2 * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of Libav.
6 * Libav 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 * Libav 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 Libav; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
28 #include "swscale_internal.h"
30 #include "libavutil/intreadwrite.h"
31 #include "libavutil/cpu.h"
32 #include "libavutil/avutil.h"
33 #include "libavutil/mathematics.h"
34 #include "libavutil/bswap.h"
35 #include "libavutil/pixdesc.h"
37 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_1)[8][8] = {
38 { 0, 1, 0, 1, 0, 1, 0, 1,},
39 { 1, 0, 1, 0, 1, 0, 1, 0,},
40 { 0, 1, 0, 1, 0, 1, 0, 1,},
41 { 1, 0, 1, 0, 1, 0, 1, 0,},
42 { 0, 1, 0, 1, 0, 1, 0, 1,},
43 { 1, 0, 1, 0, 1, 0, 1, 0,},
44 { 0, 1, 0, 1, 0, 1, 0, 1,},
45 { 1, 0, 1, 0, 1, 0, 1, 0,},
47 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_3)[8][8] = {
48 { 1, 2, 1, 2, 1, 2, 1, 2,},
49 { 3, 0, 3, 0, 3, 0, 3, 0,},
50 { 1, 2, 1, 2, 1, 2, 1, 2,},
51 { 3, 0, 3, 0, 3, 0, 3, 0,},
52 { 1, 2, 1, 2, 1, 2, 1, 2,},
53 { 3, 0, 3, 0, 3, 0, 3, 0,},
54 { 1, 2, 1, 2, 1, 2, 1, 2,},
55 { 3, 0, 3, 0, 3, 0, 3, 0,},
57 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_64)[8][8] = {
58 { 18, 34, 30, 46, 17, 33, 29, 45,},
59 { 50, 2, 62, 14, 49, 1, 61, 13,},
60 { 26, 42, 22, 38, 25, 41, 21, 37,},
61 { 58, 10, 54, 6, 57, 9, 53, 5,},
62 { 16, 32, 28, 44, 19, 35, 31, 47,},
63 { 48, 0, 60, 12, 51, 3, 63, 15,},
64 { 24, 40, 20, 36, 27, 43, 23, 39,},
65 { 56, 8, 52, 4, 59, 11, 55, 7,},
67 extern const uint8_t dither_8x8_128[8][8];
68 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_256)[8][8] = {
69 { 72, 136, 120, 184, 68, 132, 116, 180,},
70 { 200, 8, 248, 56, 196, 4, 244, 52,},
71 { 104, 168, 88, 152, 100, 164, 84, 148,},
72 { 232, 40, 216, 24, 228, 36, 212, 20,},
73 { 64, 128, 102, 176, 76, 140, 124, 188,},
74 { 192, 0, 240, 48, 204, 12, 252, 60,},
75 { 96, 160, 80, 144, 108, 172, 92, 156,},
76 { 224, 32, 208, 16, 236, 44, 220, 28,},
79 #define RGB2YUV_SHIFT 15
80 #define BY ( (int)(0.114*219/255*(1<<RGB2YUV_SHIFT)+0.5))
81 #define BV (-(int)(0.081*224/255*(1<<RGB2YUV_SHIFT)+0.5))
82 #define BU ( (int)(0.500*224/255*(1<<RGB2YUV_SHIFT)+0.5))
83 #define GY ( (int)(0.587*219/255*(1<<RGB2YUV_SHIFT)+0.5))
84 #define GV (-(int)(0.419*224/255*(1<<RGB2YUV_SHIFT)+0.5))
85 #define GU (-(int)(0.331*224/255*(1<<RGB2YUV_SHIFT)+0.5))
86 #define RY ( (int)(0.299*219/255*(1<<RGB2YUV_SHIFT)+0.5))
87 #define RV ( (int)(0.500*224/255*(1<<RGB2YUV_SHIFT)+0.5))
88 #define RU (-(int)(0.169*224/255*(1<<RGB2YUV_SHIFT)+0.5))
90 static void fillPlane(uint8_t* plane, int stride, int width, int height, int y, uint8_t val)
93 uint8_t *ptr = plane + stride*y;
94 for (i=0; i<height; i++) {
95 memset(ptr, val, width);
100 static void copyPlane(const uint8_t *src, int srcStride,
101 int srcSliceY, int srcSliceH, int width,
102 uint8_t *dst, int dstStride)
104 dst += dstStride * srcSliceY;
105 if (dstStride == srcStride && srcStride > 0) {
106 memcpy(dst, src, srcSliceH * dstStride);
109 for (i=0; i<srcSliceH; i++) {
110 memcpy(dst, src, width);
117 static int planarToNv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
118 int srcSliceH, uint8_t* dstParam[], int dstStride[])
120 uint8_t *dst = dstParam[1] + dstStride[1]*srcSliceY/2;
122 copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
123 dstParam[0], dstStride[0]);
125 if (c->dstFormat == PIX_FMT_NV12)
126 interleaveBytes(src[1], src[2], dst, c->srcW/2, srcSliceH/2, srcStride[1], srcStride[2], dstStride[0]);
