Implement shuffle_bytes_abcd() functions and use them for shuffling
[libav.git] / libswscale / swscale.c
CommitLineData
fe8054c0 1/*
d026b45e
DB
2 * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
3 *
4 * This file is part of FFmpeg.
5 *
6 * FFmpeg is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
10 *
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
15 *
16 * You should have received a copy of the GNU General Public License
17 * along with FFmpeg; if not, write to the Free Software
b19bcbaa 18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
d026b45e 19 *
807e0c66
LA
20 * the C code (not assembly, mmx, ...) of this file can be used
21 * under the LGPL license too
d026b45e 22 */
783e9cc9 23
28bf81c9 24/*
9990e426 25 supported Input formats: YV12, I420/IYUV, YUY2, UYVY, BGR32, BGR32_1, BGR24, BGR16, BGR15, RGB32, RGB32_1, RGB24, Y8/Y800, YVU9/IF09, PAL8
caeaabe7 26 supported output formats: YV12, I420/IYUV, YUY2, UYVY, {BGR,RGB}{1,4,8,15,16,24,32}, Y8/Y800, YVU9/IF09
e09d12f4 27 {BGR,RGB}{1,4,8,15,16} support dithering
6a4970ab 28
e09d12f4
MN
29 unscaled special converters (YV12=I420=IYUV, Y800=Y8)
30 YV12 -> {BGR,RGB}{1,4,8,15,16,24,32}
31 x -> x
32 YUV9 -> YV12
33 YUV9/YV12 -> Y800
34 Y800 -> YUV9/YV12
b935781b
MN
35 BGR24 -> BGR32 & RGB24 -> RGB32
36 BGR32 -> BGR24 & RGB32 -> RGB24
4bb3fa5e 37 BGR15 -> BGR16
b935781b
MN
38*/
39
6a4970ab 40/*
a6f6b237 41tested special converters (most are tested actually, but I did not write it down ...)
e09d12f4 42 YV12 -> BGR16
b935781b 43 YV12 -> YV12
4bb3fa5e 44 BGR15 -> BGR16
1e1c4fe9 45 BGR16 -> BGR16
e09d12f4 46 YVU9 -> YV12
b935781b
MN
47
48untested special converters
f40c7dbb 49 YV12/I420 -> BGR15/BGR24/BGR32 (it is the yuv2rgb stuff, so it should be OK)
1e1c4fe9
MN
50 YV12/I420 -> YV12/I420
51 YUY2/BGR15/BGR24/BGR32/RGB24/RGB32 -> same format
b935781b
MN
52 BGR24 -> BGR32 & RGB24 -> RGB32
53 BGR32 -> BGR24 & RGB32 -> RGB24
ec22603f 54 BGR24 -> YV12
28bf81c9
MN
55*/
56
d3f41512 57#include <inttypes.h>
dda87e9f 58#include <string.h>
077ea8a7 59#include <math.h>
c1b0bfb4 60#include <stdio.h>
b2d374c9 61#include "config.h"
81b7c056 62#include <assert.h>
d604bab9 63#include "swscale.h"
5427e242 64#include "swscale_internal.h"
37079906 65#include "rgb2rgb.h"
52154148 66#include "libavutil/intreadwrite.h"
83da2c6f 67#include "libavutil/x86_cpu.h"
7248797c 68#include "libavutil/avutil.h"
83da2c6f 69#include "libavutil/bswap.h"
a9af75ae 70#include "libavutil/pixdesc.h"
0d9f3d85 71
541c4eb9 72#undef MOVNTQ
7d7f78b5 73#undef PAVGB
d3f41512 74
783e9cc9 75//#undef HAVE_MMX2
f4406ec1 76//#define HAVE_AMD3DNOW
d3f41512 77//#undef HAVE_MMX
783e9cc9 78//#undef ARCH_X86
d604bab9 79#define DITHER1XBPP
d3f41512 80
f40c7dbb 81#define FAST_BGR2YV12 // use 7 bit coefficients instead of 15 bit
ac6a2e45 82
28bf81c9
MN
83#ifdef M_PI
84#define PI M_PI
85#else
86#define PI 3.14159265358979323846
87#endif
c1b0bfb4 88
9d9de37d
IP
89#define isPacked(x) ( \
90 (x)==PIX_FMT_PAL8 \
91 || (x)==PIX_FMT_YUYV422 \
92 || (x)==PIX_FMT_UYVY422 \
60222557 93 || isAnyRGB(x) \
9d9de37d 94 )
f412d5e9 95#define usePal(x) (av_pix_fmt_descriptors[x].flags & PIX_FMT_PAL)
6ff0ad6b 96
6b79dbce 97#define RGB2YUV_SHIFT 15
7b5d7b9e
MN
98#define BY ( (int)(0.114*219/255*(1<<RGB2YUV_SHIFT)+0.5))
99#define BV (-(int)(0.081*224/255*(1<<RGB2YUV_SHIFT)+0.5))
100#define BU ( (int)(0.500*224/255*(1<<RGB2YUV_SHIFT)+0.5))
101#define GY ( (int)(0.587*219/255*(1<<RGB2YUV_SHIFT)+0.5))
102#define GV (-(int)(0.419*224/255*(1<<RGB2YUV_SHIFT)+0.5))
103#define GU (-(int)(0.331*224/255*(1<<RGB2YUV_SHIFT)+0.5))
104#define RY ( (int)(0.299*219/255*(1<<RGB2YUV_SHIFT)+0.5))
105#define RV ( (int)(0.500*224/255*(1<<RGB2YUV_SHIFT)+0.5))
106#define RU (-(int)(0.169*224/255*(1<<RGB2YUV_SHIFT)+0.5))
6c7506de 107
0f5d4aa8
MN
108static const double rgb2yuv_table[8][9]={
109 {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5},
110 {0.7152, 0.0722, 0.2126, -0.386, 0.5, -0.115, -0.454, -0.046, 0.5},
111 {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5},
112 {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5},
113 {0.59 , 0.11 , 0.30 , -0.331, 0.5, -0.169, -0.421, -0.079, 0.5}, //FCC
114 {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5},
115 {0.587 , 0.114 , 0.299 , -0.331, 0.5, -0.169, -0.419, -0.081, 0.5}, //SMPTE 170M
116 {0.701 , 0.087 , 0.212 , -0.384, 0.5 -0.116, -0.445, -0.055, 0.5}, //SMPTE 240M
117};
118
783e9cc9
MN
119/*
120NOTES
d604bab9 121Special versions: fast Y 1:1 scaling (no interpolation in y direction)
31190492 122
783e9cc9 123TODO
bd7c6fd5 124more intelligent misalignment avoidance for the horizontal scaler
c1b0bfb4 125write special vertical cubic upscale version
f40c7dbb
DB
126optimize C code (YV12 / minmax)
127add support for packed pixel YUV input & output
6ff0ad6b 128add support for Y8 output
f40c7dbb 129optimize BGR24 & BGR32
ff7ba856 130add BGR4 output support
1e621b18 131write special BGR->BGR scaler
783e9cc9 132*/
31190492 133
b63f641e 134#if ARCH_X86 && CONFIG_GPL
d334c7c2
RD
135DECLARE_ASM_CONST(8, uint64_t, bF8)= 0xF8F8F8F8F8F8F8F8LL;
136DECLARE_ASM_CONST(8, uint64_t, bFC)= 0xFCFCFCFCFCFCFCFCLL;
137DECLARE_ASM_CONST(8, uint64_t, w10)= 0x0010001000100010LL;
138DECLARE_ASM_CONST(8, uint64_t, w02)= 0x0002000200020002LL;
139DECLARE_ASM_CONST(8, uint64_t, bm00001111)=0x00000000FFFFFFFFLL;
140DECLARE_ASM_CONST(8, uint64_t, bm00000111)=0x0000000000FFFFFFLL;
141DECLARE_ASM_CONST(8, uint64_t, bm11111000)=0xFFFFFFFFFF000000LL;
142DECLARE_ASM_CONST(8, uint64_t, bm01010101)=0x00FF00FF00FF00FFLL;
d604bab9 143
4a888526 144const DECLARE_ALIGNED(8, uint64_t, ff_dither4)[2] = {
221b804f
DB
145 0x0103010301030103LL,
146 0x0200020002000200LL,};
d8fa3c54 147
4a888526 148const DECLARE_ALIGNED(8, uint64_t, ff_dither8)[2] = {
221b804f
DB
149 0x0602060206020602LL,
150 0x0004000400040004LL,};
d604bab9 151
d334c7c2
RD
152DECLARE_ASM_CONST(8, uint64_t, b16Mask)= 0x001F001F001F001FLL;
153DECLARE_ASM_CONST(8, uint64_t, g16Mask)= 0x07E007E007E007E0LL;
154DECLARE_ASM_CONST(8, uint64_t, r16Mask)= 0xF800F800F800F800LL;
155DECLARE_ASM_CONST(8, uint64_t, b15Mask)= 0x001F001F001F001FLL;
156DECLARE_ASM_CONST(8, uint64_t, g15Mask)= 0x03E003E003E003E0LL;
157DECLARE_ASM_CONST(8, uint64_t, r15Mask)= 0x7C007C007C007C00LL;
d604bab9 158
5802683a
RD
159DECLARE_ALIGNED(8, const uint64_t, ff_M24A) = 0x00FF0000FF0000FFLL;
160DECLARE_ALIGNED(8, const uint64_t, ff_M24B) = 0xFF0000FF0000FF00LL;
161DECLARE_ALIGNED(8, const uint64_t, ff_M24C) = 0x0000FF0000FF0000LL;
99d2cb72 162
ac6a2e45 163#ifdef FAST_BGR2YV12
5802683a
RD
164DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000000210041000DULL;
165DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000FFEEFFDC0038ULL;
166DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00000038FFD2FFF8ULL;
ac6a2e45 167#else
5802683a
RD
168DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff) = 0x000020E540830C8BULL;
169DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff) = 0x0000ED0FDAC23831ULL;
170DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff) = 0x00003831D0E6F6EAULL;
69796008 171#endif /* FAST_BGR2YV12 */
5802683a
RD
172DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset) = 0x1010101010101010ULL;
173DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;
174DECLARE_ALIGNED(8, const uint64_t, ff_w1111) = 0x0001000100010001ULL;
dfb09bd1 175
b5c44b1a
DP
176DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY1Coeff) = 0x0C88000040870C88ULL;
177DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toY2Coeff) = 0x20DE4087000020DEULL;
178DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY1Coeff) = 0x20DE0000408720DEULL;
179DECLARE_ASM_CONST(8, uint64_t, ff_rgb24toY2Coeff) = 0x0C88408700000C88ULL;
180DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toYOffset) = 0x0008400000084000ULL;
dfb09bd1 181
4a888526 182DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUV)[2][4] = {
dfb09bd1
MN
183 {0x38380000DAC83838ULL, 0xECFFDAC80000ECFFULL, 0xF6E40000D0E3F6E4ULL, 0x3838D0E300003838ULL},
184 {0xECFF0000DAC8ECFFULL, 0x3838DAC800003838ULL, 0x38380000D0E33838ULL, 0xF6E4D0E30000F6E4ULL},
185};
186
b5c44b1a 187DECLARE_ASM_CONST(8, uint64_t, ff_bgr24toUVOffset)= 0x0040400000404000ULL;
dfb09bd1 188
7a24ec50 189#endif /* ARCH_X86 && CONFIG_GPL */
783e9cc9 190
4a888526 191DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_4)[2][8]={
45e18be8
MN
192{ 1, 3, 1, 3, 1, 3, 1, 3, },
193{ 2, 0, 2, 0, 2, 0, 2, 0, },
194};
195
4a888526 196DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_8)[2][8]={
45e18be8
MN
197{ 6, 2, 6, 2, 6, 2, 6, 2, },
198{ 0, 4, 0, 4, 0, 4, 0, 4, },
199};
200
4a888526 201DECLARE_ALIGNED(8, const uint8_t, dither_8x8_32)[8][8]={
45e18be8
MN
202{ 17, 9, 23, 15, 16, 8, 22, 14, },
203{ 5, 29, 3, 27, 4, 28, 2, 26, },
204{ 21, 13, 19, 11, 20, 12, 18, 10, },
205{ 0, 24, 6, 30, 1, 25, 7, 31, },
206{ 16, 8, 22, 14, 17, 9, 23, 15, },
207{ 4, 28, 2, 26, 5, 29, 3, 27, },
208{ 20, 12, 18, 10, 21, 13, 19, 11, },
209{ 1, 25, 7, 31, 0, 24, 6, 30, },
210};
211
4a888526 212DECLARE_ALIGNED(8, const uint8_t, dither_8x8_73)[8][8]={
45e18be8
MN
213{ 0, 55, 14, 68, 3, 58, 17, 72, },
214{ 37, 18, 50, 32, 40, 22, 54, 35, },
215{ 9, 64, 5, 59, 13, 67, 8, 63, },
216{ 46, 27, 41, 23, 49, 31, 44, 26, },
217{ 2, 57, 16, 71, 1, 56, 15, 70, },
218{ 39, 21, 52, 34, 38, 19, 51, 33, },
219{ 11, 66, 7, 62, 10, 65, 6, 60, },
220{ 48, 30, 43, 25, 47, 29, 42, 24, },
221};
222
45e18be8 223#if 1
4a888526 224DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
45e18be8
MN
225{117, 62, 158, 103, 113, 58, 155, 100, },
226{ 34, 199, 21, 186, 31, 196, 17, 182, },
227{144, 89, 131, 76, 141, 86, 127, 72, },
228{ 0, 165, 41, 206, 10, 175, 52, 217, },
229{110, 55, 151, 96, 120, 65, 162, 107, },
230{ 28, 193, 14, 179, 38, 203, 24, 189, },
231{138, 83, 124, 69, 148, 93, 134, 79, },
232{ 7, 172, 48, 213, 3, 168, 45, 210, },
233};
234#elif 1
235// tries to correct a gamma of 1.5
4a888526 236DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
45e18be8
MN
237{ 0, 143, 18, 200, 2, 156, 25, 215, },
238{ 78, 28, 125, 64, 89, 36, 138, 74, },
239{ 10, 180, 3, 161, 16, 195, 8, 175, },
240{109, 51, 93, 38, 121, 60, 105, 47, },
241{ 1, 152, 23, 210, 0, 147, 20, 205, },
242{ 85, 33, 134, 71, 81, 30, 130, 67, },
243{ 14, 190, 6, 171, 12, 185, 5, 166, },
244{117, 57, 101, 44, 113, 54, 97, 41, },
245};
246#elif 1
247// tries to correct a gamma of 2.0
4a888526 248DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
45e18be8
MN
249{ 0, 124, 8, 193, 0, 140, 12, 213, },
250{ 55, 14, 104, 42, 66, 19, 119, 52, },
251{ 3, 168, 1, 145, 6, 187, 3, 162, },
252{ 86, 31, 70, 21, 99, 39, 82, 28, },
253{ 0, 134, 11, 206, 0, 129, 9, 200, },
254{ 62, 17, 114, 48, 58, 16, 109, 45, },
255{ 5, 181, 2, 157, 4, 175, 1, 151, },
256{ 95, 36, 78, 26, 90, 34, 74, 24, },
257};
258#else
259// tries to correct a gamma of 2.5
4a888526 260DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
45e18be8
MN
261{ 0, 107, 3, 187, 0, 125, 6, 212, },
262{ 39, 7, 86, 28, 49, 11, 102, 36, },
263{ 1, 158, 0, 131, 3, 180, 1, 151, },
264{ 68, 19, 52, 12, 81, 25, 64, 17, },
265{ 0, 119, 5, 203, 0, 113, 4, 195, },
266{ 45, 9, 96, 33, 42, 8, 91, 30, },
267{ 2, 172, 1, 144, 2, 165, 0, 137, },
268{ 77, 23, 60, 15, 72, 21, 56, 14, },
269};
270#endif
5cebb24b 271
52154148
RP
272static av_always_inline void yuv2yuvX16inC_template(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
273 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
274 const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest,
275 int dstW, int chrDstW, int big_endian)
276{
277 //FIXME Optimize (just quickly written not optimized..)
278 int i;
279
280 for (i = 0; i < dstW; i++) {
281 int val = 1 << 10;
282 int j;
283
284 for (j = 0; j < lumFilterSize; j++)
285 val += lumSrc[j][i] * lumFilter[j];
286
287 if (big_endian) {
288 AV_WB16(&dest[i], av_clip_uint16(val >> 11));
289 } else {
290 AV_WL16(&dest[i], av_clip_uint16(val >> 11));
291 }
292 }
293
294 if (uDest) {
295 for (i = 0; i < chrDstW; i++) {
296 int u = 1 << 10;
297 int v = 1 << 10;
298 int j;
299
300 for (j = 0; j < chrFilterSize; j++) {
301 u += chrSrc[j][i ] * chrFilter[j];
302 v += chrSrc[j][i + VOFW] * chrFilter[j];
303 }
304
305 if (big_endian) {
306 AV_WB16(&uDest[i], av_clip_uint16(u >> 11));
307 AV_WB16(&vDest[i], av_clip_uint16(v >> 11));
308 } else {
309 AV_WL16(&uDest[i], av_clip_uint16(u >> 11));
310 AV_WL16(&vDest[i], av_clip_uint16(v >> 11));
311 }
312 }
313 }
314
315 if (CONFIG_SWSCALE_ALPHA && aDest) {
316 for (i = 0; i < dstW; i++) {
317 int val = 1 << 10;
318 int j;
319
320 for (j = 0; j < lumFilterSize; j++)
321 val += alpSrc[j][i] * lumFilter[j];
322
323 if (big_endian) {
324 AV_WB16(&aDest[i], av_clip_uint16(val >> 11));
325 } else {
326 AV_WL16(&aDest[i], av_clip_uint16(val >> 11));
327 }
328 }
329 }
330}
331
332static inline void yuv2yuvX16inC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
333 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
334 const int16_t **alpSrc, uint16_t *dest, uint16_t *uDest, uint16_t *vDest, uint16_t *aDest, int dstW, int chrDstW,
335 enum PixelFormat dstFormat)
336{
337 if (isBE(dstFormat)) {
338 yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
339 chrFilter, chrSrc, chrFilterSize,
340 alpSrc,
341 dest, uDest, vDest, aDest,
342 dstW, chrDstW, 1);
343 } else {
344 yuv2yuvX16inC_template(lumFilter, lumSrc, lumFilterSize,
345 chrFilter, chrSrc, chrFilterSize,
346 alpSrc,
347 dest, uDest, vDest, aDest,
348 dstW, chrDstW, 0);
349 }
350}
351
7ac40327
RP
352static inline void yuv2yuvXinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
353 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
354 const int16_t **alpSrc, uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, int dstW, int chrDstW)
e3d2500f 355{
f40c7dbb 356 //FIXME Optimize (just quickly written not optimized..)
221b804f 357 int i;
dd68318c 358 for (i=0; i<dstW; i++) {
221b804f
DB
359 int val=1<<18;
360 int j;
361 for (j=0; j<lumFilterSize; j++)
362 val += lumSrc[j][i] * lumFilter[j];
363
364 dest[i]= av_clip_uint8(val>>19);
365 }
366
1b0a4572 367 if (uDest)
dd68318c 368 for (i=0; i<chrDstW; i++) {
221b804f
DB
369 int u=1<<18;
370 int v=1<<18;
371 int j;
dd68318c 372 for (j=0; j<chrFilterSize; j++) {
221b804f 373 u += chrSrc[j][i] * chrFilter[j];
8b2fce0d 374 v += chrSrc[j][i + VOFW] * chrFilter[j];
221b804f
DB
375 }
376
377 uDest[i]= av_clip_uint8(u>>19);
378 vDest[i]= av_clip_uint8(v>>19);
379 }
6858492e
CS
380
381 if (CONFIG_SWSCALE_ALPHA && aDest)
dd68318c 382 for (i=0; i<dstW; i++) {
6858492e
CS
383 int val=1<<18;
384 int j;
385 for (j=0; j<lumFilterSize; j++)
386 val += alpSrc[j][i] * lumFilter[j];
387
388 aDest[i]= av_clip_uint8(val>>19);
389 }
390
e3d2500f
MN
391}
392
7ac40327
RP
393static inline void yuv2nv12XinC(const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
394 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
221b804f 395 uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, int dstFormat)
6118e52e 396{
f40c7dbb 397 //FIXME Optimize (just quickly written not optimized..)
