2 * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
27 #include "swscale_internal.h"
29 #include "libavutil/intreadwrite.h"
30 #include "libavutil/cpu.h"
31 #include "libavutil/avutil.h"
32 #include "libavutil/mathematics.h"
33 #include "libavutil/bswap.h"
34 #include "libavutil/pixdesc.h"
35 #include "libavutil/avassert.h"
37 DECLARE_ALIGNED(8, const uint8_t, dithers)[8][8][8]={
39 { 0, 1, 0, 1, 0, 1, 0, 1,},
40 { 1, 0, 1, 0, 1, 0, 1, 0,},
41 { 0, 1, 0, 1, 0, 1, 0, 1,},
42 { 1, 0, 1, 0, 1, 0, 1, 0,},
43 { 0, 1, 0, 1, 0, 1, 0, 1,},
44 { 1, 0, 1, 0, 1, 0, 1, 0,},
45 { 0, 1, 0, 1, 0, 1, 0, 1,},
46 { 1, 0, 1, 0, 1, 0, 1, 0,},
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 { 2, 4, 3, 5, 2, 4, 3, 5,},
58 { 6, 0, 7, 1, 6, 0, 7, 1,},
59 { 3, 5, 2, 4, 3, 5, 2, 4,},
60 { 7, 1, 6, 0, 7, 1, 6, 0,},
61 { 2, 4, 3, 5, 2, 4, 3, 5,},
62 { 6, 0, 7, 1, 6, 0, 7, 1,},
63 { 3, 5, 2, 4, 3, 5, 2, 4,},
64 { 7, 1, 6, 0, 7, 1, 6, 0,},
66 { 4, 8, 7, 11, 4, 8, 7, 11,},
67 { 12, 0, 15, 3, 12, 0, 15, 3,},
68 { 6, 10, 5, 9, 6, 10, 5, 9,},
69 { 14, 2, 13, 1, 14, 2, 13, 1,},
70 { 4, 8, 7, 11, 4, 8, 7, 11,},
71 { 12, 0, 15, 3, 12, 0, 15, 3,},
72 { 6, 10, 5, 9, 6, 10, 5, 9,},
73 { 14, 2, 13, 1, 14, 2, 13, 1,},
75 { 9, 17, 15, 23, 8, 16, 14, 22,},
76 { 25, 1, 31, 7, 24, 0, 30, 6,},
77 { 13, 21, 11, 19, 12, 20, 10, 18,},
78 { 29, 5, 27, 3, 28, 4, 26, 2,},
79 { 8, 16, 14, 22, 9, 17, 15, 23,},
80 { 24, 0, 30, 6, 25, 1, 31, 7,},
81 { 12, 20, 10, 18, 13, 21, 11, 19,},
82 { 28, 4, 26, 2, 29, 5, 27, 3,},
84 { 18, 34, 30, 46, 17, 33, 29, 45,},
85 { 50, 2, 62, 14, 49, 1, 61, 13,},
86 { 26, 42, 22, 38, 25, 41, 21, 37,},
87 { 58, 10, 54, 6, 57, 9, 53, 5,},
88 { 16, 32, 28, 44, 19, 35, 31, 47,},
89 { 48, 0, 60, 12, 51, 3, 63, 15,},
90 { 24, 40, 20, 36, 27, 43, 23, 39,},
91 { 56, 8, 52, 4, 59, 11, 55, 7,},
93 { 18, 34, 30, 46, 17, 33, 29, 45,},
94 { 50, 2, 62, 14, 49, 1, 61, 13,},
95 { 26, 42, 22, 38, 25, 41, 21, 37,},
96 { 58, 10, 54, 6, 57, 9, 53, 5,},
97 { 16, 32, 28, 44, 19, 35, 31, 47,},
98 { 48, 0, 60, 12, 51, 3, 63, 15,},
99 { 24, 40, 20, 36, 27, 43, 23, 39,},
100 { 56, 8, 52, 4, 59, 11, 55, 7,},
102 { 36, 68, 60, 92, 34, 66, 58, 90,},
103 { 100, 4,124, 28, 98, 2,122, 26,},
104 { 52, 84, 44, 76, 50, 82, 42, 74,},
105 { 116, 20,108, 12,114, 18,106, 10,},
106 { 32, 64, 56, 88, 38, 70, 62, 94,},
107 { 96, 0,120, 24,102, 6,126, 30,},
108 { 48, 80, 40, 72, 54, 86, 46, 78,},
109 { 112, 16,104, 8,118, 22,110, 14,},
112 const uint16_t dither_scale[15][16]={
113 { 2, 3, 3, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5,},
114 { 2, 3, 7, 7, 13, 13, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25,},
115 { 3, 3, 4, 15, 15, 29, 57, 57, 57, 113, 113, 113, 113, 113, 113, 113,},
116 { 3, 4, 4, 5, 31, 31, 61, 121, 241, 241, 241, 241, 481, 481, 481, 481,},
117 { 3, 4, 5, 5, 6, 63, 63, 125, 249, 497, 993, 993, 993, 993, 993, 1985,},
118 { 3, 5, 6, 6, 6, 7, 127, 127, 253, 505, 1009, 2017, 4033, 4033, 4033, 4033,},
119 { 3, 5, 6, 7, 7, 7, 8, 255, 255, 509, 1017, 2033, 4065, 8129,16257,16257,},
120 { 3, 5, 6, 8, 8, 8, 8, 9, 511, 511, 1021, 2041, 4081, 8161,16321,32641,},
121 { 3, 5, 7, 8, 9, 9, 9, 9, 10, 1023, 1023, 2045, 4089, 8177,16353,32705,},
122 { 3, 5, 7, 8, 10, 10, 10, 10, 10, 11, 2047, 2047, 4093, 8185,16369,32737,},
123 { 3, 5, 7, 8, 10, 11, 11, 11, 11, 11, 12, 4095, 4095, 8189,16377,32753,},
124 { 3, 5, 7, 9, 10, 12, 12, 12, 12, 12, 12, 13, 8191, 8191,16381,32761,},
125 { 3, 5, 7, 9, 10, 12, 13, 13, 13, 13, 13, 13, 14,16383,16383,32765,},
126 { 3, 5, 7, 9, 10, 12, 14, 14, 14, 14, 14, 14, 14, 15,32767,32767,},
127 { 3, 5, 7, 9, 11, 12, 14, 15, 15, 15, 15, 15, 15, 15, 16,65535,},
131 static void fillPlane(uint8_t *plane, int stride, int width, int height, int y,
135 uint8_t *ptr = plane + stride * y;
136 for (i = 0; i < height; i++) {
137 memset(ptr, val, width);
142 static void fillPlane16(uint8_t *plane, int stride, int width, int height, int y,
146 uint8_t *ptr = plane + stride * y;
147 int v = alpha ? -1 : (1<<bits);
