blob: 26afdc666b4304604e0bfead205a30c72b72d8c1
1 | /* |
2 | * DVD subtitle encoding |
3 | * Copyright (c) 2005 Wolfram Gloger |
4 | * |
5 | * This file is part of FFmpeg. |
6 | * |
7 | * FFmpeg is free software; you can redistribute it and/or |
8 | * modify it under the terms of the GNU Lesser General Public |
9 | * License as published by the Free Software Foundation; either |
10 | * version 2.1 of the License, or (at your option) any later version. |
11 | * |
12 | * FFmpeg is distributed in the hope that it will be useful, |
13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
15 | * Lesser General Public License for more details. |
16 | * |
17 | * You should have received a copy of the GNU Lesser General Public |
18 | * License along with FFmpeg; if not, write to the Free Software |
19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
20 | */ |
21 | #include "avcodec.h" |
22 | #include "bytestream.h" |
23 | #include "internal.h" |
24 | #include "libavutil/avassert.h" |
25 | #include "libavutil/bprint.h" |
26 | #include "libavutil/imgutils.h" |
27 | #include "libavutil/opt.h" |
28 | |
29 | typedef struct { |
30 | AVClass *class; |
31 | uint32_t global_palette[16]; |
32 | int even_rows_fix; |
33 | } DVDSubtitleContext; |
34 | |
35 | // ncnt is the nibble counter |
36 | #define PUTNIBBLE(val)\ |
37 | do {\ |
38 | if (ncnt++ & 1)\ |
39 | *q++ = bitbuf | ((val) & 0x0f);\ |
40 | else\ |
41 | bitbuf = (val) << 4;\ |
42 | } while(0) |
43 | |
44 | static void dvd_encode_rle(uint8_t **pq, |
45 | const uint8_t *bitmap, int linesize, |
46 | int w, int h, |
47 | const int cmap[256]) |
48 | { |
49 | uint8_t *q; |
50 | unsigned int bitbuf = 0; |
51 | int ncnt; |
52 | int x, y, len, color; |
53 | |
54 | q = *pq; |
55 | |
56 | for (y = 0; y < h; ++y) { |
57 | ncnt = 0; |
58 | for(x = 0; x < w; x += len) { |
59 | color = bitmap[x]; |
60 | for (len=1; x+len < w; ++len) |
61 | if (bitmap[x+len] != color) |
62 | break; |
63 | color = cmap[color]; |
64 | av_assert0(color < 4); |
65 | if (len < 0x04) { |
66 | PUTNIBBLE((len << 2)|color); |
67 | } else if (len < 0x10) { |
68 | PUTNIBBLE(len >> 2); |
69 | PUTNIBBLE((len << 2)|color); |
70 | } else if (len < 0x40) { |
71 | PUTNIBBLE(0); |
72 | PUTNIBBLE(len >> 2); |
73 | PUTNIBBLE((len << 2)|color); |
74 | } else if (x+len == w) { |
75 | PUTNIBBLE(0); |
76 | PUTNIBBLE(0); |
77 | PUTNIBBLE(0); |
78 | PUTNIBBLE(color); |
79 | } else { |
80 | if (len > 0xff) |
81 | len = 0xff; |
82 | PUTNIBBLE(0); |
83 | PUTNIBBLE(len >> 6); |
84 | PUTNIBBLE(len >> 2); |
85 | PUTNIBBLE((len << 2)|color); |
86 | } |
87 | } |
88 | /* end of line */ |
89 | if (ncnt & 1) |
90 | PUTNIBBLE(0); |
91 | bitmap += linesize; |
92 | } |
93 | |
94 | *pq = q; |
95 | } |
96 | |
97 | static int color_distance(uint32_t a, uint32_t b) |
98 | { |
99 | int r = 0, d, i; |
100 | int alpha_a = 8, alpha_b = 8; |
101 | |
102 | for (i = 24; i >= 0; i -= 8) { |
103 | d = alpha_a * (int)((a >> i) & 0xFF) - |
