1 /*
2 * software RGB to RGB converter
3 * pluralize by software PAL8 to RGB converter
4 * software YUV to YUV converter
5 * software YUV to RGB converter
6 * Written by Nick Kurshev.
7 * palette & YUV & runtime CPU stuff by Michael (michaelni@gmx.at)
8 *
9 * This file is part of FFmpeg.
10 *
11 * FFmpeg is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public
13 * License as published by the Free Software Foundation; either
14 * version 2.1 of the License, or (at your option) any later version.
15 *
16 * FFmpeg is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
20 *
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with FFmpeg; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 */
25
26 #include <inttypes.h>
27
28 #include "libavutil/attributes.h"
29 #include "libavutil/bswap.h"
30 #include "config.h"
31 #include "rgb2rgb.h"
32 #include "swscale.h"
33 #include "swscale_internal.h"
34
35 void (*rgb32tobgr24)(const uint8_t *src, uint8_t *dst, int src_size);
36 void (*rgb32tobgr16)(const uint8_t *src, uint8_t *dst, int src_size);
37 void (*rgb32tobgr15)(const uint8_t *src, uint8_t *dst, int src_size);
38 void (*rgb24tobgr32)(const uint8_t *src, uint8_t *dst, int src_size);
39 void (*rgb24tobgr24)(const uint8_t *src, uint8_t *dst, int src_size);
40 void (*rgb24tobgr16)(const uint8_t *src, uint8_t *dst, int src_size);
41 void (*rgb24tobgr15)(const uint8_t *src, uint8_t *dst, int src_size);
42 void (*rgb16tobgr24)(const uint8_t *src, uint8_t *dst, int src_size);
43 void (*rgb15tobgr24)(const uint8_t *src, uint8_t *dst, int src_size);
44
45 void (*rgb32to16)(const uint8_t *src, uint8_t *dst, int src_size);
46 void (*rgb32to15)(const uint8_t *src, uint8_t *dst, int src_size);
47 void (*rgb24to16)(const uint8_t *src, uint8_t *dst, int src_size);
48 void (*rgb24to15)(const uint8_t *src, uint8_t *dst, int src_size);
49 void (*rgb16to32)(const uint8_t *src, uint8_t *dst, int src_size);
50 void (*rgb16to15)(const uint8_t *src, uint8_t *dst, int src_size);
51 void (*rgb15to16)(const uint8_t *src, uint8_t *dst, int src_size);
52 void (*rgb15to32)(const uint8_t *src, uint8_t *dst, int src_size);
53
54 void (*shuffle_bytes_0321)(const uint8_t *src, uint8_t *dst, int src_size);
55 void (*shuffle_bytes_2103)(const uint8_t *src, uint8_t *dst, int src_size);
56 void (*shuffle_bytes_1230)(const uint8_t *src, uint8_t *dst, int src_size);
57 void (*shuffle_bytes_3012)(const uint8_t *src, uint8_t *dst, int src_size);
58 void (*shuffle_bytes_3210)(const uint8_t *src, uint8_t *dst, int src_size);
59
60
61 void (*yv12toyuy2)(const uint8_t *ysrc, const uint8_t *usrc,
62 const uint8_t *vsrc, uint8_t *dst,
63 int width, int height,
64 int lumStride, int chromStride, int dstStride);
65 void (*yv12touyvy)(const uint8_t *ysrc, const uint8_t *usrc,
66 const uint8_t *vsrc, uint8_t *dst,
67 int width, int height,
68 int lumStride, int chromStride, int dstStride);
69 void (*yuv422ptoyuy2)(const uint8_t *ysrc, const uint8_t *usrc,
70 const uint8_t *vsrc, uint8_t *dst,
71 int width, int height,