221b804f 398 int i;
dd68318c 399 for (i=0; i<dstW; i++) {
221b804f
DB
400 int val=1<<18;
401 int j;
402 for (j=0; j<lumFilterSize; j++)
403 val += lumSrc[j][i] * lumFilter[j];
404
405 dest[i]= av_clip_uint8(val>>19);
406 }
407
1b0a4572 408 if (!uDest)
221b804f
DB
409 return;
410
411 if (dstFormat == PIX_FMT_NV12)
dd68318c 412 for (i=0; i<chrDstW; i++) {
221b804f
DB
413 int u=1<<18;
414 int v=1<<18;
415 int j;
dd68318c 416 for (j=0; j<chrFilterSize; j++) {
221b804f 417 u += chrSrc[j][i] * chrFilter[j];
8b2fce0d 418 v += chrSrc[j][i + VOFW] * chrFilter[j];
221b804f
DB
419 }
420
421 uDest[2*i]= av_clip_uint8(u>>19);
422 uDest[2*i+1]= av_clip_uint8(v>>19);
423 }
424 else
dd68318c 425 for (i=0; i<chrDstW; i++) {
221b804f
DB
426 int u=1<<18;
427 int v=1<<18;
428 int j;
dd68318c 429 for (j=0; j<chrFilterSize; j++) {
221b804f 430 u += chrSrc[j][i] * chrFilter[j];
8b2fce0d 431 v += chrSrc[j][i + VOFW] * chrFilter[j];
221b804f
DB
432 }
433
434 uDest[2*i]= av_clip_uint8(v>>19);
435 uDest[2*i+1]= av_clip_uint8(u>>19);
436 }
6118e52e 437}
46de8b73 438
6858492e 439#define YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha) \
dd68318c 440 for (i=0; i<(dstW>>1); i++) {\
221b804f
DB
441 int j;\
442 int Y1 = 1<<18;\
443 int Y2 = 1<<18;\
444 int U = 1<<18;\
445 int V = 1<<18;\
6858492e 446 int av_unused A1, A2;\
2db27aad 447 type av_unused *r, *b, *g;\
221b804f
DB
448 const int i2= 2*i;\
449 \
dd68318c 450 for (j=0; j<lumFilterSize; j++) {\
221b804f
DB
451 Y1 += lumSrc[j][i2] * lumFilter[j];\
452 Y2 += lumSrc[j][i2+1] * lumFilter[j];\
453 }\
dd68318c 454 for (j=0; j<chrFilterSize; j++) {\
221b804f 455 U += chrSrc[j][i] * chrFilter[j];\
8b2fce0d 456 V += chrSrc[j][i+VOFW] * chrFilter[j];\
221b804f
DB
457 }\
458 Y1>>=19;\
459 Y2>>=19;\
460 U >>=19;\
461 V >>=19;\
dd68318c 462 if (alpha) {\
6858492e
CS
463 A1 = 1<<18;\
464 A2 = 1<<18;\
dd68318c 465 for (j=0; j<lumFilterSize; j++) {\
6858492e
CS
466 A1 += alpSrc[j][i2 ] * lumFilter[j];\
467 A2 += alpSrc[j][i2+1] * lumFilter[j];\
468 }\
469 A1>>=19;\
470 A2>>=19;\
c36a4294 471 }
bdf397ba 472
6858492e
CS
473#define YSCALE_YUV_2_PACKEDX_C(type,alpha) \
474 YSCALE_YUV_2_PACKEDX_NOCLIP_C(type,alpha)\
dd68318c 475 if ((Y1|Y2|U|V)&256) {\
221b804f
DB
476 if (Y1>255) Y1=255; \
477 else if (Y1<0)Y1=0; \
478 if (Y2>255) Y2=255; \
479 else if (Y2<0)Y2=0; \
480 if (U>255) U=255; \
481 else if (U<0) U=0; \
482 if (V>255) V=255; \
483 else if (V<0) V=0; \
6858492e 484 }\
dd68318c 485 if (alpha && ((A1|A2)&256)) {\
6858492e
CS
486 A1=av_clip_uint8(A1);\
487 A2=av_clip_uint8(A2);\
221b804f 488 }
6a4970ab 489
6858492e 490#define YSCALE_YUV_2_PACKEDX_FULL_C(rnd,alpha) \
dd68318c 491 for (i=0; i<dstW; i++) {\
f0faee4c
MN
492 int j;\
493 int Y = 0;\
494 int U = -128<<19;\
495 int V = -128<<19;\
6858492e 496 int av_unused A;\
f0faee4c
MN
497 int R,G,B;\
498 \
dd68318c 499 for (j=0; j<lumFilterSize; j++) {\
f0faee4c
MN
500 Y += lumSrc[j][i ] * lumFilter[j];\
501 }\
dd68318c 502 for (j=0; j<chrFilterSize; j++) {\
f0faee4c
MN
503 U += chrSrc[j][i ] * chrFilter[j];\
504 V += chrSrc[j][i+VOFW] * chrFilter[j];\
505 }\
506 Y >>=10;\
507 U >>=10;\
508 V >>=10;\
dd68318c 509 if (alpha) {\
6858492e
CS
510 A = rnd;\
511 for (j=0; j<lumFilterSize; j++)\
512 A += alpSrc[j][i ] * lumFilter[j];\
513 A >>=19;\
514 if (A&256)\
515 A = av_clip_uint8(A);\
c36a4294 516 }
f0faee4c 517
6858492e
CS
518#define YSCALE_YUV_2_RGBX_FULL_C(rnd,alpha) \
519 YSCALE_YUV_2_PACKEDX_FULL_C(rnd>>3,alpha)\
43c16478
MN
520 Y-= c->yuv2rgb_y_offset;\
521 Y*= c->yuv2rgb_y_coeff;\
f0faee4c 522 Y+= rnd;\
43c16478
MN
523 R= Y + V*c->yuv2rgb_v2r_coeff;\
524 G= Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;\
525 B= Y + U*c->yuv2rgb_u2b_coeff;\
dd68318c 526 if ((R|G|B)&(0xC0000000)) {\
f0faee4c
MN
527 if (R>=(256<<22)) R=(256<<22)-1; \
528 else if (R<0)R=0; \
529 if (G>=(256<<22)) G=(256<<22)-1; \
530 else if (G<0)G=0; \
531 if (B>=(256<<22)) B=(256<<22)-1; \
532 else if (B<0)B=0; \
c36a4294 533 }
f0faee4c 534
e69bd294 535#define YSCALE_YUV_2_GRAY16_C \
dd68318c 536 for (i=0; i<(dstW>>1); i++) {\
b0880d5d
MN
537 int j;\
538 int Y1 = 1<<18;\
539 int Y2 = 1<<18;\
540 int U = 1<<18;\
541 int V = 1<<18;\
e69bd294 542 \
b0880d5d
MN
543 const int i2= 2*i;\
544 \
dd68318c 545 for (j=0; j<lumFilterSize; j++) {\
b0880d5d
MN
546 Y1 += lumSrc[j][i2] * lumFilter[j];\
547 Y2 += lumSrc[j][i2+1] * lumFilter[j];\
548 }\
549 Y1>>=11;\
550 Y2>>=11;\
dd68318c 551 if ((Y1|Y2|U|V)&65536) {\
b0880d5d
MN
552 if (Y1>65535) Y1=65535; \
553 else if (Y1<0)Y1=0; \
554 if (Y2>65535) Y2=65535; \
555 else if (Y2<0)Y2=0; \
556 }
557
6858492e
CS
558#define YSCALE_YUV_2_RGBX_C(type,alpha) \
559 YSCALE_YUV_2_PACKEDX_C(type,alpha) /* FIXME fix tables so that clipping is not needed and then use _NOCLIP*/\
221b804f
DB
560 r = (type *)c->table_rV[V]; \
561 g = (type *)(c->table_gU[U] + c->table_gV[V]); \
c36a4294 562 b = (type *)c->table_bU[U];
221b804f 563
6858492e 564#define YSCALE_YUV_2_PACKED2_C(type,alpha) \
dd68318c 565 for (i=0; i<(dstW>>1); i++) { \
221b804f
DB
566 const int i2= 2*i; \
567 int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>19; \
568 int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19; \
569 int U= (uvbuf0[i ]*uvalpha1+uvbuf1[i ]*uvalpha)>>19; \
8b2fce0d 570 int V= (uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19; \
6858492e
CS
571 type av_unused *r, *b, *g; \
572 int av_unused A1, A2; \
dd68318c 573 if (alpha) {\
6858492e
CS
574 A1= (abuf0[i2 ]*yalpha1+abuf1[i2 ]*yalpha)>>19; \
575 A2= (abuf0[i2+1]*yalpha1+abuf1[i2+1]*yalpha)>>19; \
c36a4294 576 }
46de8b73 577
b0880d5d 578#define YSCALE_YUV_2_GRAY16_2_C \
dd68318c 579 for (i=0; i<(dstW>>1); i++) { \
b0880d5d
MN
580 const int i2= 2*i; \
581 int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>11; \
c36a4294 582 int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>11;
b0880d5d 583
6858492e
CS
584#define YSCALE_YUV_2_RGB2_C(type,alpha) \
585 YSCALE_YUV_2_PACKED2_C(type,alpha)\
221b804f
DB
586 r = (type *)c->table_rV[V];\
587 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
c36a4294 588 b = (type *)c->table_bU[U];
cf7d1c1a 589
6858492e 590#define YSCALE_YUV_2_PACKED1_C(type,alpha) \
dd68318c 591 for (i=0; i<(dstW>>1); i++) {\
221b804f
DB
592 const int i2= 2*i;\
593 int Y1= buf0[i2 ]>>7;\
594 int Y2= buf0[i2+1]>>7;\
595 int U= (uvbuf1[i ])>>7;\
8b2fce0d 596 int V= (uvbuf1[i+VOFW])>>7;\
6858492e
CS
597 type av_unused *r, *b, *g;\
598 int av_unused A1, A2;\
dd68318c 599 if (alpha) {\
6858492e
CS
600 A1= abuf0[i2 ]>>7;\
601 A2= abuf0[i2+1]>>7;\
c36a4294 602 }
46de8b73 603
b0880d5d 604#define YSCALE_YUV_2_GRAY16_1_C \
dd68318c 605 for (i=0; i<(dstW>>1); i++) {\
b0880d5d
MN
606 const int i2= 2*i;\
607 int Y1= buf0[i2 ]<<1;\
c36a4294 608 int Y2= buf0[i2+1]<<1;
b0880d5d 609
6858492e
CS
610#define YSCALE_YUV_2_RGB1_C(type,alpha) \
611 YSCALE_YUV_2_PACKED1_C(type,alpha)\
221b804f
DB
612 r = (type *)c->table_rV[V];\
613 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
c36a4294 614 b = (type *)c->table_bU[U];
cf7d1c1a 615
6858492e 616#define YSCALE_YUV_2_PACKED1B_C(type,alpha) \
dd68318c 617 for (i=0; i<(dstW>>1); i++) {\
221b804f
DB
618 const int i2= 2*i;\
619 int Y1= buf0[i2 ]>>7;\
620 int Y2= buf0[i2+1]>>7;\
621 int U= (uvbuf0[i ] + uvbuf1[i ])>>8;\
8b2fce0d 622 int V= (uvbuf0[i+VOFW] + uvbuf1[i+VOFW])>>8;\
6858492e
CS
623 type av_unused *r, *b, *g;\
624 int av_unused A1, A2;\
dd68318c 625 if (alpha) {\
6858492e
CS
626 A1= abuf0[i2 ]>>7;\
627 A2= abuf0[i2+1]>>7;\
c36a4294 628 }
46de8b73 629
6858492e
CS
630#define YSCALE_YUV_2_RGB1B_C(type,alpha) \
631 YSCALE_YUV_2_PACKED1B_C(type,alpha)\
221b804f
DB
632 r = (type *)c->table_rV[V];\
633 g = (type *)(c->table_gU[U] + c->table_gV[V]);\
c36a4294 634 b = (type *)c->table_bU[U];
cf7d1c1a 635
ec1bca2a 636#define YSCALE_YUV_2_MONO2_C \
e69bd294
MN
637 const uint8_t * const d128=dither_8x8_220[y&7];\
638 uint8_t *g= c->table_gU[128] + c->table_gV[128];\
dd68318c 639 for (i=0; i<dstW-7; i+=8) {\
e69bd294
MN
640 int acc;\
641 acc = g[((buf0[i ]*yalpha1+buf1[i ]*yalpha)>>19) + d128[0]];\
642 acc+= acc + g[((buf0[i+1]*yalpha1+buf1[i+1]*yalpha)>>19) + d128[1]];\
643 acc+= acc + g[((buf0[i+2]*yalpha1+buf1[i+2]*yalpha)>>19) + d128[2]];\
644 acc+= acc + g[((buf0[i+3]*yalpha1+buf1[i+3]*yalpha)>>19) + d128[3]];\
645 acc+= acc + g[((buf0[i+4]*yalpha1+buf1[i+4]*yalpha)>>19) + d128[4]];\
646 acc+= acc + g[((buf0[i+5]*yalpha1+buf1[i+5]*yalpha)>>19) + d128[5]];\
647 acc+= acc + g[((buf0[i+6]*yalpha1+buf1[i+6]*yalpha)>>19) + d128[6]];\
648 acc+= acc + g[((buf0[i+7]*yalpha1+buf1[i+7]*yalpha)>>19) + d128[7]];\
ec1bca2a 649 ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
e69bd294 650 dest++;\
c36a4294 651 }
e69bd294 652
ec1bca2a 653#define YSCALE_YUV_2_MONOX_C \
e69bd294
MN
654 const uint8_t * const d128=dither_8x8_220[y&7];\
655 uint8_t *g= c->table_gU[128] + c->table_gV[128];\
656 int acc=0;\
dd68318c 657 for (i=0; i<dstW-1; i+=2) {\
e69bd294
MN
658 int j;\
659 int Y1=1<<18;\
660 int Y2=1<<18;\
661\
dd68318c 662 for (j=0; j<lumFilterSize; j++) {\
e69bd294
MN
663 Y1 += lumSrc[j][i] * lumFilter[j];\
664 Y2 += lumSrc[j][i+1] * lumFilter[j];\
665 }\
666 Y1>>=19;\
667 Y2>>=19;\
dd68318c 668 if ((Y1|Y2)&256) {\
e69bd294
MN
669 if (Y1>255) Y1=255;\
670 else if (Y1<0)Y1=0;\
671 if (Y2>255) Y2=255;\
672 else if (Y2<0)Y2=0;\
673 }\
