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, int output_bits)
367 //FIXME Optimize (just quickly written not optimized..)
369 int shift = 11 + 16 - output_bits;
371 #define output_pixel(pos, val) \
373 if (output_bits == 16) { \
374 AV_WB16(pos, av_clip_uint16(val >> shift)); \
376 AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
379 if (output_bits == 16) { \
380 AV_WL16(pos, av_clip_uint16(val >> shift)); \
382 AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
385 for (i = 0; i < dstW; i++) {
386 int val = 1 << (26-output_bits);
389 for (j = 0; j < lumFilterSize; j++)
390 val += lumSrc[j][i] * lumFilter[j];
392 output_pixel(&dest[i], val);
396 for (i = 0; i < chrDstW; i++) {
397 int u = 1 << (26-output_bits);
398 int v = 1 << (26-output_bits);
401 for (j = 0; j < chrFilterSize; j++) {
402 u += chrSrc[j][i ] * chrFilter[j];
403 v += chrSrc[j][i + VOFW] * chrFilter[j];
406 output_pixel(&uDest[i], u);
407 output_pixel(&vDest[i], v);
411 if (CONFIG_SWSCALE_ALPHA && aDest) {
412 for (i = 0; i < dstW; i++) {
413 int val = 1 << (26-output_bits);
416 for (j = 0; j < lumFilterSize; j++)
417 val += alpSrc[j][i] * lumFilter[j];
419 output_pixel(&aDest[i], val);
424 static inline void yuv2yuvX16inC(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, int dstW, int chrDstW,
427 enum PixelFormat dstFormat)
429 if (isNBPS(dstFormat)) {
430 const int depth = av_pix_fmt_descriptors[dstFormat].comp[0].depth_minus1+1;
431 yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
432 chrFilter, chrSrc, chrFilterSize,
434 dest, uDest, vDest, aDest,
435 dstW, chrDstW, isBE(dstFormat), depth);
437 if (isBE(dstFormat)) {
438 yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
439 chrFilter, chrSrc, chrFilterSize,
441 dest, uDest, vDest, aDest,
442 dstW, chrDstW, 1, 16);
444 yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
445 chrFilter, chrSrc, chrFilterSize,
447 dest, uDest, vDest, aDest,
448 dstW, chrDstW, 0, 16);
453 static inline void yuv2yuvXinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
454 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
455 const int16_t **alpSrc, uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, int dstW, int chrDstW)
457 //FIXME Optimize (just quickly written not optimized..)
459 for (i=0; i<dstW; i++) {
462 for (j=0; j<lumFilterSize; j++)
463 val += lumSrc[j][i] * lumFilter[j];
465 dest[i]= av_clip_uint8(val>>19);
469 for (i=0; i<chrDstW; i++) {
473 for (j=0; j<chrFilterSize; j++) {
474 u += chrSrc[j][i] * chrFilter[j];
475 v += chrSrc[j][i + VOFW] * chrFilter[j];
478 uDest[i]= av_clip_uint8(u>>19);
479 vDest[i]= av_clip_uint8(v>>19);
482 if (CONFIG_SWSCALE_ALPHA && aDest)
483 for (i=0; i<dstW; i++) {
486 for (j=0; j<lumFilterSize; j++)
487 val += alpSrc[j][i] * lumFilter[j];
489 aDest[i]= av_clip_uint8(val>>19);
494 static inline void yuv2nv12XinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
495 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
496 uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat)
498 //FIXME Optimize (just quickly written not optimized..)
500 for (i=0; i<dstW; i++) {
503 for (j=0; j<lumFilterSize; j++)
504 val += lumSrc[j][i] * lumFilter[j];
506 dest[i]= av_clip_uint8(val>>19);
512 if (dstFormat == PIX_FMT_NV12)
513 for (i=0; i<chrDstW; i++) {
517 for (j=0; j<chrFilterSize; j++) {
518 u += chrSrc[j][i] * chrFilter[j];
519 v += chrSrc[j][i + VOFW] * chrFilter[j];
522 uDest[2*i]= av_clip_uint8(u>>19);
523 uDest[2*i+1]= av_clip_uint8(v>>19);
526 for (i=0; i<chrDstW; i++) {
530 for (j=0; j<chrFilterSize; j++) {
531 u += chrSrc[j][i] * chrFilter[j];
532 v += chrSrc[j][i + VOFW] * chrFilter[j];
535 uDest[2*i]= av_clip_uint8(v>>19);
536 uDest[2*i+1]= av_clip_uint8(u>>19);
540 #define YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha) \
541 for (i=0; i<(dstW>>1); i++) {\
547 int av_unused A1, A2;\
548 type av_unused *r, *b, *g;\
551 for (j=0; j<lumFilterSize; j++) {\
552 Y1 += lumSrc[j][i2] * lumFilter[j];\
553 Y2 += lumSrc[j][i2+1] * lumFilter[j];\
555 for (j=0; j<chrFilterSize; j++) {\
556 U += chrSrc[j][i] * chrFilter[j];\
557 V += chrSrc[j][i+VOFW] * chrFilter[j];\
566 for (j=0; j<lumFilterSize; j++) {\
567 A1 += alpSrc[j][i2 ] * lumFilter[j];\
568 A2 += alpSrc[j][i2+1] * lumFilter[j];\
574 #define YSCALE_YUV_2_PACKEDX_C(type,alpha) \
575 YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha)\
576 if ((Y1|Y2|U|V)&256) {\
577 if (Y1>255) Y1=255; \
578 else if (Y1<0)Y1=0; \
579 if (Y2>255) Y2=255; \
580 else if (Y2<0)Y2=0; \
586 if (alpha && ((A1|A2)&256)) {\
587 A1=av_clip_uint8(A1);\
588 A2=av_clip_uint8(A2);\
591 #define YSCALE_YUV_2_PACKEDX_FULL_C(rnd,alpha) \
592 for (i=0; i<dstW; i++) {\
600 for (j=0; j<lumFilterSize; j++) {\
601 Y += lumSrc[j][i ] * lumFilter[j];\
603 for (j=0; j<chrFilterSize; j++) {\
604 U += chrSrc[j][i ] * chrFilter[j];\
605 V += chrSrc[j][i+VOFW] * chrFilter[j];\
612 for (j=0; j<lumFilterSize; j++)\
613 A += alpSrc[j][i ] * lumFilter[j];\
616 A = av_clip_uint8(A);\
619 #define YSCALE_YUV_2_RGBX_FULL_C(rnd,alpha) \
620 YSCALE_YUV_2_PACKEDX_FULL_C(rnd>>3,alpha)\
621 Y-= c->yuv2rgb_y_offset;\
622 Y*= c->yuv2rgb_y_coeff;\
624 R= Y + V*c->yuv2rgb_v2r_coeff;\
625 G= Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;\
626 B= Y + U*c->yuv2rgb_u2b_coeff;\
627 if ((R|G|B)&(0xC0000000)) {\
628 if (R>=(256<<22)) R=(256<<22)-1; \
630 if (G>=(256<<22)) G=(256<<22)-1; \
632 if (B>=(256<<22)) B=(256<<22)-1; \
636 #define YSCALE_YUV_2_GRAY16_C \
637 for (i=0; i<(dstW>>1); i++) {\
646 for (j=0; j<lumFilterSize; j++) {\
647 Y1 += lumSrc[j][i2] * lumFilter[j];\
648 Y2 += lumSrc[j][i2+1] * lumFilter[j];\
652 if ((Y1|Y2|U|V)&65536) {\
653 if (Y1>65535) Y1=65535; \
654 else if (Y1<0)Y1=0; \
655 if (Y2>65535) Y2=65535; \
656 else if (Y2<0)Y2=0; \
659 #define YSCALE_YUV_2_RGBX_C(type,alpha) \
660 YSCALE_YUV_2_PACKEDX_C(type,alpha) /* FIXME fix tables so that clipping is not needed and then use _NOCLIP*/\
661 r = (type *)c->table_rV[V]; \
662 g = (type *)(c->table_gU[U] + c->table_gV[V]); \
663 b = (type *)c->table_bU[U];
665 #define YSCALE_YUV_2_PACKED2_C(type,alpha) \
666 for (i=0; i<(dstW>>1); i++) { \
668 int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>19; \
669 int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19; \