128 interleaveBytes(src[2], src[1], dst, c->srcW/2, srcSliceH/2, srcStride[2], srcStride[1], dstStride[0]);
133 static int planarToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
134 int srcSliceH, uint8_t* dstParam[], int dstStride[])
136 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
138 yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
143 static int planarToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
144 int srcSliceH, uint8_t* dstParam[], int dstStride[])
146 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
148 yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
153 static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
154 int srcSliceH, uint8_t* dstParam[], int dstStride[])
156 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
158 yuv422ptoyuy2(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
163 static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
164 int srcSliceH, uint8_t* dstParam[], int dstStride[])
166 uint8_t *dst=dstParam[0] + dstStride[0]*srcSliceY;
168 yuv422ptouyvy(src[0],src[1],src[2],dst,c->srcW,srcSliceH,srcStride[0],srcStride[1],dstStride[0]);
173 static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
174 int srcSliceH, uint8_t* dstParam[], int dstStride[])
176 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
177 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
178 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
180 yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
183 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
188 static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
189 int srcSliceH, uint8_t* dstParam[], int dstStride[])
191 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
192 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
193 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
195 yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
200 static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
201 int srcSliceH, uint8_t* dstParam[], int dstStride[])
203 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
204 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY/2;
205 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY/2;
207 uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
210 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
215 static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
216 int srcSliceH, uint8_t* dstParam[], int dstStride[])
218 uint8_t *ydst=dstParam[0] + dstStride[0]*srcSliceY;
219 uint8_t *udst=dstParam[1] + dstStride[1]*srcSliceY;
220 uint8_t *vdst=dstParam[2] + dstStride[2]*srcSliceY;
222 uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0], dstStride[1], srcStride[0]);
227 static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)
230 for (i=0; i<num_pixels; i++)
231 ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | (src[(i<<1)+1] << 24);
234 static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)
238 for (i=0; i<num_pixels; i++)
239 ((uint32_t *) dst)[i] = ((const uint32_t *)palette)[src[i<<1]] | src[(i<<1)+1];
242 static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)
246 for (i=0; i<num_pixels; i++) {
248 dst[0]= palette[src[i<<1]*4+0];
249 dst[1]= palette[src[i<<1]*4+1];
250 dst[2]= palette[src[i<<1]*4+2];
255 static int palToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
256 int srcSliceH, uint8_t* dst[], int dstStride[])
