2 * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of FFmpeg.
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.
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.
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
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
26 unscaled special converters (YV12=I420=IYUV, Y800=Y8)
27 YV12 -> {BGR,RGB}{1,4,8,12,15,16,24,32}
32 BGR24 -> BGR32 & RGB24 -> RGB32
33 BGR32 -> BGR24 & RGB32 -> RGB24
38 tested special converters (most are tested actually, but I did not write it down ...)
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
61 #include "swscale_internal.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"
74 //#define HAVE_AMD3DNOW
79 #define isPacked(x) ( \
81 || (x)==PIX_FMT_YUYV422 \
82 || (x)==PIX_FMT_UYVY422 \
83 || (x)==PIX_FMT_GRAY8A \
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))
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
111 Special versions: fast Y 1:1 scaling (no interpolation in y direction)
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
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;
134 const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {
135 0x0103010301030103LL,
136 0x0200020002000200LL,};
138 const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {
139 0x0602060206020602LL,
140 0x0004000400040004LL,};
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;
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;
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;
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;
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;
172 DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUV)[2][4] = {
173 {0x38380000DAC83838ULL, 0xECFFDAC80000ECFFULL, 0xF6E40000D0E3F6E4ULL, 0x3838D0E300003838ULL},
174 {0xECFF0000DAC8ECFFULL, 0x3838DAC800003838ULL, 0x38380000D0E33838ULL, 0xF6E4D0E30000F6E4ULL},
177 DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUVOffset)= 0x0040400000404000ULL;
179 #endif /* ARCH_X86 */
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, },
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, },
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, },
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, },
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, },
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, },
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, },
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, },
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, },
269 DECLARE_ALIGNED(8, const uint8_t, dithers)[8][8][8]={
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,},
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,},
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,},
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,},
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,},
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,},
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,},
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,},
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,},
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)
367 //FIXME Optimize (just quickly written not optimized..)
370 for (i = 0; i < dstW; i++) {
374 for (j = 0; j < lumFilterSize; j++)
375 val += lumSrc[j][i] * lumFilter[j];
378 AV_WB16(&dest[i], av_clip_uint16(val >> 11));
380 AV_WL16(&dest[i], av_clip_uint16(val >> 11));
385 for (i = 0; i < chrDstW; i++) {
390 for (j = 0; j < chrFilterSize; j++) {
391 u += chrSrc[j][i ] * chrFilter[j];
392 v += chrSrc[j][i + VOFW] * chrFilter[j];
396 AV_WB16(&uDest[i], av_clip_uint16(u >> 11));
397 AV_WB16(&vDest[i], av_clip_uint16(v >> 11));
399 AV_WL16(&uDest[i], av_clip_uint16(u >> 11));
400 AV_WL16(&vDest[i], av_clip_uint16(v >> 11));
405 if (CONFIG_SWSCALE_ALPHA && aDest) {
406 for (i = 0; i < dstW; i++) {
410 for (j = 0; j < lumFilterSize; j++)
411 val += alpSrc[j][i] * lumFilter[j];
414 AV_WB16(&aDest[i], av_clip_uint16(val >> 11));
416 AV_WL16(&aDest[i], av_clip_uint16(val >> 11));
422 static av_always_inline void yuv2yuvXNinC_template(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
423 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
424 const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest,
425 int dstW, int chrDstW, int big_endian, int depth)
427 //FIXME Optimize (just quickly written not optimized..)
430 for (i = 0; i < dstW; i++) {
431 int val = 1 << (26-depth);
434 for (j = 0; j < lumFilterSize; j++)
435 val += lumSrc[j][i] * lumFilter[j];
438 AV_WB16(&dest[i], av_clip(val >> (27-depth), 0, (1<<depth)-1));
440 AV_WL16(&dest[i], av_clip(val >> (27-depth), 0, (1<<depth)-1));
445 for (i = 0; i < chrDstW; i++) {
446 int u = 1 << (26-depth);
447 int v = 1 << (26-depth);
450 for (j = 0; j < chrFilterSize; j++) {
451 u += chrSrc[j][i ] * chrFilter[j];
452 v += chrSrc[j][i + VOFW] * chrFilter[j];
456 AV_WB16(&uDest[i], av_clip(u >> (27-depth), 0, (1<<depth)-1));
457 AV_WB16(&vDest[i], av_clip(v >> (27-depth), 0, (1<<depth)-1));
459 AV_WL16(&uDest[i], av_clip(u >> (27-depth), 0, (1<<depth)-1));
460 AV_WL16(&vDest[i], av_clip(v >> (27-depth), 0, (1<<depth)-1));
466 static inline void yuv2yuvX16inC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
467 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
468 const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest, int dstW, int chrDstW,
469 enum PixelFormat dstFormat)
471 if (isNBPS(dstFormat)) {
472 const int depth = av_pix_fmt_descriptors[dstFormat].comp[0].depth_minus1+1;
473 yuv2yuvXNinC_template(lumFilter, lumSrc, lumFilterSize,
474 chrFilter, chrSrc, chrFilterSize,
476 dest, uDest, vDest, aDest,
477 dstW, chrDstW, isBE(dstFormat), depth);
479 if (isBE(dstFormat)) {
480 yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
481 chrFilter, chrSrc, chrFilterSize,
483 dest, uDest, vDest, aDest,
486 yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
487 chrFilter, chrSrc, chrFilterSize,
489 dest, uDest, vDest, aDest,
495 static inline void yuv2yuvXinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
496 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
497 const int16_t **alpSrc, uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, int dstW, int chrDstW)
499 //FIXME Optimize (just quickly written not optimized..)
501 for (i=0; i<dstW; i++) {
504 for (j=0; j<lumFilterSize; j++)
505 val += lumSrc[j][i] * lumFilter[j];
507 dest[i]= av_clip_uint8(val>>19);
511 for (i=0; i<chrDstW; i++) {
515 for (j=0; j<chrFilterSize; j++) {
516 u += chrSrc[j][i] * chrFilter[j];
517 v += chrSrc[j][i + VOFW] * chrFilter[j];
520 uDest[i]= av_clip_uint8(u>>19);
521 vDest[i]= av_clip_uint8(v>>19);
524 if (CONFIG_SWSCALE_ALPHA && aDest)
525 for (i=0; i<dstW; i++) {
528 for (j=0; j<lumFilterSize; j++)
529 val += alpSrc[j][i] * lumFilter[j];
531 aDest[i]= av_clip_uint8(val>>19);
536 static inline void yuv2nv12XinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
537 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
538 uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat)
540 //FIXME Optimize (just quickly written not optimized..)