670 int U= (uvbuf0[i ]*uvalpha1+uvbuf1[i ]*uvalpha)>>19; \
671 int V= (uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19; \
672 type av_unused *r, *b, *g; \
673 int av_unused A1, A2; \
675 A1= (abuf0[i2 ]*yalpha1+abuf1[i2 ]*yalpha)>>19; \
676 A2= (abuf0[i2+1]*yalpha1+abuf1[i2+1]*yalpha)>>19; \
679 #define YSCALE_YUV_2_GRAY16_2_C \
680 for (i=0; i<(dstW>>1); i++) { \
682 int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>11; \
683 int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>11;
685 #define YSCALE_YUV_2_RGB2_C(type,alpha) \
686 YSCALE_YUV_2_PACKED2_C(type,alpha)\
687 r = (type *)c->table_rV[V];\
688 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
689 b = (type *)c->table_bU[U];
691 #define YSCALE_YUV_2_PACKED1_C(type,alpha) \
692 for (i=0; i<(dstW>>1); i++) {\
694 int Y1= buf0[i2 ]>>7;\
695 int Y2= buf0[i2+1]>>7;\
696 int U= (uvbuf1[i ])>>7;\
697 int V= (uvbuf1[i+VOFW])>>7;\
698 type av_unused *r, *b, *g;\
699 int av_unused A1, A2;\
705 #define YSCALE_YUV_2_GRAY16_1_C \
706 for (i=0; i<(dstW>>1); i++) {\
708 int Y1= buf0[i2 ]<<1;\
709 int Y2= buf0[i2+1]<<1;
711 #define YSCALE_YUV_2_RGB1_C(type,alpha) \
712 YSCALE_YUV_2_PACKED1_C(type,alpha)\
713 r = (type *)c->table_rV[V];\
714 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
715 b = (type *)c->table_bU[U];
717 #define YSCALE_YUV_2_PACKED1B_C(type,alpha) \
718 for (i=0; i<(dstW>>1); i++) {\
720 int Y1= buf0[i2 ]>>7;\
721 int Y2= buf0[i2+1]>>7;\
722 int U= (uvbuf0[i ] + uvbuf1[i ])>>8;\
723 int V= (uvbuf0[i+VOFW] + uvbuf1[i+VOFW])>>8;\
724 type av_unused *r, *b, *g;\
725 int av_unused A1, A2;\
731 #define YSCALE_YUV_2_RGB1B_C(type,alpha) \
732 YSCALE_YUV_2_PACKED1B_C(type,alpha)\
733 r = (type *)c->table_rV[V];\
734 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
735 b = (type *)c->table_bU[U];
737 #define YSCALE_YUV_2_MONO2_C \
738 const uint8_t * const d128=dither_8x8_220[y&7];\
739 uint8_t *g= c->table_gU[128] + c->table_gV[128];\
740 for (i=0; i<dstW-7; i+=8) {\
742 acc = g[((buf0[i ]*yalpha1+buf1[i ]*yalpha)>>19) + d128[0]];\
743 acc+= acc + g[((buf0[i+1]*yalpha1+buf1[i+1]*yalpha)>>19) + d128[1]];\
744 acc+= acc + g[((buf0[i+2]*yalpha1+buf1[i+2]*yalpha)>>19) + d128[2]];\
745 acc+= acc + g[((buf0[i+3]*yalpha1+buf1[i+3]*yalpha)>>19) + d128[3]];\
746 acc+= acc + g[((buf0[i+4]*yalpha1+buf1[i+4]*yalpha)>>19) + d128[4]];\
747 acc+= acc + g[((buf0[i+5]*yalpha1+buf1[i+5]*yalpha)>>19) + d128[5]];\
748 acc+= acc + g[((buf0[i+6]*yalpha1+buf1[i+6]*yalpha)>>19) + d128[6]];\
749 acc+= acc + g[((buf0[i+7]*yalpha1+buf1[i+7]*yalpha)>>19) + d128[7]];\
750 ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
754 #define YSCALE_YUV_2_MONOX_C \
755 const uint8_t * const d128=dither_8x8_220[y&7];\
756 uint8_t *g= c->table_gU[128] + c->table_gV[128];\
758 for (i=0; i<dstW-1; i+=2) {\
763 for (j=0; j<lumFilterSize; j++) {\
764 Y1 += lumSrc[j][i] * lumFilter[j];\
765 Y2 += lumSrc[j][i+1] * lumFilter[j];\
775 acc+= acc + g[Y1+d128[(i+0)&7]];\
776 acc+= acc + g[Y2+d128[(i+1)&7]];\
778 ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
783 #define YSCALE_YUV_2_ANYRGB_C(func, func2, func_g16, func_monoblack)\
784 switch(c->dstFormat) {\
785 case PIX_FMT_RGB48BE:\
786 case PIX_FMT_RGB48LE:\
788 ((uint8_t*)dest)[ 0]= r[Y1];\
789 ((uint8_t*)dest)[ 1]= r[Y1];\
790 ((uint8_t*)dest)[ 2]= g[Y1];\
791 ((uint8_t*)dest)[ 3]= g[Y1];\
792 ((uint8_t*)dest)[ 4]= b[Y1];\
793 ((uint8_t*)dest)[ 5]= b[Y1];\
794 ((uint8_t*)dest)[ 6]= r[Y2];\
795 ((uint8_t*)dest)[ 7]= r[Y2];\
796 ((uint8_t*)dest)[ 8]= g[Y2];\
797 ((uint8_t*)dest)[ 9]= g[Y2];\
798 ((uint8_t*)dest)[10]= b[Y2];\
799 ((uint8_t*)dest)[11]= b[Y2];\
803 case PIX_FMT_BGR48BE:\
804 case PIX_FMT_BGR48LE:\
806 ((uint8_t*)dest)[ 0] = ((uint8_t*)dest)[ 1] = b[Y1];\
807 ((uint8_t*)dest)[ 2] = ((uint8_t*)dest)[ 3] = g[Y1];\
808 ((uint8_t*)dest)[ 4] = ((uint8_t*)dest)[ 5] = r[Y1];\
809 ((uint8_t*)dest)[ 6] = ((uint8_t*)dest)[ 7] = b[Y2];\
810 ((uint8_t*)dest)[ 8] = ((uint8_t*)dest)[ 9] = g[Y2];\
811 ((uint8_t*)dest)[10] = ((uint8_t*)dest)[11] = r[Y2];\
818 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\
819 func(uint32_t,needAlpha)\
820 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? (A1<<24) : 0);\
821 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? (A2<<24) : 0);\
824 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\
826 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (A1<<24);\
827 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (A2<<24);\
831 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
832 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
840 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\
841 func(uint32_t,needAlpha)\
842 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? A1 : 0);\
843 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? A2 : 0);\
846 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\
848 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + A1;\
849 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + A2;\
853 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
854 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
861 ((uint8_t*)dest)[0]= r[Y1];\
862 ((uint8_t*)dest)[1]= g[Y1];\
863 ((uint8_t*)dest)[2]= b[Y1];\
864 ((uint8_t*)dest)[3]= r[Y2];\
865 ((uint8_t*)dest)[4]= g[Y2];\
866 ((uint8_t*)dest)[5]= b[Y2];\
872 ((uint8_t*)dest)[0]= b[Y1];\
873 ((uint8_t*)dest)[1]= g[Y1];\
874 ((uint8_t*)dest)[2]= r[Y1];\
875 ((uint8_t*)dest)[3]= b[Y2];\
876 ((uint8_t*)dest)[4]= g[Y2];\
877 ((uint8_t*)dest)[5]= r[Y2];\
881 case PIX_FMT_RGB565BE:\
882 case PIX_FMT_RGB565LE:\
883 case PIX_FMT_BGR565BE:\
884 case PIX_FMT_BGR565LE:\
886 const int dr1= dither_2x2_8[y&1 ][0];\
887 const int dg1= dither_2x2_4[y&1 ][0];\
888 const int db1= dither_2x2_8[(y&1)^1][0];\
889 const int dr2= dither_2x2_8[y&1 ][1];\
890 const int dg2= dither_2x2_4[y&1 ][1];\
891 const int db2= dither_2x2_8[(y&1)^1][1];\
893 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
894 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
898 case PIX_FMT_RGB555BE:\
899 case PIX_FMT_RGB555LE:\
900 case PIX_FMT_BGR555BE:\
901 case PIX_FMT_BGR555LE:\
903 const int dr1= dither_2x2_8[y&1 ][0];\
904 const int dg1= dither_2x2_8[y&1 ][1];\
905 const int db1= dither_2x2_8[(y&1)^1][0];\
906 const int dr2= dither_2x2_8[y&1 ][1];\
