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.
30184 Commits
32 Branches
291MB
Tree: 040e1c67ed
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '040e1c67ed'
${ noResults }
FFmpeg/libswscale/swscale.c

2343 lines
84KB
Raw Blame History 

  1. /*

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

  3.  *

  4.  * This file is part of FFmpeg.

  5.  *

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

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

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

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

  10.  *

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

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

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

  14.  * Lesser General Public License for more details.

  15.  *

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

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

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

  19.  */

  20. 

  21. /*

  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/intreadwrite.h"

  64. #include "libavutil/x86_cpu.h"

  65. #include "libavutil/avutil.h"

  66. #include "libavutil/mathematics.h"

  67. #include "libavutil/bswap.h"

  68. #include "libavutil/pixdesc.h"

  69. 

  70. #undef MOVNTQ

  71. #undef PAVGB

  72. 

  73. //#undef HAVE_MMX2

  74. //#define HAVE_AMD3DNOW

  75. //#undef HAVE_MMX

  76. //#undef ARCH_X86

  77. #define DITHER1XBPP

  78. 

  79. #define isPacked(x)         (       \

  80.            (x)==PIX_FMT_PAL8        \

  81.         || (x)==PIX_FMT_YUYV422     \

  82.         || (x)==PIX_FMT_UYVY422     \

  83.         || (x)==PIX_FMT_GRAY8A       \

  84.         || isAnyRGB(x)              \

  85.     )

  86. 

  87. #define RGB2YUV_SHIFT 15

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

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

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

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

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

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

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

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

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

  97. 

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

  99.     {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5}, //ITU709

  100.     {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5}, //ITU709

  101.     {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M

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

  104.     {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M

  105.     {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //DEFAULT / ITU601 / ITU624 / SMPTE 170M

  106.     {0.701 , 0.087 , 0.212 , -0.384, 0.5, -0.116, -0.445, -0.055, 0.5}, //SMPTE 240M

  107. };

  108. 

  109. /*

  110. NOTES

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

  112. 

  113. TODO

  114. more intelligent misalignment avoidance for the horizontal scaler

  115. write special vertical cubic upscale version

  116. optimize C code (YV12 / minmax)

  117. add support for packed pixel YUV input & output

  118. add support for Y8 output

  119. optimize BGR24 & BGR32

  120. add BGR4 output support

  121. write special BGR->BGR scaler

  122. */

  123. 

  124. #if ARCH_X86

  125. DECLARE_ASM_CONST(8, uint64_t, bF8)=       0xF8F8F8F8F8F8F8F8LL;

  126. DECLARE_ASM_CONST(8, uint64_t, bFC)=       0xFCFCFCFCFCFCFCFCLL;

  127. DECLARE_ASM_CONST(8, uint64_t, w10)=       0x0010001000100010LL;

  128. DECLARE_ASM_CONST(8, uint64_t, w02)=       0x0002000200020002LL;

  129. DECLARE_ASM_CONST(8, uint64_t, bm00001111)=0x00000000FFFFFFFFLL;

  130. DECLARE_ASM_CONST(8, uint64_t, bm00000111)=0x0000000000FFFFFFLL;

  131. DECLARE_ASM_CONST(8, uint64_t, bm11111000)=0xFFFFFFFFFF000000LL;

  132. DECLARE_ASM_CONST(8, uint64_t, bm01010101)=0x00FF00FF00FF00FFLL;

  133. 

  134. const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {

  135.         0x0103010301030103LL,

  136.         0x0200020002000200LL,};

  137. 

  138. const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {

  139.         0x0602060206020602LL,

  140.         0x0004000400040004LL,};

  141. 

  142. DECLARE_ASM_CONST(8, uint64_t, b16Mask)=   0x001F001F001F001FLL;

  143. DECLARE_ASM_CONST(8, uint64_t, g16Mask)=   0x07E007E007E007E0LL;

  144. DECLARE_ASM_CONST(8, uint64_t, r16Mask)=   0xF800F800F800F800LL;

  145. DECLARE_ASM_CONST(8, uint64_t, b15Mask)=   0x001F001F001F001FLL;

  146. DECLARE_ASM_CONST(8, uint64_t, g15Mask)=   0x03E003E003E003E0LL;

  147. DECLARE_ASM_CONST(8, uint64_t, r15Mask)=   0x7C007C007C007C00LL;

  148. 

  149. DECLARE_ALIGNED(8, const uint64_t, ff_M24A)         = 0x00FF0000FF0000FFLL;

  150. DECLARE_ALIGNED(8, const uint64_t, ff_M24B)         = 0xFF0000FF0000FF00LL;

  151. DECLARE_ALIGNED(8, const uint64_t, ff_M24C)         = 0x0000FF0000FF0000LL;

  152. 

  153. #ifdef FAST_BGR2YV12

  154. DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff)   = 0x000000210041000DULL;

  155. DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff)   = 0x0000FFEEFFDC0038ULL;

  156. DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff)   = 0x00000038FFD2FFF8ULL;

  157. #else

  158. DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff)   = 0x000020E540830C8BULL;

  159. DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff)   = 0x0000ED0FDAC23831ULL;

  160. DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff)   = 0x00003831D0E6F6EAULL;

  161. #endif /* FAST_BGR2YV12 */

  162. DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset)  = 0x1010101010101010ULL;

  163. DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;

  164. DECLARE_ALIGNED(8, const uint64_t, ff_w1111)        = 0x0001000100010001ULL;

  165. 

  166. DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY1Coeff) = 0x0C88000040870C88ULL;

  167. DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY2Coeff) = 0x20DE4087000020DEULL;

  168. DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY1Coeff) = 0x20DE0000408720DEULL;

  169. DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY2Coeff) = 0x0C88408700000C88ULL;

  170. DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toYOffset) = 0x0008400000084000ULL;

  171. 

  172. DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUV)[2][4] = {

  173.     {0x38380000DAC83838ULL, 0xECFFDAC80000ECFFULL, 0xF6E40000D0E3F6E4ULL, 0x3838D0E300003838ULL},

  174.     {0xECFF0000DAC8ECFFULL, 0x3838DAC800003838ULL, 0x38380000D0E33838ULL, 0xF6E4D0E30000F6E4ULL},

  175. };

  176. 

  177. DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUVOffset)= 0x0040400000404000ULL;

  178. 

  179. #endif /* ARCH_X86 */

  180. 

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

  182. {  1,   3,   1,   3,   1,   3,   1,   3, },

  183. {  2,   0,   2,   0,   2,   0,   2,   0, },

  184. };

  185. 

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

  187. {  6,   2,   6,   2,   6,   2,   6,   2, },

  188. {  0,   4,   0,   4,   0,   4,   0,   4, },

  189. };

  190. 

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

  192. {  8,   4,  11,   7,   8,   4,  11,   7, },

  193. {  2,  14,   1,  13,   2,  14,   1,  13, },

  194. { 10,   6,   9,   5,  10,   6,   9,   5, },

  195. {  0,  12,   3,  15,   0,  12,   3,  15, },

  196. };

  197. 

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

  199. { 17,   9,  23,  15,  16,   8,  22,  14, },

  200. {  5,  29,   3,  27,   4,  28,   2,  26, },

  201. { 21,  13,  19,  11,  20,  12,  18,  10, },

  202. {  0,  24,   6,  30,   1,  25,   7,  31, },

  203. { 16,   8,  22,  14,  17,   9,  23,  15, },

  204. {  4,  28,   2,  26,   5,  29,   3,  27, },

  205. { 20,  12,  18,  10,  21,  13,  19,  11, },

  206. {  1,  25,   7,  31,   0,  24,   6,  30, },

  207. };

  208. 

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

  210. {  0,  55,  14,  68,   3,  58,  17,  72, },

  211. { 37,  18,  50,  32,  40,  22,  54,  35, },

  212. {  9,  64,   5,  59,  13,  67,   8,  63, },

  213. { 46,  27,  41,  23,  49,  31,  44,  26, },

  214. {  2,  57,  16,  71,   1,  56,  15,  70, },

  215. { 39,  21,  52,  34,  38,  19,  51,  33, },

  216. { 11,  66,   7,  62,  10,  65,   6,  60, },

  217. { 48,  30,  43,  25,  47,  29,  42,  24, },

  218. };

  219. 

  220. #if 1

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

  222. {117,  62, 158, 103, 113,  58, 155, 100, },

  223. { 34, 199,  21, 186,  31, 196,  17, 182, },

  224. {144,  89, 131,  76, 141,  86, 127,  72, },

  225. {  0, 165,  41, 206,  10, 175,  52, 217, },

  226. {110,  55, 151,  96, 120,  65, 162, 107, },

  227. { 28, 193,  14, 179,  38, 203,  24, 189, },

  228. {138,  83, 124,  69, 148,  93, 134,  79, },

  229. {  7, 172,  48, 213,   3, 168,  45, 210, },

  230. };

  231. #elif 1

  232. // tries to correct a gamma of 1.5

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

  234. {  0, 143,  18, 200,   2, 156,  25, 215, },

  235. { 78,  28, 125,  64,  89,  36, 138,  74, },

  236. { 10, 180,   3, 161,  16, 195,   8, 175, },

  237. {109,  51,  93,  38, 121,  60, 105,  47, },

  238. {  1, 152,  23, 210,   0, 147,  20, 205, },

  239. { 85,  33, 134,  71,  81,  30, 130,  67, },

  240. { 14, 190,   6, 171,  12, 185,   5, 166, },

  241. {117,  57, 101,  44, 113,  54,  97,  41, },

  242. };

  243. #elif 1

  244. // tries to correct a gamma of 2.0

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

  246. {  0, 124,   8, 193,   0, 140,  12, 213, },

  247. { 55,  14, 104,  42,  66,  19, 119,  52, },

  248. {  3, 168,   1, 145,   6, 187,   3, 162, },

  249. { 86,  31,  70,  21,  99,  39,  82,  28, },

  250. {  0, 134,  11, 206,   0, 129,   9, 200, },

  251. { 62,  17, 114,  48,  58,  16, 109,  45, },

  252. {  5, 181,   2, 157,   4, 175,   1, 151, },

  253. { 95,  36,  78,  26,  90,  34,  74,  24, },

  254. };

  255. #else

  256. // tries to correct a gamma of 2.5

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

  258. {  0, 107,   3, 187,   0, 125,   6, 212, },

  259. { 39,   7,  86,  28,  49,  11, 102,  36, },

  260. {  1, 158,   0, 131,   3, 180,   1, 151, },

  261. { 68,  19,  52,  12,  81,  25,  64,  17, },

  262. {  0, 119,   5, 203,   0, 113,   4, 195, },

  263. { 45,   9,  96,  33,  42,   8,  91,  30, },

  264. {  2, 172,   1, 144,   2, 165,   0, 137, },

  265. { 77,  23,  60,  15,  72,  21,  56,  14, },

  266. };

  267. #endif

  268. 