542 for (i=0; i<dstW; i++) {
545 for (j=0; j<lumFilterSize; j++)
546 val += lumSrc[j][i] * lumFilter[j];
548 dest[i]= av_clip_uint8(val>>19);
554 if (dstFormat == PIX_FMT_NV12)
555 for (i=0; i<chrDstW; i++) {
559 for (j=0; j<chrFilterSize; j++) {
560 u += chrSrc[j][i] * chrFilter[j];
561 v += chrSrc[j][i + VOFW] * chrFilter[j];
564 uDest[2*i]= av_clip_uint8(u>>19);
565 uDest[2*i+1]= av_clip_uint8(v>>19);
568 for (i=0; i<chrDstW; i++) {
572 for (j=0; j<chrFilterSize; j++) {
573 u += chrSrc[j][i] * chrFilter[j];
574 v += chrSrc[j][i + VOFW] * chrFilter[j];
577 uDest[2*i]= av_clip_uint8(v>>19);
578 uDest[2*i+1]= av_clip_uint8(u>>19);
582 #define YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha) \
583 for (i=0; i<(dstW>>1); i++) {\
589 int av_unused A1, A2;\
590 type av_unused *r, *b, *g;\
593 for (j=0; j<lumFilterSize; j++) {\
594 Y1 += lumSrc[j][i2] * lumFilter[j];\
595 Y2 += lumSrc[j][i2+1] * lumFilter[j];\
597 for (j=0; j<chrFilterSize; j++) {\
598 U += chrSrc[j][i] * chrFilter[j];\
599 V += chrSrc[j][i+VOFW] * chrFilter[j];\
608 for (j=0; j<lumFilterSize; j++) {\
609 A1 += alpSrc[j][i2 ] * lumFilter[j];\
610 A2 += alpSrc[j][i2+1] * lumFilter[j];\
616 #define YSCALE_YUV_2_PACKEDX_C(type,alpha) \
617 YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha)\
618 if ((Y1|Y2|U|V)&256) {\
619 if (Y1>255) Y1=255; \
620 else if (Y1<0)Y1=0; \
621 if (Y2>255) Y2=255; \
622 else if (Y2<0)Y2=0; \
628 if (alpha && ((A1|A2)&256)) {\
629 A1=av_clip_uint8(A1);\
630 A2=av_clip_uint8(A2);\
633 #define YSCALE_YUV_2_PACKEDX_FULL_C(rnd,alpha) \
634 for (i=0; i<dstW; i++) {\
642 for (j=0; j<lumFilterSize; j++) {\
643 Y += lumSrc[j][i ] * lumFilter[j];\
645 for (j=0; j<chrFilterSize; j++) {\
646 U += chrSrc[j][i ] * chrFilter[j];\
647 V += chrSrc[j][i+VOFW] * chrFilter[j];\
654 for (j=0; j<lumFilterSize; j++)\
655 A += alpSrc[j][i ] * lumFilter[j];\
658 A = av_clip_uint8(A);\
661 #define YSCALE_YUV_2_RGBX_FULL_C(rnd,alpha) \
662 YSCALE_YUV_2_PACKEDX_FULL_C(rnd>>3,alpha)\
663 Y-= c->yuv2rgb_y_offset;\
664 Y*= c->yuv2rgb_y_coeff;\
666 R= Y + V*c->yuv2rgb_v2r_coeff;\
667 G= Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;\
668 B= Y + U*c->yuv2rgb_u2b_coeff;\
669 if ((R|G|B)&(0xC0000000)) {\
670 if (R>=(256<<22)) R=(256<<22)-1; \
672 if (G>=(256<<22)) G=(256<<22)-1; \
674 if (B>=(256<<22)) B=(256<<22)-1; \
678 #define YSCALE_YUV_2_GRAY16_C \
679 for (i=0; i<(dstW>>1); i++) {\
688 for (j=0; j<lumFilterSize; j++) {\
689 Y1 += lumSrc[j][i2] * lumFilter[j];\
690 Y2 += lumSrc[j][i2+1] * lumFilter[j];\
694 if ((Y1|Y2|U|V)&65536) {\
695 if (Y1>65535) Y1=65535; \
696 else if (Y1<0)Y1=0; \
697 if (Y2>65535) Y2=65535; \
698 else if (Y2<0)Y2=0; \
701 #define YSCALE_YUV_2_RGBX_C(type,alpha) \
702 YSCALE_YUV_2_PACKEDX_C(type,alpha) /* FIXME fix tables so that clipping is not needed and then use _NOCLIP*/\
703 r = (type *)c->table_rV[V]; \
704 g = (type *)(c->table_gU[U] + c->table_gV[V]); \
705 b = (type *)c->table_bU[U];
707 #define YSCALE_YUV_2_PACKED2_C(type,alpha) \
708 for (i=0; i<(dstW>>1); i++) { \
710 int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>19; \
711 int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19; \
712 int U= (uvbuf0[i ]*uvalpha1+uvbuf1[i ]*uvalpha)>>19; \
713 int V= (uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19; \
714 type av_unused *r, *b, *g; \
715 int av_unused A1, A2; \
717 A1= (abuf0[i2 ]*yalpha1+abuf1[i2 ]*yalpha)>>19; \
718 A2= (abuf0[i2+1]*yalpha1+abuf1[i2+1]*yalpha)>>19; \
721 #define YSCALE_YUV_2_GRAY16_2_C \
722 for (i=0; i<(dstW>>1); i++) { \
724 int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>11; \
725 int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>11;
727 #define YSCALE_YUV_2_RGB2_C(type,alpha) \
728 YSCALE_YUV_2_PACKED2_C(type,alpha)\
729 r = (type *)c->table_rV[V];\
730 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
731 b = (type *)c->table_bU[U];
733 #define YSCALE_YUV_2_PACKED1_C(type,alpha) \
734 for (i=0; i<(dstW>>1); i++) {\
736 int Y1= buf0[i2 ]>>7;\
737 int Y2= buf0[i2+1]>>7;\
738 int U= (uvbuf1[i ])>>7;\
739 int V= (uvbuf1[i+VOFW])>>7;\
740 type av_unused *r, *b, *g;\
741 int av_unused A1, A2;\
747 #define YSCALE_YUV_2_GRAY16_1_C \
748 for (i=0; i<(dstW>>1); i++) {\
750 int Y1= buf0[i2 ]<<1;\
751 int Y2= buf0[i2+1]<<1;
753 #define YSCALE_YUV_2_RGB1_C(type,alpha) \
754 YSCALE_YUV_2_PACKED1_C(type,alpha)\
755 r = (type *)c->table_rV[V];\
756 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
757 b = (type *)c->table_bU[U];
759 #define YSCALE_YUV_2_PACKED1B_C(type,alpha) \
760 for (i=0; i<(dstW>>1); i++) {\
762 int Y1= buf0[i2 ]>>7;\
763 int Y2= buf0[i2+1]>>7;\
764 int U= (uvbuf0[i ] + uvbuf1[i ])>>8;\
765 int V= (uvbuf0[i+VOFW] + uvbuf1[i+VOFW])>>8;\
766 type av_unused *r, *b, *g;\
767 int av_unused A1, A2;\
773 #define YSCALE_YUV_2_RGB1B_C(type,alpha) \
774 YSCALE_YUV_2_PACKED1B_C(type,alpha)\
775 r = (type *)c->table_rV[V];\
776 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
777 b = (type *)c->table_bU[U];
779 #define YSCALE_YUV_2_MONO2_C \
780 const uint8_t * const d128=dither_8x8_220[y&7];\
781 uint8_t *g= c->table_gU[128] + c->table_gV[128];\
782 for (i=0; i<dstW-7; i+=8) {\
784 acc = g[((buf0[i ]*yalpha1+buf1[i ]*yalpha)>>19) + d128[0]];\
785 acc+= acc + g[((buf0[i+1]*yalpha1+buf1[i+1]*yalpha)>>19) + d128[1]];\
786 acc+= acc + g[((buf0[i+2]*yalpha1+buf1[i+2]*yalpha)>>19) + d128[2]];\
787 acc+= acc + g[((buf0[i+3]*yalpha1+buf1[i+3]*yalpha)>>19) + d128[3]];\
788 acc+= acc + g[((buf0[i+4]*yalpha1+buf1[i+4]*yalpha)>>19) + d128[4]];\
789 acc+= acc + g[((buf0[i+5]*yalpha1+buf1[i+5]*yalpha)>>19) + d128[5]];\
790 acc+= acc + g[((buf0[i+6]*yalpha1+buf1[i+6]*yalpha)>>19) + d128[6]];\
791 acc+= acc + g[((buf0[i+7]*yalpha1+buf1[i+7]*yalpha)>>19) + d128[7]];\
792 ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
796 #define YSCALE_YUV_2_MONOX_C \
797 const uint8_t * const d128=dither_8x8_220[y&7];\
798 uint8_t *g= c->table_gU[128] + c->table_gV[128];\
800 for (i=0; i<dstW-1; i+=2) {\
805 for (j=0; j<lumFilterSize; j++) {\
806 Y1 += lumSrc[j][i] * lumFilter[j];\
807 Y2 += lumSrc[j][i+1] * lumFilter[j];\
817 acc+= acc + g[Y1+d128[(i+0)&7]];\
818 acc+= acc + g[Y2+d128[(i+1)&7]];\
820 ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
825 #define YSCALE_YUV_2_ANYRGB_C(func, func2, func_g16, func_monoblack)\
826 switch(c->dstFormat) {\
827 case PIX_FMT_RGB48BE:\
828 case PIX_FMT_RGB48LE:\
830 ((uint8_t*)dest)[ 0]= r[Y1];\
831 ((uint8_t*)dest)[ 1]= r[Y1];\
832 ((uint8_t*)dest)[ 2]= g[Y1];\
833 ((uint8_t*)dest)[ 3]= g[Y1];\
834 ((uint8_t*)dest)[ 4]= b[Y1];\
835 ((uint8_t*)dest)[ 5]= b[Y1];\
836 ((uint8_t*)dest)[ 6]= r[Y2];\
837 ((uint8_t*)dest)[ 7]= r[Y2];\
838 ((uint8_t*)dest)[ 8]= g[Y2];\
839 ((uint8_t*)dest)[ 9]= g[Y2];\
840 ((uint8_t*)dest)[10]= b[Y2];\