907 const int dg2= dither_2x2_8[y&1 ][0];\
908 const int db2= dither_2x2_8[(y&1)^1][1];\
910 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
911 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
915 case PIX_FMT_RGB444BE:\
916 case PIX_FMT_RGB444LE:\
917 case PIX_FMT_BGR444BE:\
918 case PIX_FMT_BGR444LE:\
920 const int dr1= dither_4x4_16[y&3 ][0];\
921 const int dg1= dither_4x4_16[y&3 ][1];\
922 const int db1= dither_4x4_16[(y&3)^3][0];\
923 const int dr2= dither_4x4_16[y&3 ][1];\
924 const int dg2= dither_4x4_16[y&3 ][0];\
925 const int db2= dither_4x4_16[(y&3)^3][1];\
927 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
928 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
935 const uint8_t * const d64= dither_8x8_73[y&7];\
936 const uint8_t * const d32= dither_8x8_32[y&7];\
938 ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]];\
939 ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]];\
946 const uint8_t * const d64= dither_8x8_73 [y&7];\
947 const uint8_t * const d128=dither_8x8_220[y&7];\
949 ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]\
950 + ((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4);\
954 case PIX_FMT_RGB4_BYTE:\
955 case PIX_FMT_BGR4_BYTE:\
957 const uint8_t * const d64= dither_8x8_73 [y&7];\
958 const uint8_t * const d128=dither_8x8_220[y&7];\
960 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\
961 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\
965 case PIX_FMT_MONOBLACK:\
966 case PIX_FMT_MONOWHITE:\
971 case PIX_FMT_YUYV422:\
973 ((uint8_t*)dest)[2*i2+0]= Y1;\
974 ((uint8_t*)dest)[2*i2+1]= U;\
975 ((uint8_t*)dest)[2*i2+2]= Y2;\
976 ((uint8_t*)dest)[2*i2+3]= V;\
979 case PIX_FMT_UYVY422:\
981 ((uint8_t*)dest)[2*i2+0]= U;\
982 ((uint8_t*)dest)[2*i2+1]= Y1;\
983 ((uint8_t*)dest)[2*i2+2]= V;\
984 ((uint8_t*)dest)[2*i2+3]= Y2;\
987 case PIX_FMT_GRAY16BE:\
989 ((uint8_t*)dest)[2*i2+0]= Y1>>8;\
990 ((uint8_t*)dest)[2*i2+1]= Y1;\
991 ((uint8_t*)dest)[2*i2+2]= Y2>>8;\
992 ((uint8_t*)dest)[2*i2+3]= Y2;\
995 case PIX_FMT_GRAY16LE:\
997 ((uint8_t*)dest)[2*i2+0]= Y1;\
998 ((uint8_t*)dest)[2*i2+1]= Y1>>8;\
999 ((uint8_t*)dest)[2*i2+2]= Y2;\
1000 ((uint8_t*)dest)[2*i2+3]= Y2>>8;\
1005 static inline void yuv2packedXinC(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
1006 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
1007 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
1010 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)
1013 static inline void yuv2rgbXinC_full(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
1014 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
1015 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
1018 int step= c->dstFormatBpp/8;
1021 switch(c->dstFormat) {
1029 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;
1030 YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)
1031 dest[aidx]= needAlpha ? A : 255;
1038 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1039 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)
1047 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)
1064 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;
1065 YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)
1066 dest[aidx]= needAlpha ? A : 255;
1073 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1074 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)
1082 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)
1097 static void fillPlane(uint8_t* plane, int stride, int width, int height, int y, uint8_t val)
1100 uint8_t *ptr = plane + stride*y;
1101 for (i=0; i<height; i++) {
1102 memset(ptr, val, width);
1107 static inline void rgb48ToY(uint8_t *dst, const uint8_t *src, long width,
1111 for (i = 0; i < width; i++) {
1116 dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1120 static inline void rgb48ToUV(uint8_t *dstU, uint8_t *dstV,
1121 const uint8_t *src1, const uint8_t *src2,
1122 long width, uint32_t *unused)
1126 for (i = 0; i < width; i++) {
1127 int r = src1[6*i + 0];
1128 int g = src1[6*i + 2];
1129 int b = src1[6*i + 4];
1131 dstU[i] = (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1132 dstV[i] = (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1136 static inline void rgb48ToUV_half(uint8_t *dstU, uint8_t *dstV,
1137 const uint8_t *src1, const uint8_t *src2,
1138 long width, uint32_t *unused)
1142 for (i = 0; i < width; i++) {
1143 int r= src1[12*i + 0] + src1[12*i + 6];
1144 int g= src1[12*i + 2] + src1[12*i + 8];
1145 int b= src1[12*i + 4] + src1[12*i + 10];
1147 dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1148 dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1152 static inline void bgr48ToY(uint8_t *dst, const uint8_t *src, long width,
1156 for (i = 0; i < width; i++) {
1161 dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
1165 static inline void bgr48ToUV(uint8_t *dstU, uint8_t *dstV,
1166 const uint8_t *src1, const uint8_t *src2,
1167 long width, uint32_t *unused)
1170 for (i = 0; i < width; i++) {
1171 int b = src1[6*i + 0];
1172 int g = src1[6*i + 2];
1173 int r = src1[6*i + 4];
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;
1180 static inline void bgr48ToUV_half(uint8_t *dstU, uint8_t *dstV,
1181 const uint8_t *src1, const uint8_t *src2,
1182 long width, uint32_t *unused)
1185 for (i = 0; i < width; i++) {
1186 int b= src1[12*i + 0] + src1[12*i + 6];
1187 int g= src1[12*i + 2] + src1[12*i + 8];
1188 int r= src1[12*i + 4] + src1[12*i + 10];
1190 dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1191 dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1195 #define BGR2Y(type, name, shr, shg, shb, maskr, maskg, maskb, RY, GY, BY, S)\
1196 static inline void name(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)\
1199 for (i=0; i<width; i++) {\
1200 int b= (((const type*)src)[i]>>shb)&maskb;\
1201 int g= (((const type*)src)[i]>>shg)&maskg;\
1202 int r= (((const type*)src)[i]>>shr)&maskr;\
1204 dst[i]= (((RY)*r + (GY)*g + (BY)*b + (33<<((S)-1)))>>(S));\
1208 BGR2Y(uint32_t, bgr32ToY,16, 0, 0, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY , BY<< 8, RGB2YUV_SHIFT+8)
1209 BGR2Y(uint32_t,bgr321ToY,16,16, 0, 0xFF00, 0x00FF, 0xFF00, RY , GY<<8, BY , RGB2YUV_SHIFT+8)
1210 BGR2Y(uint32_t, rgb32ToY, 0, 0,16, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY , BY<< 8, RGB2YUV_SHIFT+8)