72 int lumStride, int chromStride, int dstStride);
73 void (*yuv422ptouyvy)(const uint8_t *ysrc, const uint8_t *usrc,
74 const uint8_t *vsrc, uint8_t *dst,
75 int width, int height,
76 int lumStride, int chromStride, int dstStride);
77 void (*yuy2toyv12)(const uint8_t *src, uint8_t *ydst,
78 uint8_t *udst, uint8_t *vdst,
79 int width, int height,
80 int lumStride, int chromStride, int srcStride);
81 void (*ff_rgb24toyv12)(const uint8_t *src, uint8_t *ydst,
82 uint8_t *udst, uint8_t *vdst,
83 int width, int height,
84 int lumStride, int chromStride, int srcStride,
85 int32_t *rgb2yuv);
86 void (*planar2x)(const uint8_t *src, uint8_t *dst, int width, int height,
87 int srcStride, int dstStride);
88 void (*interleaveBytes)(const uint8_t *src1, const uint8_t *src2, uint8_t *dst,
89 int width, int height, int src1Stride,
90 int src2Stride, int dstStride);
91 void (*deinterleaveBytes)(const uint8_t *src, uint8_t *dst1, uint8_t *dst2,
92 int width, int height, int srcStride,
93 int dst1Stride, int dst2Stride);
94 void (*vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
95 uint8_t *dst1, uint8_t *dst2,
96 int width, int height,
97 int srcStride1, int srcStride2,
98 int dstStride1, int dstStride2);
99 void (*yvu9_to_yuy2)(const uint8_t *src1, const uint8_t *src2,
100 const uint8_t *src3, uint8_t *dst,
101 int width, int height,
102 int srcStride1, int srcStride2,
103 int srcStride3, int dstStride);
104 void (*uyvytoyuv420)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
105 const uint8_t *src, int width, int height,
106 int lumStride, int chromStride, int srcStride);
107 void (*uyvytoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
108 const uint8_t *src, int width, int height,
109 int lumStride, int chromStride, int srcStride);
110 void (*yuyvtoyuv420)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
111 const uint8_t *src, int width, int height,
112 int lumStride, int chromStride, int srcStride);
113 void (*yuyvtoyuv422)(uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
114 const uint8_t *src, int width, int height,
115 int lumStride, int chromStride, int srcStride);
116
117 #define BY ((int)( 0.098 * (1 << RGB2YUV_SHIFT) + 0.5))
118 #define BV ((int)(-0.071 * (1 << RGB2YUV_SHIFT) + 0.5))
119 #define BU ((int)( 0.439 * (1 << RGB2YUV_SHIFT) + 0.5))
120 #define GY ((int)( 0.504 * (1 << RGB2YUV_SHIFT) + 0.5))
121 #define GV ((int)(-0.368 * (1 << RGB2YUV_SHIFT) + 0.5))
122 #define GU ((int)(-0.291 * (1 << RGB2YUV_SHIFT) + 0.5))
123 #define RY ((int)( 0.257 * (1 << RGB2YUV_SHIFT) + 0.5))
124 #define RV ((int)( 0.439 * (1 << RGB2YUV_SHIFT) + 0.5))
125 #define RU ((int)(-0.148 * (1 << RGB2YUV_SHIFT) + 0.5))
126
127 //plain C versions
128 #include "rgb2rgb_template.c"
129
130 /*
131 * RGB15->RGB16 original by Strepto/Astral
132 * ported to gcc & bugfixed : A'rpi
133 * MMXEXT, 3DNOW optimization by Nick Kurshev
134 * 32-bit C version, and and&add trick by Michael Niedermayer
135 */
136
ff_sws_rgb2rgb_init(void)137 av_cold void ff_sws_rgb2rgb_init(void)
138 {
139 rgb2rgb_init_c();