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

  270. {

  271.   {   0,  1,  0,  1,  0,  1,  0,  1,},

  272.   {   1,  0,  1,  0,  1,  0,  1,  0,},

  273.   {   0,  1,  0,  1,  0,  1,  0,  1,},

  274.   {   1,  0,  1,  0,  1,  0,  1,  0,},

  275.   {   0,  1,  0,  1,  0,  1,  0,  1,},

  276.   {   1,  0,  1,  0,  1,  0,  1,  0,},

  277.   {   0,  1,  0,  1,  0,  1,  0,  1,},

  278.   {   1,  0,  1,  0,  1,  0,  1,  0,},

  279. },{

  280.   {   1,  2,  1,  2,  1,  2,  1,  2,},

  281.   {   3,  0,  3,  0,  3,  0,  3,  0,},

  282.   {   1,  2,  1,  2,  1,  2,  1,  2,},

  283.   {   3,  0,  3,  0,  3,  0,  3,  0,},

  284.   {   1,  2,  1,  2,  1,  2,  1,  2,},

  285.   {   3,  0,  3,  0,  3,  0,  3,  0,},

  286.   {   1,  2,  1,  2,  1,  2,  1,  2,},

  287.   {   3,  0,  3,  0,  3,  0,  3,  0,},

  288. },{

  289.   {   2,  4,  3,  5,  2,  4,  3,  5,},

  290.   {   6,  0,  7,  1,  6,  0,  7,  1,},

  291.   {   3,  5,  2,  4,  3,  5,  2,  4,},

  292.   {   7,  1,  6,  0,  7,  1,  6,  0,},

  293.   {   2,  4,  3,  5,  2,  4,  3,  5,},

  294.   {   6,  0,  7,  1,  6,  0,  7,  1,},

  295.   {   3,  5,  2,  4,  3,  5,  2,  4,},

  296.   {   7,  1,  6,  0,  7,  1,  6,  0,},

  297. },{

  298.   {   4,  8,  7, 11,  4,  8,  7, 11,},

  299.   {  12,  0, 15,  3, 12,  0, 15,  3,},

  300.   {   6, 10,  5,  9,  6, 10,  5,  9,},

  301.   {  14,  2, 13,  1, 14,  2, 13,  1,},

  302.   {   4,  8,  7, 11,  4,  8,  7, 11,},

  303.   {  12,  0, 15,  3, 12,  0, 15,  3,},

  304.   {   6, 10,  5,  9,  6, 10,  5,  9,},

  305.   {  14,  2, 13,  1, 14,  2, 13,  1,},

  306. },{

  307.   {   9, 17, 15, 23,  8, 16, 14, 22,},

  308.   {  25,  1, 31,  7, 24,  0, 30,  6,},

  309.   {  13, 21, 11, 19, 12, 20, 10, 18,},

  310.   {  29,  5, 27,  3, 28,  4, 26,  2,},

  311.   {   8, 16, 14, 22,  9, 17, 15, 23,},

  312.   {  24,  0, 30,  6, 25,  1, 31,  7,},

  313.   {  12, 20, 10, 18, 13, 21, 11, 19,},

  314.   {  28,  4, 26,  2, 29,  5, 27,  3,},

  315. },{

  316.   {  18, 34, 30, 46, 17, 33, 29, 45,},

  317.   {  50,  2, 62, 14, 49,  1, 61, 13,},

  318.   {  26, 42, 22, 38, 25, 41, 21, 37,},

  319.   {  58, 10, 54,  6, 57,  9, 53,  5,},

  320.   {  16, 32, 28, 44, 19, 35, 31, 47,},

  321.   {  48,  0, 60, 12, 51,  3, 63, 15,},

  322.   {  24, 40, 20, 36, 27, 43, 23, 39,},

  323.   {  56,  8, 52,  4, 59, 11, 55,  7,},

  324. },{

  325.   {  18, 34, 30, 46, 17, 33, 29, 45,},

  326.   {  50,  2, 62, 14, 49,  1, 61, 13,},

  327.   {  26, 42, 22, 38, 25, 41, 21, 37,},

  328.   {  58, 10, 54,  6, 57,  9, 53,  5,},

  329.   {  16, 32, 28, 44, 19, 35, 31, 47,},

  330.   {  48,  0, 60, 12, 51,  3, 63, 15,},

  331.   {  24, 40, 20, 36, 27, 43, 23, 39,},

  332.   {  56,  8, 52,  4, 59, 11, 55,  7,},

  333. },{

  334.   {  36, 68, 60, 92, 34, 66, 58, 90,},

  335.   { 100,  4,124, 28, 98,  2,122, 26,},

  336.   {  52, 84, 44, 76, 50, 82, 42, 74,},

  337.   { 116, 20,108, 12,114, 18,106, 10,},

  338.   {  32, 64, 56, 88, 38, 70, 62, 94,},

  339.   {  96,  0,120, 24,102,  6,126, 30,},

  340.   {  48, 80, 40, 72, 54, 86, 46, 78,},

  341.   { 112, 16,104,  8,118, 22,110, 14,},

  342. }};

  343. 

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

  345. {    2,    3,    3,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,    5,},

  346. {    2,    3,    7,    7,   13,   13,   25,   25,   25,   25,   25,   25,   25,   25,   25,   25,},

  347. {    3,    3,    4,   15,   15,   29,   57,   57,   57,  113,  113,  113,  113,  113,  113,  113,},

  348. {    3,    4,    4,    5,   31,   31,   61,  121,  241,  241,  241,  241,  481,  481,  481,  481,},

  349. {    3,    4,    5,    5,    6,   63,   63,  125,  249,  497,  993,  993,  993,  993,  993, 1985,},

  350. {    3,    5,    6,    6,    6,    7,  127,  127,  253,  505, 1009, 2017, 4033, 4033, 4033, 4033,},

  351. {    3,    5,    6,    7,    7,    7,    8,  255,  255,  509, 1017, 2033, 4065, 8129,16257,16257,},

  352. {    3,    5,    6,    8,    8,    8,    8,    9,  511,  511, 1021, 2041, 4081, 8161,16321,32641,},

  353. {    3,    5,    7,    8,    9,    9,    9,    9,   10, 1023, 1023, 2045, 4089, 8177,16353,32705,},

  354. {    3,    5,    7,    8,   10,   10,   10,   10,   10,   11, 2047, 2047, 4093, 8185,16369,32737,},

  355. {    3,    5,    7,    8,   10,   11,   11,   11,   11,   11,   12, 4095, 4095, 8189,16377,32753,},

  356. {    3,    5,    7,    9,   10,   12,   12,   12,   12,   12,   12,   13, 8191, 8191,16381,32761,},

  357. {    3,    5,    7,    9,   10,   12,   13,   13,   13,   13,   13,   13,   14,16383,16383,32765,},

  358. {    3,    5,    7,    9,   10,   12,   14,   14,   14,   14,   14,   14,   14,   15,32767,32767,},

  359. {    3,    5,    7,    9,   11,   12,   14,   15,   15,   15,   15,   15,   15,   15,   16,65535,},

  360. };

  361. 

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

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

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

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

  366. {

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

  368.     int i;

  369.     int shift = 11 + 16 - output_bits;

  370. 

  371. #define output_pixel(pos, val) \

  372.     if (big_endian) { \

  373.         if (output_bits == 16) { \

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

  375.         } else { \

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

  377.         } \

  378.     } else { \

  379.         if (output_bits == 16) { \

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

  381.         } else { \

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

  383.         } \

  384.     }

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

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

  387.         int j;

  388. 

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

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

  391. 

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

  393.     }

  394. 

  395.     if (uDest) {

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

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

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

  399.             int j;

  400. 

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

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

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

  404.             }

  405. 

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

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

  408.         }

  409.     }

  410. 

  411.     if (CONFIG_SWSCALE_ALPHA && aDest) {

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

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

  414.             int j;

  415. 

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

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

  418. 

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

  420.         }

  421.     }

  422. }

  423. 

  424. static av_always_inline void yuv2yuvXNinC_template(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

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

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

  427.                                                    int dstW, int chrDstW, int big_endian, int depth)

  428. {

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

  430.     int i;

  431. 

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

  433.         int val = 1 << (26-depth);

  434.         int j;

  435. 

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

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

  438. 

  439.         if (big_endian) {

  440.             AV_WB16(&dest[i], av_clip(val >> (27-depth), 0, (1<<depth)-1));

  441.         } else {

  442.             AV_WL16(&dest[i], av_clip(val >> (27-depth), 0, (1<<depth)-1));

  443.         }

  444.     }

  445. 

  446.     if (uDest) {

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

  448.             int u = 1 << (26-depth);

  449.             int v = 1 << (26-depth);

  450.             int j;

  451. 

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

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

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

  455.             }

  456. 

  457.             if (big_endian) {

  458.                 AV_WB16(&uDest[i], av_clip(u >> (27-depth), 0, (1<<depth)-1));

  459.                 AV_WB16(&vDest[i], av_clip(v >> (27-depth), 0, (1<<depth)-1));

  460.             } else {

  461.                 AV_WL16(&uDest[i], av_clip(u >> (27-depth), 0, (1<<depth)-1));

  462.                 AV_WL16(&vDest[i], av_clip(v >> (27-depth), 0, (1<<depth)-1));

  463.             }

  464.         }

  465.     }

  466. }

  467. 

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

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

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

  471.                                  enum PixelFormat dstFormat)

  472. {

  473.     if (isNBPS(dstFormat)) {

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

  475.         yuv2yuvXNinC_template(lumFilter, lumSrc, lumFilterSize,

  476.                               chrFilter, chrSrc, chrFilterSize,

  477.                               alpSrc,

  478.                               dest, uDest, vDest, aDest,

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

  480.     } else {

  481.         if (isBE(dstFormat)) {

  482.             yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,

  483.                                    chrFilter, chrSrc, chrFilterSize,

  484.                                    alpSrc,

  485.                                    dest, uDest, vDest, aDest,

  486.                                    dstW, chrDstW, 1, 16);

  487.         } else {

  488.             yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,

  489.                                    chrFilter, chrSrc, chrFilterSize,

  490.                                    alpSrc,

  491.                                    dest, uDest, vDest, aDest,

  492.                                    dstW, chrDstW, 0, 16);

  493.         }

  494.     }

  495. }

  496. 

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

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

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

  500. {

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

  502.     int i;

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

  504.         int val=1<<18;

  505.         int j;

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

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

  508. 

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

  510.     }

  511. 

  512.     if (uDest)

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

  514.             int u=1<<18;

  515.             int v=1<<18;

  516.             int j;

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

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

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

  520.             }

  521. 

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

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

  524.         }

  525. 

  526.     if (CONFIG_SWSCALE_ALPHA && aDest)

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

  528.             int val=1<<18;

  529.             int j;

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

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

  532. 

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

  534.         }

  535. 

  536. }

  537. 

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

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

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

  541. {

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

  543.     int i;

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

  545.         int val=1<<18;

  546.         int j;

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

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

  549. 

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

  551.     }

  552. 

  553.     if (!uDest)

  554.         return;

  555. 

  556.     if (dstFormat == PIX_FMT_NV12)

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

  558.             int u=1<<18;

  559.             int v=1<<18;

  560.             int j;

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

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

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

  564.             }

  565. 

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

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

  568.         }

  569.     else

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

  571.             int u=1<<18;

  572.             int v=1<<18;

  573.             int j;

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

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

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

  577.             }

  578. 

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

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

  581.         }

  582. }

  583. 