1211 BGR2Y(uint32_t,rgb321ToY, 0,16,16, 0xFF00, 0x00FF, 0xFF00, RY , GY<<8, BY , RGB2YUV_SHIFT+8)
1212 BGR2Y(uint16_t, bgr16ToY, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RY<<11, GY<<5, BY , RGB2YUV_SHIFT+8)
1213 BGR2Y(uint16_t, bgr15ToY, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RY<<10, GY<<5, BY , RGB2YUV_SHIFT+7)
1214 BGR2Y(uint16_t, rgb16ToY, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RY , GY<<5, BY<<11, RGB2YUV_SHIFT+8)
1215 BGR2Y(uint16_t, rgb15ToY, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RY , GY<<5, BY<<10, RGB2YUV_SHIFT+7)
1217 static inline void abgrToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1220 for (i=0; i<width; i++) {
1225 #define BGR2UV(type, name, shr, shg, shb, shp, maskr, maskg, maskb, RU, GU, BU, RV, GV, BV, S) \
1226 static inline void name(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\
1229 for (i=0; i<width; i++) {\
1230 int b= ((((const type*)src)[i]>>shp)&maskb)>>shb;\
1231 int g= ((((const type*)src)[i]>>shp)&maskg)>>shg;\
1232 int r= ((((const type*)src)[i]>>shp)&maskr)>>shr;\
1234 dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<((S)-1)))>>(S);\
1235 dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<((S)-1)))>>(S);\
1238 static inline void name ## _half(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\
1241 for (i=0; i<width; i++) {\
1242 int pix0= ((const type*)src)[2*i+0]>>shp;\
1243 int pix1= ((const type*)src)[2*i+1]>>shp;\
1244 int g= (pix0&~(maskr|maskb))+(pix1&~(maskr|maskb));\
1245 int b= ((pix0+pix1-g)&(maskb|(2*maskb)))>>shb;\
1246 int r= ((pix0+pix1-g)&(maskr|(2*maskr)))>>shr;\
1247 g&= maskg|(2*maskg);\
1251 dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<(S)))>>((S)+1);\
1252 dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<(S)))>>((S)+1);\
1256 BGR2UV(uint32_t, bgr32ToUV,16, 0, 0, 0, 0xFF0000, 0xFF00, 0x00FF, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1257 BGR2UV(uint32_t,bgr321ToUV,16, 0, 0, 8, 0xFF0000, 0xFF00, 0x00FF, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1258 BGR2UV(uint32_t, rgb32ToUV, 0, 0,16, 0, 0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1259 BGR2UV(uint32_t,rgb321ToUV, 0, 0,16, 8, 0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1260 BGR2UV(uint16_t, bgr16ToUV, 0, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RU<<11, GU<<5, BU , RV<<11, GV<<5, BV , RGB2YUV_SHIFT+8)
1261 BGR2UV(uint16_t, bgr15ToUV, 0, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RU<<10, GU<<5, BU , RV<<10, GV<<5, BV , RGB2YUV_SHIFT+7)
1262 BGR2UV(uint16_t, rgb16ToUV, 0, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RU , GU<<5, BU<<11, RV , GV<<5, BV<<11, RGB2YUV_SHIFT+8)
1263 BGR2UV(uint16_t, rgb15ToUV, 0, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RU , GU<<5, BU<<10, RV , GV<<5, BV<<10, RGB2YUV_SHIFT+7)
1265 static inline void palToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)
1268 for (i=0; i<width; i++) {
1271 dst[i]= pal[d] >> 24;
1275 static inline void palToY(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)
1278 for (i=0; i<width; i++) {
1281 dst[i]= pal[d] & 0xFF;
1285 static inline void palToUV(uint8_t *dstU, uint8_t *dstV,
1286 const uint8_t *src1, const uint8_t *src2,
1287 long width, uint32_t *pal)
1290 assert(src1 == src2);
1291 for (i=0; i<width; i++) {
1292 int p= pal[src1[i]];
1299 static inline void monowhite2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1302 for (i=0; i<width/8; i++) {
1305 dst[8*i+j]= ((d>>(7-j))&1)*255;
1309 static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1312 for (i=0; i<width/8; i++) {
1315 dst[8*i+j]= ((d>>(7-j))&1)*255;
1319 //Note: we have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW+MMX2 one
1321 #if CONFIG_RUNTIME_CPUDETECT
1322 # define COMPILE_C 1
1324 # define COMPILE_MMX 1
1325 # define COMPILE_MMX2 1
1326 # define COMPILE_3DNOW 1
1328 # define COMPILE_ALTIVEC HAVE_ALTIVEC
1330 #else /* CONFIG_RUNTIME_CPUDETECT */
1333 # define COMPILE_MMX2 1
1334 # elif HAVE_AMD3DNOW
1335 # define COMPILE_3DNOW 1
1337 # define COMPILE_MMX 1
1339 # define COMPILE_C 1
1341 # elif ARCH_PPC && HAVE_ALTIVEC
1342 # define COMPILE_ALTIVEC 1
1344 # define COMPILE_C 1
1349 # define COMPILE_C 0
1352 # define COMPILE_MMX 0
1354 #ifndef COMPILE_MMX2
1355 # define COMPILE_MMX2 0
1357 #ifndef COMPILE_3DNOW
1358 # define COMPILE_3DNOW 0
1360 #ifndef COMPILE_ALTIVEC
1361 # define COMPILE_ALTIVEC 0
1364 #define COMPILE_TEMPLATE_MMX 0
1365 #define COMPILE_TEMPLATE_MMX2 0
1366 #define COMPILE_TEMPLATE_AMD3DNOW 0
1367 #define COMPILE_TEMPLATE_ALTIVEC 0
1370 #define RENAME(a) a ## _C
1371 #include "swscale_template.c"
1376 #undef COMPILE_TEMPLATE_ALTIVEC
1377 #define COMPILE_TEMPLATE_ALTIVEC 1
1378 #define RENAME(a) a ## _altivec
1379 #include "swscale_template.c"
1387 #undef COMPILE_TEMPLATE_MMX
1388 #undef COMPILE_TEMPLATE_MMX2
1389 #undef COMPILE_TEMPLATE_AMD3DNOW
1390 #define COMPILE_TEMPLATE_MMX 1
1391 #define COMPILE_TEMPLATE_MMX2 0
1392 #define COMPILE_TEMPLATE_AMD3DNOW 0
1393 #define RENAME(a) a ## _MMX
1394 #include "swscale_template.c"
1400 #undef COMPILE_TEMPLATE_MMX
1401 #undef COMPILE_TEMPLATE_MMX2
1402 #undef COMPILE_TEMPLATE_AMD3DNOW
1403 #define COMPILE_TEMPLATE_MMX 1
1404 #define COMPILE_TEMPLATE_MMX2 1
1405 #define COMPILE_TEMPLATE_AMD3DNOW 0
1406 #define RENAME(a) a ## _MMX2
1407 #include "swscale_template.c"
1413 #undef COMPILE_TEMPLATE_MMX
1414 #undef COMPILE_TEMPLATE_MMX2
1415 #undef COMPILE_TEMPLATE_AMD3DNOW
1416 #define COMPILE_TEMPLATE_MMX 1
1417 #define COMPILE_TEMPLATE_MMX2 0
1418 #define COMPILE_TEMPLATE_AMD3DNOW 1
1419 #define RENAME(a) a ## _3DNow
1420 #include "swscale_template.c"
1425 SwsFunc ff_getSwsFunc(SwsContext *c)
1427 #if CONFIG_RUNTIME_CPUDETECT
1428 int flags = c->flags;
1431 // ordered per speed fastest first
1432 if (flags & SWS_CPU_CAPS_MMX2) {
1433 sws_init_swScale_MMX2(c);
1434 return swScale_MMX2;
1435 } else if (flags & SWS_CPU_CAPS_3DNOW) {
1436 sws_init_swScale_3DNow(c);
1437 return swScale_3DNow;
1438 } else if (flags & SWS_CPU_CAPS_MMX) {
1439 sws_init_swScale_MMX(c);
1442 sws_init_swScale_C(c);
1448 if (flags & SWS_CPU_CAPS_ALTIVEC) {
1449 sws_init_swScale_altivec(c);
1450 return swScale_altivec;
1452 sws_init_swScale_C(c);
1456 sws_init_swScale_C(c);
1458 #endif /* ARCH_X86 */