104 | alpha_b * (int)((b >> i) & 0xFF); |
105 | r += d * d; |
106 | alpha_a = a >> 28; |
107 | alpha_b = b >> 28; |
108 | } |
109 | return r; |
110 | } |
111 | |
112 | /** |
113 | * Count colors used in a rectangle, quantizing alpha and grouping by |
114 | * nearest global palette entry. |
115 | */ |
116 | static void count_colors(AVCodecContext *avctx, unsigned hits[33], |
117 | const AVSubtitleRect *r) |
118 | { |
119 | DVDSubtitleContext *dvdc = avctx->priv_data; |
120 | unsigned count[256] = { 0 }; |
121 | uint32_t *palette = (uint32_t *)r->data[1]; |
122 | uint32_t color; |
123 | int x, y, i, j, match, d, best_d, av_uninit(best_j); |
124 | uint8_t *p = r->data[0]; |
125 | |
126 | for (y = 0; y < r->h; y++) { |
127 | for (x = 0; x < r->w; x++) |
128 | count[*(p++)]++; |
129 | p += r->linesize[0] - r->w; |
130 | } |
131 | for (i = 0; i < 256; i++) { |
132 | if (!count[i]) /* avoid useless search */ |
133 | continue; |
134 | color = palette[i]; |
135 | /* 0: transparent, 1-16: semi-transparent, 17-33 opaque */ |
136 | match = color < 0x33000000 ? 0 : color < 0xCC000000 ? 1 : 17; |
137 | if (match) { |
138 | best_d = INT_MAX; |
139 | for (j = 0; j < 16; j++) { |
140 | d = color_distance(0xFF000000 | color, |
141 | 0xFF000000 | dvdc->global_palette[j]); |
142 | if (d < best_d) { |
143 | best_d = d; |
144 | best_j = j; |
145 | } |
146 | } |
147 | match += best_j; |
148 | } |
149 | hits[match] += count[i]; |
150 | } |
151 | } |
152 | |
153 | static void select_palette(AVCodecContext *avctx, int out_palette[4], |
154 | int out_alpha[4], unsigned hits[33]) |
155 | { |
156 | DVDSubtitleContext *dvdc = avctx->priv_data; |
157 | int i, j, bright, mult; |
158 | uint32_t color; |
159 | int selected[4] = { 0 }; |
160 | uint32_t pseudopal[33] = { 0 }; |
161 | uint32_t refcolor[3] = { 0x00000000, 0xFFFFFFFF, 0xFF000000 }; |
162 | |
163 | /* Bonus for transparent: if the rectangle fits tightly the text, the |
164 | background color can be quite rare, but it would be ugly without it */ |
165 | hits[0] *= 16; |
166 | /* Bonus for bright colors */ |
167 | for (i = 0; i < 16; i++) { |
168 | if (!(hits[1 + i] + hits[17 + i])) |
169 | continue; /* skip unused colors to gain time */ |
170 | color = dvdc->global_palette[i]; |
171 | bright = 0; |
172 | for (j = 0; j < 3; j++, color >>= 8) |
173 | bright += (color & 0xFF) < 0x40 || (color & 0xFF) >= 0xC0; |
174 | mult = 2 + FFMIN(bright, 2); |
175 | hits[ 1 + i] *= mult; |
176 | hits[17 + i] *= mult; |
177 | } |
178 | |
179 | /* Select four most frequent colors */ |
180 | for (i = 0; i < 4; i++) { |
181 | for (j = 0; j < 33; j++) |
182 | if (hits[j] > hits[selected[i]]) |
183 | selected[i] = j; |
184 | hits[selected[i]] = 0; |
185 | } |
186 | |
187 | /* Order the colors like in most DVDs: |
188 | 0: background, 1: foreground, 2: outline */ |
189 | for (i = 0; i < 16; i++) { |
190 | pseudopal[ 1 + i] = 0x80000000 | dvdc->global_palette[i]; |