841 ((uint8_t*)dest)[11]= b[Y2];\
845 case PIX_FMT_BGR48BE:\
846 case PIX_FMT_BGR48LE:\
848 ((uint8_t*)dest)[ 0] = ((uint8_t*)dest)[ 1] = b[Y1];\
849 ((uint8_t*)dest)[ 2] = ((uint8_t*)dest)[ 3] = g[Y1];\
850 ((uint8_t*)dest)[ 4] = ((uint8_t*)dest)[ 5] = r[Y1];\
851 ((uint8_t*)dest)[ 6] = ((uint8_t*)dest)[ 7] = b[Y2];\
852 ((uint8_t*)dest)[ 8] = ((uint8_t*)dest)[ 9] = g[Y2];\
853 ((uint8_t*)dest)[10] = ((uint8_t*)dest)[11] = r[Y2];\
860 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\
861 func(uint32_t,needAlpha)\
862 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? (A1<<24) : 0);\
863 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? (A2<<24) : 0);\
866 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\
868 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (A1<<24);\
869 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (A2<<24);\
873 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
874 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
882 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\
883 func(uint32_t,needAlpha)\
884 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? A1 : 0);\
885 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? A2 : 0);\
888 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\
890 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + A1;\
891 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + A2;\
895 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
896 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
903 ((uint8_t*)dest)[0]= r[Y1];\
904 ((uint8_t*)dest)[1]= g[Y1];\
905 ((uint8_t*)dest)[2]= b[Y1];\
906 ((uint8_t*)dest)[3]= r[Y2];\
907 ((uint8_t*)dest)[4]= g[Y2];\
908 ((uint8_t*)dest)[5]= b[Y2];\
914 ((uint8_t*)dest)[0]= b[Y1];\
915 ((uint8_t*)dest)[1]= g[Y1];\
916 ((uint8_t*)dest)[2]= r[Y1];\
917 ((uint8_t*)dest)[3]= b[Y2];\
918 ((uint8_t*)dest)[4]= g[Y2];\
919 ((uint8_t*)dest)[5]= r[Y2];\
923 case PIX_FMT_RGB565BE:\
924 case PIX_FMT_RGB565LE:\
925 case PIX_FMT_BGR565BE:\
926 case PIX_FMT_BGR565LE:\
928 const int dr1= dither_2x2_8[y&1 ][0];\
929 const int dg1= dither_2x2_4[y&1 ][0];\
930 const int db1= dither_2x2_8[(y&1)^1][0];\
931 const int dr2= dither_2x2_8[y&1 ][1];\
932 const int dg2= dither_2x2_4[y&1 ][1];\
933 const int db2= dither_2x2_8[(y&1)^1][1];\
935 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
936 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
940 case PIX_FMT_RGB555BE:\
941 case PIX_FMT_RGB555LE:\
942 case PIX_FMT_BGR555BE:\
943 case PIX_FMT_BGR555LE:\
945 const int dr1= dither_2x2_8[y&1 ][0];\
946 const int dg1= dither_2x2_8[y&1 ][1];\
947 const int db1= dither_2x2_8[(y&1)^1][0];\
948 const int dr2= dither_2x2_8[y&1 ][1];\
949 const int dg2= dither_2x2_8[y&1 ][0];\
950 const int db2= dither_2x2_8[(y&1)^1][1];\
952 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
953 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
957 case PIX_FMT_RGB444BE:\
958 case PIX_FMT_RGB444LE:\
959 case PIX_FMT_BGR444BE:\
960 case PIX_FMT_BGR444LE:\
962 const int dr1= dither_4x4_16[y&3 ][0];\
963 const int dg1= dither_4x4_16[y&3 ][1];\
964 const int db1= dither_4x4_16[(y&3)^3][0];\
965 const int dr2= dither_4x4_16[y&3 ][1];\
966 const int dg2= dither_4x4_16[y&3 ][0];\
967 const int db2= dither_4x4_16[(y&3)^3][1];\
969 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
970 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
977 const uint8_t * const d64= dither_8x8_73[y&7];\
978 const uint8_t * const d32= dither_8x8_32[y&7];\
980 ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]];\
981 ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]];\
988 const uint8_t * const d64= dither_8x8_73 [y&7];\
989 const uint8_t * const d128=dither_8x8_220[y&7];\
991 ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]\
992 + ((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4);\
996 case PIX_FMT_RGB4_BYTE:\
997 case PIX_FMT_BGR4_BYTE:\
999 const uint8_t * const d64= dither_8x8_73 [y&7];\
1000 const uint8_t * const d128=dither_8x8_220[y&7];\
1002 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\
1003 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\
1007 case PIX_FMT_MONOBLACK:\
1008 case PIX_FMT_MONOWHITE:\
1013 case PIX_FMT_YUYV422:\
1015 ((uint8_t*)dest)[2*i2+0]= Y1;\
1016 ((uint8_t*)dest)[2*i2+1]= U;\
1017 ((uint8_t*)dest)[2*i2+2]= Y2;\
1018 ((uint8_t*)dest)[2*i2+3]= V;\
1021 case PIX_FMT_UYVY422:\
1023 ((uint8_t*)dest)[2*i2+0]= U;\
1024 ((uint8_t*)dest)[2*i2+1]= Y1;\
1025 ((uint8_t*)dest)[2*i2+2]= V;\
1026 ((uint8_t*)dest)[2*i2+3]= Y2;\
1029 case PIX_FMT_GRAY16BE:\
1031 ((uint8_t*)dest)[2*i2+0]= Y1>>8;\
1032 ((uint8_t*)dest)[2*i2+1]= Y1;\
1033 ((uint8_t*)dest)[2*i2+2]= Y2>>8;\
1034 ((uint8_t*)dest)[2*i2+3]= Y2;\
1037 case PIX_FMT_GRAY16LE:\
1039 ((uint8_t*)dest)[2*i2+0]= Y1;\
1040 ((uint8_t*)dest)[2*i2+1]= Y1>>8;\
1041 ((uint8_t*)dest)[2*i2+2]= Y2;\
1042 ((uint8_t*)dest)[2*i2+3]= Y2>>8;\
1047 static inline void yuv2packedXinC(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
1048 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
1049 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
1052 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 static inline void yuv2rgbXinC_full(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
1056 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
1057 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
1060 int step= c->dstFormatBpp/8;
1063 switch(c->dstFormat) {
1071 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;
1072 YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)
1073 dest[aidx]= needAlpha ? A : 255;
1080 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1081 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)
1089 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)
1106 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;
1107 YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)
1108 dest[aidx]= needAlpha ? A : 255;
1115 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1116 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)
1124 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)
1139 static void fillPlane(uint8_t* plane, int stride, int width, int height, int y, uint8_t val)
1142 uint8_t *ptr = plane + stride*y;
1143 for (i=0; i<height; i++) {
1144 memset(ptr, val, width);
1149 static inline void rgb48ToY(uint8_t *dst, const uint8_t *src, long width,
1153 for (i = 0; i < width; i++) {
1158 dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1162 static inline void rgb48ToUV(uint8_t *dstU, uint8_t *dstV,
1163 const uint8_t *src1, const uint8_t *src2,
1164 long width, uint32_t *unused)
1168 for (i = 0; i < width; i++) {
1169 int r = src1[6*i + 0];
1170 int g = src1[6*i + 2];
1171 int b = src1[6*i + 4];