1459 #else //CONFIG_RUNTIME_CPUDETECT
1460 #if COMPILE_TEMPLATE_MMX2
1461 sws_init_swScale_MMX2(c);
1462 return swScale_MMX2;
1463 #elif COMPILE_TEMPLATE_AMD3DNOW
1464 sws_init_swScale_3DNow(c);
1465 return swScale_3DNow;
1466 #elif COMPILE_TEMPLATE_MMX
1467 sws_init_swScale_MMX(c);
1469 #elif COMPILE_TEMPLATE_ALTIVEC
1470 sws_init_swScale_altivec(c);
1471 return swScale_altivec;
1473 sws_init_swScale_C(c);
1476 #endif //!CONFIG_RUNTIME_CPUDETECT
1479 static void copyPlane(const uint8_t *src, int srcStride,
1480 int srcSliceY, int srcSliceH, int width,
1481 uint8_t *dst, int dstStride)
1483 dst += dstStride * srcSliceY;
1484 if (dstStride == srcStride && srcStride > 0) {
1485 memcpy(dst, src, srcSliceH * dstStride);
1488 for (i=0; i<srcSliceH; i++) {
1489 memcpy(dst, src, width);
1496 static int planarToNv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1497 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1499 uint8_t *dst = dstParam[1] + dstStride[1]*srcSliceY/2;
1501 copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
1502 dstParam[0], dstStride[0]);
1504 if (c->dstFormat == PIX_FMT_NV12)
1505 interleaveBytes(src[1], src[2], dst, c->srcW/2, srcSliceH/2, srcStride[1], srcStride[2], dstStride[0]);
1507 interleaveBytes(src[2], src[1], dst, c->srcW/2, srcSliceH/2, srcStride[2], srcStride[1], dstStride[0]);
1512 static int planarToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1513 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1515 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1517 yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
1522 static int planarToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1523 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1525 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1527 yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
1532 static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1533 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1535 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1537 yuv422ptoyuy2(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
1542 static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1543 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1545 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1547 yuv422ptouyvy(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
1552 static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1553 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1555 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1556 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
1557 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
1559 yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1562 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1567 static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1568 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1570 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1571 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
1572 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
1574 yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1579 static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1580 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1582 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1583 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
1584 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
1586 uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1589 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1594 static int uyvyToYuv422Wrapper(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;
1599 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
1601 uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1606 static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1609 for (i=0; i<num_pixels; i++)
1610 ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | (src[(i<<1)+1] << 24);
1613 static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1617 for (i=0; i<num_pixels; i++)
1618 ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | src[(i<<1)+1];
1621 static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
1625 for (i=0; i<num_pixels; i++) {
1627 dst[0]= palette[src[i<<1]*4+0];
1628 dst[1]= palette[src[i<<1]*4+1];
1629 dst[2]= palette[src[i<<1]*4+2];
1634 static int palToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1635 int srcSliceH, uint8_t* dst[], int dstStride[])
1637 const enum PixelFormat srcFormat= c->srcFormat;
1638 const enum PixelFormat dstFormat= c->dstFormat;
1639 void (*conv)(const uint8_t *src, uint8_t *dst, long num_pixels,
1640 const uint8_t *palette)=NULL;
1642 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
1643 const uint8_t *srcPtr= src[0];
1645 if (srcFormat == PIX_FMT_GRAY8A) {
1646 switch (dstFormat) {
1647 case PIX_FMT_RGB32 : conv = gray8aToPacked32; break;
1648 case PIX_FMT_BGR32 : conv = gray8aToPacked32; break;
1649 case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;
1650 case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;
1651 case PIX_FMT_RGB24 : conv = gray8aToPacked24; break;
1652 case PIX_FMT_BGR24 : conv = gray8aToPacked24; break;
1654 } else if (usePal(srcFormat)) {
1655 switch (dstFormat) {
1656 case PIX_FMT_RGB32 : conv = sws_convertPalette8ToPacked32; break;
1657 case PIX_FMT_BGR32 : conv = sws_convertPalette8ToPacked32; break;
1658 case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;
1659 case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;
1660 case PIX_FMT_RGB24 : conv = sws_convertPalette8ToPacked24; break;
1661 case PIX_FMT_BGR24 : conv = sws_convertPalette8ToPacked24; break;
1666 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
1667 sws_format_name(srcFormat), sws_format_name(dstFormat));
1669 for (i=0; i<srcSliceH; i++) {
1670 conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);
1671 srcPtr+= srcStride[0];
1672 dstPtr+= dstStride[0];
1679 #define isRGBA32(x) ( \
1680 (x) == PIX_FMT_ARGB \
1681 || (x) == PIX_FMT_RGBA \
1682 || (x) == PIX_FMT_BGRA \
1683 || (x) == PIX_FMT_ABGR \
1686 /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */
1687 static int rgbToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1688 int srcSliceH, uint8_t* dst[], int dstStride[])