148 for (i = 0; i < height; i++) {
149 for (j = 0; j < width; j++) {
156 static void copyPlane(const uint8_t *src, int srcStride,
157 int srcSliceY, int srcSliceH, int width,
158 uint8_t *dst, int dstStride)
160 dst += dstStride * srcSliceY;
161 if (dstStride == srcStride && srcStride > 0) {
162 memcpy(dst, src, srcSliceH * dstStride);
165 for (i = 0; i < srcSliceH; i++) {
166 memcpy(dst, src, width);
173 static int planarToNv12Wrapper(SwsContext *c, const uint8_t *src[],
174 int srcStride[], int srcSliceY,
175 int srcSliceH, uint8_t *dstParam[],
178 uint8_t *dst = dstParam[1] + dstStride[1] * srcSliceY / 2;
180 copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
181 dstParam[0], dstStride[0]);
183 if (c->dstFormat == PIX_FMT_NV12)
184 interleaveBytes(src[1], src[2], dst, c->srcW / 2, srcSliceH / 2,
185 srcStride[1], srcStride[2], dstStride[0]);
187 interleaveBytes(src[2], src[1], dst, c->srcW / 2, srcSliceH / 2,
188 srcStride[2], srcStride[1], dstStride[0]);
193 static int planarToYuy2Wrapper(SwsContext *c, const uint8_t *src[],
194 int srcStride[], int srcSliceY, int srcSliceH,
195 uint8_t *dstParam[], int dstStride[])
197 uint8_t *dst = dstParam[0] + dstStride[0] * srcSliceY;
199 yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0],
200 srcStride[1], dstStride[0]);
205 static int planarToUyvyWrapper(SwsContext *c, const uint8_t *src[],
206 int srcStride[], int srcSliceY, int srcSliceH,
207 uint8_t *dstParam[], int dstStride[])
209 uint8_t *dst = dstParam[0] + dstStride[0] * srcSliceY;
211 yv12touyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0],
212 srcStride[1], dstStride[0]);
217 static int yuv422pToYuy2Wrapper(SwsContext *c, const uint8_t *src[],
218 int srcStride[], int srcSliceY, int srcSliceH,
219 uint8_t *dstParam[], int dstStride[])
221 uint8_t *dst = dstParam[0] + dstStride[0] * srcSliceY;
223 yuv422ptoyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0],
224 srcStride[1], dstStride[0]);
229 static int yuv422pToUyvyWrapper(SwsContext *c, const uint8_t *src[],
230 int srcStride[], int srcSliceY, int srcSliceH,
231 uint8_t *dstParam[], int dstStride[])
233 uint8_t *dst = dstParam[0] + dstStride[0] * srcSliceY;
235 yuv422ptouyvy(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0],
236 srcStride[1], dstStride[0]);
241 static int yuyvToYuv420Wrapper(SwsContext *c, const uint8_t *src[],
242 int srcStride[], int srcSliceY, int srcSliceH,
243 uint8_t *dstParam[], int dstStride[])
245 uint8_t *ydst = dstParam[0] + dstStride[0] * srcSliceY;
246 uint8_t *udst = dstParam[1] + dstStride[1] * srcSliceY / 2;
247 uint8_t *vdst = dstParam[2] + dstStride[2] * srcSliceY / 2;
249 yuyvtoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0],
250 dstStride[1], srcStride[0]);
253 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
258 static int yuyvToYuv422Wrapper(SwsContext *c, const uint8_t *src[],
259 int srcStride[], int srcSliceY, int srcSliceH,
260 uint8_t *dstParam[], int dstStride[])
262 uint8_t *ydst = dstParam[0] + dstStride[0] * srcSliceY;
263 uint8_t *udst = dstParam[1] + dstStride[1] * srcSliceY;
264 uint8_t *vdst = dstParam[2] + dstStride[2] * srcSliceY;
266 yuyvtoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0],
267 dstStride[1], srcStride[0]);
272 static int uyvyToYuv420Wrapper(SwsContext *c, const uint8_t *src[],
273 int srcStride[], int srcSliceY, int srcSliceH,
274 uint8_t *dstParam[], int dstStride[])
276 uint8_t *ydst = dstParam[0] + dstStride[0] * srcSliceY;
277 uint8_t *udst = dstParam[1] + dstStride[1] * srcSliceY / 2;
278 uint8_t *vdst = dstParam[2] + dstStride[2] * srcSliceY / 2;
280 uyvytoyuv420(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0],
281 dstStride[1], srcStride[0]);
284 fillPlane(dstParam[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
289 static int uyvyToYuv422Wrapper(SwsContext *c, const uint8_t *src[],
290 int srcStride[], int srcSliceY, int srcSliceH,
291 uint8_t *dstParam[], int dstStride[])