674 acc+= acc + g[Y1+d128[(i+0)&7]];\
675 acc+= acc + g[Y2+d128[(i+1)&7]];\
dd68318c 676 if ((i&7)==6) {\
ec1bca2a 677 ((uint8_t*)dest)[0]= c->dstFormat == PIX_FMT_MONOBLACK ? acc : ~acc;\
e69bd294
MN
678 dest++;\
679 }\
680 }
681
e69bd294 682#define YSCALE_YUV_2_ANYRGB_C(func, func2, func_g16, func_monoblack)\
dd68318c 683 switch(c->dstFormat) {\
68e7f482
KS
684 case PIX_FMT_RGB48BE:\
685 case PIX_FMT_RGB48LE:\
686 func(uint8_t,0)\
687 ((uint8_t*)dest)[ 0]= r[Y1];\
688 ((uint8_t*)dest)[ 1]= r[Y1];\
689 ((uint8_t*)dest)[ 2]= g[Y1];\
690 ((uint8_t*)dest)[ 3]= g[Y1];\
691 ((uint8_t*)dest)[ 4]= b[Y1];\
692 ((uint8_t*)dest)[ 5]= b[Y1];\
693 ((uint8_t*)dest)[ 6]= r[Y2];\
694 ((uint8_t*)dest)[ 7]= r[Y2];\
695 ((uint8_t*)dest)[ 8]= g[Y2];\
696 ((uint8_t*)dest)[ 9]= g[Y2];\
697 ((uint8_t*)dest)[10]= b[Y2];\
698 ((uint8_t*)dest)[11]= b[Y2];\
699 dest+=12;\
700 }\
701 break;\
6858492e
CS
702 case PIX_FMT_RGBA:\
703 case PIX_FMT_BGRA:\
dd68318c 704 if (CONFIG_SMALL) {\
6858492e
CS
705 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\
706 func(uint32_t,needAlpha)\
707 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? (A1<<24) : 0);\
708 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? (A2<<24) : 0);\
709 }\
dd68318c
RP
710 } else {\
711 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\
6858492e
CS
712 func(uint32_t,1)\
713 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (A1<<24);\
714 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (A2<<24);\
715 }\
dd68318c 716 } else {\
6858492e
CS
717 func(uint32_t,0)\
718 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
719 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
720 }\
721 }\
722 }\
723 break;\
724 case PIX_FMT_ARGB:\
725 case PIX_FMT_ABGR:\
dd68318c 726 if (CONFIG_SMALL) {\
6858492e
CS
727 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;\
728 func(uint32_t,needAlpha)\
729 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + (needAlpha ? A1 : 0);\
730 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + (needAlpha ? A2 : 0);\
731 }\
dd68318c
RP
732 } else {\
733 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {\
6858492e
CS
734 func(uint32_t,1)\
735 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1] + A1;\
736 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2] + A2;\
737 }\
dd68318c 738 } else {\
6858492e
CS
739 func(uint32_t,0)\
740 ((uint32_t*)dest)[i2+0]= r[Y1] + g[Y1] + b[Y1];\
741 ((uint32_t*)dest)[i2+1]= r[Y2] + g[Y2] + b[Y2];\
742 }\
743 }\
221b804f
DB
744 } \
745 break;\
746 case PIX_FMT_RGB24:\
6858492e 747 func(uint8_t,0)\
221b804f
DB
748 ((uint8_t*)dest)[0]= r[Y1];\
749 ((uint8_t*)dest)[1]= g[Y1];\
750 ((uint8_t*)dest)[2]= b[Y1];\
751 ((uint8_t*)dest)[3]= r[Y2];\
752 ((uint8_t*)dest)[4]= g[Y2];\
753 ((uint8_t*)dest)[5]= b[Y2];\
754 dest+=6;\
755 }\
756 break;\
757 case PIX_FMT_BGR24:\
6858492e 758 func(uint8_t,0)\
221b804f
DB
759 ((uint8_t*)dest)[0]= b[Y1];\
760 ((uint8_t*)dest)[1]= g[Y1];\
761 ((uint8_t*)dest)[2]= r[Y1];\
762 ((uint8_t*)dest)[3]= b[Y2];\
763 ((uint8_t*)dest)[4]= g[Y2];\
764 ((uint8_t*)dest)[5]= r[Y2];\
765 dest+=6;\
766 }\
767 break;\
768 case PIX_FMT_RGB565:\
769 case PIX_FMT_BGR565:\
770 {\
771 const int dr1= dither_2x2_8[y&1 ][0];\
772 const int dg1= dither_2x2_4[y&1 ][0];\
773 const int db1= dither_2x2_8[(y&1)^1][0];\
774 const int dr2= dither_2x2_8[y&1 ][1];\
775 const int dg2= dither_2x2_4[y&1 ][1];\
776 const int db2= dither_2x2_8[(y&1)^1][1];\
6858492e 777 func(uint16_t,0)\
221b804f
DB
778 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
779 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
780 }\
781 }\
782 break;\
783 case PIX_FMT_RGB555:\
784 case PIX_FMT_BGR555:\
785 {\
786 const int dr1= dither_2x2_8[y&1 ][0];\
787 const int dg1= dither_2x2_8[y&1 ][1];\
788 const int db1= dither_2x2_8[(y&1)^1][0];\
789 const int dr2= dither_2x2_8[y&1 ][1];\
790 const int dg2= dither_2x2_8[y&1 ][0];\
791 const int db2= dither_2x2_8[(y&1)^1][1];\
6858492e 792 func(uint16_t,0)\
221b804f
DB
793 ((uint16_t*)dest)[i2+0]= r[Y1+dr1] + g[Y1+dg1] + b[Y1+db1];\
794 ((uint16_t*)dest)[i2+1]= r[Y2+dr2] + g[Y2+dg2] + b[Y2+db2];\
795 }\
796 }\
797 break;\
798 case PIX_FMT_RGB8:\
799 case PIX_FMT_BGR8:\
800 {\
801 const uint8_t * const d64= dither_8x8_73[y&7];\
802 const uint8_t * const d32= dither_8x8_32[y&7];\
6858492e 803 func(uint8_t,0)\
221b804f
DB
804 ((uint8_t*)dest)[i2+0]= r[Y1+d32[(i2+0)&7]] + g[Y1+d32[(i2+0)&7]] + b[Y1+d64[(i2+0)&7]];\
805 ((uint8_t*)dest)[i2+1]= r[Y2+d32[(i2+1)&7]] + g[Y2+d32[(i2+1)&7]] + b[Y2+d64[(i2+1)&7]];\
806 }\
807 }\
808 break;\
809 case PIX_FMT_RGB4:\
810 case PIX_FMT_BGR4:\
811 {\
812 const uint8_t * const d64= dither_8x8_73 [y&7];\
813 const uint8_t * const d128=dither_8x8_220[y&7];\
6858492e 814 func(uint8_t,0)\
221b804f
DB
815 ((uint8_t*)dest)[i]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]]\
816 + ((r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]])<<4);\
817 }\
818 }\
819 break;\
820 case PIX_FMT_RGB4_BYTE:\
821 case PIX_FMT_BGR4_BYTE:\
822 {\
823 const uint8_t * const d64= dither_8x8_73 [y&7];\
824 const uint8_t * const d128=dither_8x8_220[y&7];\
6858492e 825 func(uint8_t,0)\
221b804f
DB
826 ((uint8_t*)dest)[i2+0]= r[Y1+d128[(i2+0)&7]] + g[Y1+d64[(i2+0)&7]] + b[Y1+d128[(i2+0)&7]];\
827 ((uint8_t*)dest)[i2+1]= r[Y2+d128[(i2+1)&7]] + g[Y2+d64[(i2+1)&7]] + b[Y2+d128[(i2+1)&7]];\
828 }\
829 }\
830 break;\
831 case PIX_FMT_MONOBLACK:\
ec1bca2a 832 case PIX_FMT_MONOWHITE:\
221b804f 833 {\
e69bd294 834 func_monoblack\
221b804f
DB
835 }\
836 break;\
837 case PIX_FMT_YUYV422:\
838 func2\
839 ((uint8_t*)dest)[2*i2+0]= Y1;\
840 ((uint8_t*)dest)[2*i2+1]= U;\
841 ((uint8_t*)dest)[2*i2+2]= Y2;\
842 ((uint8_t*)dest)[2*i2+3]= V;\
843 } \
844 break;\
845 case PIX_FMT_UYVY422:\
846 func2\
847 ((uint8_t*)dest)[2*i2+0]= U;\
848 ((uint8_t*)dest)[2*i2+1]= Y1;\
849 ((uint8_t*)dest)[2*i2+2]= V;\
850 ((uint8_t*)dest)[2*i2+3]= Y2;\
851 } \
852 break;\
b0880d5d
MN
853 case PIX_FMT_GRAY16BE:\
854 func_g16\
855 ((uint8_t*)dest)[2*i2+0]= Y1>>8;\
856 ((uint8_t*)dest)[2*i2+1]= Y1;\
857 ((uint8_t*)dest)[2*i2+2]= Y2>>8;\
858 ((uint8_t*)dest)[2*i2+3]= Y2;\
859 } \
860 break;\
861 case PIX_FMT_GRAY16LE:\
862 func_g16\
863 ((uint8_t*)dest)[2*i2+0]= Y1;\
864 ((uint8_t*)dest)[2*i2+1]= Y1>>8;\
865 ((uint8_t*)dest)[2*i2+2]= Y2;\
866 ((uint8_t*)dest)[2*i2+3]= Y2>>8;\
867 } \
868 break;\
c36a4294 869 }
cf7d1c1a 870
7ac40327
RP
871static inline void yuv2packedXinC(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
872 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
873 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
e3d2500f 874{
221b804f 875 int i;
6858492e 876 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)
e3d2500f
MN
877}
878
7ac40327
RP
879static inline void yuv2rgbXinC_full(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
880 const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
881 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
f0faee4c
MN
882{
883 int i;
30b61475 884 int step= c->dstFormatBpp/8;
d616c8ae 885 int aidx= 3;
f0faee4c 886
dd68318c 887 switch(c->dstFormat) {
f0faee4c
MN
888 case PIX_FMT_ARGB:
889 dest++;
a3398feb 890 aidx= 0;
f0faee4c 891 case PIX_FMT_RGB24:
d616c8ae 892 aidx--;
f0faee4c 893 case PIX_FMT_RGBA:
dd68318c 894 if (CONFIG_SMALL) {
6858492e
CS
895 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;
896 YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)
897 dest[aidx]= needAlpha ? A : 255;
898 dest[0]= R>>22;
899 dest[1]= G>>22;
900 dest[2]= B>>22;
901 dest+= step;
902 }
dd68318c
RP
903 } else {
904 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
6858492e
CS
905 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)
906 dest[aidx]= A;
907 dest[0]= R>>22;
908 dest[1]= G>>22;
909 dest[2]= B>>22;
910 dest+= step;
911 }
dd68318c 912 } else {
6858492e
CS
913 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)
914 dest[aidx]= 255;
915 dest[0]= R>>22;
916 dest[1]= G>>22;
917 dest[2]= B>>22;
918 dest+= step;
919 }
920 }
f0faee4c
MN
921 }
922 break;
923 case PIX_FMT_ABGR:
924 dest++;
a3398feb 925 aidx= 0;
f0faee4c 926 case PIX_FMT_BGR24:
d616c8ae 927 aidx--;
f0faee4c 928 case PIX_FMT_BGRA:
dd68318c 929 if (CONFIG_SMALL) {
6858492e
CS
930 int needAlpha = CONFIG_SWSCALE_ALPHA && c->alpPixBuf;
931 YSCALE_YUV_2_RGBX_FULL_C(1<<21, needAlpha)
932 dest[aidx]= needAlpha ? A : 255;
933 dest[0]= B>>22;
934 dest[1]= G>>22;
935 dest[2]= R>>22;
936 dest+= step;
937 }
dd68318c
RP
938 } else {
939 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
6858492e
CS
940 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 1)
941 dest[aidx]= A;
942 dest[0]= B>>22;
943 dest[1]= G>>22;
944 dest[2]= R>>22;
945 dest+= step;
946 }
dd68318c 947 } else {
6858492e
CS
948 YSCALE_YUV_2_RGBX_FULL_C(1<<21, 0)
949 dest[aidx]= 255;
950 dest[0]= B>>22;
951 dest[1]= G>>22;
952 dest[2]= R>>22;
953 dest+= step;
954 }
955 }
f0faee4c
MN
956 }
957 break;
958 default:
959 assert(0);
960 }
961}
e3d2500f 962
dd68318c
RP
963static void fillPlane(uint8_t* plane, int stride, int width, int height, int y, uint8_t val)
964{
d4da3e47
CS
965 int i;
966 uint8_t *ptr = plane + stride*y;
dd68318c 967 for (i=0; i<height; i++) {
d4da3e47
CS
968 memset(ptr, val, width);
969 ptr += stride;
970 }
971}
972
a3e35e28
RP