140 if (ARCH_AARCH64)
141 rgb2rgb_init_aarch64();
142 if (ARCH_X86)
143 rgb2rgb_init_x86();
144 }
145
rgb32to24(const uint8_t * src,uint8_t * dst,int src_size)146 void rgb32to24(const uint8_t *src, uint8_t *dst, int src_size)
147 {
148 int i, num_pixels = src_size >> 2;
149
150 for (i = 0; i < num_pixels; i++) {
151 #if HAVE_BIGENDIAN
152 /* RGB32 (= A,B,G,R) -> BGR24 (= B,G,R) */
153 dst[3 * i + 0] = src[4 * i + 1];
154 dst[3 * i + 1] = src[4 * i + 2];
155 dst[3 * i + 2] = src[4 * i + 3];
156 #else
157 dst[3 * i + 0] = src[4 * i + 2];
158 dst[3 * i + 1] = src[4 * i + 1];
159 dst[3 * i + 2] = src[4 * i + 0];
160 #endif
161 }
162 }
163
rgb24to32(const uint8_t * src,uint8_t * dst,int src_size)164 void rgb24to32(const uint8_t *src, uint8_t *dst, int src_size)
165 {
166 int i;
167
168 for (i = 0; 3 * i < src_size; i++) {
169 #if HAVE_BIGENDIAN
170 /* RGB24 (= R, G, B) -> BGR32 (= A, R, G, B) */
171 dst[4 * i + 0] = 255;
172 dst[4 * i + 1] = src[3 * i + 0];
173 dst[4 * i + 2] = src[3 * i + 1];
174 dst[4 * i + 3] = src[3 * i + 2];
175 #else
176 dst[4 * i + 0] = src[3 * i + 2];
177 dst[4 * i + 1] = src[3 * i + 1];
178 dst[4 * i + 2] = src[3 * i + 0];
179 dst[4 * i + 3] = 255;
180 #endif
181 }
182 }
183
rgb16tobgr32(const uint8_t * src,uint8_t * dst,int src_size)184 void rgb16tobgr32(const uint8_t *src, uint8_t *dst, int src_size)
185 {
186 uint8_t *d = dst;
187 const uint16_t *s = (const uint16_t *)src;
188 const uint16_t *end = s + src_size / 2;
189
190 while (s < end) {
191 register uint16_t bgr = *s++;
192 #if HAVE_BIGENDIAN
193 *d++ = 255;
194 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
195 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
196 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
197 #else
198 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
199 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
200 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
201 *d++ = 255;
202 #endif
203 }
204 }
205
rgb12to15(const uint8_t * src,uint8_t * dst,int src_size)206 void rgb12to15(const uint8_t *src, uint8_t *dst, int src_size)
207 {
208 uint16_t rgb, r, g, b;
209 uint16_t *d = (uint16_t *)dst;
210 const uint16_t *s = (const uint16_t *)src;
211 const uint16_t *end = s + src_size / 2;
212
213 while (s < end) {
214 rgb = *s++;
215 r = rgb & 0xF00;
216 g = rgb & 0x0F0;
217 b = rgb & 0x00F;
218 r = (r << 3) | ((r & 0x800) >> 1);
219 g = (g << 2) | ((g & 0x080) >> 2);
220 b = (b << 1) | ( b >> 3);
221 *d++ = r | g | b;
222 }
223 }
224
rgb16to24(const uint8_t * src,uint8_t * dst,int src_size)225 void rgb16to24(const uint8_t *src, uint8_t *dst, int src_size)
226 {
227 uint8_t *d = dst;
228 const uint16_t *s = (const uint16_t *)src;
229 const uint16_t *end = s + src_size / 2;
230
231 while (s < end) {
232 register uint16_t bgr = *s++;
233 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
234 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
235 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
236 }
237 }
238