  584. #define YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha) \

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

  586.         int j;\

  587.         int Y1 = 1<<18;\

  588.         int Y2 = 1<<18;\

  589.         int U  = 1<<18;\

  590.         int V  = 1<<18;\

  591.         int av_unused A1, A2;\

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

  593.         const int i2= 2*i;\

  594.         \

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

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

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

  598.         }\

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

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

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

  602.         }\

  603.         Y1>>=19;\

  604.         Y2>>=19;\

  605.         U >>=19;\

  606.         V >>=19;\

  607.         if (alpha) {\

  608.             A1 = 1<<18;\

  609.             A2 = 1<<18;\

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

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

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

  613.             }\

  614.             A1>>=19;\

  615.             A2>>=19;\

  616.         }

  617. 

  618. #define YSCALE_YUV_2_PACKEDX_C(type,alpha) \

  619.         YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha)\

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

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

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

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

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

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

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

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

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

  629.         }\

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

  631.             A1=av_clip_uint8(A1);\

  632.             A2=av_clip_uint8(A2);\

  633.         }

  634. 

  635. #define YSCALE_YUV_2_PACKEDX_FULL_C(rnd,alpha) \

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

  637.         int j;\

  638.         int Y = 0;\

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

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

  641.         int av_unused A;\

  642.         int R,G,B;\

  643.         \

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

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

  646.         }\

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

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

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

  650.         }\

  651.         Y >>=10;\

  652.         U >>=10;\

  653.         V >>=10;\

  654.         if (alpha) {\

  655.             A = rnd;\

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

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

  658.             A >>=19;\

  659.             if (A&256)\

  660.                 A = av_clip_uint8(A);\

  661.         }

  662. 

  663. #define YSCALE_YUV_2_RGBX_FULL_C(rnd,alpha) \

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

  665.         Y-= c->yuv2rgb_y_offset;\

  666.         Y*= c->yuv2rgb_y_coeff;\

  667.         Y+= rnd;\

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

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

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

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

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

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

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

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

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

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

  678.         }

  679. 

  680. #define YSCALE_YUV_2_GRAY16_C \

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

  682.         int j;\

  683.         int Y1 = 1<<18;\

  684.         int Y2 = 1<<18;\

  685.         int U  = 1<<18;\

  686.         int V  = 1<<18;\

  687.         \

  688.         const int i2= 2*i;\

  689.         \

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

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

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

  693.         }\

  694.         Y1>>=11;\

  695.         Y2>>=11;\

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

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

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

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

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

  701.         }

  702. 

  703. #define YSCALE_YUV_2_RGBX_C(type,alpha) \

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

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

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

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

  708. 

  709. #define YSCALE_YUV_2_PACKED2_C(type,alpha)   \

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

  711.         const int i2= 2*i;       \

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

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

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

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

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

  717.         int av_unused A1, A2;                                         \

  718.         if (alpha) {\

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

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

  721.         }

  722. 

  723. #define YSCALE_YUV_2_GRAY16_2_C   \

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

  725.         const int i2= 2*i;       \

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

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

  728. 

  729. #define YSCALE_YUV_2_RGB2_C(type,alpha) \

  730.     YSCALE_YUV_2_PACKED2_C(type,alpha)\

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

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

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

  734. 

  735. #define YSCALE_YUV_2_PACKED1_C(type,alpha) \

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

  737.         const int i2= 2*i;\

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

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

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

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

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

  743.         int av_unused A1, A2;\

  744.         if (alpha) {\

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

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

  747.         }

  748. 

  749. #define YSCALE_YUV_2_GRAY16_1_C \

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

  751.         const int i2= 2*i;\

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

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

  754. 

  755. #define YSCALE_YUV_2_RGB1_C(type,alpha) \

  756.     YSCALE_YUV_2_PACKED1_C(type,alpha)\

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

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

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

  760. 

  761. #define YSCALE_YUV_2_PACKED1B_C(type,alpha) \

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

  763.         const int i2= 2*i;\

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

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

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

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

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

  769.         int av_unused A1, A2;\

  770.         if (alpha) {\

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

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

  773.         }

  774. 

  775. #define YSCALE_YUV_2_RGB1B_C(type,alpha) \

  776.     YSCALE_YUV_2_PACKED1B_C(type,alpha)\

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

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

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

  780. 

  781. #define YSCALE_YUV_2_MONO2_C \

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

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

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

  785.         int acc;\

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

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

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

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

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

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

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

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

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

  795.         dest++;\

  796.     }

  797. 

  798. #define YSCALE_YUV_2_MONOX_C \

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

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

  801.     int acc=0;\

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

  803.         int j;\

  804.         int Y1=1<<18;\

  805.         int Y2=1<<18;\

  806. \

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

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

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

  810.         }\

  811.         Y1>>=19;\

  812.         Y2>>=19;\

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

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

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

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

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

  818.         }\

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

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

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

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

  823.             dest++;\

  824.         }\

  825.     }

  826. 

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

  828.     switch(c->dstFormat) {\

  829.     case PIX_FMT_RGB48BE:\

  830.     case PIX_FMT_RGB48LE:\

  831.         func(uint8_t,0)\

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

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

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

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

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

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

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

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

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

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

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

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

  844.             dest+=12;\

  845.         }\

  846.         break;\

  847.     case PIX_FMT_BGR48BE:\

  848.     case PIX_FMT_BGR48LE:\

  849.         func(uint8_t,0)\

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

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

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

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

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

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

  856.             dest+=12;\

  857.         }\

  858.         break;\

  859.     case PIX_FMT_RGBA:\

  860.     case PIX_FMT_BGRA:\

  861.         if (CONFIG_SMALL) {\

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

  863.             func(uint32_t,needAlpha)\

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

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

  866.             }\

  867.         } else {\

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

  869.                 func(uint32_t,1)\

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

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

  872.                 }\

  873.             } else {\

  874.                 func(uint32_t,0)\

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

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

  877.                 }\

  878.             }\

  879.         }\

  880.         break;\

  881.     case PIX_FMT_ARGB:\

  882.     case PIX_FMT_ABGR:\

  883.         if (CONFIG_SMALL) {\

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

  885.             func(uint32_t,needAlpha)\

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

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

  888.             }\

  889.         } else {\

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

  891.                 func(uint32_t,1)\

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

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

  894.                 }\

  895.             } else {\

  896.                 func(uint32_t,0)\

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

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

  899.                 }\

  900.             }\

  901.         }                \

  902.         break;\

  903.     case PIX_FMT_RGB24:\

  904.         func(uint8_t,0)\

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

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

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

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

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

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

  911.             dest+=6;\

  912.         }\

  913.         break;\

  914.     case PIX_FMT_BGR24:\

  915.         func(uint8_t,0)\

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

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

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

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

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

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

  922.             dest+=6;\

  923.         }\

  924.         break;\

  925.     case PIX_FMT_RGB565BE:\

  926.     case PIX_FMT_RGB565LE:\

  927.     case PIX_FMT_BGR565BE:\

  928.     case PIX_FMT_BGR565LE:\

  929.         {\

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

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

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

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

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

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

  936.             func(uint16_t,0)\

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

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

  939.             }\

  940.         }\

  941.         break;\

  942.     case PIX_FMT_RGB555BE:\

  943.     case PIX_FMT_RGB555LE:\

  944.     case PIX_FMT_BGR555BE:\

  945.     case PIX_FMT_BGR555LE:\

  946.         {\

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

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

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

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

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

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

  953.             func(uint16_t,0)\

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

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

  956.             }\

  957.         }\

  958.         break;\

  959.     case PIX_FMT_RGB444BE:\

  960.     case PIX_FMT_RGB444LE:\

  961.     case PIX_FMT_BGR444BE:\

  962.     case PIX_FMT_BGR444LE:\

  963.         {\

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

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

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

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

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

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

  970.             func(uint16_t,0)\

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

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

  973.             }\

  974.         }\

  975.         break;\

  976.     case PIX_FMT_RGB8:\

  977.     case PIX_FMT_BGR8:\

  978.         {\

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

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

  981.             func(uint8_t,0)\

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

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

  984.             }\

  985.         }\

  986.         break;\

  987.     case PIX_FMT_RGB4:\

  988.     case PIX_FMT_BGR4:\

  989.         {\

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

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

  992.             func(uint8_t,0)\

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

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

  995.             }\

  996.         }\

  997.         break;\

  998.     case PIX_FMT_RGB4_BYTE:\

  999.     case PIX_FMT_BGR4_BYTE:\

  1000.         {\

  1001.             const uint8_t * const d64= dither_8x8_73 [y&7];\

  1002.             const uint8_t * const d128=dither_8x8_220[y&7];\

  1003.             func(uint8_t,0)\

  1004.                 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\

  1005.                 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\

  1006.             }\

  1007.         }\

  1008.         break;\

  1009.     case PIX_FMT_MONOBLACK:\

  1010.     case PIX_FMT_MONOWHITE:\

  1011.         {\

  1012.             func_monoblack\

  1013.         }\

  1014.         break;\

  1015.     case PIX_FMT_YUYV422:\

  1016.         func2\

  1017.             ((uint8_t*)dest)[2*i2+0]= Y1;\

  1018.             ((uint8_t*)dest)[2*i2+1]= U;\

  1019.             ((uint8_t*)dest)[2*i2+2]= Y2;\

  1020.             ((uint8_t*)dest)[2*i2+3]= V;\

  1021.         }                \

  1022.         break;\

  1023.     case PIX_FMT_UYVY422:\

  1024.         func2\

  1025.             ((uint8_t*)dest)[2*i2+0]= U;\

  1026.             ((uint8_t*)dest)[2*i2+1]= Y1;\

  1027.             ((uint8_t*)dest)[2*i2+2]= V;\

  1028.             ((uint8_t*)dest)[2*i2+3]= Y2;\

  1029.         }                \

  1030.         break;\

  1031.     case PIX_FMT_GRAY16BE:\

  1032.         func_g16\

  1033.             ((uint8_t*)dest)[2*i2+0]= Y1>>8;\

  1034.             ((uint8_t*)dest)[2*i2+1]= Y1;\

  1035.             ((uint8_t*)dest)[2*i2+2]= Y2>>8;\

  1036.             ((uint8_t*)dest)[2*i2+3]= Y2;\

  1037.         }                \

  1038.         break;\

  1039.     case PIX_FMT_GRAY16LE:\

  1040.         func_g16\

  1041.             ((uint8_t*)dest)[2*i2+0]= Y1;\

  1042.             ((uint8_t*)dest)[2*i2+1]= Y1>>8;\

  1043.             ((uint8_t*)dest)[2*i2+2]= Y2;\

  1044.             ((uint8_t*)dest)[2*i2+3]= Y2>>8;\

  1045.         }                \

  1046.         break;\

  1047.     }

  1048. 

  1049. static inline void yuv2packedXinC(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

  1050.                                   const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,

  1051.                                   const int16_t **alpSrc, uint8_t *dest, int dstW, int y)

  1052. {

  1053.     int i;

  1054.     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)

  1055. }

  1056. 

  1057. static inline void yuv2rgbXinC_full(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,

  1058.                                     const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,

  1059.                                     const int16_t **alpSrc, uint8_t *dest, int dstW, int y)

  1060. {

  1061.     int i;

  1062.     int step= c->dstFormatBpp/8;

  1063.     int aidx= 3;

  1064. 