258 const enum PixelFormat srcFormat= c->srcFormat;
259 const enum PixelFormat dstFormat= c->dstFormat;
260 void (*conv)(const uint8_t *src, uint8_t *dst, int num_pixels,
261 const uint8_t *palette)=NULL;
263 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
264 const uint8_t *srcPtr= src[0];
266 if (srcFormat == PIX_FMT_Y400A) {
268 case PIX_FMT_RGB32 : conv = gray8aToPacked32; break;
269 case PIX_FMT_BGR32 : conv = gray8aToPacked32; break;
270 case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;
271 case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;
272 case PIX_FMT_RGB24 : conv = gray8aToPacked24; break;
273 case PIX_FMT_BGR24 : conv = gray8aToPacked24; break;
275 } else if (usePal(srcFormat)) {
277 case PIX_FMT_RGB32 : conv = sws_convertPalette8ToPacked32; break;
278 case PIX_FMT_BGR32 : conv = sws_convertPalette8ToPacked32; break;
279 case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;
280 case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;
281 case PIX_FMT_RGB24 : conv = sws_convertPalette8ToPacked24; break;
282 case PIX_FMT_BGR24 : conv = sws_convertPalette8ToPacked24; break;
287 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
288 sws_format_name(srcFormat), sws_format_name(dstFormat));
290 for (i=0; i<srcSliceH; i++) {
291 conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);
292 srcPtr+= srcStride[0];
293 dstPtr+= dstStride[0];
300 #define isRGBA32(x) ( \
301 (x) == PIX_FMT_ARGB \
302 || (x) == PIX_FMT_RGBA \
303 || (x) == PIX_FMT_BGRA \
304 || (x) == PIX_FMT_ABGR \
307 /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */
308 static int rgbToRgbWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
309 int srcSliceH, uint8_t* dst[], int dstStride[])
311 const enum PixelFormat srcFormat= c->srcFormat;
312 const enum PixelFormat dstFormat= c->dstFormat;
313 const int srcBpp= (c->srcFormatBpp + 7) >> 3;
314 const int dstBpp= (c->dstFormatBpp + 7) >> 3;
315 const int srcId= c->srcFormatBpp >> 2; /* 1:0, 4:1, 8:2, 15:3, 16:4, 24:6, 32:8 */
316 const int dstId= c->dstFormatBpp >> 2;
317 void (*conv)(const uint8_t *src, uint8_t *dst, int src_size)=NULL;
319 #define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)
321 if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {
322 if ( CONV_IS(ABGR, RGBA)
323 || CONV_IS(ARGB, BGRA)
324 || CONV_IS(BGRA, ARGB)
325 || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;
326 else if (CONV_IS(ABGR, ARGB)
327 || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;
328 else if (CONV_IS(ABGR, BGRA)
329 || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;
330 else if (CONV_IS(BGRA, RGBA)
331 || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;
332 else if (CONV_IS(BGRA, ABGR)
333 || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;
336 if ( (isBGRinInt(srcFormat) && isBGRinInt(dstFormat))
337 || (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {
338 switch(srcId | (dstId<<4)) {
339 case 0x34: conv= rgb16to15; break;
340 case 0x36: conv= rgb24to15; break;
341 case 0x38: conv= rgb32to15; break;
342 case 0x43: conv= rgb15to16; break;
343 case 0x46: conv= rgb24to16; break;
344 case 0x48: conv= rgb32to16; break;
345 case 0x63: conv= rgb15to24; break;
346 case 0x64: conv= rgb16to24; break;
347 case 0x68: conv= rgb32to24; break;
348 case 0x83: conv= rgb15to32; break;
349 case 0x84: conv= rgb16to32; break;
350 case 0x86: conv= rgb24to32; break;
352 } else if ( (isBGRinInt(srcFormat) && isRGBinInt(dstFormat))