rgb16tobgr16(const uint8_t * src,uint8_t * dst,int src_size)239 void rgb16tobgr16(const uint8_t *src, uint8_t *dst, int src_size)
240 {
241 int i, num_pixels = src_size >> 1;
242
243 for (i = 0; i < num_pixels; i++) {
244 unsigned rgb = ((const uint16_t *)src)[i];
245 ((uint16_t *)dst)[i] = (rgb >> 11) | (rgb & 0x7E0) | (rgb << 11);
246 }
247 }
248
rgb16tobgr15(const uint8_t * src,uint8_t * dst,int src_size)249 void rgb16tobgr15(const uint8_t *src, uint8_t *dst, int src_size)
250 {
251 int i, num_pixels = src_size >> 1;
252
253 for (i = 0; i < num_pixels; i++) {
254 unsigned rgb = ((const uint16_t *)src)[i];
255 ((uint16_t *)dst)[i] = (rgb >> 11) | ((rgb & 0x7C0) >> 1) | ((rgb & 0x1F) << 10);
256 }
257 }
258
rgb15tobgr32(const uint8_t * src,uint8_t * dst,int src_size)259 void rgb15tobgr32(const uint8_t *src, uint8_t *dst, int src_size)
260 {
261 uint8_t *d = dst;
262 const uint16_t *s = (const uint16_t *)src;
263 const uint16_t *end = s + src_size / 2;
264
265 while (s < end) {
266 register uint16_t bgr = *s++;
267 #if HAVE_BIGENDIAN
268 *d++ = 255;
269 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
270 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
271 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
272 #else
273 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
274 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
275 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
276 *d++ = 255;
277 #endif
278 }
279 }
280
rgb15to24(const uint8_t * src,uint8_t * dst,int src_size)281 void rgb15to24(const uint8_t *src, uint8_t *dst, int src_size)
282 {
283 uint8_t *d = dst;
284 const uint16_t *s = (const uint16_t *)src;
285 const uint16_t *end = s + src_size / 2;
286
287 while (s < end) {
288 register uint16_t bgr = *s++;
289 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
290 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
291 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
292 }
293 }
294
rgb15tobgr16(const uint8_t * src,uint8_t * dst,int src_size)295 void rgb15tobgr16(const uint8_t *src, uint8_t *dst, int src_size)
296 {
297 int i, num_pixels = src_size >> 1;
298
299 for (i = 0; i < num_pixels; i++) {
300 unsigned rgb = ((const uint16_t *)src)[i];
301 ((uint16_t *)dst)[i] = ((rgb & 0x7C00) >> 10) | ((rgb & 0x3E0) << 1) | (rgb << 11);
302 }
303 }
304
rgb15tobgr15(const uint8_t * src,uint8_t * dst,int src_size)305 void rgb15tobgr15(const uint8_t *src, uint8_t *dst, int src_size)
306 {
307 int i, num_pixels = src_size >> 1;
308
309 for (i = 0; i < num_pixels; i++) {
310 unsigned rgb = ((const uint16_t *)src)[i];
311 unsigned br = rgb & 0x7C1F;
312 ((uint16_t *)dst)[i] = (br >> 10) | (rgb & 0x3E0) | (br << 10);
313 }
314 }
315
rgb12tobgr12(const uint8_t * src,uint8_t * dst,int src_size)316 void rgb12tobgr12(const uint8_t *src, uint8_t *dst, int src_size)
317 {
318 uint16_t *d = (uint16_t *)dst;
319 uint16_t *s = (uint16_t *)src;
320 int i, num_pixels = src_size >> 1;
321
322 for (i = 0; i < num_pixels; i++) {
323 unsigned rgb = s[i];
324 d[i] = (rgb << 8 | rgb & 0xF0 | rgb >> 8) & 0xFFF;
325 }