  1065.     switch(c->dstFormat) {

  1066.     case PIX_FMT_ARGB:

  1067.         dest++;

  1068.         aidx= 0;

  1069.     case PIX_FMT_RGB24:

  1070.         aidx--;

  1071.     case PIX_FMT_RGBA:

  1072.         if (CONFIG_SMALL) {

  1073.             int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;

  1074.             YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)

  1075.                 dest[aidx]= needAlpha ? A : 255;

  1076.                 dest[0]= R>>22;

  1077.                 dest[1]= G>>22;

  1078.                 dest[2]= B>>22;

  1079.                 dest+= step;

  1080.             }

  1081.         } else {

  1082.             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  1083.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)

  1084.                     dest[aidx]= A;

  1085.                     dest[0]= R>>22;

  1086.                     dest[1]= G>>22;

  1087.                     dest[2]= B>>22;

  1088.                     dest+= step;

  1089.                 }

  1090.             } else {

  1091.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)

  1092.                     dest[aidx]= 255;

  1093.                     dest[0]= R>>22;

  1094.                     dest[1]= G>>22;

  1095.                     dest[2]= B>>22;

  1096.                     dest+= step;

  1097.                 }

  1098.             }

  1099.         }

  1100.         break;

  1101.     case PIX_FMT_ABGR:

  1102.         dest++;

  1103.         aidx= 0;

  1104.     case PIX_FMT_BGR24:

  1105.         aidx--;

  1106.     case PIX_FMT_BGRA:

  1107.         if (CONFIG_SMALL) {

  1108.             int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;

  1109.             YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)

  1110.                 dest[aidx]= needAlpha ? A : 255;

  1111.                 dest[0]= B>>22;

  1112.                 dest[1]= G>>22;

  1113.                 dest[2]= R>>22;

  1114.                 dest+= step;

  1115.             }

  1116.         } else {

  1117.             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {

  1118.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)

  1119.                     dest[aidx]= A;

  1120.                     dest[0]= B>>22;

  1121.                     dest[1]= G>>22;

  1122.                     dest[2]= R>>22;

  1123.                     dest+= step;

  1124.                 }

  1125.             } else {

  1126.                 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)

  1127.                     dest[aidx]= 255;

  1128.                     dest[0]= B>>22;

  1129.                     dest[1]= G>>22;

  1130.                     dest[2]= R>>22;

  1131.                     dest+= step;

  1132.                 }

  1133.             }

  1134.         }

  1135.         break;

  1136.     default:

  1137.         assert(0);

  1138.     }

  1139. }

  1140. 

  1141. static void fillPlane(uint8_t* plane, int stride, int width, int height, int y, uint8_t val)

  1142. {

  1143.     int i;

  1144.     uint8_t *ptr = plane + stride*y;

  1145.     for (i=0; i<height; i++) {

  1146.         memset(ptr, val, width);

  1147.         ptr += stride;

  1148.     }

  1149. }

  1150. 

  1151. static inline void rgb48ToY(uint8_t *dst, const uint8_t *src, long width,

  1152.                             uint32_t *unused)

  1153. {

  1154.     int i;

  1155.     for (i = 0; i < width; i++) {

  1156.         int r = src[i*6+0];

  1157.         int g = src[i*6+2];

  1158.         int b = src[i*6+4];

  1159. 

  1160.         dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;

  1161.     }

  1162. }

  1163. 

  1164. static inline void rgb48ToUV(uint8_t *dstU, uint8_t *dstV,

  1165.                              const uint8_t *src1, const uint8_t *src2,

  1166.                              long width, uint32_t *unused)

  1167. {

  1168.     int i;

  1169.     assert(src1==src2);

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

  1171.         int r = src1[6*i + 0];

  1172.         int g = src1[6*i + 2];

  1173.         int b = src1[6*i + 4];

  1174. 

  1175.         dstU[i] = (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;

  1176.         dstV[i] = (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;

  1177.     }

  1178. }

  1179. 

  1180. static inline void rgb48ToUV_half(uint8_t *dstU, uint8_t *dstV,

  1181.                                   const uint8_t *src1, const uint8_t *src2,

  1182.                                   long width, uint32_t *unused)

  1183. {

  1184.     int i;

  1185.     assert(src1==src2);

  1186.     for (i = 0; i < width; i++) {

  1187.         int r= src1[12*i + 0] + src1[12*i + 6];

  1188.         int g= src1[12*i + 2] + src1[12*i + 8];

  1189.         int b= src1[12*i + 4] + src1[12*i + 10];

  1190. 

  1191.         dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);

  1192.         dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);

  1193.     }

  1194. }

  1195. 

  1196. static inline void bgr48ToY(uint8_t *dst, const uint8_t *src, long width,

  1197.                             uint32_t *unused)

  1198. {

  1199.     int i;

  1200.     for (i = 0; i < width; i++) {

  1201.         int b = src[i*6+0];

  1202.         int g = src[i*6+2];

  1203.         int r = src[i*6+4];

  1204. 

  1205.         dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;

  1206.     }

  1207. }

  1208. 

  1209. static inline void bgr48ToUV(uint8_t *dstU, uint8_t *dstV,

  1210.                              const uint8_t *src1, const uint8_t *src2,

  1211.                              long width, uint32_t *unused)

  1212. {

  1213.     int i;

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

  1215.         int b = src1[6*i + 0];

  1216.         int g = src1[6*i + 2];

  1217.         int r = src1[6*i + 4];

  1218. 

  1219.         dstU[i] = (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;

  1220.         dstV[i] = (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;

  1221.     }

  1222. }

  1223. 

  1224. static inline void bgr48ToUV_half(uint8_t *dstU, uint8_t *dstV,

  1225.                                   const uint8_t *src1, const uint8_t *src2,

  1226.                                   long width, uint32_t *unused)

  1227. {

  1228.     int i;

  1229.     for (i = 0; i < width; i++) {

  1230.         int b= src1[12*i + 0] + src1[12*i + 6];

  1231.         int g= src1[12*i + 2] + src1[12*i + 8];

  1232.         int r= src1[12*i + 4] + src1[12*i + 10];

  1233. 

  1234.         dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);

  1235.         dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);

  1236.     }

  1237. }

  1238. 

  1239. #define BGR2Y(type, name, shr, shg, shb, maskr, maskg, maskb, RY, GY, BY, S)\

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

  1241. {\

  1242.     int i;\

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

  1244.         int b= (((const type*)src)[i]>>shb)&maskb;\

  1245.         int g= (((const type*)src)[i]>>shg)&maskg;\

  1246.         int r= (((const type*)src)[i]>>shr)&maskr;\

  1247. \

  1248.         dst[i]= (((RY)*r + (GY)*g + (BY)*b + (33<<((S)-1)))>>(S));\

  1249.     }\

  1250. }

  1251. 

  1252. BGR2Y(uint32_t, bgr32ToY,16, 0, 0, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY   , BY<< 8, RGB2YUV_SHIFT+8)

  1253. BGR2Y(uint32_t,bgr321ToY,16,16, 0, 0xFF00, 0x00FF, 0xFF00, RY    , GY<<8, BY    , RGB2YUV_SHIFT+8)

  1254. BGR2Y(uint32_t, rgb32ToY, 0, 0,16, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY   , BY<< 8, RGB2YUV_SHIFT+8)

  1255. BGR2Y(uint32_t,rgb321ToY, 0,16,16, 0xFF00, 0x00FF, 0xFF00, RY    , GY<<8, BY    , RGB2YUV_SHIFT+8)

  1256. BGR2Y(uint16_t, bgr16ToY, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RY<<11, GY<<5, BY    , RGB2YUV_SHIFT+8)

  1257. BGR2Y(uint16_t, bgr15ToY, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RY<<10, GY<<5, BY    , RGB2YUV_SHIFT+7)

  1258. BGR2Y(uint16_t, rgb16ToY, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RY    , GY<<5, BY<<11, RGB2YUV_SHIFT+8)

  1259. BGR2Y(uint16_t, rgb15ToY, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RY    , GY<<5, BY<<10, RGB2YUV_SHIFT+7)

  1260. 

  1261. static inline void abgrToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)

  1262. {

  1263.     int i;

  1264.     for (i=0; i<width; i++) {

  1265.         dst[i]= src[4*i];

  1266.     }

  1267. }

  1268. 

  1269. #define BGR2UV(type, name, shr, shg, shb, shp, maskr, maskg, maskb, RU, GU, BU, RV, GV, BV, S) \

  1270. static inline void name(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\

  1271. {\

  1272.     int i;\

  1273.     for (i=0; i<width; i++) {\

  1274.         int b= ((((const type*)src)[i]>>shp)&maskb)>>shb;\

  1275.         int g= ((((const type*)src)[i]>>shp)&maskg)>>shg;\

  1276.         int r= ((((const type*)src)[i]>>shp)&maskr)>>shr;\

  1277. \

  1278.         dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<((S)-1)))>>(S);\

  1279.         dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<((S)-1)))>>(S);\

  1280.     }\

  1281. }\

  1282. static inline void name ## _half(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\

  1283. {\

  1284.     int i;\

  1285.     for (i=0; i<width; i++) {\

  1286.         int pix0= ((const type*)src)[2*i+0]>>shp;\

  1287.         int pix1= ((const type*)src)[2*i+1]>>shp;\

  1288.         int g= (pix0&~(maskr|maskb))+(pix1&~(maskr|maskb));\

  1289.         int b= ((pix0+pix1-g)&(maskb|(2*maskb)))>>shb;\

  1290.         int r= ((pix0+pix1-g)&(maskr|(2*maskr)))>>shr;\

  1291.         g&= maskg|(2*maskg);\

  1292. \

  1293.         g>>=shg;\

  1294. \

  1295.         dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<(S)))>>((S)+1);\

  1296.         dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<(S)))>>((S)+1);\

  1297.     }\

  1298. }

  1299. 

  1300. BGR2UV(uint32_t, bgr32ToUV,16, 0, 0, 0, 0xFF0000, 0xFF00,   0x00FF, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1301. BGR2UV(uint32_t,bgr321ToUV,16, 0, 0, 8, 0xFF0000, 0xFF00,   0x00FF, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1302. BGR2UV(uint32_t, rgb32ToUV, 0, 0,16, 0,   0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1303. BGR2UV(uint32_t,rgb321ToUV, 0, 0,16, 8,   0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU   , BU<< 8, RV<< 8, GV   , BV<< 8, RGB2YUV_SHIFT+8)

  1304. BGR2UV(uint16_t, bgr16ToUV, 0, 0, 0, 0,   0x001F, 0x07E0,   0xF800, RU<<11, GU<<5, BU    , RV<<11, GV<<5, BV    , RGB2YUV_SHIFT+8)

  1305. BGR2UV(uint16_t, bgr15ToUV, 0, 0, 0, 0,   0x001F, 0x03E0,   0x7C00, RU<<10, GU<<5, BU    , RV<<10, GV<<5, BV    , RGB2YUV_SHIFT+7)

  1306. BGR2UV(uint16_t, rgb16ToUV, 0, 0, 0, 0,   0xF800, 0x07E0,   0x001F, RU    , GU<<5, BU<<11, RV    , GV<<5, BV<<11, RGB2YUV_SHIFT+8)

  1307. BGR2UV(uint16_t, rgb15ToUV, 0, 0, 0, 0,   0x7C00, 0x03E0,   0x001F, RU    , GU<<5, BU<<10, RV    , GV<<5, BV<<10, RGB2YUV_SHIFT+7)

  1308. 