293 uint8_t *ydst = dstParam[0] + dstStride[0] * srcSliceY;
294 uint8_t *udst = dstParam[1] + dstStride[1] * srcSliceY;
295 uint8_t *vdst = dstParam[2] + dstStride[2] * srcSliceY;
297 uyvytoyuv422(ydst, udst, vdst, src[0], c->srcW, srcSliceH, dstStride[0],
298 dstStride[1], srcStride[0]);
303 static void gray8aToPacked32(const uint8_t *src, uint8_t *dst, int num_pixels,
304 const uint8_t *palette)
307 for (i = 0; i < num_pixels; i++)
308 ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i << 1]] | (src[(i << 1) + 1] << 24);
311 static void gray8aToPacked32_1(const uint8_t *src, uint8_t *dst, int num_pixels,
312 const uint8_t *palette)
316 for (i = 0; i < num_pixels; i++)
317 ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i << 1]] | src[(i << 1) + 1];
320 static void gray8aToPacked24(const uint8_t *src, uint8_t *dst, int num_pixels,
321 const uint8_t *palette)
325 for (i = 0; i < num_pixels; i++) {
327 dst[0] = palette[src[i << 1] * 4 + 0];
328 dst[1] = palette[src[i << 1] * 4 + 1];
329 dst[2] = palette[src[i << 1] * 4 + 2];
334 static int packed_16bpc_bswap(SwsContext *c, const uint8_t *src[],
335 int srcStride[], int srcSliceY, int srcSliceH,
336 uint8_t *dst[], int dstStride[])
339 int srcstr = srcStride[0] >> 1;
340 int dststr = dstStride[0] >> 1;
341 uint16_t *dstPtr = (uint16_t *) dst[0];
342 const uint16_t *srcPtr = (const uint16_t *) src[0];
343 int min_stride = FFMIN(srcstr, dststr);
345 for (i = 0; i < srcSliceH; i++) {
346 for (j = 0; j < min_stride; j++) {
347 dstPtr[j] = av_bswap16(srcPtr[j]);
356 static int palToRgbWrapper(SwsContext *c, const uint8_t *src[], int srcStride[],
357 int srcSliceY, int srcSliceH, uint8_t *dst[],
360 const enum PixelFormat srcFormat = c->srcFormat;
361 const enum PixelFormat dstFormat = c->dstFormat;
362 void (*conv)(const uint8_t *src, uint8_t *dst, int num_pixels,
363 const uint8_t *palette) = NULL;
365 uint8_t *dstPtr = dst[0] + dstStride[0] * srcSliceY;
366 const uint8_t *srcPtr = src[0];
368 if (srcFormat == PIX_FMT_GRAY8A) {
370 case PIX_FMT_RGB32 : conv = gray8aToPacked32; break;
371 case PIX_FMT_BGR32 : conv = gray8aToPacked32; break;
372 case PIX_FMT_BGR32_1: conv = gray8aToPacked32_1; break;
373 case PIX_FMT_RGB32_1: conv = gray8aToPacked32_1; break;
374 case PIX_FMT_RGB24 : conv = gray8aToPacked24; break;
375 case PIX_FMT_BGR24 : conv = gray8aToPacked24; break;
377 } else if (usePal(srcFormat)) {
379 case PIX_FMT_RGB32 : conv = sws_convertPalette8ToPacked32; break;
380 case PIX_FMT_BGR32 : conv = sws_convertPalette8ToPacked32; break;
381 case PIX_FMT_BGR32_1: conv = sws_convertPalette8ToPacked32; break;
382 case PIX_FMT_RGB32_1: conv = sws_convertPalette8ToPacked32; break;
383 case PIX_FMT_RGB24 : conv = sws_convertPalette8ToPacked24; break;
384 case PIX_FMT_BGR24 : conv = sws_convertPalette8ToPacked24; break;
389 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
390 av_get_pix_fmt_name(srcFormat), av_get_pix_fmt_name(dstFormat));
392 for (i = 0; i < srcSliceH; i++) {
393 conv(srcPtr, dstPtr, c->srcW, (uint8_t *) c->pal_rgb);
394 srcPtr += srcStride[0];
395 dstPtr += dstStride[0];
402 static void gbr24ptopacked24(const uint8_t *src[], int srcStride[],
403 uint8_t *dst, int dstStride, int srcSliceH,
407 for (h = 0; h < srcSliceH; h++) {
408 uint8_t *dest = dst + dstStride * h;
409 for (x = 0; x < width; x++) {
415 for (i = 0; i < 3; i++)
416 src[i] += srcStride[i];
420 static void gbr24ptopacked32(const uint8_t *src[], int srcStride[],
421 uint8_t *dst, int dstStride, int srcSliceH,
422 int alpha_first, int width)
425 for (h = 0; h < srcSliceH; h++) {
426 uint8_t *dest = dst + dstStride * h;
429 for (x = 0; x < width; x++) {
436 for (x = 0; x < width; x++) {
444 for (i = 0; i < 3; i++)
445 src[i] += srcStride[i];
449 static int planarRgbToRgbWrapper(SwsContext *c, const uint8_t *src[],
450 int srcStride[], int srcSliceY, int srcSliceH,
451 uint8_t *dst[], int dstStride[])
454 if (c->srcFormat != PIX_FMT_GBRP) {