191 | pseudopal[17 + i] = 0xFF000000 | dvdc->global_palette[i]; |
192 | } |
193 | for (i = 0; i < 3; i++) { |
194 | int best_d = color_distance(refcolor[i], pseudopal[selected[i]]); |
195 | for (j = i + 1; j < 4; j++) { |
196 | int d = color_distance(refcolor[i], pseudopal[selected[j]]); |
197 | if (d < best_d) { |
198 | FFSWAP(int, selected[i], selected[j]); |
199 | best_d = d; |
200 | } |
201 | } |
202 | } |
203 | |
204 | /* Output */ |
205 | for (i = 0; i < 4; i++) { |
206 | out_palette[i] = selected[i] ? (selected[i] - 1) & 0xF : 0; |
207 | out_alpha [i] = !selected[i] ? 0 : selected[i] < 17 ? 0x80 : 0xFF; |
208 | } |
209 | } |
210 | |
211 | static void build_color_map(AVCodecContext *avctx, int cmap[], |
212 | const uint32_t palette[], |
213 | const int out_palette[], unsigned int const out_alpha[]) |
214 | { |
215 | DVDSubtitleContext *dvdc = avctx->priv_data; |
216 | int i, j, d, best_d; |
217 | uint32_t pseudopal[4]; |
218 | |
219 | for (i = 0; i < 4; i++) |
220 | pseudopal[i] = (out_alpha[i] << 24) | |
221 | dvdc->global_palette[out_palette[i]]; |
222 | for (i = 0; i < 256; i++) { |
223 | best_d = INT_MAX; |
224 | for (j = 0; j < 4; j++) { |
225 | d = color_distance(pseudopal[j], palette[i]); |
226 | if (d < best_d) { |
227 | cmap[i] = j; |
228 | best_d = d; |
229 | } |
230 | } |
231 | } |
232 | } |
233 | |
234 | static void copy_rectangle(AVSubtitleRect *dst, AVSubtitleRect *src, int cmap[]) |
235 | { |
236 | int x, y; |
237 | uint8_t *p, *q; |
238 | |
239 | p = src->data[0]; |
240 | q = dst->data[0] + (src->x - dst->x) + |
241 | (src->y - dst->y) * dst->linesize[0]; |
242 | for (y = 0; y < src->h; y++) { |
243 | for (x = 0; x < src->w; x++) |
244 | *(q++) = cmap[*(p++)]; |
245 | p += src->linesize[0] - src->w; |
246 | q += dst->linesize[0] - src->w; |
247 | } |
248 | } |
249 | |
250 | static int encode_dvd_subtitles(AVCodecContext *avctx, |
251 | uint8_t *outbuf, int outbuf_size, |
252 | const AVSubtitle *h) |
253 | { |
254 | DVDSubtitleContext *dvdc = avctx->priv_data; |
255 | uint8_t *q, *qq; |
256 | int offset1, offset2; |
257 | int i, rects = h->num_rects, ret; |
258 | unsigned global_palette_hits[33] = { 0 }; |
259 | int cmap[256]; |
260 | int out_palette[4]; |
261 | int out_alpha[4]; |
262 | AVSubtitleRect vrect; |
263 | uint8_t *vrect_data = NULL; |
264 | int x2, y2; |
265 | int forced = 0; |
266 | |
267 | if (rects == 0 || !h->rects) |
268 | return AVERROR(EINVAL); |
269 | for (i = 0; i < rects; i++) |
270 | if (h->rects[i]->type != SUBTITLE_BITMAP) { |
271 | av_log(avctx, AV_LOG_ERROR, "Bitmap subtitle required\n"); |
272 | return AVERROR(EINVAL); |
273 | } |
274 | /* Mark this subtitle forced if any of the rectangles is forced. */ |
275 | for (i = 0; i < rects; i++) |
276 | if ((h->rects[i]->flags & AV_SUBTITLE_FLAG_FORCED) != 0) { |
277 | forced = 1; |
278 | break; |
279 | } |
280 | |
281 | #if FF_API_AVPICTURE |
282 | FF_DISABLE_DEPRECATION_WARNINGS |