  1309. static inline void palToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)

  1310. {

  1311.     int i;

  1312.     for (i=0; i<width; i++) {

  1313.         int d= src[i];

  1314. 

  1315.         dst[i]= pal[d] >> 24;

  1316.     }

  1317. }

  1318. 

  1319. static inline void palToY(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)

  1320. {

  1321.     int i;

  1322.     for (i=0; i<width; i++) {

  1323.         int d= src[i];

  1324. 

  1325.         dst[i]= pal[d] & 0xFF;

  1326.     }

  1327. }

  1328. 

  1329. static inline void palToUV(uint8_t *dstU, uint8_t *dstV,

  1330.                            const uint8_t *src1, const uint8_t *src2,

  1331.                            long width, uint32_t *pal)

  1332. {

  1333.     int i;

  1334.     assert(src1 == src2);

  1335.     for (i=0; i<width; i++) {

  1336.         int p= pal[src1[i]];

  1337. 

  1338.         dstU[i]= p>>8;

  1339.         dstV[i]= p>>16;

  1340.     }

  1341. }

  1342. 

  1343. static inline void monowhite2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)

  1344. {

  1345.     int i, j;

  1346.     for (i=0; i<width/8; i++) {

  1347.         int d= ~src[i];

  1348.         for(j=0; j<8; j++)

  1349.             dst[8*i+j]= ((d>>(7-j))&1)*255;

  1350.     }

  1351. }

  1352. 

  1353. static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)

  1354. {

  1355.     int i, j;

  1356.     for (i=0; i<width/8; i++) {

  1357.         int d= src[i];

  1358.         for(j=0; j<8; j++)

  1359.             dst[8*i+j]= ((d>>(7-j))&1)*255;

  1360.     }

  1361. }

  1362. 

  1363. //Note: we have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW+MMX2 one

  1364. //Plain C versions

  1365. #if CONFIG_RUNTIME_CPUDETECT

  1366. #  define COMPILE_C 1

  1367. #  if   ARCH_X86

  1368. #    define COMPILE_MMX     1

  1369. #    define COMPILE_MMX2    1

  1370. #    define COMPILE_3DNOW   1

  1371. #  elif ARCH_PPC

  1372. #    define COMPILE_ALTIVEC HAVE_ALTIVEC

  1373. #  endif

  1374. #else /* CONFIG_RUNTIME_CPUDETECT */

  1375. #  if   ARCH_X86

  1376. #    if   HAVE_MMX2

  1377. #      define COMPILE_MMX2  1

  1378. #    elif HAVE_AMD3DNOW

  1379. #      define COMPILE_3DNOW 1

  1380. #    elif HAVE_MMX

  1381. #      define COMPILE_MMX   1

  1382. #    else

  1383. #      define COMPILE_C     1

  1384. #    endif

  1385. #  elif ARCH_PPC && HAVE_ALTIVEC

  1386. #    define COMPILE_ALTIVEC 1

  1387. #  else

  1388. #    define COMPILE_C       1

  1389. #  endif

  1390. #endif

  1391. 

  1392. #ifndef COMPILE_C

  1393. #  define COMPILE_C 0

  1394. #endif

  1395. #ifndef COMPILE_MMX

  1396. #  define COMPILE_MMX 0

  1397. #endif

  1398. #ifndef COMPILE_MMX2

  1399. #  define COMPILE_MMX2 0

  1400. #endif

  1401. #ifndef COMPILE_3DNOW

  1402. #  define COMPILE_3DNOW 0

  1403. #endif

  1404. #ifndef COMPILE_ALTIVEC

  1405. #  define COMPILE_ALTIVEC 0

  1406. #endif

  1407. 

  1408. #define COMPILE_TEMPLATE_MMX 0

  1409. #define COMPILE_TEMPLATE_MMX2 0

  1410. #define COMPILE_TEMPLATE_AMD3DNOW 0

  1411. #define COMPILE_TEMPLATE_ALTIVEC 0

  1412. 

  1413. #if COMPILE_C

  1414. #define RENAME(a) a ## _C

  1415. #include "swscale_template.c"

  1416. #endif

  1417. 

  1418. #if COMPILE_ALTIVEC

  1419. #undef RENAME

  1420. #undef COMPILE_TEMPLATE_ALTIVEC

  1421. #define COMPILE_TEMPLATE_ALTIVEC 1

  1422. #define RENAME(a) a ## _altivec

  1423. #include "swscale_template.c"

  1424. #endif

  1425. 

  1426. #if ARCH_X86

  1427. 

  1428. //MMX versions

  1429. #if COMPILE_MMX

  1430. #undef RENAME

  1431. #undef COMPILE_TEMPLATE_MMX

  1432. #undef COMPILE_TEMPLATE_MMX2

  1433. #undef COMPILE_TEMPLATE_AMD3DNOW

  1434. #define COMPILE_TEMPLATE_MMX 1

  1435. #define COMPILE_TEMPLATE_MMX2 0

  1436. #define COMPILE_TEMPLATE_AMD3DNOW 0

  1437. #define RENAME(a) a ## _MMX

  1438. #include "swscale_template.c"

  1439. #endif

  1440. 

  1441. //MMX2 versions

  1442. #if COMPILE_MMX2

  1443. #undef RENAME

  1444. #undef COMPILE_TEMPLATE_MMX

  1445. #undef COMPILE_TEMPLATE_MMX2

  1446. #undef COMPILE_TEMPLATE_AMD3DNOW

  1447. #define COMPILE_TEMPLATE_MMX 1

  1448. #define COMPILE_TEMPLATE_MMX2 1

  1449. #define COMPILE_TEMPLATE_AMD3DNOW 0

  1450. #define RENAME(a) a ## _MMX2

  1451. #include "swscale_template.c"

  1452. #endif

  1453. 

  1454. //3DNOW versions

  1455. #if COMPILE_3DNOW

  1456. #undef RENAME

  1457. #undef COMPILE_TEMPLATE_MMX

  1458. #undef COMPILE_TEMPLATE_MMX2

  1459. #undef COMPILE_TEMPLATE_AMD3DNOW

  1460. #define COMPILE_TEMPLATE_MMX 1

  1461. #define COMPILE_TEMPLATE_MMX2 0

  1462. #define COMPILE_TEMPLATE_AMD3DNOW 1

  1463. #define RENAME(a) a ## _3DNow

  1464. #include "swscale_template.c"

  1465. #endif

  1466. 

  1467. #endif //ARCH_X86

  1468. 

  1469. SwsFunc ff_getSwsFunc(SwsContext *c)

  1470. {

  1471. #if CONFIG_RUNTIME_CPUDETECT

  1472.     int flags = c->flags;

  1473. 

  1474. #if ARCH_X86

  1475.     // ordered per speed fastest first

  1476.     if (flags & SWS_CPU_CAPS_MMX2) {

  1477.         sws_init_swScale_MMX2(c);

  1478.         return swScale_MMX2;

  1479.     } else if (flags & SWS_CPU_CAPS_3DNOW) {

  1480.         sws_init_swScale_3DNow(c);

  1481.         return swScale_3DNow;

  1482.     } else if (flags & SWS_CPU_CAPS_MMX) {

  1483.         sws_init_swScale_MMX(c);

  1484.         return swScale_MMX;

  1485.     } else {

  1486.         sws_init_swScale_C(c);

  1487.         return swScale_C;

  1488.     }

  1489. 

  1490. #else

  1491. #if COMPILE_ALTIVEC

  1492.     if (flags & SWS_CPU_CAPS_ALTIVEC) {

  1493.         sws_init_swScale_altivec(c);

  1494.         return swScale_altivec;

  1495.     } else {

  1496.         sws_init_swScale_C(c);

  1497.         return swScale_C;

  1498.     }

  1499. #endif

  1500.     sws_init_swScale_C(c);

  1501.     return swScale_C;

  1502. #endif /* ARCH_X86 */

  1503. #else //CONFIG_RUNTIME_CPUDETECT

  1504. #if   COMPILE_TEMPLATE_MMX2

  1505.     sws_init_swScale_MMX2(c);

  1506.     return swScale_MMX2;

  1507. #elif COMPILE_TEMPLATE_AMD3DNOW

  1508.     sws_init_swScale_3DNow(c);

  1509.     return swScale_3DNow;

  1510. #elif COMPILE_TEMPLATE_MMX

  1511.     sws_init_swScale_MMX(c);

  1512.     return swScale_MMX;

  1513. #elif COMPILE_TEMPLATE_ALTIVEC

  1514.     sws_init_swScale_altivec(c);

  1515.     return swScale_altivec;

  1516. #else

  1517.     sws_init_swScale_C(c);

  1518.     return swScale_C;

  1519. #endif

  1520. #endif //!CONFIG_RUNTIME_CPUDETECT

  1521. }

  1522. 

  1523. static void copyPlane(const uint8_t *src, int srcStride,

  1524.                       int srcSliceY, int srcSliceH, int width,

  1525.                       uint8_t *dst, int dstStride)

  1526. {

  1527.     dst += dstStride * srcSliceY;

  1528.     if (dstStride == srcStride && srcStride > 0) {

  1529.         memcpy(dst, src, srcSliceH * dstStride);

  1530.     } else {

  1531.         int i;

  1532.         for (i=0; i<srcSliceH; i++) {

  1533.             memcpy(dst, src, width);

  1534.             src += srcStride;

  1535.             dst += dstStride;

  1536.         }

  1537.     }

  1538. }

  1539. 

  1540. static int planarToNv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1541.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1542. {

  1543.     uint8_t *dst = dstParam[1] + dstStride[1]*srcSliceY/2;

  1544. 

  1545.     copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,

  1546.               dstParam[0], dstStride[0]);

  1547. 

  1548.     if (c->dstFormat == PIX_FMT_NV12)

  1549.         interleaveBytes(src[1], src[2], dst, c->srcW/2, srcSliceH/2, srcStride[1], srcStride[2], dstStride[0]);

  1550.     else

  1551.         interleaveBytes(src[2], src[1], dst, c->srcW/2, srcSliceH/2, srcStride[2], srcStride[1], dstStride[0]);

  1552. 

  1553.     return srcSliceH;

  1554. }

  1555. 

  1556. static int planarToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1557.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1558. {

  1559.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1560. 

  1561.     yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);

  1562. 

  1563.     return srcSliceH;

  1564. }

  1565. 

  1566. static int planarToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1567.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1568. {

  1569.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1570. 

  1571.     yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);

  1572. 

  1573.     return srcSliceH;

  1574. }

  1575. 

  1576. static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1577.                                 int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1578. {

  1579.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1580. 

  1581.     yuv422ptoyuy2(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);

  1582. 

  1583.     return srcSliceH;

  1584. }

  1585. 

  1586. static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1587.                                 int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1588. {

  1589.     uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;

  1590. 