455 av_log(c, AV_LOG_ERROR, "unsupported planar RGB conversion %s -> %s\n",
456 av_get_pix_fmt_name(c->srcFormat),
457 av_get_pix_fmt_name(c->dstFormat));
461 switch (c->dstFormat) {
463 gbr24ptopacked24((const uint8_t *[]) { src[1], src[0], src[2] },
464 (int []) { srcStride[1], srcStride[0], srcStride[2] },
465 dst[0] + srcSliceY * dstStride[0], dstStride[0],
470 gbr24ptopacked24((const uint8_t *[]) { src[2], src[0], src[1] },
471 (int []) { srcStride[2], srcStride[0], srcStride[1] },
472 dst[0] + srcSliceY * dstStride[0], dstStride[0],
479 gbr24ptopacked32((const uint8_t *[]) { src[2], src[0], src[1] },
480 (int []) { srcStride[2], srcStride[0], srcStride[1] },
481 dst[0] + srcSliceY * dstStride[0], dstStride[0],
482 srcSliceH, alpha_first, c->srcW);
488 gbr24ptopacked32((const uint8_t *[]) { src[1], src[0], src[2] },
489 (int []) { srcStride[1], srcStride[0], srcStride[2] },
490 dst[0] + srcSliceY * dstStride[0], dstStride[0],
491 srcSliceH, alpha_first, c->srcW);
495 av_log(c, AV_LOG_ERROR,
496 "unsupported planar RGB conversion %s -> %s\n",
497 av_get_pix_fmt_name(c->srcFormat),
498 av_get_pix_fmt_name(c->dstFormat));
504 #define isRGBA32(x) ( \
505 (x) == PIX_FMT_ARGB \
506 || (x) == PIX_FMT_RGBA \
507 || (x) == PIX_FMT_BGRA \
508 || (x) == PIX_FMT_ABGR \
511 #define isRGBA64(x) ( \
512 (x) == PIX_FMT_RGBA64LE \
513 || (x) == PIX_FMT_RGBA64BE \
514 || (x) == PIX_FMT_BGRA64LE \
515 || (x) == PIX_FMT_BGRA64BE \
518 #define isRGB48(x) ( \
519 (x) == PIX_FMT_RGB48LE \
520 || (x) == PIX_FMT_RGB48BE \
521 || (x) == PIX_FMT_BGR48LE \
522 || (x) == PIX_FMT_BGR48BE \
525 /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */
526 typedef void (* rgbConvFn) (const uint8_t *, uint8_t *, int);
527 static rgbConvFn findRgbConvFn(SwsContext *c)
529 const enum PixelFormat srcFormat = c->srcFormat;
530 const enum PixelFormat dstFormat = c->dstFormat;
531 const int srcId = c->srcFormatBpp;
532 const int dstId = c->dstFormatBpp;
533 rgbConvFn conv = NULL;
535 #define IS_NOT_NE(bpp, fmt) \
536 (((bpp + 7) >> 3) == 2 && \
537 (!(av_pix_fmt_descriptors[fmt].flags & PIX_FMT_BE) != !HAVE_BIGENDIAN))
539 #define CONV_IS(src, dst) (srcFormat == PIX_FMT_##src && dstFormat == PIX_FMT_##dst)
541 if (isRGBA32(srcFormat) && isRGBA32(dstFormat)) {
542 if ( CONV_IS(ABGR, RGBA)
543 || CONV_IS(ARGB, BGRA)
544 || CONV_IS(BGRA, ARGB)
545 || CONV_IS(RGBA, ABGR)) conv = shuffle_bytes_3210;
546 else if (CONV_IS(ABGR, ARGB)
547 || CONV_IS(ARGB, ABGR)) conv = shuffle_bytes_0321;
548 else if (CONV_IS(ABGR, BGRA)
549 || CONV_IS(ARGB, RGBA)) conv = shuffle_bytes_1230;
550 else if (CONV_IS(BGRA, RGBA)
551 || CONV_IS(RGBA, BGRA)) conv = shuffle_bytes_2103;
552 else if (CONV_IS(BGRA, ABGR)
553 || CONV_IS(RGBA, ARGB)) conv = shuffle_bytes_3012;
554 } else if (isRGB48(srcFormat) && isRGB48(dstFormat)) {
555 if (CONV_IS(RGB48LE, BGR48LE)
556 || CONV_IS(BGR48LE, RGB48LE)
557 || CONV_IS(RGB48BE, BGR48BE)
558 || CONV_IS(BGR48BE, RGB48BE)) conv = rgb48tobgr48_nobswap;
559 else if (CONV_IS(RGB48LE, BGR48BE)
560 || CONV_IS(BGR48LE, RGB48BE)
561 || CONV_IS(RGB48BE, BGR48LE)
562 || CONV_IS(BGR48BE, RGB48LE)) conv = rgb48tobgr48_bswap;
563 } else if (isRGBA64(srcFormat) && isRGB48(dstFormat)) {
564 if (CONV_IS(RGBA64LE, BGR48LE)
565 || CONV_IS(BGRA64LE, RGB48LE)
566 || CONV_IS(RGBA64BE, BGR48BE)
567 || CONV_IS(BGRA64BE, RGB48BE)) conv = rgb64tobgr48_nobswap;
568 else if (CONV_IS(RGBA64LE, BGR48BE)
569 || CONV_IS(BGRA64LE, RGB48BE)
570 || CONV_IS(RGBA64BE, BGR48LE)
571 || CONV_IS(BGRA64BE, RGB48LE)) conv = rgb64tobgr48_bswap;
572 else if (CONV_IS(RGBA64LE, RGB48LE)
573 || CONV_IS(BGRA64LE, BGR48LE)
574 || CONV_IS(RGBA64BE, RGB48BE)
575 || CONV_IS(BGRA64BE, BGR48BE)) conv = rgb64to48_nobswap;
576 else if (CONV_IS(RGBA64LE, RGB48BE)
577 || CONV_IS(BGRA64LE, BGR48BE)
578 || CONV_IS(RGBA64BE, RGB48LE)
579 || CONV_IS(BGRA64BE, BGR48LE)) conv = rgb64to48_bswap;
582 if ((isBGRinInt(srcFormat) && isBGRinInt(dstFormat)) ||
583 (isRGBinInt(srcFormat) && isRGBinInt(dstFormat))) {
