

  1836.         && srcFormat != PIX_FMT_BGR8      && dstFormat != PIX_FMT_BGR8

  1837.         && srcFormat != PIX_FMT_RGB8      && dstFormat != PIX_FMT_RGB8

  1838.         && srcFormat != PIX_FMT_BGR4      && dstFormat != PIX_FMT_BGR4

  1839.         && srcFormat != PIX_FMT_RGB4      && dstFormat != PIX_FMT_RGB4

  1840.         && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE

  1841.         && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE

  1842.         && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK

  1843.         && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE

  1844.         && srcFormat != PIX_FMT_RGB48LE   && dstFormat != PIX_FMT_RGB48LE

  1845.         && srcFormat != PIX_FMT_RGB48BE   && dstFormat != PIX_FMT_RGB48BE

  1846.         && srcFormat != PIX_FMT_BGR48LE   && dstFormat != PIX_FMT_BGR48LE

  1847.         && srcFormat != PIX_FMT_BGR48BE   && dstFormat != PIX_FMT_BGR48BE

  1848.         && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))

  1849.         c->swScale= rgbToRgbWrapper;

  1850. 

  1851.     if ((usePal(srcFormat) && (

  1852.         dstFormat == PIX_FMT_RGB32   ||

  1853.         dstFormat == PIX_FMT_RGB32_1 ||

  1854.         dstFormat == PIX_FMT_RGB24   ||

  1855.         dstFormat == PIX_FMT_BGR32   ||

  1856.         dstFormat == PIX_FMT_BGR32_1 ||

  1857.         dstFormat == PIX_FMT_BGR24)))

  1858.         c->swScale= palToRgbWrapper;

  1859. 

  1860.     if (srcFormat == PIX_FMT_YUV422P) {

  1861.         if (dstFormat == PIX_FMT_YUYV422)

  1862.             c->swScale= yuv422pToYuy2Wrapper;

  1863.         else if (dstFormat == PIX_FMT_UYVY422)

  1864.             c->swScale= yuv422pToUyvyWrapper;

  1865.     }

  1866. 

  1867.     /* LQ converters if -sws 0 or -sws 4*/

  1868.     if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {

  1869.         /* yv12_to_yuy2 */

  1870.         if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {

  1871.             if (dstFormat == PIX_FMT_YUYV422)

  1872.                 c->swScale= planarToYuy2Wrapper;

  1873.             else if (dstFormat == PIX_FMT_UYVY422)

  1874.                 c->swScale= planarToUyvyWrapper;

  1875.         }

  1876.     }

  1877.     if(srcFormat == PIX_FMT_YUYV422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))

  1878.         c->swScale= yuyvToYuv420Wrapper;

  1879.     if(srcFormat == PIX_FMT_UYVY422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))

  1880.         c->swScale= uyvyToYuv420Wrapper;

  1881.     if(srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)

  1882.         c->swScale= yuyvToYuv422Wrapper;

  1883.     if(srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)

  1884.         c->swScale= uyvyToYuv422Wrapper;

  1885. 

  1886. #if HAVE_ALTIVEC

  1887.     if ((av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC) &&

  1888.         !(c->flags & SWS_BITEXACT) &&

  1889.         srcFormat == PIX_FMT_YUV420P) {

  1890.         // unscaled YV12 -> packed YUV, we want speed

  1891.         if (dstFormat == PIX_FMT_YUYV422)

  1892.             c->swScale= yv12toyuy2_unscaled_altivec;

  1893.         else if (dstFormat == PIX_FMT_UYVY422)

  1894.             c->swScale= yv12touyvy_unscaled_altivec;

  1895.     }

  1896. #endif

  1897. 

  1898.     /* simple copy */

  1899.     if (  srcFormat == dstFormat

  1900.         || (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P)

  1901.         || (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P)

  1902.         || (isPlanarYUV(srcFormat) && isGray(dstFormat))

  1903.         || (isPlanarYUV(dstFormat) && isGray(srcFormat))

  1904.         || (isGray(dstFormat) && isGray(srcFormat))

  1905.         || (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat)

  1906.             && c->chrDstHSubSample == c->chrSrcHSubSample

  1907.             && c->chrDstVSubSample == c->chrSrcVSubSample

  1908.             && dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21

  1909.             && srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))

  1910.     {

  1911.         if (isPacked(c->srcFormat))

  1912.             c->swScale= packedCopyWrapper;

  1913.         else /* Planar YUV or gray */

  1914.             c->swScale= planarCopyWrapper;

  1915.     }

  1916. #if ARCH_BFIN

  1917.     ff_bfin_get_unscaled_swscale (c);

  1918. #endif

  1919. }

  1920. 