1173 dstU[i] = (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1174 dstV[i] = (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1178 static inline void rgb48ToUV_half(uint8_t *dstU, uint8_t *dstV,
1179 const uint8_t *src1, const uint8_t *src2,
1180 long width, uint32_t *unused)
1184 for (i = 0; i < width; i++) {
1185 int r= src1[12*i + 0] + src1[12*i + 6];
1186 int g= src1[12*i + 2] + src1[12*i + 8];
1187 int b= src1[12*i + 4] + src1[12*i + 10];
1189 dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1190 dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1194 static inline void bgr48ToY(uint8_t *dst, const uint8_t *src, long width,
1198 for (i = 0; i < width; i++) {
1203 dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1207 static inline void bgr48ToUV(uint8_t *dstU, uint8_t *dstV,
1208 const uint8_t *src1, const uint8_t *src2,
1209 long width, uint32_t *unused)
1212 for (i = 0; i < width; i++) {
1213 int b = src1[6*i + 0];
1214 int g = src1[6*i + 2];
1215 int r = src1[6*i + 4];
1217 dstU[i] = (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1218 dstV[i] = (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1222 static inline void bgr48ToUV_half(uint8_t *dstU, uint8_t *dstV,
1223 const uint8_t *src1, const uint8_t *src2,
1224 long width, uint32_t *unused)
1227 for (i = 0; i < width; i++) {
1228 int b= src1[12*i + 0] + src1[12*i + 6];
1229 int g= src1[12*i + 2] + src1[12*i + 8];
1230 int r= src1[12*i + 4] + src1[12*i + 10];
1232 dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1233 dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1237 #define BGR2Y(type, name, shr, shg, shb, maskr, maskg, maskb, RY, GY, BY, S)\
1238 static inline void name(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)\
1241 for (i=0; i<width; i++) {\
1242 int b= (((const type*)src)[i]>>shb)&maskb;\
1243 int g= (((const type*)src)[i]>>shg)&maskg;\
1244 int r= (((const type*)src)[i]>>shr)&maskr;\
1246 dst[i]= (((RY)*r + (GY)*g + (BY)*b + (33<<((S)-1)))>>(S));\
1250 BGR2Y(uint32_t, bgr32ToY,16, 0, 0, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY , BY<< 8, RGB2YUV_SHIFT+8)
1251 BGR2Y(uint32_t,bgr321ToY,16,16, 0, 0xFF00, 0x00FF, 0xFF00, RY , GY<<8, BY , RGB2YUV_SHIFT+8)
1252 BGR2Y(uint32_t, rgb32ToY, 0, 0,16, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY , BY<< 8, RGB2YUV_SHIFT+8)
1253 BGR2Y(uint32_t,rgb321ToY, 0,16,16, 0xFF00, 0x00FF, 0xFF00, RY , GY<<8, BY , RGB2YUV_SHIFT+8)
1254 BGR2Y(uint16_t, bgr16ToY, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RY<<11, GY<<5, BY , RGB2YUV_SHIFT+8)
1255 BGR2Y(uint16_t, bgr15ToY, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RY<<10, GY<<5, BY , RGB2YUV_SHIFT+7)
1256 BGR2Y(uint16_t, rgb16ToY, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RY , GY<<5, BY<<11, RGB2YUV_SHIFT+8)
1257 BGR2Y(uint16_t, rgb15ToY, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RY , GY<<5, BY<<10, RGB2YUV_SHIFT+7)
1259 static inline void abgrToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1262 for (i=0; i<width; i++) {
1267 #define BGR2UV(type, name, shr, shg, shb, shp, maskr, maskg, maskb, RU, GU, BU, RV, GV, BV, S) \
1268 static inline void name(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\
1271 for (i=0; i<width; i++) {\
1272 int b= ((((const type*)src)[i]>>shp)&maskb)>>shb;\
1273 int g= ((((const type*)src)[i]>>shp)&maskg)>>shg;\
1274 int r= ((((const type*)src)[i]>>shp)&maskr)>>shr;\
1276 dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<((S)-1)))>>(S);\
1277 dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<((S)-1)))>>(S);\
1280 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 for (i=0; i<width; i++) {\
1284 int pix0= ((const type*)src)[2*i+0]>>shp;\
1285 int pix1= ((const type*)src)[2*i+1]>>shp;\
1286 int g= (pix0&~(maskr|maskb))+(pix1&~(maskr|maskb));\
1287 int b= ((pix0+pix1-g)&(maskb|(2*maskb)))>>shb;\
1288 int r= ((pix0+pix1-g)&(maskr|(2*maskr)))>>shr;\
1289 g&= maskg|(2*maskg);\
1293 dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<(S)))>>((S)+1);\
1294 dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<(S)))>>((S)+1);\
1298 BGR2UV(uint32_t, bgr32ToUV,16, 0, 0, 0, 0xFF0000, 0xFF00, 0x00FF, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1299 BGR2UV(uint32_t,bgr321ToUV,16, 0, 0, 8, 0xFF0000, 0xFF00, 0x00FF, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1300 BGR2UV(uint32_t, rgb32ToUV, 0, 0,16, 0, 0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1301 BGR2UV(uint32_t,rgb321ToUV, 0, 0,16, 8, 0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1302 BGR2UV(uint16_t, bgr16ToUV, 0, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RU<<11, GU<<5, BU , RV<<11, GV<<5, BV , RGB2YUV_SHIFT+8)
1303 BGR2UV(uint16_t, bgr15ToUV, 0, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RU<<10, GU<<5, BU , RV<<10, GV<<5, BV , RGB2YUV_SHIFT+7)
1304 BGR2UV(uint16_t, rgb16ToUV, 0, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RU , GU<<5, BU<<11, RV , GV<<5, BV<<11, RGB2YUV_SHIFT+8)
1305 BGR2UV(uint16_t, rgb15ToUV, 0, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RU , GU<<5, BU<<10, RV , GV<<5, BV<<10, RGB2YUV_SHIFT+7)
1307 static inline void palToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)
1310 for (i=0; i<width; i++) {
1313 dst[i]= pal[d] >> 24;
1317 static inline void palToY(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)
1320 for (i=0; i<width; i++) {
1323 dst[i]= pal[d] & 0xFF;
1327 static inline void palToUV(uint8_t *dstU, uint8_t *dstV,
1328 const uint8_t *src1, const uint8_t *src2,
1329 long width, uint32_t *pal)
1332 assert(src1 == src2);
1333 for (i=0; i<width; i++) {
1334 int p= pal[src1[i]];
1341 static inline void monowhite2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1344 for (i=0; i<width/8; i++) {
1347 dst[8*i+j]= ((d>>(7-j))&1)*255;
1351 static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1354 for (i=0; i<width/8; i++) {
1357 dst[8*i+j]= ((d>>(7-j))&1)*255;
1361 //Note: we have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW+MMX2 one
1363 #if CONFIG_RUNTIME_CPUDETECT
1364 # define COMPILE_C 1
1366 # define COMPILE_MMX 1
1367 # define COMPILE_MMX2 1
1368 # define COMPILE_3DNOW 1
1370 # define COMPILE_ALTIVEC HAVE_ALTIVEC
1372 #else /* CONFIG_RUNTIME_CPUDETECT */
1375 # define COMPILE_MMX2 1
1376 # elif HAVE_AMD3DNOW
1377 # define COMPILE_3DNOW 1
1379 # define COMPILE_MMX 1
1381 # define COMPILE_C 1
1383 # elif ARCH_PPC && HAVE_ALTIVEC
1384 # define COMPILE_ALTIVEC 1
1386 # define COMPILE_C 1
1391 # define COMPILE_C 0
1394 # define COMPILE_MMX 0
1396 #ifndef COMPILE_MMX2
1397 # define COMPILE_MMX2 0
1399 #ifndef COMPILE_3DNOW
1400 # define COMPILE_3DNOW 0
1402 #ifndef COMPILE_ALTIVEC
1403 # define COMPILE_ALTIVEC 0
1406 #define COMPILE_TEMPLATE_MMX 0