353 || (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {
354 switch(srcId | (dstId<<4)) {
355 case 0x33: conv= rgb15tobgr15; break;
356 case 0x34: conv= rgb16tobgr15; break;
357 case 0x36: conv= rgb24tobgr15; break;
358 case 0x38: conv= rgb32tobgr15; break;
359 case 0x43: conv= rgb15tobgr16; break;
360 case 0x44: conv= rgb16tobgr16; break;
361 case 0x46: conv= rgb24tobgr16; break;
362 case 0x48: conv= rgb32tobgr16; break;
363 case 0x63: conv= rgb15tobgr24; break;
364 case 0x64: conv= rgb16tobgr24; break;
365 case 0x66: conv= rgb24tobgr24; break;
366 case 0x68: conv= rgb32tobgr24; break;
367 case 0x83: conv= rgb15tobgr32; break;
368 case 0x84: conv= rgb16tobgr32; break;
369 case 0x86: conv= rgb24tobgr32; break;
374 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
375 sws_format_name(srcFormat), sws_format_name(dstFormat));
377 const uint8_t *srcPtr= src[0];
378 uint8_t *dstPtr= dst[0];
379 if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) && !isRGBA32(dstFormat))
380 srcPtr += ALT32_CORR;
382 if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) && !isRGBA32(srcFormat))
383 dstPtr += ALT32_CORR;
385 if (dstStride[0]*srcBpp == srcStride[0]*dstBpp && srcStride[0] > 0)
386 conv(srcPtr, dstPtr + dstStride[0]*srcSliceY, srcSliceH*srcStride[0]);
389 dstPtr += dstStride[0]*srcSliceY;
391 for (i=0; i<srcSliceH; i++) {
392 conv(srcPtr, dstPtr, c->srcW*srcBpp);
393 srcPtr+= srcStride[0];
394 dstPtr+= dstStride[0];
401 static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
402 int srcSliceH, uint8_t* dst[], int dstStride[])
406 dst[0]+ srcSliceY *dstStride[0],
407 dst[1]+(srcSliceY>>1)*dstStride[1],
408 dst[2]+(srcSliceY>>1)*dstStride[2],
410 dstStride[0], dstStride[1], srcStride[0]);
412 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
416 static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
417 int srcSliceH, uint8_t* dst[], int dstStride[])
419 copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
420 dst[0], dstStride[0]);
422 planar2x(src[1], dst[1] + dstStride[1]*(srcSliceY >> 1), c->chrSrcW,
423 srcSliceH >> 2, srcStride[1], dstStride[1]);
424 planar2x(src[2], dst[2] + dstStride[2]*(srcSliceY >> 1), c->chrSrcW,
425 srcSliceH >> 2, srcStride[2], dstStride[2]);
427 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
431 /* unscaled copy like stuff (assumes nearly identical formats) */
432 static int packedCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
433 int srcSliceH, uint8_t* dst[], int dstStride[])
435 if (dstStride[0]==srcStride[0] && srcStride[0] > 0)
436 memcpy(dst[0] + dstStride[0]*srcSliceY, src[0], srcSliceH*dstStride[0]);
439 const uint8_t *srcPtr= src[0];
440 uint8_t *dstPtr= dst[0] + dstStride[0]*srcSliceY;
443 /* universal length finder */
444 while(length+c->srcW <= FFABS(dstStride[0])
445 && length+c->srcW <= FFABS(srcStride[0])) length+= c->srcW;
448 for (i=0; i<srcSliceH; i++) {
449 memcpy(dstPtr, srcPtr, length);
450 srcPtr+= srcStride[0];
451 dstPtr+= dstStride[0];
457 #define DITHER_COPY(dst, dstStride, wfunc, src, srcStride, rfunc, dithers, shift) \
458 for (i = 0; i < height; i++) { \
459 const uint8_t *dither = dithers[i & 7]; \
460 for (j = 0; j < length - 7; j += 8) { \
461 wfunc(&dst[j + 0], (rfunc(&src[j + 0]) + dither[0]) >> shift); \
462 wfunc(&dst[j + 1], (rfunc(&src[j + 1]) + dither[1]) >> shift); \
463 wfunc(&dst[j + 2], (rfunc(&src[j + 2]) + dither[2]) >> shift); \
464 wfunc(&dst[j + 3], (rfunc(&src[j + 3]) + dither[3]) >> shift); \
