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 #elif ARCH_X86
143 rgb2rgb_init_x86();
144 #endif
145 }
146
rgb32to24(const uint8_t * src,uint8_t * dst,int src_size)147 void rgb32to24(const uint8_t *src, uint8_t *dst, int src_size)
148 {
149 int i, num_pixels = src_size >> 2;
150
151 for (i = 0; i < num_pixels; i++) {
152 #if HAVE_BIGENDIAN
153 /* RGB32 (= A,B,G,R) -> BGR24 (= B,G,R) */
154 dst[3 * i + 0] = src[4 * i + 1];
155 dst[3 * i + 1] = src[4 * i + 2];
156 dst[3 * i + 2] = src[4 * i + 3];
157 #else
158 dst[3 * i + 0] = src[4 * i + 2];
159 dst[3 * i + 1] = src[4 * i + 1];
160 dst[3 * i + 2] = src[4 * i + 0];
161 #endif
162 }
163 }
164
rgb24to32(const uint8_t * src,uint8_t * dst,int src_size)165 void rgb24to32(const uint8_t *src, uint8_t *dst, int src_size)
166 {
167 int i;
168
169 for (i = 0; 3 * i < src_size; i++) {
170 #if HAVE_BIGENDIAN
171 /* RGB24 (= R, G, B) -> BGR32 (= A, R, G, B) */
172 dst[4 * i + 0] = 255;
173 dst[4 * i + 1] = src[3 * i + 0];
174 dst[4 * i + 2] = src[3 * i + 1];
175 dst[4 * i + 3] = src[3 * i + 2];
176 #else
177 dst[4 * i + 0] = src[3 * i + 2];
178 dst[4 * i + 1] = src[3 * i + 1];
179 dst[4 * i + 2] = src[3 * i + 0];
180 dst[4 * i + 3] = 255;
181 #endif
182 }
183 }
184
rgb16tobgr32(const uint8_t * src,uint8_t * dst,int src_size)185 void rgb16tobgr32(const uint8_t *src, uint8_t *dst, int src_size)
186 {
187 uint8_t *d = dst;
188 const uint16_t *s = (const uint16_t *)src;
189 const uint16_t *end = s + src_size / 2;
190
191 while (s < end) {
192 register uint16_t bgr = *s++;
193 #if HAVE_BIGENDIAN
194 *d++ = 255;
195 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
196 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
197 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
198 #else
199 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
200 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
201 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
202 *d++ = 255;
203 #endif
204 }
205 }
206
rgb12to15(const uint8_t * src,uint8_t * dst,int src_size)207 void rgb12to15(const uint8_t *src, uint8_t *dst, int src_size)
208 {
209 uint16_t rgb, r, g, b;
210 uint16_t *d = (uint16_t *)dst;
211 const uint16_t *s = (const uint16_t *)src;
212 const uint16_t *end = s + src_size / 2;
213
214 while (s < end) {
215 rgb = *s++;
216 r = rgb & 0xF00;
217 g = rgb & 0x0F0;
218 b = rgb & 0x00F;
219 r = (r << 3) | ((r & 0x800) >> 1);
220 g = (g << 2) | ((g & 0x080) >> 2);
221 b = (b << 1) | ( b >> 3);
222 *d++ = r | g | b;
223 }
224 }
225
rgb16to24(const uint8_t * src,uint8_t * dst,int src_size)226 void rgb16to24(const uint8_t *src, uint8_t *dst, int src_size)
227 {
228 uint8_t *d = dst;
229 const uint16_t *s = (const uint16_t *)src;
230 const uint16_t *end = s + src_size / 2;
231
232 while (s < end) {
233 register uint16_t bgr = *s++;
234 *d++ = ((bgr&0xF800)>>8) | ((bgr&0xF800)>>13);
235 *d++ = ((bgr&0x07E0)>>3) | ((bgr&0x07E0)>> 9);
236 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
237 }
238 }
239