584 switch (srcId | (dstId << 16)) {
585 case 0x000F000C: conv = rgb12to15; break;
586 case 0x000F0010: conv = rgb16to15; break;
587 case 0x000F0018: conv = rgb24to15; break;
588 case 0x000F0020: conv = rgb32to15; break;
589 case 0x0010000F: conv = rgb15to16; break;
590 case 0x00100018: conv = rgb24to16; break;
591 case 0x00100020: conv = rgb32to16; break;
592 case 0x0018000F: conv = rgb15to24; break;
593 case 0x00180010: conv = rgb16to24; break;
594 case 0x00180020: conv = rgb32to24; break;
595 case 0x0020000F: conv = rgb15to32; break;
596 case 0x00200010: conv = rgb16to32; break;
597 case 0x00200018: conv = rgb24to32; break;
599 } else if ((isBGRinInt(srcFormat) && isRGBinInt(dstFormat)) ||
600 (isRGBinInt(srcFormat) && isBGRinInt(dstFormat))) {
601 switch (srcId | (dstId << 16)) {
602 case 0x000C000C: conv = rgb12tobgr12; break;
603 case 0x000F000F: conv = rgb15tobgr15; break;
604 case 0x000F0010: conv = rgb16tobgr15; break;
605 case 0x000F0018: conv = rgb24tobgr15; break;
606 case 0x000F0020: conv = rgb32tobgr15; break;
607 case 0x0010000F: conv = rgb15tobgr16; break;
608 case 0x00100010: conv = rgb16tobgr16; break;
609 case 0x00100018: conv = rgb24tobgr16; break;
610 case 0x00100020: conv = rgb32tobgr16; break;
611 case 0x0018000F: conv = rgb15tobgr24; break;
612 case 0x00180010: conv = rgb16tobgr24; break;
613 case 0x00180018: conv = rgb24tobgr24; break;
614 case 0x00180020: conv = rgb32tobgr24; break;
615 case 0x0020000F: conv = rgb15tobgr32; break;
616 case 0x00200010: conv = rgb16tobgr32; break;
617 case 0x00200018: conv = rgb24tobgr32; break;
624 /* {RGB,BGR}{15,16,24,32,32_1} -> {RGB,BGR}{15,16,24,32} */
625 static int rgbToRgbWrapper(SwsContext *c, const uint8_t *src[], int srcStride[],
626 int srcSliceY, int srcSliceH, uint8_t *dst[],
630 const enum PixelFormat srcFormat = c->srcFormat;
631 const enum PixelFormat dstFormat = c->dstFormat;
632 const int srcBpp = (c->srcFormatBpp + 7) >> 3;
633 const int dstBpp = (c->dstFormatBpp + 7) >> 3;
634 rgbConvFn conv = findRgbConvFn(c);
637 av_log(c, AV_LOG_ERROR, "internal error %s -> %s converter\n",
638 av_get_pix_fmt_name(srcFormat), av_get_pix_fmt_name(dstFormat));
640 const uint8_t *srcPtr = src[0];
641 uint8_t *dstPtr = dst[0];
642 int src_bswap = IS_NOT_NE(c->srcFormatBpp, srcFormat);
643 int dst_bswap = IS_NOT_NE(c->dstFormatBpp, dstFormat);
645 if ((srcFormat == PIX_FMT_RGB32_1 || srcFormat == PIX_FMT_BGR32_1) &&
646 !isRGBA32(dstFormat))
647 srcPtr += ALT32_CORR;
649 if ((dstFormat == PIX_FMT_RGB32_1 || dstFormat == PIX_FMT_BGR32_1) &&
650 !isRGBA32(srcFormat))
651 dstPtr += ALT32_CORR;
653 if (dstStride[0] * srcBpp == srcStride[0] * dstBpp && srcStride[0] > 0 &&
654 !(srcStride[0] % srcBpp) && !dst_bswap && !src_bswap)
655 conv(srcPtr, dstPtr + dstStride[0] * srcSliceY,
656 srcSliceH * srcStride[0]);
659 dstPtr += dstStride[0] * srcSliceY;
661 for (i = 0; i < srcSliceH; i++) {
663 for(j=0; j<c->srcW; j++)
664 ((uint16_t*)c->formatConvBuffer)[j] = av_bswap16(((uint16_t*)srcPtr)[j]);
665 conv(c->formatConvBuffer, dstPtr, c->srcW * srcBpp);
667 conv(srcPtr, dstPtr, c->srcW * srcBpp);
669 for(j=0; j<c->srcW; j++)
670 ((uint16_t*)dstPtr)[j] = av_bswap16(((uint16_t*)dstPtr)[j]);
671 srcPtr += srcStride[0];
672 dstPtr += dstStride[0];
679 static int bgr24ToYv12Wrapper(SwsContext *c, const uint8_t *src[],
680 int srcStride[], int srcSliceY, int srcSliceH,
681 uint8_t *dst[], int dstStride[])
685 dst[0] + srcSliceY * dstStride[0],
686 dst[1] + (srcSliceY >> 1) * dstStride[1],
687 dst[2] + (srcSliceY >> 1) * dstStride[2],
689 dstStride[0], dstStride[1], srcStride[0]);
691 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
695 static int yvu9ToYv12Wrapper(SwsContext *c, const uint8_t *src[],
696 int srcStride[], int srcSliceY, int srcSliceH,
697 uint8_t *dst[], int dstStride[])
699 copyPlane(src[0], srcStride[0], srcSliceY, srcSliceH, c->srcW,
700 dst[0], dstStride[0]);
702 planar2x(src[1], dst[1] + dstStride[1] * (srcSliceY >> 1), c->chrSrcW,