973static inline void rgb48ToY(uint8_t *dst, const uint8_t *src, int width,
974 uint32_t *unused)
e8417235
KS
975{
976 int i;
977 for (i = 0; i < width; i++) {
978 int r = src[i*6+0];
979 int g = src[i*6+2];
980 int b = src[i*6+4];
981
982 dst[i] = (RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
983 }
984}
985
986static inline void rgb48ToUV(uint8_t *dstU, uint8_t *dstV,
efc034cc
RP
987 const uint8_t *src1, const uint8_t *src2,
988 int width, uint32_t *unused)
e8417235
KS
989{
990 int i;
991 assert(src1==src2);
992 for (i = 0; i < width; i++) {
993 int r = src1[6*i + 0];
994 int g = src1[6*i + 2];
995 int b = src1[6*i + 4];
996
997 dstU[i] = (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
998 dstV[i] = (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
999 }
1000}
1001
1002static inline void rgb48ToUV_half(uint8_t *dstU, uint8_t *dstV,
efc034cc
RP
1003 const uint8_t *src1, const uint8_t *src2,
1004 int width, uint32_t *unused)
e8417235
KS
1005{
1006 int i;
1007 assert(src1==src2);
1008 for (i = 0; i < width; i++) {
1009 int r= src1[12*i + 0] + src1[12*i + 6];
1010 int g= src1[12*i + 2] + src1[12*i + 8];
1011 int b= src1[12*i + 4] + src1[12*i + 10];
1012
1013 dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1014 dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT)) >> (RGB2YUV_SHIFT+1);
1015 }
1016}
1017
80704c47
KS
1018#define BGR2Y(type, name, shr, shg, shb, maskr, maskg, maskb, RY, GY, BY, S)\
1019static inline void name(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)\
1020{\
1021 int i;\
dd68318c 1022 for (i=0; i<width; i++) {\
80704c47
KS
1023 int b= (((const type*)src)[i]>>shb)&maskb;\
1024 int g= (((const type*)src)[i]>>shg)&maskg;\
1025 int r= (((const type*)src)[i]>>shr)&maskr;\
1026\
1027 dst[i]= (((RY)*r + (GY)*g + (BY)*b + (33<<((S)-1)))>>(S));\
1028 }\
1029}
1030
1031BGR2Y(uint32_t, bgr32ToY,16, 0, 0, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY , BY<< 8, RGB2YUV_SHIFT+8)
1032BGR2Y(uint32_t, rgb32ToY, 0, 0,16, 0x00FF, 0xFF00, 0x00FF, RY<< 8, GY , BY<< 8, RGB2YUV_SHIFT+8)
1033BGR2Y(uint16_t, bgr16ToY, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RY<<11, GY<<5, BY , RGB2YUV_SHIFT+8)
1034BGR2Y(uint16_t, bgr15ToY, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RY<<10, GY<<5, BY , RGB2YUV_SHIFT+7)
1035BGR2Y(uint16_t, rgb16ToY, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RY , GY<<5, BY<<11, RGB2YUV_SHIFT+8)
1036BGR2Y(uint16_t, rgb15ToY, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RY , GY<<5, BY<<10, RGB2YUV_SHIFT+7)
1037
dd68318c
RP
1038static inline void abgrToA(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1039{
80704c47 1040 int i;
dd68318c 1041 for (i=0; i<width; i++) {
80704c47
KS
1042 dst[i]= src[4*i];
1043 }
1044}
1045
1046#define BGR2UV(type, name, shr, shg, shb, maska, maskr, maskg, maskb, RU, GU, BU, RV, GV, BV, S)\
1047static inline void name(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\
1048{\
1049 int i;\
dd68318c 1050 for (i=0; i<width; i++) {\
80704c47
KS
1051 int b= (((const type*)src)[i]&maskb)>>shb;\
1052 int g= (((const type*)src)[i]&maskg)>>shg;\
1053 int r= (((const type*)src)[i]&maskr)>>shr;\
1054\
1055 dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<((S)-1)))>>(S);\
1056 dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<((S)-1)))>>(S);\
1057 }\
1058}\
1059static inline void name ## _half(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, const uint8_t *dummy, long width, uint32_t *unused)\
1060{\
1061 int i;\
dd68318c 1062 for (i=0; i<width; i++) {\
80704c47
KS
1063 int pix0= ((const type*)src)[2*i+0];\
1064 int pix1= ((const type*)src)[2*i+1];\
1065 int g= (pix0&~(maskr|maskb))+(pix1&~(maskr|maskb));\
1066 int b= ((pix0+pix1-g)&(maskb|(2*maskb)))>>shb;\
1067 int r= ((pix0+pix1-g)&(maskr|(2*maskr)))>>shr;\
1068 g&= maskg|(2*maskg);\
1069\
1070 g>>=shg;\
1071\
1072 dstU[i]= ((RU)*r + (GU)*g + (BU)*b + (257<<(S)))>>((S)+1);\
1073 dstV[i]= ((RV)*r + (GV)*g + (BV)*b + (257<<(S)))>>((S)+1);\
1074 }\
1075}
1076
1077BGR2UV(uint32_t, bgr32ToUV,16, 0, 0, 0xFF000000, 0xFF0000, 0xFF00, 0x00FF, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1078BGR2UV(uint32_t, rgb32ToUV, 0, 0,16, 0xFF000000, 0x00FF, 0xFF00, 0xFF0000, RU<< 8, GU , BU<< 8, RV<< 8, GV , BV<< 8, RGB2YUV_SHIFT+8)
1079BGR2UV(uint16_t, bgr16ToUV, 0, 0, 0, 0, 0x001F, 0x07E0, 0xF800, RU<<11, GU<<5, BU , RV<<11, GV<<5, BV , RGB2YUV_SHIFT+8)
1080BGR2UV(uint16_t, bgr15ToUV, 0, 0, 0, 0, 0x001F, 0x03E0, 0x7C00, RU<<10, GU<<5, BU , RV<<10, GV<<5, BV , RGB2YUV_SHIFT+7)
1081BGR2UV(uint16_t, rgb16ToUV, 0, 0, 0, 0, 0xF800, 0x07E0, 0x001F, RU , GU<<5, BU<<11, RV , GV<<5, BV<<11, RGB2YUV_SHIFT+8)
1082BGR2UV(uint16_t, rgb15ToUV, 0, 0, 0, 0, 0x7C00, 0x03E0, 0x001F, RU , GU<<5, BU<<10, RV , GV<<5, BV<<10, RGB2YUV_SHIFT+7)
1083
1084static inline void palToY(uint8_t *dst, const uint8_t *src, long width, uint32_t *pal)
1085{
1086 int i;
dd68318c 1087 for (i=0; i<width; i++) {
80704c47
KS
1088 int d= src[i];
1089
1090 dst[i]= pal[d] & 0xFF;
1091 }
1092}
1093
1094static inline void palToUV(uint8_t *dstU, uint8_t *dstV,
1095 const uint8_t *src1, const uint8_t *src2,
1096 long width, uint32_t *pal)
1097{
1098 int i;
1099 assert(src1 == src2);
dd68318c 1100 for (i=0; i<width; i++) {
80704c47
KS
1101 int p= pal[src1[i]];
1102
1103 dstU[i]= p>>8;
1104 dstV[i]= p>>16;
1105 }
1106}
1107
1108static inline void monowhite2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1109{
1110 int i, j;
dd68318c 1111 for (i=0; i<width/8; i++) {
80704c47
KS
1112 int d= ~src[i];
1113 for(j=0; j<8; j++)
1114 dst[8*i+j]= ((d>>(7-j))&1)*255;
1115 }
1116}
1117
1118static inline void monoblack2Y(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1119{
1120 int i, j;
dd68318c 1121 for (i=0; i<width/8; i++) {
80704c47
KS
1122 int d= src[i];
1123 for(j=0; j<8; j++)
1124 dst[8*i+j]= ((d>>(7-j))&1)*255;
1125 }
1126}
1127
dd409025 1128//Note: we have C, MMX, MMX2, 3DNOW versions, there is no 3DNOW+MMX2 one
7630f2e0 1129//Plain C versions
8b1a6441 1130#if ((!HAVE_MMX || !CONFIG_GPL) && !HAVE_ALTIVEC) || CONFIG_RUNTIME_CPUDETECT
726a959a
MN
1131#define COMPILE_C
1132#endif
1133
b63f641e 1134#if ARCH_PPC
57f6d52a 1135#if HAVE_ALTIVEC || CONFIG_RUNTIME_CPUDETECT
a2faa401 1136#define COMPILE_ALTIVEC
7a24ec50 1137#endif
cb82a073 1138#endif //ARCH_PPC
a2faa401 1139
b63f641e 1140#if ARCH_X86
726a959a 1141
10a7d216 1142#if ((HAVE_MMX && !HAVE_AMD3DNOW && !HAVE_MMX2) || CONFIG_RUNTIME_CPUDETECT) && CONFIG_GPL
726a959a
MN
1143#define COMPILE_MMX
1144#endif
1145
10a7d216 1146#if (HAVE_MMX2 || CONFIG_RUNTIME_CPUDETECT) && CONFIG_GPL
726a959a
MN
1147#define COMPILE_MMX2
1148#endif
1149
10a7d216 1150#if ((HAVE_AMD3DNOW && !HAVE_MMX2) || CONFIG_RUNTIME_CPUDETECT) && CONFIG_GPL
726a959a
MN
1151#define COMPILE_3DNOW
1152#endif
7a24ec50 1153#endif //ARCH_X86
726a959a 1154
94daf2e9
RP
1155#define COMPILE_TEMPLATE_MMX 0
1156#define COMPILE_TEMPLATE_MMX2 0
1157#define COMPILE_TEMPLATE_AMD3DNOW 0
1158#define COMPILE_TEMPLATE_ALTIVEC 0
726a959a
MN
1159
1160#ifdef COMPILE_C
7630f2e0
MN
1161#define RENAME(a) a ## _C
1162#include "swscale_template.c"
726a959a 1163#endif
397c035e 1164
a2faa401
RD
1165#ifdef COMPILE_ALTIVEC
1166#undef RENAME
94daf2e9
RP
1167#undef COMPILE_TEMPLATE_ALTIVEC
1168#define COMPILE_TEMPLATE_ALTIVEC 1
a2faa401
RD
1169#define RENAME(a) a ## _altivec
1170#include "swscale_template.c"
1171#endif
a2faa401 1172
b63f641e 1173#if ARCH_X86
397c035e 1174
7630f2e0 1175//MMX versions
726a959a 1176#ifdef COMPILE_MMX
7630f2e0 1177#undef RENAME
94daf2e9
RP
1178#undef COMPILE_TEMPLATE_MMX
1179#undef COMPILE_TEMPLATE_MMX2
1180#undef COMPILE_TEMPLATE_AMD3DNOW
1181#define COMPILE_TEMPLATE_MMX 1
1182#define COMPILE_TEMPLATE_MMX2 0
1183#define COMPILE_TEMPLATE_AMD3DNOW 0
7630f2e0
MN
1184#define RENAME(a) a ## _MMX
1185#include "swscale_template.c"
726a959a 1186#endif
7630f2e0
MN
1187
1188//MMX2 versions
726a959a 1189#ifdef COMPILE_MMX2
7630f2e0 1190#undef RENAME
94daf2e9
RP
1191#undef COMPILE_TEMPLATE_MMX
1192#undef COMPILE_TEMPLATE_MMX2
1193#undef COMPILE_TEMPLATE_AMD3DNOW
1194#define COMPILE_TEMPLATE_MMX 1
1195#define COMPILE_TEMPLATE_MMX2 1
1196#define COMPILE_TEMPLATE_AMD3DNOW 0
7630f2e0
MN
1197#define RENAME(a) a ## _MMX2
1198#include "swscale_template.c"
726a959a 1199#endif
7630f2e0
MN
1200
1201//3DNOW versions
726a959a 1202#ifdef COMPILE_3DNOW
7630f2e0 1203#undef RENAME
94daf2e9
RP
1204#undef COMPILE_TEMPLATE_MMX
1205#undef COMPILE_TEMPLATE_MMX2
1206#undef COMPILE_TEMPLATE_AMD3DNOW
1207#define COMPILE_TEMPLATE_MMX 1
1208#define COMPILE_TEMPLATE_MMX2 0
1209#define COMPILE_TEMPLATE_AMD3DNOW 1
7630f2e0
MN
1210#define RENAME(a) a ## _3DNow
1211#include "swscale_template.c"
726a959a 1212#endif
7630f2e0 1213
7a24ec50 1214#endif //ARCH_X86
7630f2e0 1215
a4388ebd 1216SwsFunc ff_getSwsFunc(SwsContext *c)
14d5c18c 1217{
8b1a6441 1218#if CONFIG_RUNTIME_CPUDETECT
40fa5140 1219 int flags = c->flags;
6a4970ab 1220
8b1a6441 1221#if ARCH_X86 && CONFIG_GPL
c14731d8 1222 // ordered per speed fastest first
14d5c18c 1223 if (flags & SWS_CPU_CAPS_MMX2) {
40fa5140 1224 sws_init_swScale_MMX2(c);
221b804f 1225 return swScale_MMX2;
14d5c18c 1226 } else if (flags & SWS_CPU_CAPS_3DNOW) {
40fa5140 1227 sws_init_swScale_3DNow(c);
221b804f 1228 return swScale_3DNow;
14d5c18c 1229 } else if (flags & SWS_CPU_CAPS_MMX) {