465 wfunc(&dst[j + 4], (rfunc(&src[j + 4]) + dither[4]) >> shift); \
466 wfunc(&dst[j + 5], (rfunc(&src[j + 5]) + dither[5]) >> shift); \
467 wfunc(&dst[j + 6], (rfunc(&src[j + 6]) + dither[6]) >> shift); \
468 wfunc(&dst[j + 7], (rfunc(&src[j + 7]) + dither[7]) >> shift); \
470 for (; j < length; j++) \
471 wfunc(&dst[j], (rfunc(&src[j]) + dither[j & 7]) >> shift); \
476 static int planarCopyWrapper(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
477 int srcSliceH, uint8_t* dst[], int dstStride[])
480 for (plane=0; plane<4; plane++) {
481 int length= (plane==0 || plane==3) ? c->srcW : -((-c->srcW )>>c->chrDstHSubSample);
482 int y= (plane==0 || plane==3) ? srcSliceY: -((-srcSliceY)>>c->chrDstVSubSample);
483 int height= (plane==0 || plane==3) ? srcSliceH: -((-srcSliceH)>>c->chrDstVSubSample);
484 const uint8_t *srcPtr= src[plane];
485 uint8_t *dstPtr= dst[plane] + dstStride[plane]*y;
487 if (!dst[plane]) continue;
488 // ignore palette for GRAY8
489 if (plane == 1 && !dst[2]) continue;
490 if (!src[plane] || (plane == 1 && !src[2])) {
491 if(is16BPS(c->dstFormat))
493 fillPlane(dst[plane], dstStride[plane], length, height, y, (plane==3) ? 255 : 128);
495 if(is9_OR_10BPS(c->srcFormat)) {
496 const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1+1;
497 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1+1;
498 const uint16_t *srcPtr2 = (const uint16_t*)srcPtr;
500 if (is16BPS(c->dstFormat)) {
501 uint16_t *dstPtr2 = (uint16_t*)dstPtr;
502 #define COPY9_OR_10TO16(rfunc, wfunc) \
503 for (i = 0; i < height; i++) { \
504 for (j = 0; j < length; j++) { \
505 int srcpx = rfunc(&srcPtr2[j]); \
506 wfunc(&dstPtr2[j], (srcpx<<(16-src_depth)) | (srcpx>>(2*src_depth-16))); \
508 dstPtr2 += dstStride[plane]/2; \
509 srcPtr2 += srcStride[plane]/2; \
511 if (isBE(c->dstFormat)) {
512 if (isBE(c->srcFormat)) {
513 COPY9_OR_10TO16(AV_RB16, AV_WB16);
515 COPY9_OR_10TO16(AV_RL16, AV_WB16);
518 if (isBE(c->srcFormat)) {
519 COPY9_OR_10TO16(AV_RB16, AV_WL16);
521 COPY9_OR_10TO16(AV_RL16, AV_WL16);
524 } else if (is9_OR_10BPS(c->dstFormat)) {
525 uint16_t *dstPtr2 = (uint16_t*)dstPtr;
526 #define COPY9_OR_10TO9_OR_10(loop) \
527 for (i = 0; i < height; i++) { \
528 for (j = 0; j < length; j++) { \
531 dstPtr2 += dstStride[plane]/2; \
532 srcPtr2 += srcStride[plane]/2; \
534 #define COPY9_OR_10TO9_OR_10_2(rfunc, wfunc) \
535 if (dst_depth > src_depth) { \
536 COPY9_OR_10TO9_OR_10(int srcpx = rfunc(&srcPtr2[j]); \
537 wfunc(&dstPtr2[j], (srcpx << 1) | (srcpx >> 9))); \
538 } else if (dst_depth < src_depth) { \
539 DITHER_COPY(dstPtr2, dstStride[plane]/2, wfunc, \
540 srcPtr2, srcStride[plane]/2, rfunc, \
543 COPY9_OR_10TO9_OR_10(wfunc(&dstPtr2[j], rfunc(&srcPtr2[j]))); \
545 if (isBE(c->dstFormat)) {
546 if (isBE(c->srcFormat)) {
547 COPY9_OR_10TO9_OR_10_2(AV_RB16, AV_WB16);
549 COPY9_OR_10TO9_OR_10_2(AV_RL16, AV_WB16);
552 if (isBE(c->srcFormat)) {
553 COPY9_OR_10TO9_OR_10_2(AV_RB16, AV_WL16);
555 COPY9_OR_10TO9_OR_10_2(AV_RL16, AV_WL16);
559 #define W8(a, b) { *(a) = (b); }
560 #define COPY9_OR_10TO8(rfunc) \
561 if (src_depth == 9) { \
562 DITHER_COPY(dstPtr, dstStride[plane], W8, \
563 srcPtr2, srcStride[plane]/2, rfunc, \
566 DITHER_COPY(dstPtr, dstStride[plane], W8, \
567 srcPtr2, srcStride[plane]/2, rfunc, \
570 if (isBE(c->srcFormat)) {
571 COPY9_OR_10TO8(AV_RB16);
573 COPY9_OR_10TO8(AV_RL16);
576 } else if(is9_OR_10BPS(c->dstFormat)) {
577 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1+1;