  1591.     yuv422ptouyvy(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);

  1592. 

  1593.     return srcSliceH;

  1594. }

  1595. 

  1596. static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1597.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1598. {

  1599.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1600.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;

  1601.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;

  1602. 

  1603.     yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1604. 

  1605.     if (dstParam[3])

  1606.         fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1607. 

  1608.     return srcSliceH;

  1609. }

  1610. 

  1611. static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1612.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1613. {

  1614.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1615.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;

  1616.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;

  1617. 

  1618.     yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1619. 

  1620.     return srcSliceH;

  1621. }

  1622. 

  1623. static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1624.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1625. {

  1626.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1627.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;

  1628.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;

  1629. 

  1630.     uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1631. 

  1632.     if (dstParam[3])

  1633.         fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1634. 

  1635.     return srcSliceH;

  1636. }

  1637. 

  1638. static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1639.                                int srcSliceH, uint8_t* dstParam[], int dstStride[])

  1640. {

  1641.     uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;

  1642.     uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;

  1643.     uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;

  1644. 

  1645.     uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);

  1646. 

  1647.     return srcSliceH;

  1648. }

  1649. 

  1650. static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  1651. {

  1652.     long i;

  1653.     for (i=0; i<num_pixels; i++)

  1654.         ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | (src[(i<<1)+1] << 24);

  1655. }

  1656. 

  1657. static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  1658. {

  1659.     long i;

  1660. 

  1661.     for (i=0; i<num_pixels; i++)

  1662.         ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | src[(i<<1)+1];

  1663. }

  1664. 

  1665. static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  1666. {

  1667.     long i;

  1668. 

  1669.     for (i=0; i<num_pixels; i++) {

  1670.         //FIXME slow?

  1671.         dst[0]= palette[src[i<<1]*4+0];

  1672.         dst[1]= palette[src[i<<1]*4+1];

  1673.         dst[2]= palette[src[i<<1]*4+2];

  1674.         dst+= 3;

  1675.     }

  1676. }

  1677. 

  1678. static int palToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1679.                            int srcSliceH, uint8_t* dst[], int dstStride[])

  1680. {

  1681.     const enum PixelFormat srcFormat= c->srcFormat;

  1682.     const enum PixelFormat dstFormat= c->dstFormat;

  1683.     void (*conv)(const uint8_t *src, uint8_t *dst, long num_pixels,

  1684.                  const uint8_t *palette)=NULL;

  1685.     int i;

  1686.     uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;

  1687.     const uint8_t *srcPtr= src[0];

  1688. 

  1689.     if (srcFormat == PIX_FMT_GRAY8A) {

  1690.         switch (dstFormat) {

  1691.         case PIX_FMT_RGB32  : conv = gray8aToPacked32; break;

  1692.         case PIX_FMT_BGR32  : conv = gray8aToPacked32; break;

  1693.         case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;

  1694.         case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;

  1695.         case PIX_FMT_RGB24  : conv = gray8aToPacked24; break;

  1696.         case PIX_FMT_BGR24  : conv = gray8aToPacked24; break;

  1697.         }

  1698.     } else if (usePal(srcFormat)) {

  1699.         switch (dstFormat) {

  1700.         case PIX_FMT_RGB32  : conv = sws_convertPalette8ToPacked32; break;

  1701.         case PIX_FMT_BGR32  : conv = sws_convertPalette8ToPacked32; break;

  1702.         case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;

  1703.         case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;

  1704.         case PIX_FMT_RGB24  : conv = sws_convertPalette8ToPacked24; break;

  1705.         case PIX_FMT_BGR24  : conv = sws_convertPalette8ToPacked24; break;

  1706.         }

  1707.     }

  1708. 

  1709.     if (!conv)

  1710.         av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",

  1711.                sws_format_name(srcFormat), sws_format_name(dstFormat));

  1712.     else {

  1713.         for (i=0; i<srcSliceH; i++) {

  1714.             conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);

  1715.             srcPtr+= srcStride[0];

  1716.             dstPtr+= dstStride[0];

  1717.         }

  1718.     }

  1719. 

  1720.     return srcSliceH;

  1721. }

  1722. 

  1723. #define isRGBA32(x) (            \

  1724.            (x) == PIX_FMT_ARGB   \

  1725.         || (x) == PIX_FMT_RGBA   \

  1726.         || (x) == PIX_FMT_BGRA   \

  1727.         || (x) == PIX_FMT_ABGR   \

  1728.         )

  1729. 

  1730. /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */

  1731. static int rgbToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1732.                            int srcSliceH, uint8_t* dst[], int dstStride[])

  1733. {

  1734.     const enum PixelFormat srcFormat= c->srcFormat;

  1735.     const enum PixelFormat dstFormat= c->dstFormat;

  1736.     const int srcBpp= (c->srcFormatBpp + 7) >> 3;

  1737.     const int dstBpp= (c->dstFormatBpp + 7) >> 3;

  1738.     const int srcId= c->srcFormatBpp >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */

  1739.     const int dstId= c->dstFormatBpp >> 2;

  1740.     void (*conv)(const uint8_t *src, uint8_t *dst, long src_size)=NULL;

  1741. 

  1742. #define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)

  1743. 

  1744.     if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {

  1745.         if (     CONV_IS(ABGR, RGBA)

  1746.               || CONV_IS(ARGB, BGRA)

  1747.               || CONV_IS(BGRA, ARGB)

  1748.               || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;

  1749.         else if (CONV_IS(ABGR, ARGB)

  1750.               || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;

  1751.         else if (CONV_IS(ABGR, BGRA)

  1752.               || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;

  1753.         else if (CONV_IS(BGRA, RGBA)

  1754.               || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;

  1755.         else if (CONV_IS(BGRA, ABGR)

  1756.               || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;

  1757.     } else

  1758.     /* BGR -> BGR */

  1759.     if (  (isBGRinInt(srcFormat) && isBGRinInt(dstFormat))

  1760.        || (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {

  1761.         switch(srcId | (dstId<<4)) {

  1762.         case 0x34: conv= rgb16to15; break;

  1763.         case 0x36: conv= rgb24to15; break;

  1764.         case 0x38: conv= rgb32to15; break;

  1765.         case 0x43: conv= rgb15to16; break;

  1766.         case 0x46: conv= rgb24to16; break;

  1767.         case 0x48: conv= rgb32to16; break;

  1768.         case 0x63: conv= rgb15to24; break;

  1769.         case 0x64: conv= rgb16to24; break;

  1770.         case 0x68: conv= rgb32to24; break;

  1771.         case 0x83: conv= rgb15to32; break;

  1772.         case 0x84: conv= rgb16to32; break;

  1773.         case 0x86: conv= rgb24to32; break;

  1774.         }

  1775.     } else if (  (isBGRinInt(srcFormat) && isRGBinInt(dstFormat))

  1776.              || (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {

  1777.         switch(srcId | (dstId<<4)) {

  1778.         case 0x33: conv= rgb15tobgr15; break;

  1779.         case 0x34: conv= rgb16tobgr15; break;

  1780.         case 0x36: conv= rgb24tobgr15; break;

  1781.         case 0x38: conv= rgb32tobgr15; break;

  1782.         case 0x43: conv= rgb15tobgr16; break;

  1783.         case 0x44: conv= rgb16tobgr16; break;

  1784.         case 0x46: conv= rgb24tobgr16; break;

  1785.         case 0x48: conv= rgb32tobgr16; break;

  1786.         case 0x63: conv= rgb15tobgr24; break;

  1787.         case 0x64: conv= rgb16tobgr24; break;

  1788.         case 0x66: conv= rgb24tobgr24; break;

  1789.         case 0x68: conv= rgb32tobgr24; break;

  1790.         case 0x83: conv= rgb15tobgr32; break;

  1791.         case 0x84: conv= rgb16tobgr32; break;

  1792.         case 0x86: conv= rgb24tobgr32; break;

  1793.         }

  1794.     }

  1795. 

  1796.     if (!conv) {

  1797.         av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",

  1798.                sws_format_name(srcFormat), sws_format_name(dstFormat));

  1799.     } else {

  1800.         const uint8_t *srcPtr= src[0];

  1801.               uint8_t *dstPtr= dst[0];

  1802.         if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) && !isRGBA32(dstFormat))

  1803.             srcPtr += ALT32_CORR;

  1804. 

  1805.         if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) && !isRGBA32(srcFormat))

  1806.             dstPtr += ALT32_CORR;

  1807. 

  1808.         if (dstStride[0]*srcBpp == srcStride[0]*dstBpp && srcStride[0] > 0 && !(srcStride[0]%srcBpp))

  1809.             conv(srcPtr, dstPtr + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]);

  1810.         else {

  1811.             int i;

  1812.             dstPtr += dstStride[0]*srcSliceY;

  1813. 

  1814.             for (i=0; i<srcSliceH; i++) {

  1815.                 conv(srcPtr, dstPtr, c->srcW*srcBpp);

  1816.                 srcPtr+= srcStride[0];

  1817.                 dstPtr+= dstStride[0];

  1818.             }

  1819.         }

  1820.     }

  1821.     return srcSliceH;

  1822. }

  1823. 

  1824. static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1825.                               int srcSliceH, uint8_t* dst[], int dstStride[])

  1826. {

  1827.     rgb24toyv12(

  1828.         src[0],

  1829.         dst[0]+ srcSliceY    *dstStride[0],

  1830.         dst[1]+(srcSliceY>>1)*dstStride[1],

  1831.         dst[2]+(srcSliceY>>1)*dstStride[2],

  1832.         c->srcW, srcSliceH,

  1833.         dstStride[0], dstStride[1], srcStride[0]);

  1834.     if (dst[3])

  1835.         fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1836.     return srcSliceH;

  1837. }

  1838. 

  1839. static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1840.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1841. {

  1842.     copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,

  1843.               dst[0], dstStride[0]);

  1844. 

  1845.     planar2x(src[1], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,

  1846.              srcSliceH >> 2, srcStride[1], dstStride[1]);

  1847.     planar2x(src[2], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,

  1848.              srcSliceH >> 2, srcStride[2], dstStride[2]);

  1849.     if (dst[3])

  1850.         fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);

  1851.     return srcSliceH;

  1852. }

  1853. 

  1854. /* unscaled copy like stuff (assumes nearly identical formats) */

  1855. static int packedCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1856.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1857. {

  1858.     if (dstStride[0]==srcStride[0] && srcStride[0] > 0)

  1859.         memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]);

  1860.     else {

  1861.         int i;

  1862.         const uint8_t *srcPtr= src[0];

  1863.         uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;

  1864.         int length=0;

  1865. 

  1866.         /* universal length finder */

  1867.         while(length+c->srcW <= FFABS(dstStride[0])

  1868.            && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW;

  1869.         assert(length!=0);

  1870. 

  1871.         for (i=0; i<srcSliceH; i++) {

  1872.             memcpy(dstPtr, srcPtr, length);

  1873.             srcPtr+= srcStride[0];

  1874.             dstPtr+= dstStride[0];

  1875.         }

  1876.     }

  1877.     return srcSliceH;

  1878. }

  1879. 