rgb16tobgr16(const uint8_t * src,uint8_t * dst,int src_size)240 void rgb16tobgr16(const uint8_t *src, uint8_t *dst, int src_size)
241 {
242 int i, num_pixels = src_size >> 1;
243
244 for (i = 0; i < num_pixels; i++) {
245 unsigned rgb = ((const uint16_t *)src)[i];
246 ((uint16_t *)dst)[i] = (rgb >> 11) | (rgb & 0x7E0) | (rgb << 11);
247 }
248 }
249
rgb16tobgr15(const uint8_t * src,uint8_t * dst,int src_size)250 void rgb16tobgr15(const uint8_t *src, uint8_t *dst, int src_size)
251 {
252 int i, num_pixels = src_size >> 1;
253
254 for (i = 0; i < num_pixels; i++) {
255 unsigned rgb = ((const uint16_t *)src)[i];
256 ((uint16_t *)dst)[i] = (rgb >> 11) | ((rgb & 0x7C0) >> 1) | ((rgb & 0x1F) << 10);
257 }
258 }
259
rgb15tobgr32(const uint8_t * src,uint8_t * dst,int src_size)260 void rgb15tobgr32(const uint8_t *src, uint8_t *dst, int src_size)
261 {
262 uint8_t *d = dst;
263 const uint16_t *s = (const uint16_t *)src;
264 const uint16_t *end = s + src_size / 2;
265
266 while (s < end) {
267 register uint16_t bgr = *s++;
268 #if HAVE_BIGENDIAN
269 *d++ = 255;
270 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
271 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
272 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
273 #else
274 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
275 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
276 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
277 *d++ = 255;
278 #endif
279 }
280 }
281
rgb15to24(const uint8_t * src,uint8_t * dst,int src_size)282 void rgb15to24(const uint8_t *src, uint8_t *dst, int src_size)
283 {
284 uint8_t *d = dst;
285 const uint16_t *s = (const uint16_t *)src;
286 const uint16_t *end = s + src_size / 2;
287
288 while (s < end) {
289 register uint16_t bgr = *s++;
290 *d++ = ((bgr&0x7C00)>>7) | ((bgr&0x7C00)>>12);
291 *d++ = ((bgr&0x03E0)>>2) | ((bgr&0x03E0)>> 7);
292 *d++ = ((bgr&0x001F)<<3) | ((bgr&0x001F)>> 2);
293 }
294 }
295
rgb15tobgr16(const uint8_t * src,uint8_t * dst,int src_size)296 void rgb15tobgr16(const uint8_t *src, uint8_t *dst, int src_size)
297 {
298 int i, num_pixels = src_size >> 1;
299
300 for (i = 0; i < num_pixels; i++) {
301 unsigned rgb = ((const uint16_t *)src)[i];
302 ((uint16_t *)dst)[i] = ((rgb & 0x7C00) >> 10) | ((rgb & 0x3E0) << 1) | (rgb << 11);
303 }
304 }
305
rgb15tobgr15(const uint8_t * src,uint8_t * dst,int src_size)306 void rgb15tobgr15(const uint8_t *src, uint8_t *dst, int src_size)
307 {
308 int i, num_pixels = src_size >> 1;
309
310 for (i = 0; i < num_pixels; i++) {
311 unsigned rgb = ((const uint16_t *)src)[i];
312 unsigned br = rgb & 0x7C1F;
313 ((uint16_t *)dst)[i] = (br >> 10) | (rgb & 0x3E0) | (br << 10);
314 }
315 }
316
rgb12tobgr12(const uint8_t * src,uint8_t * dst,int src_size)317 void rgb12tobgr12(const uint8_t *src, uint8_t *dst, int src_size)
318 {
319 uint16_t *d = (uint16_t *)dst;
320 uint16_t *s = (uint16_t *)src;
321 int i, num_pixels = src_size >> 1;
322
323 for (i = 0; i < num_pixels; i++) {
324 unsigned rgb = s[i];
325 d[i] = (rgb << 8 | rgb & 0xF0 | rgb >> 8) & 0xFFF;
326 }
327 }