  1921. static void reset_ptr(const uint8_t* src[], int format)

  1922. {

  1923.     if(!isALPHA(format))

  1924.         src[3]=NULL;

  1925.     if(!isPlanarYUV(format)) {

  1926.         src[3]=src[2]=NULL;

  1927. 

  1928.         if (!usePal(format))

  1929.             src[1]= NULL;

  1930.     }

  1931. }

  1932. 

  1933. static int check_image_pointers(uint8_t *data[4], enum PixelFormat pix_fmt,

  1934.                                 const int linesizes[4])

  1935. {

  1936.     const AVPixFmtDescriptor *desc = &av_pix_fmt_descriptors[pix_fmt];

  1937.     int i;

  1938. 

  1939.     for (i = 0; i < 4; i++) {

  1940.         int plane = desc->comp[i].plane;

  1941.         if (!data[plane] || !linesizes[plane])

  1942.             return 0;

  1943.     }

  1944. 

  1945.     return 1;

  1946. }

  1947. 

  1948. /**

  1949.  * swscale wrapper, so we don't need to export the SwsContext.

  1950.  * Assumes planar YUV to be in YUV order instead of YVU.

  1951.  */

  1952. int sws_scale(SwsContext *c, const uint8_t* const src[], const int srcStride[], int srcSliceY,

  1953.               int srcSliceH, uint8_t* const dst[], const int dstStride[])

  1954. {

  1955.     int i;

  1956.     const uint8_t* src2[4]= {src[0], src[1], src[2], src[3]};

  1957.     uint8_t* dst2[4]= {dst[0], dst[1], dst[2], dst[3]};

  1958. 

  1959.     // do not mess up sliceDir if we have a "trailing" 0-size slice

  1960.     if (srcSliceH == 0)

  1961.         return 0;

  1962. 

  1963.     if (!check_image_pointers(src, c->srcFormat, srcStride)) {

  1964.         av_log(c, AV_LOG_ERROR, "bad src image pointers\n");

  1965.         return 0;

  1966.     }

  1967.     if (!check_image_pointers(dst, c->dstFormat, dstStride)) {

  1968.         av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");

  1969.         return 0;

  1970.     }

  1971. 

  1972.     if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {

  1973.         av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");

  1974.         return 0;

  1975.     }

  1976.     if (c->sliceDir == 0) {

  1977.         if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;

  1978.     }

  1979. 

  1980.     if (usePal(c->srcFormat)) {

  1981.         for (i=0; i<256; i++) {

  1982.             int p, r, g, b, y, u, v, a = 0xff;

  1983.             if(c->srcFormat == PIX_FMT_PAL8) {

  1984.                 p=((const uint32_t*)(src[1]))[i];

  1985.                 a= (p>>24)&0xFF;

  1986.                 r= (p>>16)&0xFF;

  1987.                 g= (p>> 8)&0xFF;

  1988.                 b=  p     &0xFF;

  1989.             } else if(c->srcFormat == PIX_FMT_RGB8) {

  1990.                 r= (i>>5    )*36;

  1991.                 g= ((i>>2)&7)*36;

  1992.                 b= (i&3     )*85;

  1993.             } else if(c->srcFormat == PIX_FMT_BGR8) {

  1994.                 b= (i>>6    )*85;

  1995.                 g= ((i>>3)&7)*36;

  1996.                 r= (i&7     )*36;

  1997.             } else if(c->srcFormat == PIX_FMT_RGB4_BYTE) {

  1998.                 r= (i>>3    )*255;

  1999.                 g= ((i>>1)&3)*85;

  2000.                 b= (i&1     )*255;

  2001.             } else if(c->srcFormat == PIX_FMT_GRAY8 || c->srcFormat == PIX_FMT_GRAY8A) {

  2002.                 r = g = b = i;

  2003.             } else {

  2004.                 assert(c->srcFormat == PIX_FMT_BGR4_BYTE);

  2005.                 b= (i>>3    )*255;

  2006.                 g= ((i>>1)&3)*85;

  2007.                 r= (i&1     )*255;

  2008.             }

  2009.             y= av_clip_uint8((RY*r + GY*g + BY*b + ( 33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2010.             u= av_clip_uint8((RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2011.             v= av_clip_uint8((RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2012.             c->pal_yuv[i]= y + (u<<8) + (v<<16) + (a<<24);

  2013. 