1407 #define COMPILE_TEMPLATE_MMX2 0
1408 #define COMPILE_TEMPLATE_AMD3DNOW 0
1409 #define COMPILE_TEMPLATE_ALTIVEC 0
1412 #define RENAME(a) a ## _C
1413 #include "swscale_template.c"
1418 #undef COMPILE_TEMPLATE_ALTIVEC
1419 #define COMPILE_TEMPLATE_ALTIVEC 1
1420 #define RENAME(a) a ## _altivec
1421 #include "swscale_template.c"
1429 #undef COMPILE_TEMPLATE_MMX
1430 #undef COMPILE_TEMPLATE_MMX2
1431 #undef COMPILE_TEMPLATE_AMD3DNOW
1432 #define COMPILE_TEMPLATE_MMX 1
1433 #define COMPILE_TEMPLATE_MMX2 0
1434 #define COMPILE_TEMPLATE_AMD3DNOW 0
1435 #define RENAME(a) a ## _MMX
1436 #include "swscale_template.c"
1442 #undef COMPILE_TEMPLATE_MMX
1443 #undef COMPILE_TEMPLATE_MMX2
1444 #undef COMPILE_TEMPLATE_AMD3DNOW
1445 #define COMPILE_TEMPLATE_MMX 1
1446 #define COMPILE_TEMPLATE_MMX2 1
1447 #define COMPILE_TEMPLATE_AMD3DNOW 0
1448 #define RENAME(a) a ## _MMX2
1449 #include "swscale_template.c"
1455 #undef COMPILE_TEMPLATE_MMX
1456 #undef COMPILE_TEMPLATE_MMX2
1457 #undef COMPILE_TEMPLATE_AMD3DNOW
1458 #define COMPILE_TEMPLATE_MMX 1
1459 #define COMPILE_TEMPLATE_MMX2 0
1460 #define COMPILE_TEMPLATE_AMD3DNOW 1
1461 #define RENAME(a) a ## _3DNow
1462 #include "swscale_template.c"
1467 SwsFunc ff_getSwsFunc(SwsContext *c)
1469 #if CONFIG_RUNTIME_CPUDETECT
1470 int flags = c->flags;
1473 // ordered per speed fastest first
1474 if (flags & SWS_CPU_CAPS_MMX2) {
1475 sws_init_swScale_MMX2(c);
1476 return swScale_MMX2;
1477 } else if (flags & SWS_CPU_CAPS_3DNOW) {
1478 sws_init_swScale_3DNow(c);
1479 return swScale_3DNow;
1480 } else if (flags & SWS_CPU_CAPS_MMX) {
1481 sws_init_swScale_MMX(c);
1484 sws_init_swScale_C(c);
1490 if (flags & SWS_CPU_CAPS_ALTIVEC) {
1491 sws_init_swScale_altivec(c);
1492 return swScale_altivec;
1494 sws_init_swScale_C(c);
1498 sws_init_swScale_C(c);
1500 #endif /* ARCH_X86 */
1501 #else //CONFIG_RUNTIME_CPUDETECT
1502 #if COMPILE_TEMPLATE_MMX2
1503 sws_init_swScale_MMX2(c);
1504 return swScale_MMX2;
1505 #elif COMPILE_TEMPLATE_AMD3DNOW
1506 sws_init_swScale_3DNow(c);
1507 return swScale_3DNow;
1508 #elif COMPILE_TEMPLATE_MMX
1509 sws_init_swScale_MMX(c);
1511 #elif COMPILE_TEMPLATE_ALTIVEC
1512 sws_init_swScale_altivec(c);
1513 return swScale_altivec;
1515 sws_init_swScale_C(c);
1518 #endif //!CONFIG_RUNTIME_CPUDETECT
1521 static void copyPlane(const uint8_t *src, int srcStride,
1522 int srcSliceY, int srcSliceH, int width,
1523 uint8_t *dst, int dstStride)
1525 dst += dstStride * srcSliceY;
1526 if (dstStride == srcStride && srcStride > 0) {
1527 memcpy(dst, src, srcSliceH * dstStride);
1530 for (i=0; i<srcSliceH; i++) {
1531 memcpy(dst, src, width);
1538 static int planarToNv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1539 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1541 uint8_t *dst = dstParam[1] + dstStride[1]*srcSliceY/2;
1543 copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
1544 dstParam[0], dstStride[0]);
1546 if (c->dstFormat == PIX_FMT_NV12)
1547 interleaveBytes(src[1], src[2], dst, c->srcW/2, srcSliceH/2, srcStride[1], srcStride[2], dstStride[0]);
1549 interleaveBytes(src[2], src[1], dst, c->srcW/2, srcSliceH/2, srcStride[2], srcStride[1], dstStride[0]);
1554 static int planarToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1555 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1557 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1559 yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
1564 static int planarToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1565 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1567 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1569 yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
1574 static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1575 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1577 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1579 yuv422ptoyuy2(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
1584 static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1585 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1587 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1589 yuv422ptouyvy(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
1594 static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1595 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1597 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1598 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
1599 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
1601 yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1604 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1609 static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1610 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1612 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1613 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
1614 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
1616 yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1621 static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1622 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1624 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1625 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
1626 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
1628 uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1631 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1636 static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1637 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1639 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1640 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
1641 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
1643 uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1648 static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1651 for (i=0; i<num_pixels; i++)
1652 ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | (src[(i<<1)+1] << 24);
1655 static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1659 for (i=0; i<num_pixels; i++)
1660 ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | src[(i<<1)+1];
1663 static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1667 for (i=0; i<num_pixels; i++) {
1669 dst[0]= palette[src[i<<1]*4+0];
1670 dst[1]= palette[src[i<<1]*4+1];
1671 dst[2]= palette[src[i<<1]*4+2];
1676 static int palToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1677 int srcSliceH, uint8_t* dst[], int dstStride[])
1679 const enum PixelFormat srcFormat= c->srcFormat;
1680 const enum PixelFormat dstFormat= c->dstFormat;
1681 void (*conv)(const uint8_t *src, uint8_t *dst, long num_pixels,
1682 const uint8_t *palette)=NULL;
1684 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
1685 const uint8_t *srcPtr= src[0];
1687 if (srcFormat == PIX_FMT_GRAY8A) {