328
329 #define DEFINE_RGB48TOBGR48(need_bswap, swap) \
330 void rgb48tobgr48_ ## need_bswap(const uint8_t *src, \
331 uint8_t *dst, int src_size) \
332 { \
333 uint16_t *d = (uint16_t *)dst; \
334 uint16_t *s = (uint16_t *)src; \
335 int i, num_pixels = src_size >> 1; \
336 \
337 for (i = 0; i < num_pixels; i += 3) { \
338 d[i ] = swap ? av_bswap16(s[i + 2]) : s[i + 2]; \
339 d[i + 1] = swap ? av_bswap16(s[i + 1]) : s[i + 1]; \
340 d[i + 2] = swap ? av_bswap16(s[i ]) : s[i ]; \
341 } \
342 }
343
344 DEFINE_RGB48TOBGR48(nobswap, 0)
345 DEFINE_RGB48TOBGR48(bswap, 1)
346
347 #define DEFINE_RGB64TOBGR48(need_bswap, swap) \
348 void rgb64tobgr48_ ## need_bswap(const uint8_t *src, \
349 uint8_t *dst, int src_size) \
350 { \
351 uint16_t *d = (uint16_t *)dst; \
352 uint16_t *s = (uint16_t *)src; \
353 int i, num_pixels = src_size >> 3; \
354 \
355 for (i = 0; i < num_pixels; i++) { \
356 d[3 * i ] = swap ? av_bswap16(s[4 * i + 2]) : s[4 * i + 2]; \
357 d[3 * i + 1] = swap ? av_bswap16(s[4 * i + 1]) : s[4 * i + 1]; \
358 d[3 * i + 2] = swap ? av_bswap16(s[4 * i ]) : s[4 * i ]; \
359 } \
360 }
361
362 DEFINE_RGB64TOBGR48(nobswap, 0)
363 DEFINE_RGB64TOBGR48(bswap, 1)
364
365 #define DEFINE_RGB64TO48(need_bswap, swap) \
366 void rgb64to48_ ## need_bswap(const uint8_t *src, \
367 uint8_t *dst, int src_size) \
368 { \
369 uint16_t *d = (uint16_t *)dst; \
370 uint16_t *s = (uint16_t *)src; \
371 int i, num_pixels = src_size >> 3; \
372 \
373 for (i = 0; i < num_pixels; i++) { \
374 d[3 * i ] = swap ? av_bswap16(s[4 * i ]) : s[4 * i ]; \
375 d[3 * i + 1] = swap ? av_bswap16(s[4 * i + 1]) : s[4 * i + 1]; \
376 d[3 * i + 2] = swap ? av_bswap16(s[4 * i + 2]) : s[4 * i + 2]; \
377 } \
378 }
379
380 DEFINE_RGB64TO48(nobswap, 0)
381 DEFINE_RGB64TO48(bswap, 1)
382
383 #define DEFINE_RGB48TOBGR64(need_bswap, swap) \
384 void rgb48tobgr64_ ## need_bswap(const uint8_t *src, \
385 uint8_t *dst, int src_size) \
386 { \
387 uint16_t *d = (uint16_t *)dst; \
388 uint16_t *s = (uint16_t *)src; \
389 int i, num_pixels = src_size / 6; \
390 \
391 for (i = 0; i < num_pixels; i++) { \
392 d[4 * i ] = swap ? av_bswap16(s[3 * i + 2]) : s[3 * i + 2]; \
393 d[4 * i + 1] = swap ? av_bswap16(s[3 * i + 1]) : s[3 * i + 1]; \
394 d[4 * i + 2] = swap ? av_bswap16(s[3 * i ]) : s[3 * i ]; \
395 d[4 * i + 3] = 0xFFFF; \
396 } \
397 }
398
399 DEFINE_RGB48TOBGR64(nobswap, 0)
400 DEFINE_RGB48TOBGR64(bswap, 1)
401
402 #define DEFINE_RGB48TO64(need_bswap, swap) \
403 void rgb48to64_ ## need_bswap(const uint8_t *src, \
404 uint8_t *dst, int src_size) \
405 { \
406 uint16_t *d = (uint16_t *)dst; \
407 uint16_t *s = (uint16_t *)src; \
408 int i, num_pixels = src_size / 6; \
409 \
410 for (i = 0; i < num_pixels; i++) { \
411 d[4 * i ] = swap ? av_bswap16(s[3 * i ]) : s[3 * i ]; \
412 d[4 * i + 1] = swap ? av_bswap16(s[3 * i + 1]) : s[3 * i + 1]; \
413 d[4 * i + 2] = swap ? av_bswap16(s[3 * i + 2]) : s[3 * i + 2]; \
414 d[4 * i + 3] = 0xFFFF; \
415 } \
416 }
417
418 DEFINE_RGB48TO64(nobswap, 0)
419 DEFINE_RGB48TO64(bswap, 1)
420