  1880. #define DITHER_COPY(dst, dstStride, src, srcStride, bswap, dbswap)\

  1881.     uint16_t scale= dither_scale[dst_depth-1][src_depth-1];\

  1882.     int shift= src_depth-dst_depth + dither_scale[src_depth-2][dst_depth-1];\

  1883.     for (i = 0; i < height; i++) {\

  1884.         uint8_t *dither= dithers[src_depth-9][i&7];\

  1885.         for (j = 0; j < length-7; j+=8){\

  1886.             dst[j+0] = dbswap((bswap(src[j+0]) + dither[0])*scale>>shift);\

  1887.             dst[j+1] = dbswap((bswap(src[j+1]) + dither[1])*scale>>shift);\

  1888.             dst[j+2] = dbswap((bswap(src[j+2]) + dither[2])*scale>>shift);\

  1889.             dst[j+3] = dbswap((bswap(src[j+3]) + dither[3])*scale>>shift);\

  1890.             dst[j+4] = dbswap((bswap(src[j+4]) + dither[4])*scale>>shift);\

  1891.             dst[j+5] = dbswap((bswap(src[j+5]) + dither[5])*scale>>shift);\

  1892.             dst[j+6] = dbswap((bswap(src[j+6]) + dither[6])*scale>>shift);\

  1893.             dst[j+7] = dbswap((bswap(src[j+7]) + dither[7])*scale>>shift);\

  1894.         }\

  1895.         for (; j < length; j++)\

  1896.             dst[j] = dbswap((bswap(src[j]) + dither[j&7])*scale>>shift);\

  1897.         dst += dstStride;\

  1898.         src += srcStride;\

  1899.     }

  1900. 

  1901. 

  1902. static int planarCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,

  1903.                              int srcSliceH, uint8_t* dst[], int dstStride[])

  1904. {

  1905.     int plane, i, j;

  1906.     for (plane=0; plane<4; plane++) {

  1907.         int length= (plane==0 || plane==3) ? c->srcW  : -((-c->srcW  )>>c->chrDstHSubSample);

  1908.         int y=      (plane==0 || plane==3) ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample);

  1909.         int height= (plane==0 || plane==3) ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample);

  1910.         const uint8_t *srcPtr= src[plane];

  1911.         uint8_t *dstPtr= dst[plane] + dstStride[plane]*y;

  1912. 

  1913.         if (!dst[plane]) continue;

  1914.         // ignore palette for GRAY8

  1915.         if (plane == 1 && !dst[2]) continue;

  1916.         if (!src[plane] || (plane == 1 && !src[2])) {

  1917.             if(is16BPS(c->dstFormat))

  1918.                 length*=2;

  1919.             fillPlane(dst[plane], dstStride[plane], length, height, y, (plane==3) ? 255 : 128);

  1920.         } else {

  1921.             if(isNBPS(c->srcFormat) || isNBPS(c->dstFormat)

  1922.                || (is16BPS(c->srcFormat) != is16BPS(c->dstFormat))

  1923.             ) {

  1924.                 const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1+1;

  1925.                 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1+1;

  1926.                 const uint16_t *srcPtr2 = (const uint16_t*)srcPtr;

  1927.                 uint16_t *dstPtr2 = (uint16_t*)dstPtr;

  1928. 

  1929.                 if (dst_depth == 8) {

  1930.                     if(isBE(c->srcFormat) == HAVE_BIGENDIAN){

  1931.                         DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, , )

  1932.                     } else {

  1933.                         DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, av_bswap16, )

  1934.                     }

  1935.                 } else if (src_depth == 8) {

  1936.                     for (i = 0; i < height; i++) {

  1937.                         if(isBE(c->dstFormat)){

  1938.                             for (j = 0; j < length; j++)

  1939.                                 AV_WB16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |

  1940.                                                      (srcPtr[j]>>(2*8-dst_depth)));

  1941.                         } else {

  1942.                             for (j = 0; j < length; j++)

  1943.                                 AV_WL16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |

  1944.                                                      (srcPtr[j]>>(2*8-dst_depth)));

  1945.                         }

  1946.                         dstPtr2 += dstStride[plane]/2;

  1947.                         srcPtr  += srcStride[plane];

  1948.                     }

  1949.                 } else if (src_depth <= dst_depth) {

  1950.                     for (i = 0; i < height; i++) {

  1951. #define COPY_UP(r,w) \

  1952.     for (j = 0; j < length; j++){ \

  1953.         unsigned int v= r(&srcPtr2[j]);\

  1954.         w(&dstPtr2[j], (v<<(dst_depth-src_depth)) | \

  1955.                        (v>>(2*src_depth-dst_depth)));\

  1956.     }

  1957.                         if(isBE(c->srcFormat)){

  1958.                             if(isBE(c->dstFormat)){

  1959.                                 COPY_UP(AV_RB16, AV_WB16)

  1960.                             } else {

  1961.                                 COPY_UP(AV_RB16, AV_WL16)

  1962.                             }

  1963.                         } else {

  1964.                             if(isBE(c->dstFormat)){

  1965.                                 COPY_UP(AV_RL16, AV_WB16)

  1966.                             } else {

  1967.                                 COPY_UP(AV_RL16, AV_WL16)

  1968.                             }

  1969.                         }

  1970.                         dstPtr2 += dstStride[plane]/2;

  1971.                         srcPtr2 += srcStride[plane]/2;

  1972.                     }

  1973.                 } else {

  1974.                     if(isBE(c->srcFormat) == HAVE_BIGENDIAN){

  1975.                         if(isBE(c->dstFormat) == HAVE_BIGENDIAN){

  1976.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , )

  1977.                         } else {

  1978.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , av_bswap16)

  1979.                         }

  1980.                     }else{

  1981.                         if(isBE(c->dstFormat) == HAVE_BIGENDIAN){

  1982.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, )

  1983.                         } else {

  1984.                             DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, av_bswap16)

  1985.                         }

  1986.                     }

  1987.                 }

  1988.             } else if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat)

  1989.                   && isBE(c->srcFormat) != isBE(c->dstFormat)) {

  1990. 

  1991.                 for (i=0; i<height; i++) {

  1992.                     for (j=0; j<length; j++)

  1993.                         ((uint16_t*)dstPtr)[j] = av_bswap16(((const uint16_t*)srcPtr)[j]);

  1994.                     srcPtr+= srcStride[plane];

  1995.                     dstPtr+= dstStride[plane];

  1996.                 }

  1997.             } else if (dstStride[plane] == srcStride[plane] &&

  1998.                        srcStride[plane] > 0 && srcStride[plane] == length) {

  1999.                 memcpy(dst[plane] + dstStride[plane]*y, src[plane],

  2000.                        height*dstStride[plane]);

  2001.             } else {

  2002.                 if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat))

  2003.                     length*=2;

  2004.                 for (i=0; i<height; i++) {

  2005.                     memcpy(dstPtr, srcPtr, length);

  2006.                     srcPtr+= srcStride[plane];

  2007.                     dstPtr+= dstStride[plane];

  2008.                 }

  2009.             }

  2010.         }

  2011.     }

  2012.     return srcSliceH;

  2013. }

  2014. 

  2015. int ff_hardcodedcpuflags(void)

  2016. {

  2017.     int flags = 0;

  2018. #if   COMPILE_TEMPLATE_MMX2

  2019.     flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2;

  2020. #elif COMPILE_TEMPLATE_AMD3DNOW

  2021.     flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_3DNOW;

  2022. #elif COMPILE_TEMPLATE_MMX

  2023.     flags |= SWS_CPU_CAPS_MMX;

  2024. #elif COMPILE_TEMPLATE_ALTIVEC

  2025.     flags |= SWS_CPU_CAPS_ALTIVEC;

  2026. #elif ARCH_BFIN

  2027.     flags |= SWS_CPU_CAPS_BFIN;

  2028. #endif

  2029.     return flags;

  2030. }

  2031. 

  2032. void ff_get_unscaled_swscale(SwsContext *c)

  2033. {

  2034.     const enum PixelFormat srcFormat = c->srcFormat;

  2035.     const enum PixelFormat dstFormat = c->dstFormat;

  2036.     const int flags = c->flags;

  2037.     const int dstH = c->dstH;

  2038.     int needsDither;

  2039. 

  2040.     needsDither= isAnyRGB(dstFormat)

  2041.         &&  c->dstFormatBpp < 24

  2042.         && (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));

  2043. 

  2044.     /* yv12_to_nv12 */

  2045.     if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {

  2046.         c->swScale= planarToNv12Wrapper;

  2047.     }

  2048.     /* yuv2bgr */

  2049.     if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P || srcFormat==PIX_FMT_YUVA420P) && isAnyRGB(dstFormat)

  2050.         && !(flags & SWS_ACCURATE_RND) && !(dstH&1)) {

  2051.         c->swScale= ff_yuv2rgb_get_func_ptr(c);

  2052.     }

  2053. 

  2054.     if (srcFormat==PIX_FMT_YUV410P && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_BITEXACT)) {

  2055.         c->swScale= yvu9ToYv12Wrapper;

  2056.     }

  2057. 

  2058.     /* bgr24toYV12 */

  2059.     if (srcFormat==PIX_FMT_BGR24 && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_ACCURATE_RND))

  2060.         c->swScale= bgr24ToYv12Wrapper;

  2061. 

  2062.     /* RGB/BGR -> RGB/BGR (no dither needed forms) */

  2063.     if (   isAnyRGB(srcFormat)

  2064.         && isAnyRGB(dstFormat)

  2065.         && srcFormat != PIX_FMT_BGR8      && dstFormat != PIX_FMT_BGR8

  2066.         && srcFormat != PIX_FMT_RGB8      && dstFormat != PIX_FMT_RGB8

  2067.         && srcFormat != PIX_FMT_BGR4      && dstFormat != PIX_FMT_BGR4

  2068.         && srcFormat != PIX_FMT_RGB4      && dstFormat != PIX_FMT_RGB4

  2069.         && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE

  2070.         && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE

  2071.         && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK

  2072.         && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE

  2073.         && srcFormat != PIX_FMT_RGB48LE   && dstFormat != PIX_FMT_RGB48LE

  2074.         && srcFormat != PIX_FMT_RGB48BE   && dstFormat != PIX_FMT_RGB48BE

  2075.         && srcFormat != PIX_FMT_BGR48LE   && dstFormat != PIX_FMT_BGR48LE

  2076.         && srcFormat != PIX_FMT_BGR48BE   && dstFormat != PIX_FMT_BGR48BE

  2077.         && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))

  2078.         c->swScale= rgbToRgbWrapper;

  2079. 

  2080.     if ((usePal(srcFormat) && (

  2081.         dstFormat == PIX_FMT_RGB32   ||

  2082.         dstFormat == PIX_FMT_RGB32_1 ||

  2083.         dstFormat == PIX_FMT_RGB24   ||

  2084.         dstFormat == PIX_FMT_BGR32   ||

  2085.         dstFormat == PIX_FMT_BGR32_1 ||

  2086.         dstFormat == PIX_FMT_BGR24)))

  2087.         c->swScale= palToRgbWrapper;

  2088. 