1690 const enum PixelFormat srcFormat= c->srcFormat;
1691 const enum PixelFormat dstFormat= c->dstFormat;
1692 const int srcBpp= (c->srcFormatBpp + 7) >> 3;
1693 const int dstBpp= (c->dstFormatBpp + 7) >> 3;
1694 const int srcId= c->srcFormatBpp >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */
1695 const int dstId= c->dstFormatBpp >> 2;
1696 void (*conv)(const uint8_t *src, uint8_t *dst, long src_size)=NULL;
1698 #define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)
1700 if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {
1701 if ( CONV_IS(ABGR, RGBA)
1702 || CONV_IS(ARGB, BGRA)
1703 || CONV_IS(BGRA, ARGB)
1704 || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;
1705 else if (CONV_IS(ABGR, ARGB)
1706 || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;
1707 else if (CONV_IS(ABGR, BGRA)
1708 || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;
1709 else if (CONV_IS(BGRA, RGBA)
1710 || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;
1711 else if (CONV_IS(BGRA, ABGR)
1712 || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;
1715 if ( (isBGRinInt(srcFormat) && isBGRinInt(dstFormat))
1716 || (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {
1717 switch(srcId | (dstId<<4)) {
1718 case 0x34: conv= rgb16to15; break;
1719 case 0x36: conv= rgb24to15; break;
1720 case 0x38: conv= rgb32to15; break;
1721 case 0x43: conv= rgb15to16; break;
1722 case 0x46: conv= rgb24to16; break;
1723 case 0x48: conv= rgb32to16; break;
1724 case 0x63: conv= rgb15to24; break;
1725 case 0x64: conv= rgb16to24; break;
1726 case 0x68: conv= rgb32to24; break;
1727 case 0x83: conv= rgb15to32; break;
1728 case 0x84: conv= rgb16to32; break;
1729 case 0x86: conv= rgb24to32; break;
1731 } else if ( (isBGRinInt(srcFormat) && isRGBinInt(dstFormat))
1732 || (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {
1733 switch(srcId | (dstId<<4)) {
1734 case 0x33: conv= rgb15tobgr15; break;
1735 case 0x34: conv= rgb16tobgr15; break;
1736 case 0x36: conv= rgb24tobgr15; break;
1737 case 0x38: conv= rgb32tobgr15; break;
1738 case 0x43: conv= rgb15tobgr16; break;
1739 case 0x44: conv= rgb16tobgr16; break;
1740 case 0x46: conv= rgb24tobgr16; break;
1741 case 0x48: conv= rgb32tobgr16; break;
1742 case 0x63: conv= rgb15tobgr24; break;
1743 case 0x64: conv= rgb16tobgr24; break;
1744 case 0x66: conv= rgb24tobgr24; break;
1745 case 0x68: conv= rgb32tobgr24; break;
1746 case 0x83: conv= rgb15tobgr32; break;
1747 case 0x84: conv= rgb16tobgr32; break;
1748 case 0x86: conv= rgb24tobgr32; break;
1753 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
1754 sws_format_name(srcFormat), sws_format_name(dstFormat));
1756 const uint8_t *srcPtr= src[0];
1757 uint8_t *dstPtr= dst[0];
1758 if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) && !isRGBA32(dstFormat))
1759 srcPtr += ALT32_CORR;
1761 if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) && !isRGBA32(srcFormat))
1762 dstPtr += ALT32_CORR;
1764 if (dstStride[0]*srcBpp == srcStride[0]*dstBpp && srcStride[0] > 0 && !(srcStride[0]%srcBpp))
1765 conv(srcPtr, dstPtr + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]);
1768 dstPtr += dstStride[0]*srcSliceY;
1770 for (i=0; i<srcSliceH; i++) {
1771 conv(srcPtr, dstPtr, c->srcW*srcBpp);
1772 srcPtr+= srcStride[0];
1773 dstPtr+= dstStride[0];
1780 static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1781 int srcSliceH, uint8_t* dst[], int dstStride[])
1785 dst[0]+ srcSliceY *dstStride[0],
1786 dst[1]+(srcSliceY>>1)*dstStride[1],
1787 dst[2]+(srcSliceY>>1)*dstStride[2],
1789 dstStride[0], dstStride[1], srcStride[0]);
1791 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1795 static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1796 int srcSliceH, uint8_t* dst[], int dstStride[])
1798 copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
1799 dst[0], dstStride[0]);
1801 planar2x(src[1], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,
1802 srcSliceH >> 2, srcStride[1], dstStride[1]);
1803 planar2x(src[2], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,
1804 srcSliceH >> 2, srcStride[2], dstStride[2]);
1806 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1810 /* unscaled copy like stuff (assumes nearly identical formats) */
1811 static int packedCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1812 int srcSliceH, uint8_t* dst[], int dstStride[])
1814 if (dstStride[0]==srcStride[0] && srcStride[0] > 0)
1815 memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]);
1818 const uint8_t *srcPtr= src[0];
1819 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
1822 /* universal length finder */
1823 while(length+c->srcW <= FFABS(dstStride[0])
1824 && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW;
1827 for (i=0; i<srcSliceH; i++) {
1828 memcpy(dstPtr, srcPtr, length);
1829 srcPtr+= srcStride[0];
1830 dstPtr+= dstStride[0];
1836 #define DITHER_COPY(dst, dstStride, src, srcStride, bswap, dbswap)\
1837 uint16_t scale= dither_scale[dst_depth-1][src_depth-1];\
1838 int shift= src_depth-dst_depth + dither_scale[src_depth-2][dst_depth-1];\
1839 for (i = 0; i < height; i++) {\
1840 uint8_t *dither= dithers[src_depth-9][i&7];\
1841 for (j = 0; j < length-7; j+=8){\
1842 dst[j+0] = dbswap((bswap(src[j+0]) + dither[0])*scale>>shift);\
1843 dst[j+1] = dbswap((bswap(src[j+1]) + dither[1])*scale>>shift);\
1844 dst[j+2] = dbswap((bswap(src[j+2]) + dither[2])*scale>>shift);\
1845 dst[j+3] = dbswap((bswap(src[j+3]) + dither[3])*scale>>shift);\
1846 dst[j+4] = dbswap((bswap(src[j+4]) + dither[4])*scale>>shift);\
1847 dst[j+5] = dbswap((bswap(src[j+5]) + dither[5])*scale>>shift);\
1848 dst[j+6] = dbswap((bswap(src[j+6]) + dither[6])*scale>>shift);\
1849 dst[j+7] = dbswap((bswap(src[j+7]) + dither[7])*scale>>shift);\
1851 for (; j < length; j++)\
1852 dst[j] = dbswap((bswap(src[j]) + dither[j&7])*scale>>shift);\
1858 static int planarCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
1859 int srcSliceH, uint8_t* dst[], int dstStride[])
1862 for (plane=0; plane<4; plane++) {
1863 int length= (plane==0 || plane==3) ? c->srcW : -((-c->srcW )>>c->chrDstHSubSample);
1864 int y= (plane==0 || plane==3) ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample);
1865 int height= (plane==0 || plane==3) ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample);