40fa5140 1230 sws_init_swScale_MMX(c);
221b804f 1231 return swScale_MMX;
730cb9da
SS
1232 } else {
1233 sws_init_swScale_C(c);
1234 return swScale_C;
14d5c18c 1235 }
730cb9da 1236
28bf81c9 1237#else
b63f641e 1238#if ARCH_PPC
14d5c18c 1239 if (flags & SWS_CPU_CAPS_ALTIVEC) {
40fa5140 1240 sws_init_swScale_altivec(c);
221b804f 1241 return swScale_altivec;
730cb9da
SS
1242 } else {
1243 sws_init_swScale_C(c);
1244 return swScale_C;
14d5c18c 1245 }
a2faa401 1246#endif
730cb9da
SS
1247 sws_init_swScale_C(c);
1248 return swScale_C;
8b1a6441 1249#endif /* ARCH_X86 && CONFIG_GPL */
10a7d216 1250#else //CONFIG_RUNTIME_CPUDETECT
94daf2e9 1251#if COMPILE_TEMPLATE_MMX2
40fa5140 1252 sws_init_swScale_MMX2(c);
221b804f 1253 return swScale_MMX2;
94daf2e9 1254#elif COMPILE_TEMPLATE_AMD3DNOW
40fa5140 1255 sws_init_swScale_3DNow(c);
221b804f 1256 return swScale_3DNow;
94daf2e9 1257#elif COMPILE_TEMPLATE_MMX
40fa5140 1258 sws_init_swScale_MMX(c);
221b804f 1259 return swScale_MMX;
94daf2e9 1260#elif COMPILE_TEMPLATE_ALTIVEC
40fa5140 1261 sws_init_swScale_altivec(c);
221b804f 1262 return swScale_altivec;
730cb9da 1263#else
80853cfc
SS
1264 sws_init_swScale_C(c);
1265 return swScale_C;
730cb9da
SS
1266#endif
1267#endif //!CONFIG_RUNTIME_CPUDETECT
31190492 1268}
7630f2e0 1269
3ece3e52 1270static int planarToNv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1271 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1272{
221b804f
DB
1273 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1274 /* Copy Y plane */
1275 if (dstStride[0]==srcStride[0] && srcStride[0] > 0)
1276 memcpy(dst, src[0], srcSliceH*dstStride[0]);
dd68318c 1277 else {
221b804f 1278 int i;
7ac40327 1279 const uint8_t *srcPtr= src[0];
221b804f 1280 uint8_t *dstPtr= dst;
dd68318c 1281 for (i=0; i<srcSliceH; i++) {
221b804f
DB
1282 memcpy(dstPtr, srcPtr, c->srcW);
1283 srcPtr+= srcStride[0];
1284 dstPtr+= dstStride[0];
1285 }
1286 }
1287 dst = dstParam[1] + dstStride[1]*srcSliceY/2;
1288 if (c->dstFormat == PIX_FMT_NV12)
30c48a0a 1289 interleaveBytes(src[1], src[2], dst, c->srcW/2, srcSliceH/2, srcStride[1], srcStride[2], dstStride[0]);
221b804f 1290 else
30c48a0a 1291 interleaveBytes(src[2], src[1], dst, c->srcW/2, srcSliceH/2, srcStride[2], srcStride[1], dstStride[0]);
221b804f
DB
1292
1293 return srcSliceH;
0d9f3d85
A
1294}
1295
3ece3e52 1296static int planarToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1297 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1298{
221b804f 1299 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
0d9f3d85 1300
30c48a0a 1301 yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
fccb9b2b 1302
221b804f 1303 return srcSliceH;
0d9f3d85
A
1304}
1305
3ece3e52 1306static int planarToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1307 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1308{
221b804f 1309 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
caeaabe7 1310
30c48a0a 1311 yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
caeaabe7 1312
221b804f 1313 return srcSliceH;
caeaabe7
AB
1314}
1315
3ece3e52 1316static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1317 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1318{
a6100f39
BC
1319 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1320
1321 yuv422ptoyuy2(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
1322
1323 return srcSliceH;
1324}
1325
3ece3e52 1326static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1327 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1328{
a6100f39
BC
1329 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
1330
1331 yuv422ptouyvy(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
1332
1333 return srcSliceH;
1334}
1335
3ece3e52 1336static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1337 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1338{
0411072e
MN
1339 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1340 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
1341 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
1342
1343 yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1344
4626ee1a
CS
1345 if (dstParam[3])
1346 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1347
0411072e
MN
1348 return srcSliceH;
1349}
1350
3ece3e52 1351static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1352 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1353{
0411072e 1354 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
72ef3dd7
MN
1355 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
1356 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
0411072e
MN
1357
1358 yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1359
1360 return srcSliceH;
1361}
1362
3ece3e52 1363static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1364 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1365{
0411072e
MN
1366 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
1367 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
1368 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
1369
1370 uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1371
4626ee1a
CS
1372 if (dstParam[3])
1373 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
1374
0411072e
MN
1375 return srcSliceH;
1376}
1377
3ece3e52 1378static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1379 int srcSliceH, uint8_t* dstParam[], int dstStride[])
1380{
0411072e 1381 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
72ef3dd7
MN
1382 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
1383 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
0411072e
MN
1384
1385 uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
1386
1387 return srcSliceH;
1388}
1389
3ece3e52 1390static int palToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
909f6594 1391 int srcSliceH, uint8_t* dst[], int dstStride[])
dd68318c 1392{
f5a2c981
CEH
1393 const enum PixelFormat srcFormat= c->srcFormat;
1394 const enum PixelFormat dstFormat= c->dstFormat;
49004617
VS
1395 void (*conv)(const uint8_t *src, uint8_t *dst, long num_pixels,
1396 const uint8_t *palette)=NULL;
1397 int i;
1398 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
a959e247 1399 const uint8_t *srcPtr= src[0];
49004617 1400
398c193d 1401 if (usePal(srcFormat)) {
e36a545d
SS
1402 switch (dstFormat) {
1403 case PIX_FMT_RGB32 : conv = palette8topacked32; break;
1404 case PIX_FMT_BGR32 : conv = palette8topacked32; break;
1405 case PIX_FMT_BGR32_1: conv = palette8topacked32; break;
1406 case PIX_FMT_RGB32_1: conv = palette8topacked32; break;
1407 case PIX_FMT_RGB24 : conv = palette8topacked24; break;
1408 case PIX_FMT_BGR24 : conv = palette8topacked24; break;
e36a545d 1409 }
c0d1dc1c 1410 }
49004617 1411
398c193d
SS
1412 if (!conv)
1413 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
1414 sws_format_name(srcFormat), sws_format_name(dstFormat));
1415 else {
e36a545d
SS
1416 for (i=0; i<srcSliceH; i++) {
1417 conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);
1418 srcPtr+= srcStride[0];
1419 dstPtr+= dstStride[0];
1420 }
398c193d 1421 }
49004617
VS
1422
1423 return srcSliceH;
1424}
1425
82e5f86b
SS
1426#define isRGBA32(x) ( \
1427 (x) == PIX_FMT_ARGB \
1428 || (x) == PIX_FMT_RGBA \
1429 || (x) == PIX_FMT_BGRA \
1430 || (x) == PIX_FMT_ABGR \
1431 )
1432
9990e426 1433/* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */
3ece3e52 1434static int rgbToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
909f6594 1435 int srcSliceH, uint8_t* dst[], int dstStride[])
dd68318c 1436{
58e4b706
CEH
1437 const enum PixelFormat srcFormat= c->srcFormat;
1438 const enum PixelFormat dstFormat= c->dstFormat;
30b61475
SS
1439 const int srcBpp= (c->srcFormatBpp + 7) >> 3;
1440 const int dstBpp= (c->dstFormatBpp + 7) >> 3;
1441 const int srcId= c->srcFormatBpp >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */
1442 const int dstId= c->dstFormatBpp >> 2;
221b804f
DB
1443 void (*conv)(const uint8_t *src, uint8_t *dst, long src_size)=NULL;
1444
82e5f86b
SS
1445#define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)
1446
1447 if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {
1448 if ( CONV_IS(ABGR, RGBA)
1449 || CONV_IS(ARGB, BGRA)
1450 || CONV_IS(BGRA, ARGB)
1451 || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;
1452 else if (CONV_IS(ABGR, ARGB)
1453 || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;
1454 else if (CONV_IS(ABGR, BGRA)
1455 || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;
1456 else if (CONV_IS(BGRA, RGBA)
1457 || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;
1458 else if (CONV_IS(BGRA, ABGR)
1459 || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;
1460 } else
221b804f 1461 /* BGR -> BGR */
3b1ee703
MN
1462 if ( (isBGRinInt(srcFormat) && isBGRinInt(dstFormat))
1463 || (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {
dd68318c 1464 switch(srcId | (dstId<<4)) {
221b804f
DB
1465 case 0x34: conv= rgb16to15; break;
1466 case 0x36: conv= rgb24to15; break;
1467 case 0x38: conv= rgb32to15; break;
1468 case 0x43: conv= rgb15to16; break;
1469 case 0x46: conv= rgb24to16; break;
1470 case 0x48: conv= rgb32to16; break;
1471 case 0x63: conv= rgb15to24; break;
1472 case 0x64: conv= rgb16to24; break;
1473 case 0x68: conv= rgb32to24; break;
1474 case 0x83: conv= rgb15to32; break;
1475 case 0x84: conv= rgb16to32; break;