703 srcSliceH >> 2, srcStride[1], dstStride[1]);
704 planar2x(src[2], dst[2] + dstStride[2] * (srcSliceY >> 1), c->chrSrcW,
705 srcSliceH >> 2, srcStride[2], dstStride[2]);
707 fillPlane(dst[3], dstStride[3], c->srcW, srcSliceH, srcSliceY, 255);
711 /* unscaled copy like stuff (assumes nearly identical formats) */
712 static int packedCopyWrapper(SwsContext *c, const uint8_t *src[],
713 int srcStride[], int srcSliceY, int srcSliceH,
714 uint8_t *dst[], int dstStride[])
716 if (dstStride[0] == srcStride[0] && srcStride[0] > 0)
717 memcpy(dst[0] + dstStride[0] * srcSliceY, src[0], srcSliceH * dstStride[0]);
720 const uint8_t *srcPtr = src[0];
721 uint8_t *dstPtr = dst[0] + dstStride[0] * srcSliceY;
724 /* universal length finder */
725 while (length + c->srcW <= FFABS(dstStride[0]) &&
726 length + c->srcW <= FFABS(srcStride[0]))
728 av_assert1(length != 0);
730 for (i = 0; i < srcSliceH; i++) {
731 memcpy(dstPtr, srcPtr, length);
732 srcPtr += srcStride[0];
733 dstPtr += dstStride[0];
739 #define DITHER_COPY(dst, dstStride, src, srcStride, bswap, dbswap)\
740 uint16_t scale= dither_scale[dst_depth-1][src_depth-1];\
741 int shift= src_depth-dst_depth + dither_scale[src_depth-2][dst_depth-1];\
742 for (i = 0; i < height; i++) {\
743 const uint8_t *dither= dithers[src_depth-9][i&7];\
744 for (j = 0; j < length-7; j+=8){\
745 dst[j+0] = dbswap((bswap(src[j+0]) + dither[0])*scale>>shift);\
746 dst[j+1] = dbswap((bswap(src[j+1]) + dither[1])*scale>>shift);\
747 dst[j+2] = dbswap((bswap(src[j+2]) + dither[2])*scale>>shift);\
748 dst[j+3] = dbswap((bswap(src[j+3]) + dither[3])*scale>>shift);\
749 dst[j+4] = dbswap((bswap(src[j+4]) + dither[4])*scale>>shift);\
750 dst[j+5] = dbswap((bswap(src[j+5]) + dither[5])*scale>>shift);\
751 dst[j+6] = dbswap((bswap(src[j+6]) + dither[6])*scale>>shift);\
752 dst[j+7] = dbswap((bswap(src[j+7]) + dither[7])*scale>>shift);\
754 for (; j < length; j++)\
755 dst[j] = dbswap((bswap(src[j]) + dither[j&7])*scale>>shift);\
760 static int planarCopyWrapper(SwsContext *c, const uint8_t *src[],
761 int srcStride[], int srcSliceY, int srcSliceH,
762 uint8_t *dst[], int dstStride[])
765 for (plane = 0; plane < 4; plane++) {
766 int length = (plane == 0 || plane == 3) ? c->srcW : -((-c->srcW ) >> c->chrDstHSubSample);
767 int y = (plane == 0 || plane == 3) ? srcSliceY: -((-srcSliceY) >> c->chrDstVSubSample);
768 int height = (plane == 0 || plane == 3) ? srcSliceH: -((-srcSliceH) >> c->chrDstVSubSample);
769 const uint8_t *srcPtr = src[plane];
770 uint8_t *dstPtr = dst[plane] + dstStride[plane] * y;
771 int shiftonly= plane==1 || plane==2 || (!c->srcRange && plane==0);
775 // ignore palette for GRAY8
776 if (plane == 1 && !dst[2]) continue;
777 if (!src[plane] || (plane == 1 && !src[2])) {
778 if (is16BPS(c->dstFormat) || isNBPS(c->dstFormat)) {
779 fillPlane16(dst[plane], dstStride[plane], length, height, y,
780 plane == 3, av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1);
782 fillPlane(dst[plane], dstStride[plane], length, height, y,
783 (plane == 3) ? 255 : 128);
786 if(isNBPS(c->srcFormat) || isNBPS(c->dstFormat)
787 || (is16BPS(c->srcFormat) != is16BPS(c->dstFormat))
789 const int src_depth = av_pix_fmt_descriptors[c->srcFormat].comp[plane].depth_minus1 + 1;
790 const int dst_depth = av_pix_fmt_descriptors[c->dstFormat].comp[plane].depth_minus1 + 1;
791 const uint16_t *srcPtr2 = (const uint16_t *) srcPtr;
792 uint16_t *dstPtr2 = (uint16_t*)dstPtr;
794 if (dst_depth == 8) {
795 if(isBE(c->srcFormat) == HAVE_BIGENDIAN){
796 DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, , )
798 DITHER_COPY(dstPtr, dstStride[plane], srcPtr2, srcStride[plane]/2, av_bswap16, )
800 } else if (src_depth == 8) {
801 for (i = 0; i < height; i++) {
804 for (j = 0; j < length; j++)\
805 w(&dstPtr2[j], srcPtr[j]<<(dst_depth-8));\
807 for (j = 0; j < length; j++)\
808 w(&dstPtr2[j], (srcPtr[j]<<(dst_depth-8)) |\
809 (srcPtr[j]>>(2*8-dst_depth)));\