1866 const uint8_t *srcPtr= src[plane];
1867 uint8_t *dstPtr= dst[plane] + dstStride[plane]*y;
1869 if (!dst[plane]) continue;
1870 // ignore palette for GRAY8
1871 if (plane == 1 && !dst[2]) continue;
1872 if (!src[plane] || (plane == 1 && !src[2])) {
1873 if(is16BPS(c->dstFormat))
1875 fillPlane(dst[plane], dstStride[plane], length, height, y, (plane==3) ? 255 : 128);
1877 if(isNBPS(c->srcFormat) || isNBPS(c->dstFormat)
1878 || (is16BPS(c->srcFormat) != is16BPS(c->dstFormat))
1880 const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1+1;
1881 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1+1;
1882 const uint16_t *srcPtr2 = (const uint16_t*)srcPtr;
1883 uint16_t *dstPtr2 = (uint16_t*)dstPtr;
1885 if (dst_depth == 8) {
1886 if(isBE(c->srcFormat) == HAVE_BIGENDIAN){
1887 DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, , )
1889 DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, av_bswap16, )
1891 } else if (src_depth == 8) {
1892 for (i = 0; i < height; i++) {
1893 if(isBE(c->dstFormat)){
1894 for (j = 0; j < length; j++)
1895 AV_WB16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |
1896 (srcPtr[j]>>(2*8-dst_depth)));
1898 for (j = 0; j < length; j++)
1899 AV_WL16(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |
1900 (srcPtr[j]>>(2*8-dst_depth)));
1902 dstPtr2 += dstStride[plane]/2;
1903 srcPtr += srcStride[plane];
1905 } else if (src_depth <= dst_depth) {
1906 for (i = 0; i < height; i++) {
1907 #define COPY_UP(r,w) \
1908 for (j = 0; j < length; j++){ \
1909 unsigned int v= r(&srcPtr2[j]);\
1910 w(&dstPtr2[j], (v<<(dst_depth-src_depth)) | \
1911 (v>>(2*src_depth-dst_depth)));\
1913 if(isBE(c->srcFormat)){
1914 if(isBE(c->dstFormat)){
1915 COPY_UP(AV_RB16, AV_WB16)
1917 COPY_UP(AV_RB16, AV_WL16)
1920 if(isBE(c->dstFormat)){
1921 COPY_UP(AV_RL16, AV_WB16)
1923 COPY_UP(AV_RL16, AV_WL16)
1926 dstPtr2 += dstStride[plane]/2;
1927 srcPtr2 += srcStride[plane]/2;
1930 if(isBE(c->srcFormat) == HAVE_BIGENDIAN){
1931 if(isBE(c->dstFormat) == HAVE_BIGENDIAN){
1932 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , )
1934 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , av_bswap16)
1937 if(isBE(c->dstFormat) == HAVE_BIGENDIAN){
1938 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, )
1940 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, av_bswap16)
1944 } else if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat)
1945 && isBE(c->srcFormat) != isBE(c->dstFormat)) {
1947 for (i=0; i<height; i++) {
1948 for (j=0; j<length; j++)
1949 ((uint16_t*)dstPtr)[j] = av_bswap16(((const uint16_t*)srcPtr)[j]);
1950 srcPtr+= srcStride[plane];
1951 dstPtr+= dstStride[plane];
1953 } else if (dstStride[plane] == srcStride[plane] &&
1954 srcStride[plane] > 0 && srcStride[plane] == length) {
1955 memcpy(dst[plane] + dstStride[plane]*y, src[plane],
1956 height*dstStride[plane]);
1958 if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat))
1960 for (i=0; i<height; i++) {
1961 memcpy(dstPtr, srcPtr, length);
1962 srcPtr+= srcStride[plane];
1963 dstPtr+= dstStride[plane];
1971 int ff_hardcodedcpuflags(void)
1974 #if COMPILE_TEMPLATE_MMX2
1975 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2;
1976 #elif COMPILE_TEMPLATE_AMD3DNOW
1977 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_3DNOW;
1978 #elif COMPILE_TEMPLATE_MMX
1979 flags |= SWS_CPU_CAPS_MMX;
1980 #elif COMPILE_TEMPLATE_ALTIVEC
1981 flags |= SWS_CPU_CAPS_ALTIVEC;
1983 flags |= SWS_CPU_CAPS_BFIN;
1988 void ff_get_unscaled_swscale(SwsContext *c)
1990 const enum PixelFormat srcFormat = c->srcFormat;
1991 const enum PixelFormat dstFormat = c->dstFormat;
1992 const int flags = c->flags;
1993 const int dstH = c->dstH;
1996 needsDither= isAnyRGB(dstFormat)
1997 && c->dstFormatBpp < 24
1998 && (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));
2001 if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {
2002 c->swScale= planarToNv12Wrapper;
2005 if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P || srcFormat==PIX_FMT_YUVA420P) && isAnyRGB(dstFormat)
2006 && !(flags & SWS_ACCURATE_RND) && !(dstH&1)) {
2007 c->swScale= ff_yuv2rgb_get_func_ptr(c);
2010 if (srcFormat==PIX_FMT_YUV410P && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_BITEXACT)) {
2011 c->swScale= yvu9ToYv12Wrapper;
2015 if (srcFormat==PIX_FMT_BGR24 && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_ACCURATE_RND))
2016 c->swScale= bgr24ToYv12Wrapper;
2018 /* RGB/BGR -> RGB/BGR (no dither needed forms) */
2019 if ( isAnyRGB(srcFormat)
2020 && isAnyRGB(dstFormat)
2021 && srcFormat != PIX_FMT_BGR8 && dstFormat != PIX_FMT_BGR8
2022 && srcFormat != PIX_FMT_RGB8 && dstFormat != PIX_FMT_RGB8
2023 && srcFormat != PIX_FMT_BGR4 && dstFormat != PIX_FMT_BGR4
2024 && srcFormat != PIX_FMT_RGB4 && dstFormat != PIX_FMT_RGB4
2025 && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE
2026 && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE
2027 && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK
2028 && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE
2029 && srcFormat != PIX_FMT_RGB48LE && dstFormat != PIX_FMT_RGB48LE
2030 && srcFormat != PIX_FMT_RGB48BE && dstFormat != PIX_FMT_RGB48BE
2031 && srcFormat != PIX_FMT_BGR48LE && dstFormat != PIX_FMT_BGR48LE
2032 && srcFormat != PIX_FMT_BGR48BE && dstFormat != PIX_FMT_BGR48BE
2033 && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))
2034 c->swScale= rgbToRgbWrapper;
2036 if ((usePal(srcFormat) && (
2037 dstFormat == PIX_FMT_RGB32 ||
2038 dstFormat == PIX_FMT_RGB32_1 ||
2039 dstFormat == PIX_FMT_RGB24 ||
2040 dstFormat == PIX_FMT_BGR32 ||
2041 dstFormat == PIX_FMT_BGR32_1 ||
2042 dstFormat == PIX_FMT_BGR24)))
2043 c->swScale= palToRgbWrapper;
2045 if (srcFormat == PIX_FMT_YUV422P) {
2046 if (dstFormat == PIX_FMT_YUYV422)
2047 c->swScale= yuv422pToYuy2Wrapper;
2048 else if (dstFormat == PIX_FMT_UYVY422)
2049 c->swScale= yuv422pToUyvyWrapper;
2052 /* LQ converters if -sws 0 or -sws 4*/
2053 if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {
2055 if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {
2056 if (dstFormat == PIX_FMT_YUYV422)
2057 c->swScale= planarToYuy2Wrapper;
2058 else if (dstFormat == PIX_FMT_UYVY422)
2059 c->swScale= planarToUyvyWrapper;
2062 if(srcFormat == PIX_FMT_YUYV422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