1476 case 0x86: conv= rgb24to32; break;
221b804f 1477 }
3b1ee703
MN
1478 } else if ( (isBGRinInt(srcFormat) && isRGBinInt(dstFormat))
1479 || (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {
dd68318c 1480 switch(srcId | (dstId<<4)) {
221b804f
DB
1481 case 0x33: conv= rgb15tobgr15; break;
1482 case 0x34: conv= rgb16tobgr15; break;
1483 case 0x36: conv= rgb24tobgr15; break;
1484 case 0x38: conv= rgb32tobgr15; break;
1485 case 0x43: conv= rgb15tobgr16; break;
1486 case 0x44: conv= rgb16tobgr16; break;
1487 case 0x46: conv= rgb24tobgr16; break;
1488 case 0x48: conv= rgb32tobgr16; break;
1489 case 0x63: conv= rgb15tobgr24; break;
1490 case 0x64: conv= rgb16tobgr24; break;
1491 case 0x66: conv= rgb24tobgr24; break;
1492 case 0x68: conv= rgb32tobgr24; break;
1493 case 0x83: conv= rgb15tobgr32; break;
1494 case 0x84: conv= rgb16tobgr32; break;
1495 case 0x86: conv= rgb24tobgr32; break;
221b804f 1496 }
221b804f
DB
1497 }
1498
9a180390
SS
1499 if (!conv) {
1500 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
1501 sws_format_name(srcFormat), sws_format_name(dstFormat));
1502 } else {
a959e247 1503 const uint8_t *srcPtr= src[0];
82e5f86b
SS
1504 uint8_t *dstPtr= dst[0];
1505 if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) && !isRGBA32(dstFormat))
9990e426
MN
1506 srcPtr += ALT32_CORR;
1507
82e5f86b
SS
1508 if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) && !isRGBA32(srcFormat))
1509 dstPtr += ALT32_CORR;
1510
5efaf000 1511 if (dstStride[0]*srcBpp == srcStride[0]*dstBpp && srcStride[0] > 0)
82e5f86b 1512 conv(srcPtr, dstPtr + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]);
dd68318c 1513 else {
c4ca31d0 1514 int i;
82e5f86b 1515 dstPtr += dstStride[0]*srcSliceY;
c4ca31d0 1516
dd68318c 1517 for (i=0; i<srcSliceH; i++) {
c4ca31d0
BF
1518 conv(srcPtr, dstPtr, c->srcW*srcBpp);
1519 srcPtr+= srcStride[0];
1520 dstPtr+= dstStride[0];
1521 }
221b804f
DB
1522 }
1523 }
1524 return srcSliceH;
0d9f3d85
A
1525}
1526
3ece3e52 1527static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1528 int srcSliceH, uint8_t* dst[], int dstStride[])
1529{
221b804f
DB
1530 rgb24toyv12(
1531 src[0],
1532 dst[0]+ srcSliceY *dstStride[0],
1533 dst[1]+(srcSliceY>>1)*dstStride[1],
1534 dst[2]+(srcSliceY>>1)*dstStride[2],
1535 c->srcW, srcSliceH,
1536 dstStride[0], dstStride[1], srcStride[0]);
08218e6d
CS
1537 if (dst[3])
1538 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
221b804f 1539 return srcSliceH;
ec22603f
MN
1540}
1541
3ece3e52 1542static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
dd68318c
RP
1543 int srcSliceH, uint8_t* dst[], int dstStride[])
1544{
221b804f
DB
1545 int i;
1546
1547 /* copy Y */
1548 if (srcStride[0]==dstStride[0] && srcStride[0] > 0)
1549 memcpy(dst[0]+ srcSliceY*dstStride[0], src[0], srcStride[0]*srcSliceH);
dd68318c 1550 else {
a959e247 1551 const uint8_t *srcPtr= src[0];
221b804f
DB
1552 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
1553
dd68318c 1554 for (i=0; i<srcSliceH; i++) {
221b804f
DB
1555 memcpy(dstPtr, srcPtr, c->srcW);
1556 srcPtr+= srcStride[0];
1557 dstPtr+= dstStride[0];
1558 }
1559 }
1560
dd68318c 1561 if (c->dstFormat==PIX_FMT_YUV420P || c->dstFormat==PIX_FMT_YUVA420P) {
5f9ae198
VS
1562 planar2x(src[1], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,
1563 srcSliceH >> 2, srcStride[1], dstStride[1]);
1564 planar2x(src[2], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,
1565 srcSliceH >> 2, srcStride[2], dstStride[2]);
dd68318c 1566 } else {
5f9ae198
VS
1567 planar2x(src[1], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,
1568 srcSliceH >> 2, srcStride[1], dstStride[2]);
1569 planar2x(src[2], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,
1570 srcSliceH >> 2, srcStride[2], dstStride[1]);
221b804f 1571 }
08218e6d
CS
1572 if (dst[3])
1573 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
221b804f 1574 return srcSliceH;
b241cbf2
MN
1575}
1576
b6654a54 1577/* unscaled copy like stuff (assumes nearly identical formats) */
3ece3e52 1578static int packedCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
909f6594 1579 int srcSliceH, uint8_t* dst[], int dstStride[])
bc5a0444
LB
1580{
1581 if (dstStride[0]==srcStride[0] && srcStride[0] > 0)
1582 memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]);
dd68318c 1583 else {
bc5a0444 1584 int i;
a959e247 1585 const uint8_t *srcPtr= src[0];
bc5a0444
LB
1586 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
1587 int length=0;
221b804f 1588
bc5a0444
LB
1589 /* universal length finder */
1590 while(length+c->srcW <= FFABS(dstStride[0])
1591 && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW;
fcc402b1 1592 assert(length!=0);
2d35ae56 1593
dd68318c 1594 for (i=0; i<srcSliceH; i++) {
bc5a0444
LB
1595 memcpy(dstPtr, srcPtr, length);
1596 srcPtr+= srcStride[0];
1597 dstPtr+= dstStride[0];
221b804f 1598 }
bc5a0444 1599 }
2d35ae56
LB
1600 return srcSliceH;
1601}
bc5a0444 1602
3ece3e52 1603static int planarCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
909f6594 1604 int srcSliceH, uint8_t* dst[], int dstStride[])
2d35ae56 1605{
61a3f379 1606 int plane, i, j;
dd68318c 1607 for (plane=0; plane<4; plane++) {
6268f55b
CS
1608 int length= (plane==0 || plane==3) ? c->srcW : -((-c->srcW )>>c->chrDstHSubSample);
1609 int y= (plane==0 || plane==3) ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample);
1610 int height= (plane==0 || plane==3) ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample);
a959e247 1611 const uint8_t *srcPtr= src[plane];
61a3f379 1612 uint8_t *dstPtr= dst[plane] + dstStride[plane]*y;
2d35ae56 1613
20ddf5a8 1614 if (!dst[plane]) continue;
6f348086
RD
1615 // ignore palette for GRAY8
1616 if (plane == 1 && !dst[2]) continue;
dd68318c 1617 if (!src[plane] || (plane == 1 && !src[2])) {
61a3f379
MN
1618 if(is16BPS(c->dstFormat))
1619 length*=2;
6268f55b 1620 fillPlane(dst[plane], dstStride[plane], length, height, y, (plane==3) ? 255 : 128);
dd68318c
RP
1621 } else {
1622 if(is16BPS(c->srcFormat) && !is16BPS(c->dstFormat)) {
61a3f379 1623 if (!isBE(c->srcFormat)) srcPtr++;
dd68318c 1624 for (i=0; i<height; i++) {
61a3f379
MN
1625 for (j=0; j<length; j++) dstPtr[j] = srcPtr[j<<1];
1626 srcPtr+= srcStride[plane];
1627 dstPtr+= dstStride[plane];
1628 }
dd68318c
RP
1629 } else if(!is16BPS(c->srcFormat) && is16BPS(c->dstFormat)) {
1630 for (i=0; i<height; i++) {
1631 for (j=0; j<length; j++) {
61a3f379
MN
1632 dstPtr[ j<<1 ] = srcPtr[j];
1633 dstPtr[(j<<1)+1] = srcPtr[j];
1634 }
1635 srcPtr+= srcStride[plane];
1636 dstPtr+= dstStride[plane];
1637 }
dd68318c
RP
1638 } else if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat)
1639 && isBE(c->srcFormat) != isBE(c->dstFormat)) {
61a3f379 1640
dd68318c 1641 for (i=0; i<height; i++) {
61a3f379 1642 for (j=0; j<length; j++)
5090d8e1 1643 ((uint16_t*)dstPtr)[j] = bswap_16(((const uint16_t*)srcPtr)[j]);
61a3f379
MN
1644 srcPtr+= srcStride[plane];
1645 dstPtr+= dstStride[plane];
1646 }
1647 } else if (dstStride[plane]==srcStride[plane] && srcStride[plane] > 0)
bc5a0444 1648 memcpy(dst[plane] + dstStride[plane]*y, src[plane], height*dstStride[plane]);
dd68318c 1649 else {
61a3f379
MN
1650 if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat))
1651 length*=2;
dd68318c 1652 for (i=0; i<height; i++) {
bc5a0444
LB
1653 memcpy(dstPtr, srcPtr, length);
1654 srcPtr+= srcStride[plane];
1655 dstPtr+= dstStride[plane];
221b804f
DB
1656 }
1657 }
1658 }
bc5a0444 1659 }
221b804f 1660 return srcSliceH;
37079906 1661}
28bf81c9 1662
a4388ebd 1663int ff_hardcodedcpuflags(void)
dd68318c 1664{
a4388ebd 1665 int flags = 0;
94daf2e9 1666#if COMPILE_TEMPLATE_MMX2
221b804f 1667 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_MMX2;
94daf2e9 1668#elif COMPILE_TEMPLATE_AMD3DNOW
221b804f 1669 flags |= SWS_CPU_CAPS_MMX|SWS_CPU_CAPS_3DNOW;
94daf2e9 1670#elif COMPILE_TEMPLATE_MMX
221b804f 1671 flags |= SWS_CPU_CAPS_MMX;
94daf2e9 1672#elif COMPILE_TEMPLATE_ALTIVEC
221b804f 1673 flags |= SWS_CPU_CAPS_ALTIVEC;
b63f641e 1674#elif ARCH_BFIN
d3f3eea9 1675 flags |= SWS_CPU_CAPS_BFIN;
516b1f82 1676#endif
a4388ebd
RP
1677 return flags;
1678}
1679
1680void ff_get_unscaled_swscale(SwsContext *c)
1681{
1682 const enum PixelFormat srcFormat = c->srcFormat;
1683 const enum PixelFormat dstFormat = c->dstFormat;
1684 const int flags = c->flags;
1685 const int dstH = c->dstH;
1686 int needsDither;
221b804f 1687
60222557 1688 needsDither= isAnyRGB(dstFormat)
aa2ead82 1689 && c->dstFormatBpp < 24
60222557 1690 && (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));
221b804f 1691
59398ea9
RP
1692 /* yv12_to_nv12 */
1693 if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {
3ece3e52 1694 c->swScale= planarToNv12Wrapper;
59398ea9
RP
1695 }
1696 /* yuv2bgr */
60222557 1697 if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P || srcFormat==PIX_FMT_YUVA420P) && isAnyRGB(dstFormat)
59398ea9
RP
1698 && !(flags & SWS_ACCURATE_RND) && !(dstH&1)) {
1699 c->swScale= ff_yuv2rgb_get_func_ptr(c);
1700 }
6a4970ab 1701
59398ea9 1702 if (srcFormat==PIX_FMT_YUV410P && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_BITEXACT)) {
3ece3e52 1703 c->swScale= yvu9ToYv12Wrapper;
59398ea9
RP
1704 }
1705
1706 /* bgr24toYV12 */
1707 if (srcFormat==PIX_FMT_BGR24 && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_ACCURATE_RND))