  2014.             switch(c->dstFormat) {

  2015.             case PIX_FMT_BGR32:

  2016. #if !HAVE_BIGENDIAN

  2017.             case PIX_FMT_RGB24:

  2018. #endif

  2019.                 c->pal_rgb[i]=  r + (g<<8) + (b<<16) + (a<<24);

  2020.                 break;

  2021.             case PIX_FMT_BGR32_1:

  2022. #if HAVE_BIGENDIAN

  2023.             case PIX_FMT_BGR24:

  2024. #endif

  2025.                 c->pal_rgb[i]= a + (r<<8) + (g<<16) + (b<<24);

  2026.                 break;

  2027.             case PIX_FMT_RGB32_1:

  2028. #if HAVE_BIGENDIAN

  2029.             case PIX_FMT_RGB24:

  2030. #endif

  2031.                 c->pal_rgb[i]= a + (b<<8) + (g<<16) + (r<<24);

  2032.                 break;

  2033.             case PIX_FMT_RGB32:

  2034. #if !HAVE_BIGENDIAN

  2035.             case PIX_FMT_BGR24:

  2036. #endif

  2037.             default:

  2038.                 c->pal_rgb[i]=  b + (g<<8) + (r<<16) + (a<<24);

  2039.             }

  2040.         }

  2041.     }

  2042. 

  2043.     // copy strides, so they can safely be modified

  2044.     if (c->sliceDir == 1) {

  2045.         // slices go from top to bottom

  2046.         int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2], srcStride[3]};

  2047.         int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2], dstStride[3]};

  2048. 

  2049.         reset_ptr(src2, c->srcFormat);

  2050.         reset_ptr((const uint8_t**)dst2, c->dstFormat);

  2051. 

  2052.         /* reset slice direction at end of frame */

  2053.         if (srcSliceY + srcSliceH == c->srcH)

  2054.             c->sliceDir = 0;

  2055. 

  2056.         return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2, dstStride2);

  2057.     } else {

  2058.         // slices go from bottom to top => we flip the image internally

  2059.         int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2], -srcStride[3]};

  2060.         int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2], -dstStride[3]};

  2061. 

  2062.         src2[0] += (srcSliceH-1)*srcStride[0];

  2063.         if (!usePal(c->srcFormat))

  2064.             src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1];

  2065.         src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2];

  2066.         src2[3] += (srcSliceH-1)*srcStride[3];

  2067.         dst2[0] += ( c->dstH                      -1)*dstStride[0];

  2068.         dst2[1] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1];

  2069.         dst2[2] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2];

  2070.         dst2[3] += ( c->dstH                      -1)*dstStride[3];

  2071. 

  2072.         reset_ptr(src2, c->srcFormat);

  2073.         reset_ptr((const uint8_t**)dst2, c->dstFormat);

  2074. 

  2075.         /* reset slice direction at end of frame */

  2076.         if (!srcSliceY)

  2077.             c->sliceDir = 0;

  2078. 

  2079.         return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2);

  2080.     }

  2081. }

  2082. 

  2083. #if LIBSWSCALE_VERSION_MAJOR < 1

  2084. int sws_scale_ordered(SwsContext *c, const uint8_t* const src[], int srcStride[], int srcSliceY,

  2085.                       int srcSliceH, uint8_t* dst[], int dstStride[])

  2086. {

  2087.     return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride);

  2088. }

  2089. #endif

  2090. 

  2091. /* Convert the palette to the same packed 32-bit format as the palette */

  2092. void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)

  2093. {

  2094.     int i;

  2095. 

  2096.     for (i=0; i<num_pixels; i++)

  2097.         ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];

  2098. }

  2099. 

  2100. /* Palette format: ABCD -> dst format: ABC */

  2101. void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)

  2102. {

  2103.     int i;

  2104. 

  2105.     for (i=0; i<num_pixels; i++) {

  2106.         //FIXME slow?

  2107.         dst[0]= palette[src[i]*4+0];

  2108.         dst[1]= palette[src[i]*4+1];

  2109.         dst[2]= palette[src[i]*4+2];

  2110.         dst+= 3;

  2111.     }

  2112. }