2063 c->swScale= yuyvToYuv420Wrapper;
2064 if(srcFormat == PIX_FMT_UYVY422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
2065 c->swScale= uyvyToYuv420Wrapper;
2066 if(srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)
2067 c->swScale= yuyvToYuv422Wrapper;
2068 if(srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)
2069 c->swScale= uyvyToYuv422Wrapper;
2072 if ((c->flags & SWS_CPU_CAPS_ALTIVEC) &&
2073 !(c->flags & SWS_BITEXACT) &&
2074 srcFormat == PIX_FMT_YUV420P) {
2075 // unscaled YV12 -> packed YUV, we want speed
2076 if (dstFormat == PIX_FMT_YUYV422)
2077 c->swScale= yv12toyuy2_unscaled_altivec;
2078 else if (dstFormat == PIX_FMT_UYVY422)
2079 c->swScale= yv12touyvy_unscaled_altivec;
2084 if ( srcFormat == dstFormat
2085 || (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P)
2086 || (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P)
2087 || (isPlanarYUV(srcFormat) && isGray(dstFormat))
2088 || (isPlanarYUV(dstFormat) && isGray(srcFormat))
2089 || (isGray(dstFormat) && isGray(srcFormat))
2090 || (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat)
2091 && c->chrDstHSubSample == c->chrSrcHSubSample
2092 && c->chrDstVSubSample == c->chrSrcVSubSample
2093 && dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21
2094 && srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))
2096 if (isPacked(c->srcFormat))
2097 c->swScale= packedCopyWrapper;
2098 else /* Planar YUV or gray */
2099 c->swScale= planarCopyWrapper;
2102 if (flags & SWS_CPU_CAPS_BFIN)
2103 ff_bfin_get_unscaled_swscale (c);
2107 static void reset_ptr(const uint8_t* src[], int format)
2109 if(!isALPHA(format))
2111 if(!isPlanarYUV(format)) {
2114 if (!usePal(format))
2119 static int check_image_pointers(uint8_t *data[4], enum PixelFormat pix_fmt,
2120 const int linesizes[4])
2122 const AVPixFmtDescriptor *desc = &av_pix_fmt_descriptors[pix_fmt];
2125 for (i = 0; i < 4; i++) {
2126 int plane = desc->comp[i].plane;
2127 if (!data[plane] || !linesizes[plane])
2135 * swscale wrapper, so we don't need to export the SwsContext.
2136 * Assumes planar YUV to be in YUV order instead of YVU.
2138 int sws_scale(SwsContext *c, const uint8_t* const src[], const int srcStride[], int srcSliceY,
2139 int srcSliceH, uint8_t* const dst[], const int dstStride[])
2142 const uint8_t* src2[4]= {src[0], src[1], src[2], src[3]};
2143 uint8_t* dst2[4]= {dst[0], dst[1], dst[2], dst[3]};
2145 // do not mess up sliceDir if we have a "trailing" 0-size slice
2149 if (!check_image_pointers(src, c->srcFormat, srcStride)) {
2150 av_log(c, AV_LOG_ERROR, "bad src image pointers\n");
2153 if (!check_image_pointers(dst, c->dstFormat, dstStride)) {
2154 av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");
2158 if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {
2159 av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");
2162 if (c->sliceDir == 0) {
2163 if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;
2166 if (usePal(c->srcFormat)) {
2167 for (i=0; i<256; i++) {
2168 int p, r, g, b, y, u, v, a = 0xff;
2169 if(c->srcFormat == PIX_FMT_PAL8) {
2170 p=((const uint32_t*)(src[1]))[i];
2175 } else if(c->srcFormat == PIX_FMT_RGB8) {
2179 } else if(c->srcFormat == PIX_FMT_BGR8) {
2183 } else if(c->srcFormat == PIX_FMT_RGB4_BYTE) {
2187 } else if(c->srcFormat == PIX_FMT_GRAY8 || c->srcFormat == PIX_FMT_GRAY8A) {
2190 assert(c->srcFormat == PIX_FMT_BGR4_BYTE);
2195 y= av_clip_uint8((RY*r + GY*g + BY*b + ( 33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2196 u= av_clip_uint8((RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2197 v= av_clip_uint8((RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2198 c->pal_yuv[i]= y + (u<<8) + (v<<16) + (a<<24);
2200 switch(c->dstFormat) {
2205 c->pal_rgb[i]= r + (g<<8) + (b<<16) + (a<<24);
2207 case PIX_FMT_BGR32_1:
2211 c->pal_rgb[i]= a + (r<<8) + (g<<16) + (b<<24);
2213 case PIX_FMT_RGB32_1:
2217 c->pal_rgb[i]= a + (b<<8) + (g<<16) + (r<<24);
2224 c->pal_rgb[i]= b + (g<<8) + (r<<16) + (a<<24);
2229 // copy strides, so they can safely be modified
2230 if (c->sliceDir == 1) {
2231 // slices go from top to bottom
2232 int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2], srcStride[3]};
2233 int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2], dstStride[3]};
2235 reset_ptr(src2, c->srcFormat);
2236 reset_ptr((const uint8_t**)dst2, c->dstFormat);
2238 /* reset slice direction at end of frame */
2239 if (srcSliceY + srcSliceH == c->srcH)
2242 return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2, dstStride2);
2244 // slices go from bottom to top => we flip the image internally
2245 int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2], -srcStride[3]};
2246 int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2], -dstStride[3]};
2248 src2[0] += (srcSliceH-1)*srcStride[0];
2249 if (!usePal(c->srcFormat))
2250 src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1];
2251 src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2];
2252 src2[3] += (srcSliceH-1)*srcStride[3];
2253 dst2[0] += ( c->dstH -1)*dstStride[0];
2254 dst2[1] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1];
2255 dst2[2] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2];
2256 dst2[3] += ( c->dstH -1)*dstStride[3];
2258 reset_ptr(src2, c->srcFormat);
2259 reset_ptr((const uint8_t**)dst2, c->dstFormat);
2261 /* reset slice direction at end of frame */
2265 return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2);
2269 #if LIBSWSCALE_VERSION_MAJOR < 1
2270 int sws_scale_ordered(SwsContext *c, const uint8_t* const src[], int srcStride[], int srcSliceY,
2271 int srcSliceH, uint8_t* dst[], int dstStride[])
2273 return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride);
2277 /* Convert the palette to the same packed 32-bit format as the palette */
2278 void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
2282 for (i=0; i<num_pixels; i++)
2283 ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];
2286 /* Palette format: ABCD -> dst format: ABC */
2287 void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst, long num_pixels, const uint8_t *palette)
2291 for (i=0; i<num_pixels; i++) {
2293 dst[0]= palette[src[i]*4+0];
2294 dst[1]= palette[src[i]*4+1];
2295 dst[2]= palette[src[i]*4+2];