811 if(isBE(c->dstFormat)){
816 dstPtr2 += dstStride[plane]/2;
817 srcPtr += srcStride[plane];
819 } else if (src_depth <= dst_depth) {
820 int orig_length = length;
821 for (i = 0; i < height; i++) {
822 if(isBE(c->srcFormat) == HAVE_BIGENDIAN &&
823 isBE(c->dstFormat) == HAVE_BIGENDIAN &&
825 unsigned shift = dst_depth - src_depth;
826 length = orig_length;
828 #define FAST_COPY_UP(shift) \
829 for (j = 0; j < length - 3; j += 4) { \
830 uint64_t v = AV_RN64A(srcPtr2 + j); \
831 AV_WN64A(dstPtr2 + j, v << shift); \
835 #define FAST_COPY_UP(shift) \
836 for (j = 0; j < length - 1; j += 2) { \
837 uint32_t v = AV_RN32A(srcPtr2 + j); \
838 AV_WN32A(dstPtr2 + j, v << shift); \
844 case 6: FAST_COPY_UP(6); break;
845 case 7: FAST_COPY_UP(7); break;
848 #define COPY_UP(r,w) \
850 for (j = 0; j < length; j++){ \
851 unsigned int v= r(&srcPtr2[j]);\
852 w(&dstPtr2[j], v<<(dst_depth-src_depth));\
855 for (j = 0; j < length; j++){ \
856 unsigned int v= r(&srcPtr2[j]);\
857 w(&dstPtr2[j], (v<<(dst_depth-src_depth)) | \
858 (v>>(2*src_depth-dst_depth)));\
861 if(isBE(c->srcFormat)){
862 if(isBE(c->dstFormat)){
863 COPY_UP(AV_RB16, AV_WB16)
865 COPY_UP(AV_RB16, AV_WL16)
868 if(isBE(c->dstFormat)){
869 COPY_UP(AV_RL16, AV_WB16)
871 COPY_UP(AV_RL16, AV_WL16)
874 dstPtr2 += dstStride[plane]/2;
875 srcPtr2 += srcStride[plane]/2;
878 if(isBE(c->srcFormat) == HAVE_BIGENDIAN){
879 if(isBE(c->dstFormat) == HAVE_BIGENDIAN){
880 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , )
882 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, , av_bswap16)
885 if(isBE(c->dstFormat) == HAVE_BIGENDIAN){
886 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, )
888 DITHER_COPY(dstPtr2, dstStride[plane]/2, srcPtr2, srcStride[plane]/2, av_bswap16, av_bswap16)
892 } else if (is16BPS(c->srcFormat) && is16BPS(c->dstFormat) &&
893 isBE(c->srcFormat) != isBE(c->dstFormat)) {
895 for (i = 0; i < height; i++) {
896 for (j = 0; j < length; j++)
897 ((uint16_t *) dstPtr)[j] = av_bswap16(((const uint16_t *) srcPtr)[j]);
898 srcPtr += srcStride[plane];
899 dstPtr += dstStride[plane];
901 } else if (dstStride[plane] == srcStride[plane] &&
902 srcStride[plane] > 0 && srcStride[plane] == length) {
903 memcpy(dst[plane] + dstStride[plane] * y, src[plane],
904 height * dstStride[plane]);
906 if (is16BPS(c->srcFormat) && is16BPS(c->dstFormat))
908 else if (!av_pix_fmt_descriptors[c->srcFormat].comp[0].depth_minus1)
909 length >>= 3; // monowhite/black
910 for (i = 0; i < height; i++) {
911 memcpy(dstPtr, srcPtr, length);
912 srcPtr += srcStride[plane];
913 dstPtr += dstStride[plane];
922 #define IS_DIFFERENT_ENDIANESS(src_fmt, dst_fmt, pix_fmt) \
923 ((src_fmt == pix_fmt ## BE && dst_fmt == pix_fmt ## LE) || \
924 (src_fmt == pix_fmt ## LE && dst_fmt == pix_fmt ## BE))
927 void ff_get_unscaled_swscale(SwsContext *c)
929 const enum PixelFormat srcFormat = c->srcFormat;
930 const enum PixelFormat dstFormat = c->dstFormat;
931 const int flags = c->flags;
932 const int dstH = c->dstH;
935 needsDither = isAnyRGB(dstFormat) &&
936 c->dstFormatBpp < 24 &&
937 (c->dstFormatBpp < c->srcFormatBpp || (!isAnyRGB(srcFormat)));
940 if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) &&
941 (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)) {
942 c->swScale = planarToNv12Wrapper;
945 if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUV422P ||
946 srcFormat == PIX_FMT_YUVA420P) && isAnyRGB(dstFormat) &&
947 !(flags & SWS_ACCURATE_RND) && !(dstH & 1)) {
948 c->swScale = ff_yuv2rgb_get_func_ptr(c);
951 if (srcFormat == PIX_FMT_YUV410P &&
952 (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P) &&
953 !(flags & SWS_BITEXACT)) {
954 c->swScale = yvu9ToYv12Wrapper;
958 if (srcFormat == PIX_FMT_BGR24 &&
959 (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P) &&
960 !(flags & SWS_ACCURATE_RND))
961 c->swScale = bgr24ToYv12Wrapper;
963 /* RGB/BGR -> RGB/BGR (no dither needed forms) */
964 if (isAnyRGB(srcFormat) && isAnyRGB(dstFormat) && findRgbConvFn(c)