1688 switch (dstFormat) {
1689 case PIX_FMT_RGB32 : conv = gray8aToPacked32; break;
1690 case PIX_FMT_BGR32 : conv = gray8aToPacked32; break;
1691 case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;
1692 case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;
1693 case PIX_FMT_RGB24 : conv = gray8aToPacked24; break;
1694 case PIX_FMT_BGR24 : conv = gray8aToPacked24; break;
1696 } else if (usePal(srcFormat)) {
1697 switch (dstFormat) {
1698 case PIX_FMT_RGB32 : conv = sws_convertPalette8ToPacked32; break;
1699 case PIX_FMT_BGR32 : conv = sws_convertPalette8ToPacked32; break;
1700 case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;
1701 case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;
1702 case PIX_FMT_RGB24 : conv = sws_convertPalette8ToPacked24; break;
1703 case PIX_FMT_BGR24 : conv = sws_convertPalette8ToPacked24; break;
1708 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
1709 sws_format_name(srcFormat), sws_format_name(dstFormat));
1711 for (i=0; i<srcSliceH; i++) {
1712 conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);
1713 srcPtr+= srcStride[0];
1714 dstPtr+= dstStride[0];
1721 #define isRGBA32(x) ( \
1722 (x) == PIX_FMT_ARGB \
1723 || (x) == PIX_FMT_RGBA \
1724 || (x) == PIX_FMT_BGRA \
1725 || (x) == PIX_FMT_ABGR \
1728 /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */
1729 static int rgbToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1730 int srcSliceH, uint8_t* dst[], int dstStride[])
1732 const enum PixelFormat srcFormat= c->srcFormat;
1733 const enum PixelFormat dstFormat= c->dstFormat;
1734 const int srcBpp= (c->srcFormatBpp + 7) >> 3;
1735 const int dstBpp= (c->dstFormatBpp + 7) >> 3;
1736 const int srcId= c->srcFormatBpp >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */
1737 const int dstId= c->dstFormatBpp >> 2;
1738 void (*conv)(const uint8_t *src, uint8_t *dst, long src_size)=NULL;
1740 #define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)
1742 if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {
1743 if ( CONV_IS(ABGR, RGBA)
1744 || CONV_IS(ARGB, BGRA)
1745 || CONV_IS(BGRA, ARGB)
1746 || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;
1747 else if (CONV_IS(ABGR, ARGB)
1748 || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;
1749 else if (CONV_IS(ABGR, BGRA)
1750 || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;
1751 else if (CONV_IS(BGRA, RGBA)
1752 || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;
1753 else if (CONV_IS(BGRA, ABGR)
1754 || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;
1757 if ( (isBGRinInt(srcFormat) && isBGRinInt(dstFormat))
1758 || (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {
1759 switch(srcId | (dstId<<4)) {
1760 case 0x34: conv= rgb16to15; break;
1761 case 0x36: conv= rgb24to15; break;
1762 case 0x38: conv= rgb32to15; break;
1763 case 0x43: conv= rgb15to16; break;
1764 case 0x46: conv= rgb24to16; break;
1765 case 0x48: conv= rgb32to16; break;
1766 case 0x63: conv= rgb15to24; break;
1767 case 0x64: conv= rgb16to24; break;
1768 case 0x68: conv= rgb32to24; break;
1769 case 0x83: conv= rgb15to32; break;
1770 case 0x84: conv= rgb16to32; break;
1771 case 0x86: conv= rgb24to32; break;
1773 } else if ( (isBGRinInt(srcFormat) && isRGBinInt(dstFormat))
1774 || (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {
1775 switch(srcId | (dstId<<4)) {
1776 case 0x33: conv= rgb15tobgr15; break;
1777 case 0x34: conv= rgb16tobgr15; break;
1778 case 0x36: conv= rgb24tobgr15; break;
1779 case 0x38: conv= rgb32tobgr15; break;
1780 case 0x43: conv= rgb15tobgr16; break;
1781 case 0x44: conv= rgb16tobgr16; break;
1782 case 0x46: conv= rgb24tobgr16; break;
1783 case 0x48: conv= rgb32tobgr16; break;
1784 case 0x63: conv= rgb15tobgr24; break;
1785 case 0x64: conv= rgb16tobgr24; break;
1786 case 0x66: conv= rgb24tobgr24; break;
1787 case 0x68: conv= rgb32tobgr24; break;
1788 case 0x83: conv= rgb15tobgr32; break;
1789 case 0x84: conv= rgb16tobgr32; break;
1790 case 0x86: conv= rgb24tobgr32; break;
1795 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
1796 sws_format_name(srcFormat), sws_format_name(dstFormat));
1798 const uint8_t *srcPtr= src[0];
1799 uint8_t *dstPtr= dst[0];
1800 if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) && !isRGBA32(dstFormat))
1801 srcPtr += ALT32_CORR;
1803 if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) && !isRGBA32(srcFormat))
1804 dstPtr += ALT32_CORR;
1806 if (dstStride[0]*srcBpp == srcStride[0]*dstBpp && srcStride[0] > 0 && !(srcStride[0]%srcBpp))
1807 conv(srcPtr, dstPtr + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]);
1810 dstPtr += dstStride[0]*srcSliceY;
1812 for (i=0; i<srcSliceH; i++) {
1813 conv(srcPtr, dstPtr, c->srcW*srcBpp);
1814 srcPtr+= srcStride[0];
1815 dstPtr+= dstStride[0];
1822 static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1823 int srcSliceH, uint8_t* dst[], int dstStride[])
1827 dst[0]+ srcSliceY *dstStride[0],
1828 dst[1]+(srcSliceY>>1)*dstStride[1],
1829 dst[2]+(srcSliceY>>1)*dstStride[2],
1831 dstStride[0], dstStride[1], srcStride[0]);
1833 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1837 static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1838 int srcSliceH, uint8_t* dst[], int dstStride[])
1840 copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
1841 dst[0], dstStride[0]);
1843 planar2x(src[1], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,
1844 srcSliceH >> 2, srcStride[1], dstStride[1]);
1845 planar2x(src[2], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,
1846 srcSliceH >> 2, srcStride[2], dstStride[2]);
1848 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1852 /* unscaled copy like stuff (assumes nearly identical formats) */
1853 static int packedCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1854 int srcSliceH, uint8_t* dst[], int dstStride[])
1856 if (dstStride[0]==srcStride[0] && srcStride[0] > 0)
1857 memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]);
1860 const uint8_t *srcPtr= src[0];
1861 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
1864 /* universal length finder */
1865 while(length+c->srcW <= FFABS(dstStride[0])
1866 && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW;
1869 for (i=0; i<srcSliceH; i++) {
1870 memcpy(dstPtr, srcPtr, length);
1871 srcPtr+= srcStride[0];
1872 dstPtr+= dstStride[0];
1878 #define DITHER_COPY(dst, dstStride, src, srcStride, bswap, dbswap)\
1879 uint16_t scale= dither_scale[dst_depth-1][src_depth-1];\
1880 int shift= src_depth-dst_depth + dither_scale[src_depth-2][dst_depth-1];\
1881 for (i = 0; i < height; i++) {\
1882 uint8_t *dither= dithers[src_depth-9][i&7];\
1883 for (j = 0; j < length-7; j+=8){\
1884 dst[j+0] = dbswap((bswap(src[j+0]) + dither[0])*scale>>shift);\