3ece3e52 1708 c->swScale= bgr24ToYv12Wrapper;
59398ea9
RP
1709
1710 /* RGB/BGR -> RGB/BGR (no dither needed forms) */
60222557
MN
1711 if ( isAnyRGB(srcFormat)
1712 && isAnyRGB(dstFormat)
59398ea9
RP
1713 && srcFormat != PIX_FMT_BGR8 && dstFormat != PIX_FMT_BGR8
1714 && srcFormat != PIX_FMT_RGB8 && dstFormat != PIX_FMT_RGB8
1715 && srcFormat != PIX_FMT_BGR4 && dstFormat != PIX_FMT_BGR4
1716 && srcFormat != PIX_FMT_RGB4 && dstFormat != PIX_FMT_RGB4
1717 && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE
1718 && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE
1719 && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK
1720 && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE
59398ea9
RP
1721 && srcFormat != PIX_FMT_RGB48LE && dstFormat != PIX_FMT_RGB48LE
1722 && srcFormat != PIX_FMT_RGB48BE && dstFormat != PIX_FMT_RGB48BE
1723 && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))
3ece3e52 1724 c->swScale= rgbToRgbWrapper;
59398ea9
RP
1725
1726 if ((usePal(srcFormat) && (
1727 dstFormat == PIX_FMT_RGB32 ||
1728 dstFormat == PIX_FMT_RGB32_1 ||
1729 dstFormat == PIX_FMT_RGB24 ||
1730 dstFormat == PIX_FMT_BGR32 ||
1731 dstFormat == PIX_FMT_BGR32_1 ||
1732 dstFormat == PIX_FMT_BGR24)))
3ece3e52 1733 c->swScale= palToRgbWrapper;
59398ea9
RP
1734
1735 if (srcFormat == PIX_FMT_YUV422P) {
1736 if (dstFormat == PIX_FMT_YUYV422)
3ece3e52 1737 c->swScale= yuv422pToYuy2Wrapper;
59398ea9 1738 else if (dstFormat == PIX_FMT_UYVY422)
3ece3e52 1739 c->swScale= yuv422pToUyvyWrapper;
59398ea9 1740 }
221b804f 1741
59398ea9
RP
1742 /* LQ converters if -sws 0 or -sws 4*/
1743 if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {
1744 /* yv12_to_yuy2 */
1745 if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {
a6100f39 1746 if (dstFormat == PIX_FMT_YUYV422)
3ece3e52 1747 c->swScale= planarToYuy2Wrapper;
a6100f39 1748 else if (dstFormat == PIX_FMT_UYVY422)
3ece3e52 1749 c->swScale= planarToUyvyWrapper;
a6100f39 1750 }
59398ea9
RP
1751 }
1752 if(srcFormat == PIX_FMT_YUYV422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
3ece3e52 1753 c->swScale= yuyvToYuv420Wrapper;
59398ea9 1754 if(srcFormat == PIX_FMT_UYVY422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
3ece3e52 1755 c->swScale= uyvyToYuv420Wrapper;
59398ea9 1756 if(srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)
3ece3e52 1757 c->swScale= yuyvToYuv422Wrapper;
59398ea9 1758 if(srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)
3ece3e52 1759 c->swScale= uyvyToYuv422Wrapper;
ec22603f 1760
6634d0ef 1761#ifdef COMPILE_ALTIVEC
59398ea9
RP
1762 if ((c->flags & SWS_CPU_CAPS_ALTIVEC) &&
1763 !(c->flags & SWS_BITEXACT) &&
1764 srcFormat == PIX_FMT_YUV420P) {
1765 // unscaled YV12 -> packed YUV, we want speed
1766 if (dstFormat == PIX_FMT_YUYV422)
1767 c->swScale= yv12toyuy2_unscaled_altivec;
1768 else if (dstFormat == PIX_FMT_UYVY422)
1769 c->swScale= yv12touyvy_unscaled_altivec;
1770 }
b71cf33c
RD
1771#endif
1772
59398ea9
RP
1773 /* simple copy */
1774 if ( srcFormat == dstFormat
1775 || (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P)
1776 || (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P)
1777 || (isPlanarYUV(srcFormat) && isGray(dstFormat))
1778 || (isPlanarYUV(dstFormat) && isGray(srcFormat))
1779 || (isGray(dstFormat) && isGray(srcFormat))
1780 || (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat)
1781 && c->chrDstHSubSample == c->chrSrcHSubSample
1782 && c->chrDstVSubSample == c->chrSrcVSubSample
1783 && dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21
1784 && srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))
1785 {
1786 if (isPacked(c->srcFormat))
3ece3e52 1787 c->swScale= packedCopyWrapper;
59398ea9 1788 else /* Planar YUV or gray */
3ece3e52 1789 c->swScale= planarCopyWrapper;
59398ea9 1790 }
b63f641e 1791#if ARCH_BFIN
59398ea9
RP
1792 if (flags & SWS_CPU_CAPS_BFIN)
1793 ff_bfin_get_unscaled_swscale (c);
1ebbfe15 1794#endif
28bf81c9
MN
1795}
1796
a959e247 1797static void reset_ptr(const uint8_t* src[], int format)
dd68318c 1798{
b87fae9f
MN
1799 if(!isALPHA(format))
1800 src[3]=NULL;
dd68318c 1801 if(!isPlanarYUV(format)) {
b87fae9f 1802 src[3]=src[2]=NULL;
3e906f4c
SS
1803
1804 if (!usePal(format))
b87fae9f
MN
1805 src[1]= NULL;
1806 }
1807}
1808
28bf81c9 1809/**
56b69633 1810 * swscale wrapper, so we don't need to export the SwsContext.
f40c7dbb 1811 * Assumes planar YUV to be in YUV order instead of YVU.
fccb9b2b 1812 */
01cc47dc
DC
1813int sws_scale(SwsContext *c, const uint8_t* const src[], const int srcStride[], int srcSliceY,
1814 int srcSliceH, uint8_t* const dst[], const int dstStride[])
dd68318c 1815{
221b804f 1816 int i;
a959e247 1817 const uint8_t* src2[4]= {src[0], src[1], src[2], src[3]};
b87fae9f 1818 uint8_t* dst2[4]= {dst[0], dst[1], dst[2], dst[3]};
14623020 1819
1b168a42
RD
1820 // do not mess up sliceDir if we have a "trailing" 0-size slice
1821 if (srcSliceH == 0)
1822 return 0;
1823
221b804f 1824 if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {
3f0bc115 1825 av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");
221b804f
DB
1826 return 0;
1827 }
1828 if (c->sliceDir == 0) {
1829 if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;
1830 }
1831
dd68318c
RP
1832 if (usePal(c->srcFormat)) {
1833 for (i=0; i<256; i++) {
14623020 1834 int p, r, g, b,y,u,v;
dd68318c 1835 if(c->srcFormat == PIX_FMT_PAL8) {
e922e34c 1836 p=((const uint32_t*)(src[1]))[i];
14623020
MN
1837 r= (p>>16)&0xFF;
1838 g= (p>> 8)&0xFF;
1839 b= p &0xFF;
dd68318c 1840 } else if(c->srcFormat == PIX_FMT_RGB8) {
14623020
MN
1841 r= (i>>5 )*36;
1842 g= ((i>>2)&7)*36;
1843 b= (i&3 )*85;
dd68318c 1844 } else if(c->srcFormat == PIX_FMT_BGR8) {
14623020
MN
1845 b= (i>>6 )*85;
1846 g= ((i>>3)&7)*36;
1847 r= (i&7 )*36;
dd68318c 1848 } else if(c->srcFormat == PIX_FMT_RGB4_BYTE) {
14623020
MN
1849 r= (i>>3 )*255;
1850 g= ((i>>1)&3)*85;
1851 b= (i&1 )*255;
dd68318c 1852 } else {
75add6fb 1853 assert(c->srcFormat == PIX_FMT_BGR4_BYTE);
14623020
MN
1854 b= (i>>3 )*255;
1855 g= ((i>>1)&3)*85;
1856 r= (i&1 )*255;
1857 }
cbcb408f
MN
1858 y= av_clip_uint8((RY*r + GY*g + BY*b + ( 33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
1859 u= av_clip_uint8((RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
1860 v= av_clip_uint8((RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
49004617 1861 c->pal_yuv[i]= y + (u<<8) + (v<<16);
522ce957 1862
522ce957
VS
1863 switch(c->dstFormat) {
1864 case PIX_FMT_BGR32:
a898cdc9 1865#if !HAVE_BIGENDIAN
522ce957
VS
1866 case PIX_FMT_RGB24:
1867#endif
1868 c->pal_rgb[i]= r + (g<<8) + (b<<16);
1869 break;
1870 case PIX_FMT_BGR32_1:
a898cdc9 1871#if HAVE_BIGENDIAN
522ce957
VS
1872 case PIX_FMT_BGR24:
1873#endif
1874 c->pal_rgb[i]= (r + (g<<8) + (b<<16)) << 8;
1875 break;
1876 case PIX_FMT_RGB32_1:
a898cdc9 1877#if HAVE_BIGENDIAN
522ce957
VS
1878 case PIX_FMT_RGB24:
1879#endif
1880 c->pal_rgb[i]= (b + (g<<8) + (r<<16)) << 8;
1881 break;
1882 case PIX_FMT_RGB32:
a898cdc9 1883#if !HAVE_BIGENDIAN
522ce957
VS
1884 case PIX_FMT_BGR24:
1885#endif
1886 default:
1887 c->pal_rgb[i]= b + (g<<8) + (r<<16);
1888 }
21c08a3f 1889 }
221b804f 1890 }
21c08a3f 1891
221b804f
DB
1892 // copy strides, so they can safely be modified
1893 if (c->sliceDir == 1) {
1894 // slices go from top to bottom
6208d676
CS
1895 int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2], srcStride[3]};
1896 int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2], dstStride[3]};
b87fae9f
MN
1897
1898 reset_ptr(src2, c->srcFormat);
a959e247 1899 reset_ptr((const uint8_t**)dst2, c->dstFormat);
b87fae9f 1900
749fac31
RP
1901 /* reset slice direction at end of frame */
1902 if (srcSliceY + srcSliceH == c->srcH)
1903 c->sliceDir = 0;
1904
b87fae9f 1905 return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2, dstStride2);
221b804f
DB
1906 } else {
1907 // slices go from bottom to top => we flip the image internally
6208d676
CS
1908 int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2], -srcStride[3]};
1909 int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2], -dstStride[3]};
221b804f
DB
1910
1911 src2[0] += (srcSliceH-1)*srcStride[0];
49004617 1912 if (!usePal(c->srcFormat))
221b804f
DB
1913 src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1];
1914 src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2];
6208d676 1915 src2[3] += (srcSliceH-1)*srcStride[3];
b87fae9f
MN
1916 dst2[0] += ( c->dstH -1)*dstStride[0];
1917 dst2[1] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1];
1918 dst2[2] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2];
1919 dst2[3] += ( c->dstH -1)*dstStride[3];
1920
1921 reset_ptr(src2, c->srcFormat);
a959e247 1922 reset_ptr((const uint8_t**)dst2, c->dstFormat);
221b804f 1923
749fac31
RP
1924 /* reset slice direction at end of frame */
1925 if (!srcSliceY)
1926 c->sliceDir = 0;
1927
221b804f
DB
1928 return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2);
1929 }
fccb9b2b
MN
1930}
1931
c099766f 1932#if LIBSWSCALE_VERSION_MAJOR < 1
f28cc577 1933int sws_scale_ordered(SwsContext *c, const uint8_t* const src[], int srcStride[], int srcSliceY,
dd68318c
RP
1934 int srcSliceH, uint8_t* dst[], int dstStride[])
1935{
221b804f 1936 return sws_scale(c, src, srcStride, srcSliceY, srcSliceH, dst, dstStride);
d4e24275 1937}
c099766f 1938#endif