965 && (!needsDither || (c->flags&(SWS_FAST_BILINEAR|SWS_POINT))))
966 c->swScale= rgbToRgbWrapper;
968 #define isByteRGB(f) (\
969 f == PIX_FMT_RGB32 ||\
970 f == PIX_FMT_RGB32_1 ||\
971 f == PIX_FMT_RGB24 ||\
972 f == PIX_FMT_BGR32 ||\
973 f == PIX_FMT_BGR32_1 ||\
976 if (isAnyRGB(srcFormat) && isPlanar(srcFormat) && isByteRGB(dstFormat))
977 c->swScale = planarRgbToRgbWrapper;
979 /* bswap 16 bits per pixel/component packed formats */
980 if (IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGR444) ||
981 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGR48) ||
982 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGRA64) ||
983 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGR555) ||
984 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_BGR565) ||
985 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_GRAY16) ||
986 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGB444) ||
987 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGB48) ||
988 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGBA64) ||
989 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGB555) ||
990 IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, PIX_FMT_RGB565))
991 c->swScale = packed_16bpc_bswap;
993 if (usePal(srcFormat) && isByteRGB(dstFormat))
994 c->swScale = palToRgbWrapper;
996 if (srcFormat == PIX_FMT_YUV422P) {
997 if (dstFormat == PIX_FMT_YUYV422)
998 c->swScale = yuv422pToYuy2Wrapper;
999 else if (dstFormat == PIX_FMT_UYVY422)
1000 c->swScale = yuv422pToUyvyWrapper;
1003 /* LQ converters if -sws 0 or -sws 4*/
1004 if (c->flags&(SWS_FAST_BILINEAR|SWS_POINT)) {
1006 if (srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUVA420P) {
1007 if (dstFormat == PIX_FMT_YUYV422)
1008 c->swScale = planarToYuy2Wrapper;
1009 else if (dstFormat == PIX_FMT_UYVY422)
1010 c->swScale = planarToUyvyWrapper;
1013 if (srcFormat == PIX_FMT_YUYV422 &&
1014 (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
1015 c->swScale = yuyvToYuv420Wrapper;
1016 if (srcFormat == PIX_FMT_UYVY422 &&
1017 (dstFormat == PIX_FMT_YUV420P || dstFormat == PIX_FMT_YUVA420P))
1018 c->swScale = uyvyToYuv420Wrapper;
1019 if (srcFormat == PIX_FMT_YUYV422 && dstFormat == PIX_FMT_YUV422P)
1020 c->swScale = yuyvToYuv422Wrapper;
1021 if (srcFormat == PIX_FMT_UYVY422 && dstFormat == PIX_FMT_YUV422P)
1022 c->swScale = uyvyToYuv422Wrapper;
1024 #define isPlanarGray(x) (isGray(x) && (x) != PIX_FMT_GRAY8A)
1026 if ( srcFormat == dstFormat ||
1027 (srcFormat == PIX_FMT_YUVA420P && dstFormat == PIX_FMT_YUV420P) ||
1028 (srcFormat == PIX_FMT_YUV420P && dstFormat == PIX_FMT_YUVA420P) ||
1029 (isPlanarYUV(srcFormat) && isPlanarGray(dstFormat)) ||
1030 (isPlanarYUV(dstFormat) && isPlanarGray(srcFormat)) ||
1031 (isPlanarGray(dstFormat) && isPlanarGray(srcFormat)) ||
1032 (isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat) &&
1033 c->chrDstHSubSample == c->chrSrcHSubSample &&
1034 c->chrDstVSubSample == c->chrSrcVSubSample &&
1035 dstFormat != PIX_FMT_NV12 && dstFormat != PIX_FMT_NV21 &&
1036 srcFormat != PIX_FMT_NV12 && srcFormat != PIX_FMT_NV21))
1038 if (isPacked(c->srcFormat))
1039 c->swScale = packedCopyWrapper;
1040 else /* Planar YUV or gray */
1041 c->swScale = planarCopyWrapper;
1045 ff_bfin_get_unscaled_swscale(c);
1047 ff_swscale_get_unscaled_altivec(c);
1050 /* Convert the palette to the same packed 32-bit format as the palette */
1051 void sws_convertPalette8ToPacked32(const uint8_t *src, uint8_t *dst,
1052 int num_pixels, const uint8_t *palette)
1056 for (i = 0; i < num_pixels; i++)
1057 ((uint32_t *) dst)[i] = ((const uint32_t *) palette)[src[i]];
1060 /* Palette format: ABCD -> dst format: ABC */
1061 void sws_convertPalette8ToPacked24(const uint8_t *src, uint8_t *dst,
1062 int num_pixels, const uint8_t *palette)
1066 for (i = 0; i < num_pixels; i++) {
1068 dst[0] = palette[src[i] * 4 + 0];
1069 dst[1] = palette[src[i] * 4 + 1];
1070 dst[2] = palette[src[i] * 4 + 2];