578 uint16_t *dstPtr2 = (uint16_t*)dstPtr;
580 if (is16BPS(c->srcFormat)) {
581 const uint16_t *srcPtr2 = (const uint16_t*)srcPtr;
582 #define COPY16TO9_OR_10(rfunc, wfunc) \
583 if (dst_depth == 9) { \
584 DITHER_COPY(dstPtr2, dstStride[plane]/2, wfunc, \
585 srcPtr2, srcStride[plane]/2, rfunc, \
586 dither_8x8_128, 7); \
588 DITHER_COPY(dstPtr2, dstStride[plane]/2, wfunc, \
589 srcPtr2, srcStride[plane]/2, rfunc, \
592 if (isBE(c->dstFormat)) {
593 if (isBE(c->srcFormat)) {
594 COPY16TO9_OR_10(AV_RB16, AV_WB16);
596 COPY16TO9_OR_10(AV_RL16, AV_WB16);
599 if (isBE(c->srcFormat)) {
600 COPY16TO9_OR_10(AV_RB16, AV_WL16);
602 COPY16TO9_OR_10(AV_RL16, AV_WL16);
606 #define COPY8TO9_OR_10(wfunc) \
607 for (i = 0; i < height; i++) { \
608 for (j = 0; j < length; j++) { \
609 const int srcpx = srcPtr[j]; \
610 wfunc(&dstPtr2[j], (srcpx<<(dst_depth-8)) | (srcpx >> (16-dst_depth))); \
612 dstPtr2 += dstStride[plane]/2; \
613 srcPtr += srcStride[plane]; \
615 if (isBE(c->dstFormat)) {
616 COPY8TO9_OR_10(AV_WB16);
618 COPY8TO9_OR_10(AV_WL16);
621 } else if(is16BPS(c->srcFormat) && !is16BPS(c->dstFormat)) {
622 const uint16_t *srcPtr2 = (const uint16_t*)srcPtr;
623 #define COPY16TO8(rfunc) \
624 DITHER_COPY(dstPtr, dstStride[plane], W8, \
625 srcPtr2, srcStride[plane]/2, rfunc, \
627 if (isBE(c->srcFormat)) {
632 } else if(!is16BPS(c->srcFormat) && is16BPS(c->dstFormat)) {
633 for (i=0; i<height; i++) {
634 for (j=0; j<length; j++) {
635 dstPtr[ j<<1 ] = srcPtr[j];
636 dstPtr[(j<<1)+1] = srcPtr[j];
638 srcPtr+= srcStride[plane];
639 dstPtr+= dstStride[plane];
641 } else if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat)
642 && isBE(c->srcFormat) != isBE(c->dstFormat)) {
644 for (i=0; i<height; i++) {
645 for (j=0; j<length; j++)
646 ((uint16_t*)dstPtr)[j] = av_bswap16(((const uint16_t*)srcPtr)[j]);
647 srcPtr+= srcStride[plane];
648 dstPtr+= dstStride[plane];
650 } else if (dstStride[plane] == srcStride[plane] &&
651 srcStride[plane] > 0 && srcStride[plane] == length) {
652 memcpy(dst[plane] + dstStride[plane]*y, src[plane],
653 height*dstStride[plane]);
655 if(is16BPS(c->srcFormat) && is16BPS(c->dstFormat))
657 for (i=0; i<height; i++) {
658 memcpy(dstPtr, srcPtr, length);
659 srcPtr+= srcStride[plane];
660 dstPtr+= dstStride[plane];
668 void ff_get_unscaled_swscale(SwsContext *c)
670 const enum PixelFormat srcFormat = c->srcFormat;
671 const enum PixelFormat dstFormat = c->dstFormat;
672 const int flags = c->flags;
673 const int dstH = c->dstH;
676 needsDither= isAnyRGB(dstFormat)
677 && c->dstFormatBpp < 24
678 && (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));
681 if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) && (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {
682 c->swScale= planarToNv12Wrapper;
685 if ((srcFormat==PIX_FMT_YUV420P || srcFormat==PIX_FMT_YUV422P || srcFormat==PIX_FMT_YUVA420P) && isAnyRGB(dstFormat)
686 && !(flags & SWS_ACCURATE_RND) && !(dstH&1)) {
687 c->swScale= ff_yuv2rgb_get_func_ptr(c);
690 if (srcFormat==PIX_FMT_YUV410P && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_BITEXACT)) {
691 c->swScale= yvu9ToYv12Wrapper;
695 if (srcFormat==PIX_FMT_BGR24 && (dstFormat==PIX_FMT_YUV420P || dstFormat==PIX_FMT_YUVA420P) && !(flags & SWS_ACCURATE_RND))
696 c->swScale= bgr24ToYv12Wrapper;
698 /* RGB/BGR -> RGB/BGR (no dither needed forms) */
699 if ( isAnyRGB(srcFormat)
700 && isAnyRGB(dstFormat)