326 }
327
328 #define DEFINE_RGB48TOBGR48(need_bswap, swap) \
329 void rgb48tobgr48_ ## need_bswap(const uint8_t *src, \
330 uint8_t *dst, int src_size) \
331 { \
332 uint16_t *d = (uint16_t *)dst; \
333 uint16_t *s = (uint16_t *)src; \
334 int i, num_pixels = src_size >> 1; \
335 \
336 for (i = 0; i < num_pixels; i += 3) { \
337 d[i ] = swap ? av_bswap16(s[i + 2]) : s[i + 2]; \
338 d[i + 1] = swap ? av_bswap16(s[i + 1]) : s[i + 1]; \
339 d[i + 2] = swap ? av_bswap16(s[i ]) : s[i ]; \
340 } \
341 }
342
343 DEFINE_RGB48TOBGR48(nobswap, 0)
344 DEFINE_RGB48TOBGR48(bswap, 1)
345
346 #define DEFINE_RGB64TOBGR48(need_bswap, swap) \
347 void rgb64tobgr48_ ## need_bswap(const uint8_t *src, \
348 uint8_t *dst, int src_size) \
349 { \
350 uint16_t *d = (uint16_t *)dst; \
351 uint16_t *s = (uint16_t *)src; \
352 int i, num_pixels = src_size >> 3; \
353 \
354 for (i = 0; i < num_pixels; i++) { \
355 d[3 * i ] = swap ? av_bswap16(s[4 * i + 2]) : s[4 * i + 2]; \
356 d[3 * i + 1] = swap ? av_bswap16(s[4 * i + 1]) : s[4 * i + 1]; \
357 d[3 * i + 2] = swap ? av_bswap16(s[4 * i ]) : s[4 * i ]; \
358 } \
359 }
360
361 DEFINE_RGB64TOBGR48(nobswap, 0)
362 DEFINE_RGB64TOBGR48(bswap, 1)
363
364 #define DEFINE_RGB64TO48(need_bswap, swap) \
365 void rgb64to48_ ## need_bswap(const uint8_t *src, \
366 uint8_t *dst, int src_size) \
367 { \
368 uint16_t *d = (uint16_t *)dst; \
369 uint16_t *s = (uint16_t *)src; \
370 int i, num_pixels = src_size >> 3; \
371 \
372 for (i = 0; i < num_pixels; i++) { \
373 d[3 * i ] = swap ? av_bswap16(s[4 * i ]) : s[4 * i ]; \
374 d[3 * i + 1] = swap ? av_bswap16(s[4 * i + 1]) : s[4 * i + 1]; \
375 d[3 * i + 2] = swap ? av_bswap16(s[4 * i + 2]) : s[4 * i + 2]; \
376 } \
377 }
378
379 DEFINE_RGB64TO48(nobswap, 0)
380 DEFINE_RGB64TO48(bswap, 1)
381
382 #define DEFINE_RGB48TOBGR64(need_bswap, swap) \
383 void rgb48tobgr64_ ## need_bswap(const uint8_t *src, \
384 uint8_t *dst, int src_size) \
385 { \
386 uint16_t *d = (uint16_t *)dst; \
387 uint16_t *s = (uint16_t *)src; \
388 int i, num_pixels = src_size / 6; \
389 \
390 for (i = 0; i < num_pixels; i++) { \
391 d[4 * i ] = swap ? av_bswap16(s[3 * i + 2]) : s[3 * i + 2]; \
392 d[4 * i + 1] = swap ? av_bswap16(s[3 * i + 1]) : s[3 * i + 1]; \
393 d[4 * i + 2] = swap ? av_bswap16(s[3 * i ]) : s[3 * i ]; \
394 d[4 * i + 3] = 0xFFFF; \
395 } \
396 }
397
398 DEFINE_RGB48TOBGR64(nobswap, 0)
399 DEFINE_RGB48TOBGR64(bswap, 1)
400
401 #define DEFINE_RGB48TO64(need_bswap, swap) \
402 void rgb48to64_ ## need_bswap(const uint8_t *src, \
403 uint8_t *dst, int src_size) \
404 { \
405 uint16_t *d = (uint16_t *)dst; \
406 uint16_t *s = (uint16_t *)src; \
407 int i, num_pixels = src_size / 6; \
408 \
409 for (i = 0; i < num_pixels; i++) { \
410 d[4 * i ] = swap ? av_bswap16(s[3 * i ]) : s[3 * i ]; \
411 d[4 * i + 1] = swap ? av_bswap16(s[3 * i + 1]) : s[3 * i + 1]; \
412 d[4 * i + 2] = swap ? av_bswap16(s[3 * i + 2]) : s[3 * i + 2]; \
413 d[4 * i + 3] = 0xFFFF; \
414 } \
415 }
416
417 DEFINE_RGB48TO64(nobswap, 0)
418 DEFINE_RGB48TO64(bswap, 1)
419