1885 dst[j+1] = dbswap((bswap(src[j+1]) + dither[1])*scale>>shift);\
1886 dst[j+2] = dbswap((bswap(src[j+2]) + dither[2])*scale>>shift);\
1887 dst[j+3] = dbswap((bswap(src[j+3]) + dither[3])*scale>>shift);\
1888 dst[j+4] = dbswap((bswap(src[j+4]) + dither[4])*scale>>shift);\
1889 dst[j+5] = dbswap((bswap(src[j+5]) + dither[5])*scale>>shift);\
1890 dst[j+6] = dbswap((bswap(src[j+6]) + dither[6])*scale>>shift);\
1891 dst[j+7] = dbswap((bswap(src[j+7]) + dither[7])*scale>>shift);\
1893 for (; j < length; j++)\
1894 dst[j] = dbswap((bswap(src[j]) + dither[j&7])*scale>>shift);\
1900 static int planarCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1901 int srcSliceH, uint8_t* dst[], int dstStride[])
1904 for (plane=0; plane<4; plane++) {
1905 int length= (plane==0 || plane==3) ? c->srcW : -((-c->srcW )>>c->chrDstHSubSample);
1906 int y= (plane==0 || plane==3) ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample);
1907 int height= (plane==0 || plane==3) ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample);
1908 const uint8_t *srcPtr= src[plane];
1909 uint8_t *dstPtr= dst[plane] + dstStride[plane]*y;
1911 if (!dst[plane]) continue;
1912 // ignore palette for GRAY8
1913 if (plane == 1 && !dst[2]) continue;
1914 if (!src[plane] || (plane == 1 && !src[2])) {
1915 if(is16BPS(c->dstFormat))
1917 fillPlane(dst[plane], dstStride[plane], length, height, y, (plane==3) ? 255 : 128);
1919 if(isNBPS(c->srcFormat) || isNBPS(c->dstFormat)
1920 || (is16BPS(c->srcFormat) != is16BPS(c->dstFormat))
1922 const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1+1;
1923 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1+1;
1924 const uint16_t *srcPtr2 = (const uint16_t*)srcPtr;
1925 uint16_t *dstPtr2 = (uint16_t*)dstPtr;
1927 if (dst_depth == 8) {
1928 if(isBE(c->srcFormat) == HAVE_BIGENDIAN){
1929 DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, , )
1931 DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, av_bswap16, )
1933 } else if (src_depth == 8) {
1934 for (i = 0; i < height; i++) {
1935 if(isBE(c->dstFormat)){
1936 for (j = 0; j < length; j++)
1937 AV_WB16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |
1938 (srcPtr[j]>>(2*8-dst_depth)));
1940 for (j = 0; j < length; j++)
1941 AV_WL16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |
1942 (srcPtr[j]>>(2*8-dst_depth)));
1944 dstPtr2 += dstStride[plane]/2;
1945 srcPtr += srcStride[plane];
1947 } else if (src_depth <= dst_depth) {
1948 for (i = 0; i < height; i++) {
1949 #define COPY_UP(r,w) \
1950 for (j = 0; j < length; j++){ \
1951 unsigned int v= r(&srcPtr2[j]);\
1952 w(&dstPtr2[j], (v<<(dst_depth-src_depth)) | \
1953 (v>>(2*src_depth-dst_depth)));\
1955 if(isBE(c->srcFormat)){
1956 if(isBE(c->dstFormat)){
1957 COPY_UP(AV_RB16, AV_WB16)
1959 COPY_UP(AV_RB16, AV_WL16)
1962 if(isBE(c->dstFormat)){
1963 COPY_UP(AV_RL16, AV_WB16)
1965 COPY_UP(AV_RL16, AV_WL16)
1968 dstPtr2 += dstStride[plane]/2;
1969 srcPtr2 += srcStride[plane]/2;
1972 if(isBE(c->srcFormat) == HAVE_BIGENDIAN){
1973 if(isBE(c->dstFormat) == HAVE_BIGENDIAN){
1974 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , )
1976 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , av_bswap16)
1979 if(isBE(c->dstFormat) == HAVE_BIGENDIAN){
1980 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, )
1982 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, av_bswap16)
1986 } else if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat)
1987 && isBE(c->srcFormat) != isBE(c->dstFormat)) {
1989 for (i=0; i<height; i++) {
1990 for (j=0; j<length; j++)
1991 ((uint16_t*)dstPtr)[j] = av_bswap16(((const uint16_t*)srcPtr)[j]);
1992 srcPtr+= srcStride[plane];
1993 dstPtr+= dstStride[plane];
1995 } else if (dstStride[plane] == srcStride[plane] &&
1996 srcStride[plane] > 0 && srcStride[plane] == length) {
1997 memcpy(dst[plane] + dstStride[plane]*y, src[plane],
1998 height*dstStride[plane]);
2000 if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat))
2002 for (i=0; i<height; i++) {
2003 memcpy(dstPtr, srcPtr, length);
2004 srcPtr+= srcStride[plane];
2005 dstPtr+= dstStride[plane];
2013 int ff_hardcodedcpuflags(void)
2016 #if COMPILE_TEMPLATE_MMX2
2017 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2;
2018 #elif COMPILE_TEMPLATE_AMD3DNOW
2019 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_3DNOW;
2020 #elif COMPILE_TEMPLATE_MMX
2021 flags |= SWS_CPU_CAPS_MMX;
2022 #elif COMPILE_TEMPLATE_ALTIVEC
2023 flags |= SWS_CPU_CAPS_ALTIVEC;
2025 flags |= SWS_CPU_CAPS_BFIN;
2030 void ff_get_unscaled_swscale(SwsContext *c)
2032 const enum PixelFormat srcFormat = c->srcFormat;
2033 const enum PixelFormat dstFormat = c->dstFormat;
2034 const int flags = c->flags;
2035 const int dstH = c->dstH;
2038 needsDither= isAnyRGB(dstFormat)
2039 && c->dstFormatBpp < 24
2040 && (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));
2043 if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {
2044 c->swScale= planarToNv12Wrapper;
2047 if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P || srcFormat==PIX_FMT_YUVA420P) && isAnyRGB(dstFormat)
2048 && !(flags & SWS_ACCURATE_RND) && !(dstH&1)) {
2049 c->swScale= ff_yuv2rgb_get_func_ptr(c);
2052 if (srcFormat==PIX_FMT_YUV410P && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_BITEXACT)) {
2053 c->swScale= yvu9ToYv12Wrapper;
2057 if (srcFormat==PIX_FMT_BGR24 && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_ACCURATE_RND))
2058 c->swScale= bgr24ToYv12Wrapper;
2060 /* RGB/BGR -> RGB/BGR (no dither needed forms) */
2061 if ( isAnyRGB(srcFormat)
2062 && isAnyRGB(dstFormat)
2063 && srcFormat != PIX_FMT_BGR8 && dstFormat != PIX_FMT_BGR8
2064 && srcFormat != PIX_FMT_RGB8 && dstFormat != PIX_FMT_RGB8
2065 && srcFormat != PIX_FMT_BGR4 && dstFormat != PIX_FMT_BGR4
2066 && srcFormat != PIX_FMT_RGB4 && dstFormat != PIX_FMT_RGB4
2067 && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE
2068 && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE
2069 && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK
2070 && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE
2071 && srcFormat != PIX_FMT_RGB48LE && dstFormat != PIX_FMT_RGB48LE
2072 && srcFormat != PIX_FMT_RGB48BE && dstFormat != PIX_FMT_RGB48BE
2073 && srcFormat != PIX_FMT_BGR48LE && dstFormat != PIX_FMT_BGR48LE
2074 && srcFormat != PIX_FMT_BGR48BE && dstFormat != PIX_FMT_BGR48BE
2075 && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))
2076 c->swScale= rgbToRgbWrapper;
2078 if ((usePal(srcFormat) && (
2079 dstFormat == PIX_FMT_RGB32 ||
2080 dstFormat == PIX_FMT_RGB32_1 ||
2081 dstFormat == PIX_FMT_RGB24 ||
2082 dstFormat == PIX_FMT_BGR32 ||
2083 dstFormat == PIX_FMT_BGR32_1 ||
2084 dstFormat == PIX_FMT_BGR24)))
2085 c->swScale= palToRgbWrapper;
2087 if (srcFormat == PIX_FMT_YUV422P) {
2088 if (dstFormat == PIX_FMT_YUYV422)