701 && srcFormat != PIX_FMT_BGR8 && dstFormat != PIX_FMT_BGR8
702 && srcFormat != PIX_FMT_RGB8 && dstFormat != PIX_FMT_RGB8
703 && srcFormat != PIX_FMT_BGR4 && dstFormat != PIX_FMT_BGR4
704 && srcFormat != PIX_FMT_RGB4 && dstFormat != PIX_FMT_RGB4
705 && srcFormat != PIX_FMT_BGR4_BYTE && dstFormat != PIX_FMT_BGR4_BYTE
706 && srcFormat != PIX_FMT_RGB4_BYTE && dstFormat != PIX_FMT_RGB4_BYTE
707 && srcFormat != PIX_FMT_MONOBLACK && dstFormat != PIX_FMT_MONOBLACK
708 && srcFormat != PIX_FMT_MONOWHITE && dstFormat != PIX_FMT_MONOWHITE
709 && srcFormat != PIX_FMT_RGB48LE && dstFormat != PIX_FMT_RGB48LE
710 && srcFormat != PIX_FMT_RGB48BE && dstFormat != PIX_FMT_RGB48BE
711 && srcFormat != PIX_FMT_BGR48LE && dstFormat != PIX_FMT_BGR48LE
712 && srcFormat != PIX_FMT_BGR48BE && dstFormat != PIX_FMT_BGR48BE
713 && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))
714 c->swScale= rgbToRgbWrapper;
716 if ((usePal(srcFormat) && (
717 dstFormat == PIX_FMT_RGB32 ||
718 dstFormat == PIX_FMT_RGB32_1 ||
719 dstFormat == PIX_FMT_RGB24 ||
720 dstFormat == PIX_FMT_BGR32 ||
721 dstFormat == PIX_FMT_BGR32_1 ||
722 dstFormat == PIX_FMT_BGR24)))
723 c->swScale= palToRgbWrapper;
725 if (srcFormat == PIX_FMT_YUV422P) {
726 if (dstFormat == PIX_FMT_YUYV422)
727 c->swScale= yuv422pToYuy2Wrapper;
728 else if (dstFormat == PIX_FMT_UYVY422)
729 c->swScale= yuv422pToUyvyWrapper;
732 /* LQ converters if -sws 0 or -sws 4*/
733 if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {
735 if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {
736 if (dstFormat == PIX_FMT_YUYV422)
737 c->swScale= planarToYuy2Wrapper;
738 else if (dstFormat == PIX_FMT_UYVY422)
739 c->swScale= planarToUyvyWrapper;
742 if(srcFormat == PIX_FMT_YUYV422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
743 c->swScale= yuyvToYuv420Wrapper;
744 if(srcFormat == PIX_FMT_UYVY422 && (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
745 c->swScale= uyvyToYuv420Wrapper;
746 if(srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)
747 c->swScale= yuyvToYuv422Wrapper;
748 if(srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)
749 c->swScale= uyvyToYuv422Wrapper;
752 if ( srcFormat == dstFormat
753 || (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P)
754 || (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P)
755 || (isPlanarYUV(srcFormat) && isGray(dstFormat))
756 || (isPlanarYUV(dstFormat) && isGray(srcFormat))
757 || (isGray(dstFormat) && isGray(srcFormat))
758 || (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat)
759 && c->chrDstHSubSample == c->chrSrcHSubSample
760 && c->chrDstVSubSample == c->chrSrcVSubSample
761 && dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21
762 && srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))
764 if (isPacked(c->srcFormat))
765 c->swScale= packedCopyWrapper;
766 else /* Planar YUV or gray */
767 c->swScale= planarCopyWrapper;
771 ff_bfin_get_unscaled_swscale(c);
773 ff_swscale_get_unscaled_altivec(c);
776 static void reset_ptr(const uint8_t* src[], int format)
780 if(!isPlanarYUV(format)) {
788 static int check_image_pointers(uint8_t *data[4], enum PixelFormat pix_fmt,
789 const int linesizes[4])
791 const AVPixFmtDescriptor *desc = &av_pix_fmt_descriptors[pix_fmt];
794 for (i = 0; i < 4; i++) {
795 int plane = desc->comp[i].plane;
796 if (!data[plane] || !linesizes[plane])
804 * swscale wrapper, so we don't need to export the SwsContext.
805 * Assumes planar YUV to be in YUV order instead of YVU.