  2089.     if (srcFormat == PIX_FMT_YUV422P) {

  2090.         if (dstFormat == PIX_FMT_YUYV422)

  2091.             c->swScale= yuv422pToYuy2Wrapper;

  2092.         else if (dstFormat == PIX_FMT_UYVY422)

  2093.             c->swScale= yuv422pToUyvyWrapper;

  2094.     }

  2095. 

  2096.     /* LQ converters if -sws 0 or -sws 4*/

  2097.     if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {

  2098.         /* yv12_to_yuy2 */

  2099.         if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {

  2100.             if (dstFormat == PIX_FMT_YUYV422)

  2101.                 c->swScale= planarToYuy2Wrapper;

  2102.             else if (dstFormat == PIX_FMT_UYVY422)

  2103.                 c->swScale= planarToUyvyWrapper;

  2104.         }

  2105.     }

  2106.     if(srcFormat == PIX_FMT_YUYV422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))

  2107.         c->swScale= yuyvToYuv420Wrapper;

  2108.     if(srcFormat == PIX_FMT_UYVY422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))

  2109.         c->swScale= uyvyToYuv420Wrapper;

  2110.     if(srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)

  2111.         c->swScale= yuyvToYuv422Wrapper;

  2112.     if(srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)

  2113.         c->swScale= uyvyToYuv422Wrapper;

  2114. 

  2115. #if COMPILE_ALTIVEC

  2116.     if ((c->flags & SWS_CPU_CAPS_ALTIVEC) &&

  2117.         !(c->flags & SWS_BITEXACT) &&

  2118.         srcFormat == PIX_FMT_YUV420P) {

  2119.         // unscaled YV12 -> packed YUV, we want speed

  2120.         if (dstFormat == PIX_FMT_YUYV422)

  2121.             c->swScale= yv12toyuy2_unscaled_altivec;

  2122.         else if (dstFormat == PIX_FMT_UYVY422)

  2123.             c->swScale= yv12touyvy_unscaled_altivec;

  2124.     }

  2125. #endif

  2126. 

  2127.     /* simple copy */

  2128.     if (  srcFormat == dstFormat

  2129.         || (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P)

  2130.         || (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P)

  2131.         || (isPlanarYUV(srcFormat) && isGray(dstFormat))

  2132.         || (isPlanarYUV(dstFormat) && isGray(srcFormat))

  2133.         || (isGray(dstFormat) && isGray(srcFormat))

  2134.         || (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat)

  2135.             && c->chrDstHSubSample == c->chrSrcHSubSample

  2136.             && c->chrDstVSubSample == c->chrSrcVSubSample

  2137.             && dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21

  2138.             && srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))

  2139.     {

  2140.         if (isPacked(c->srcFormat))

  2141.             c->swScale= packedCopyWrapper;

  2142.         else /* Planar YUV or gray */

  2143.             c->swScale= planarCopyWrapper;

  2144.     }

  2145. #if ARCH_BFIN

  2146.     if (flags & SWS_CPU_CAPS_BFIN)

  2147.         ff_bfin_get_unscaled_swscale (c);

  2148. #endif

  2149. }

  2150. 

  2151. static void reset_ptr(const uint8_t* src[], int format)

  2152. {

  2153.     if(!isALPHA(format))

  2154.         src[3]=NULL;

  2155.     if(!isPlanarYUV(format)) {

  2156.         src[3]=src[2]=NULL;

  2157. 

  2158.         if (!usePal(format))

  2159.             src[1]= NULL;

  2160.     }

  2161. }

  2162. 

  2163. static int check_image_pointers(uint8_t *data[4], enum PixelFormat pix_fmt,

  2164.                                 const int linesizes[4])

  2165. {

  2166.     const AVPixFmtDescriptor *desc = &av_pix_fmt_descriptors[pix_fmt];

  2167.     int i;

  2168. 

  2169.     for (i = 0; i < 4; i++) {

  2170.         int plane = desc->comp[i].plane;

  2171.         if (!data[plane] || !linesizes[plane])

  2172.             return 0;

  2173.     }

  2174. 

  2175.     return 1;

  2176. }

  2177. 

  2178. /**

  2179.  * swscale wrapper, so we don't need to export the SwsContext.

  2180.  * Assumes planar YUV to be in YUV order instead of YVU.

  2181.  */

  2182. int sws_scale(SwsContext *c, const uint8_t* const src[], const int srcStride[], int srcSliceY,

  2183.               int srcSliceH, uint8_t* const dst[], const int dstStride[])

  2184. {

  2185.     int i;

  2186.     const uint8_t* src2[4]= {src[0], src[1], src[2], src[3]};

  2187.     uint8_t* dst2[4]= {dst[0], dst[1], dst[2], dst[3]};

  2188. 

  2189.     // do not mess up sliceDir if we have a "trailing" 0-size slice

  2190.     if (srcSliceH == 0)

  2191.         return 0;

  2192. 

  2193.     if (!check_image_pointers(src, c->srcFormat, srcStride)) {

  2194.         av_log(c, AV_LOG_ERROR, "bad src image pointers\n");

  2195.         return 0;

  2196.     }

  2197.     if (!check_image_pointers(dst, c->dstFormat, dstStride)) {

  2198.         av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");

  2199.         return 0;

  2200.     }

  2201. 

  2202.     if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {

  2203.         av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");

  2204.         return 0;

  2205.     }

  2206.     if (c->sliceDir == 0) {

  2207.         if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;

  2208.     }

  2209. 

  2210.     if (usePal(c->srcFormat)) {

  2211.         for (i=0; i<256; i++) {

  2212.             int p, r, g, b, y, u, v, a = 0xff;

  2213.             if(c->srcFormat == PIX_FMT_PAL8) {

  2214.                 p=((const uint32_t*)(src[1]))[i];

  2215.                 a= (p>>24)&0xFF;

  2216.                 r= (p>>16)&0xFF;

  2217.                 g= (p>> 8)&0xFF;

  2218.                 b=  p     &0xFF;

  2219.             } else if(c->srcFormat == PIX_FMT_RGB8) {

  2220.                 r= (i>>5    )*36;

  2221.                 g= ((i>>2)&7)*36;

  2222.                 b= (i&3     )*85;

  2223.             } else if(c->srcFormat == PIX_FMT_BGR8) {

  2224.                 b= (i>>6    )*85;

  2225.                 g= ((i>>3)&7)*36;

  2226.                 r= (i&7     )*36;

  2227.             } else if(c->srcFormat == PIX_FMT_RGB4_BYTE) {

  2228.                 r= (i>>3    )*255;

  2229.                 g= ((i>>1)&3)*85;

  2230.                 b= (i&1     )*255;

  2231.             } else if(c->srcFormat == PIX_FMT_GRAY8 || c->srcFormat == PIX_FMT_GRAY8A) {

  2232.                 r = g = b = i;

  2233.             } else {

  2234.                 assert(c->srcFormat == PIX_FMT_BGR4_BYTE);

  2235.                 b= (i>>3    )*255;

  2236.                 g= ((i>>1)&3)*85;

  2237.                 r= (i&1     )*255;

  2238.             }

  2239.             y= av_clip_uint8((RY*r + GY*g + BY*b + ( 33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2240.             u= av_clip_uint8((RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2241.             v= av_clip_uint8((RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);

  2242.             c->pal_yuv[i]= y + (u<<8) + (v<<16) + (a<<24);

  2243. 

  2244.             switch(c->dstFormat) {

  2245.             case PIX_FMT_BGR32:

  2246. #if !HAVE_BIGENDIAN

  2247.             case PIX_FMT_RGB24:

  2248. #endif

  2249.                 c->pal_rgb[i]=  r + (g<<8) + (b<<16) + (a<<24);

  2250.                 break;

  2251.             case PIX_FMT_BGR32_1:

  2252. #if HAVE_BIGENDIAN

  2253.             case PIX_FMT_BGR24:

  2254. #endif

  2255.                 c->pal_rgb[i]= a + (r<<8) + (g<<16) + (b<<24);

  2256.                 break;

  2257.             case PIX_FMT_RGB32_1:

  2258. #if HAVE_BIGENDIAN

  2259.             case PIX_FMT_RGB24:

  2260. #endif

  2261.                 c->pal_rgb[i]= a + (b<<8) + (g<<16) + (r<<24);

  2262.                 break;

  2263.             case PIX_FMT_RGB32:

  2264. #if !HAVE_BIGENDIAN

  2265.             case PIX_FMT_BGR24:

  2266. #endif

  2267.             default:

  2268.                 c->pal_rgb[i]=  b + (g<<8) + (r<<16) + (a<<24);

  2269.             }

  2270.         }

  2271.     }

  2272. 

  2273.     // copy strides, so they can safely be modified

  2274.     if (c->sliceDir == 1) {

  2275.         // slices go from top to bottom

  2276.         int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2], srcStride[3]};

  2277.         int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2], dstStride[3]};

  2278. 

  2279.         reset_ptr(src2, c->srcFormat);

  2280.         reset_ptr((const uint8_t**)dst2, c->dstFormat);

  2281. 

  2282.         /* reset slice direction at end of frame */

  2283.         if (srcSliceY + srcSliceH == c->srcH)

  2284.             c->sliceDir = 0;

  2285. 

  2286.         return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2, dstStride2);

  2287.     } else {

  2288.         // slices go from bottom to top => we flip the image internally

  2289.         int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2], -srcStride[3]};

  2290.         int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2], -dstStride[3]};

  2291. 

  2292.         src2[0] += (srcSliceH-1)*srcStride[0];

  2293.         if (!usePal(c->srcFormat))

  2294.             src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1];

  2295.         src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2];

  2296.         src2[3] += (srcSliceH-1)*srcStride[3];

  2297.         dst2[0] += ( c->dstH                      -1)*dstStride[0];

  2298.         dst2[1] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1];

  2299.         dst2[2] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2];

  2300.         dst2[3] += ( c->dstH                      -1)*dstStride[3];

  2301. 

  2302.         reset_ptr(src2, c->srcFormat);

  2303.         reset_ptr((const uint8_t**)dst2, c->dstFormat);

  2304. 

  2305.         /* reset slice direction at end of frame */

  2306.         if (!srcSliceY)

  2307.             c->sliceDir = 0;

  2308. 

  2309.         return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2);

  2310.     }

  2311. }

  2312. 

  2313. #if LIBSWSCALE_VERSION_MAJOR < 1

  2314. int sws_scale_ordered(SwsContext *c, const uint8_t* const src[], int srcStride[], int srcSliceY,

  2315.                       int srcSliceH, uint8_t* dst[], int dstStride[])

  2316. {

  2317.     return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride);

  2318. }

  2319. #endif

  2320. 

  2321. /* Convert the palette to the same packed 32-bit format as the palette */

  2322. void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  2323. {

  2324.     long i;

  2325. 

  2326.     for (i=0; i<num_pixels; i++)

  2327.         ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];

  2328. }

  2329. 

  2330. /* Palette format: ABCD -> dst format: ABC */

  2331. void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)

  2332. {

  2333.     long i;

  2334. 

  2335.     for (i=0; i<num_pixels; i++) {

  2336.         //FIXME slow?

  2337.         dst[0]= palette[src[i]*4+0];

  2338.         dst[1]= palette[src[i]*4+1];

  2339.         dst[2]= palette[src[i]*4+2];

  2340.         dst+= 3;

  2341.     }

  2342. }