2089 c->swScale= yuv422pToYuy2Wrapper;
2090 else if (dstFormat == PIX_FMT_UYVY422)
2091 c->swScale= yuv422pToUyvyWrapper;
2094 /* LQ converters if -sws 0 or -sws 4*/
2095 if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {
2097 if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {
2098 if (dstFormat == PIX_FMT_YUYV422)
2099 c->swScale= planarToYuy2Wrapper;
2100 else if (dstFormat == PIX_FMT_UYVY422)
2101 c->swScale= planarToUyvyWrapper;
2104 if(srcFormat == PIX_FMT_YUYV422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
2105 c->swScale= yuyvToYuv420Wrapper;
2106 if(srcFormat == PIX_FMT_UYVY422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
2107 c->swScale= uyvyToYuv420Wrapper;
2108 if(srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)
2109 c->swScale= yuyvToYuv422Wrapper;
2110 if(srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)
2111 c->swScale= uyvyToYuv422Wrapper;
2114 if ((c->flags & SWS_CPU_CAPS_ALTIVEC) &&
2115 !(c->flags & SWS_BITEXACT) &&
2116 srcFormat == PIX_FMT_YUV420P) {
2117 // unscaled YV12 -> packed YUV, we want speed
2118 if (dstFormat == PIX_FMT_YUYV422)
2119 c->swScale= yv12toyuy2_unscaled_altivec;
2120 else if (dstFormat == PIX_FMT_UYVY422)
2121 c->swScale= yv12touyvy_unscaled_altivec;
2126 if ( srcFormat == dstFormat
2127 || (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P)
2128 || (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P)
2129 || (isPlanarYUV(srcFormat) && isGray(dstFormat))
2130 || (isPlanarYUV(dstFormat) && isGray(srcFormat))
2131 || (isGray(dstFormat) && isGray(srcFormat))
2132 || (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat)
2133 && c->chrDstHSubSample == c->chrSrcHSubSample
2134 && c->chrDstVSubSample == c->chrSrcVSubSample
2135 && dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21
2136 && srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))
2138 if (isPacked(c->srcFormat))
2139 c->swScale= packedCopyWrapper;
2140 else /* Planar YUV or gray */
2141 c->swScale= planarCopyWrapper;
2144 if (flags & SWS_CPU_CAPS_BFIN)
2145 ff_bfin_get_unscaled_swscale (c);
2149 static void reset_ptr(const uint8_t* src[], int format)
2151 if(!isALPHA(format))
2153 if(!isPlanarYUV(format)) {
2156 if (!usePal(format))
2161 static int check_image_pointers(uint8_t *data[4], enum PixelFormat pix_fmt,
2162 const int linesizes[4])
2164 const AVPixFmtDescriptor *desc = &av_pix_fmt_descriptors[pix_fmt];
2167 for (i = 0; i < 4; i++) {
2168 int plane = desc->comp[i].plane;
2169 if (!data[plane] || !linesizes[plane])
2177 * swscale wrapper, so we don't need to export the SwsContext.
2178 * Assumes planar YUV to be in YUV order instead of YVU.
2180 int sws_scale(SwsContext *c, const uint8_t* const src[], const int srcStride[], int srcSliceY,
2181 int srcSliceH, uint8_t* const dst[], const int dstStride[])
2184 const uint8_t* src2[4]= {src[0], src[1], src[2], src[3]};
2185 uint8_t* dst2[4]= {dst[0], dst[1], dst[2], dst[3]};
2187 // do not mess up sliceDir if we have a "trailing" 0-size slice
2191 if (!check_image_pointers(src, c->srcFormat, srcStride)) {
2192 av_log(c, AV_LOG_ERROR, "bad src image pointers\n");
2195 if (!check_image_pointers(dst, c->dstFormat, dstStride)) {
2196 av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");
2200 if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {
2201 av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");
2204 if (c->sliceDir == 0) {
2205 if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;
2208 if (usePal(c->srcFormat)) {
2209 for (i=0; i<256; i++) {
2210 int p, r, g, b, y, u, v, a = 0xff;
2211 if(c->srcFormat == PIX_FMT_PAL8) {
2212 p=((const uint32_t*)(src[1]))[i];
2217 } else if(c->srcFormat == PIX_FMT_RGB8) {
2221 } else if(c->srcFormat == PIX_FMT_BGR8) {
2225 } else if(c->srcFormat == PIX_FMT_RGB4_BYTE) {
2229 } else if(c->srcFormat == PIX_FMT_GRAY8 || c->srcFormat == PIX_FMT_GRAY8A) {
2232 assert(c->srcFormat == PIX_FMT_BGR4_BYTE);
2237 y= av_clip_uint8((RY*r + GY*g + BY*b + ( 33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2238 u= av_clip_uint8((RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2239 v= av_clip_uint8((RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2240 c->pal_yuv[i]= y + (u<<8) + (v<<16) + (a<<24);
2242 switch(c->dstFormat) {
2247 c->pal_rgb[i]= r + (g<<8) + (b<<16) + (a<<24);
2249 case PIX_FMT_BGR32_1:
2253 c->pal_rgb[i]= a + (r<<8) + (g<<16) + (b<<24);
2255 case PIX_FMT_RGB32_1:
2259 c->pal_rgb[i]= a + (b<<8) + (g<<16) + (r<<24);
2266 c->pal_rgb[i]= b + (g<<8) + (r<<16) + (a<<24);
2271 // copy strides, so they can safely be modified
2272 if (c->sliceDir == 1) {
2273 // slices go from top to bottom
2274 int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2], srcStride[3]};
2275 int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2], dstStride[3]};
2277 reset_ptr(src2, c->srcFormat);
2278 reset_ptr((const uint8_t**)dst2, c->dstFormat);
2280 /* reset slice direction at end of frame */
2281 if (srcSliceY + srcSliceH == c->srcH)
2284 return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2, dstStride2);
2286 // slices go from bottom to top => we flip the image internally
2287 int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2], -srcStride[3]};
2288 int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2], -dstStride[3]};
2290 src2[0] += (srcSliceH-1)*srcStride[0];
2291 if (!usePal(c->srcFormat))
2292 src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1];
2293 src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2];
2294 src2[3] += (srcSliceH-1)*srcStride[3];
2295 dst2[0] += ( c->dstH -1)*dstStride[0];
2296 dst2[1] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1];
2297 dst2[2] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2];
2298 dst2[3] += ( c->dstH -1)*dstStride[3];
2300 reset_ptr(src2, c->srcFormat);
2301 reset_ptr((const uint8_t**)dst2, c->dstFormat);
2303 /* reset slice direction at end of frame */
2307 return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2);
2311 #if LIBSWSCALE_VERSION_MAJOR < 1
2312 int sws_scale_ordered(SwsContext *c, const uint8_t* const src[], int srcStride[], int srcSliceY,
2313 int srcSliceH, uint8_t* dst[], int dstStride[])
2315 return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride);
2319 /* Convert the palette to the same packed 32-bit format as the palette */
2320 void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
2324 for (i=0; i<num_pixels; i++)
2325 ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];
2328 /* Palette format: ABCD -> dst format: ABC */
2329 void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
2333 for (i=0; i<num_pixels; i++) {
2335 dst[0]= palette[src[i]*4+0];
2336 dst[1]= palette[src[i]*4+1];
2337 dst[2]= palette[src[i]*4+2];