807 int sws_scale(struct SwsContext *c, const uint8_t* const srcSlice[],
808 const int srcStride[], int srcSliceY, int srcSliceH,
809 uint8_t* const dst[], const int dstStride[])
812 const uint8_t* src2[4]= {srcSlice[0], srcSlice[1], srcSlice[2], srcSlice[3]};
813 uint8_t* dst2[4]= {dst[0], dst[1], dst[2], dst[3]};
815 // do not mess up sliceDir if we have a "trailing" 0-size slice
819 if (!check_image_pointers(srcSlice, c->srcFormat, srcStride)) {
820 av_log(c, AV_LOG_ERROR, "bad src image pointers\n");
823 if (!check_image_pointers(dst, c->dstFormat, dstStride)) {
824 av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");
828 if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {
829 av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");
832 if (c->sliceDir == 0) {
833 if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;
836 if (usePal(c->srcFormat)) {
837 for (i=0; i<256; i++) {
838 int p, r, g, b,y,u,v;
839 if(c->srcFormat == PIX_FMT_PAL8) {
840 p=((const uint32_t*)(srcSlice[1]))[i];
844 } else if(c->srcFormat == PIX_FMT_RGB8) {
848 } else if(c->srcFormat == PIX_FMT_BGR8) {
852 } else if(c->srcFormat == PIX_FMT_RGB4_BYTE) {
856 } else if(c->srcFormat == PIX_FMT_GRAY8 || c->srcFormat == PIX_FMT_Y400A) {
859 assert(c->srcFormat == PIX_FMT_BGR4_BYTE);
864 y= av_clip_uint8((RY*r + GY*g + BY*b + ( 33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
865 u= av_clip_uint8((RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
866 v= av_clip_uint8((RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
867 c->pal_yuv[i]= y + (u<<8) + (v<<16);
869 switch(c->dstFormat) {
874 c->pal_rgb[i]= r + (g<<8) + (b<<16);
876 case PIX_FMT_BGR32_1:
880 c->pal_rgb[i]= (r + (g<<8) + (b<<16)) << 8;
882 case PIX_FMT_RGB32_1:
886 c->pal_rgb[i]= (b + (g<<8) + (r<<16)) << 8;
893 c->pal_rgb[i]= b + (g<<8) + (r<<16);
898 // copy strides, so they can safely be modified
899 if (c->sliceDir == 1) {
900 // slices go from top to bottom
901 int srcStride2[4]= {srcStride[0], srcStride[1], srcStride[2], srcStride[3]};
902 int dstStride2[4]= {dstStride[0], dstStride[1], dstStride[2], dstStride[3]};
904 reset_ptr(src2, c->srcFormat);
905 reset_ptr((const uint8_t**)dst2, c->dstFormat);
907 /* reset slice direction at end of frame */
908 if (srcSliceY + srcSliceH == c->srcH)
911 return c->swScale(c, src2, srcStride2, srcSliceY, srcSliceH, dst2, dstStride2);
913 // slices go from bottom to top => we flip the image internally
914 int srcStride2[4]= {-srcStride[0], -srcStride[1], -srcStride[2], -srcStride[3]};
915 int dstStride2[4]= {-dstStride[0], -dstStride[1], -dstStride[2], -dstStride[3]};
917 src2[0] += (srcSliceH-1)*srcStride[0];
918 if (!usePal(c->srcFormat))
919 src2[1] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[1];
920 src2[2] += ((srcSliceH>>c->chrSrcVSubSample)-1)*srcStride[2];
921 src2[3] += (srcSliceH-1)*srcStride[3];
922 dst2[0] += ( c->dstH -1)*dstStride[0];
923 dst2[1] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[1];
924 dst2[2] += ((c->dstH>>c->chrDstVSubSample)-1)*dstStride[2];
925 dst2[3] += ( c->dstH -1)*dstStride[3];
927 reset_ptr(src2, c->srcFormat);
928 reset_ptr((const uint8_t**)dst2, c->dstFormat);
930 /* reset slice direction at end of frame */
934 return c->swScale(c, src2, srcStride2, c->srcH-srcSliceY-srcSliceH, srcSliceH, dst2, dstStride2);
938 /* Convert the palette to the same packed 32-bit format as the palette */
939 void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)
943 for (i=0; i<num_pixels; i++)
944 ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];
947 /* Palette format: ABCD -> dst format: ABC */
948 void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst, int num_pixels, const uint8_t *palette)
952 for (i=0; i<num_pixels; i++) {
954 dst[0]= palette[src[i]*4+0];
955 dst[1]= palette[src[i]*4+1];
956 dst[2]= palette[src[i]*4+2];