283 | for (i = 0; i < rects; i++) |
284 | if (!h->rects[i]->data[0]) { |
285 | AVSubtitleRect *rect = h->rects[i]; |
286 | int j; |
287 | for (j = 0; j < 4; j++) { |
288 | rect->data[j] = rect->pict.data[j]; |
289 | rect->linesize[j] = rect->pict.linesize[j]; |
290 | } |
291 | } |
292 | FF_ENABLE_DEPRECATION_WARNINGS |
293 | #endif |
294 | |
295 | vrect = *h->rects[0]; |
296 | |
297 | if (rects > 1) { |
298 | /* DVD subtitles can have only one rectangle: build a virtual |
299 | rectangle containing all actual rectangles. |
300 | The data of the rectangles will be copied later, when the palette |
301 | is decided, because the rectangles may have different palettes. */ |
302 | int xmin = h->rects[0]->x, xmax = xmin + h->rects[0]->w; |
303 | int ymin = h->rects[0]->y, ymax = ymin + h->rects[0]->h; |
304 | for (i = 1; i < rects; i++) { |
305 | xmin = FFMIN(xmin, h->rects[i]->x); |
306 | ymin = FFMIN(ymin, h->rects[i]->y); |
307 | xmax = FFMAX(xmax, h->rects[i]->x + h->rects[i]->w); |
308 | ymax = FFMAX(ymax, h->rects[i]->y + h->rects[i]->h); |
309 | } |
310 | vrect.x = xmin; |
311 | vrect.y = ymin; |
312 | vrect.w = xmax - xmin; |
313 | vrect.h = ymax - ymin; |
314 | if ((ret = av_image_check_size(vrect.w, vrect.h, 0, avctx)) < 0) |
315 | return ret; |
316 | |
317 | /* Count pixels outside the virtual rectangle as transparent */ |
318 | global_palette_hits[0] = vrect.w * vrect.h; |
319 | for (i = 0; i < rects; i++) |
320 | global_palette_hits[0] -= h->rects[i]->w * h->rects[i]->h; |
321 | } |
322 | |
323 | for (i = 0; i < rects; i++) |
324 | count_colors(avctx, global_palette_hits, h->rects[i]); |
325 | select_palette(avctx, out_palette, out_alpha, global_palette_hits); |
326 | |
327 | if (rects > 1) { |
328 | if (!(vrect_data = av_calloc(vrect.w, vrect.h))) |
329 | return AVERROR(ENOMEM); |
330 | vrect.data [0] = vrect_data; |
331 | vrect.linesize[0] = vrect.w; |
332 | for (i = 0; i < rects; i++) { |
333 | build_color_map(avctx, cmap, (uint32_t *)h->rects[i]->data[1], |
334 | out_palette, out_alpha); |
335 | copy_rectangle(&vrect, h->rects[i], cmap); |
336 | } |
337 | for (i = 0; i < 4; i++) |
338 | cmap[i] = i; |
339 | } else { |
340 | build_color_map(avctx, cmap, (uint32_t *)h->rects[0]->data[1], |
341 | out_palette, out_alpha); |
342 | } |
343 | |
344 | av_log(avctx, AV_LOG_DEBUG, "Selected palette:"); |
345 | for (i = 0; i < 4; i++) |
346 | av_log(avctx, AV_LOG_DEBUG, " 0x%06"PRIx32"@@%02x (0x%x,0x%x)", |
347 | dvdc->global_palette[out_palette[i]], out_alpha[i], |
348 | out_palette[i], out_alpha[i] >> 4); |
349 | av_log(avctx, AV_LOG_DEBUG, "\n"); |
350 | |
351 | // encode data block |
352 | q = outbuf + 4; |
353 | offset1 = q - outbuf; |
354 | // worst case memory requirement: 1 nibble per pixel.. |
355 | if ((q - outbuf) + vrect.w * vrect.h / 2 + 17 + 21 > outbuf_size) { |
356 | av_log(NULL, AV_LOG_ERROR, "dvd_subtitle too big\n"); |
357 | ret = AVERROR_BUFFER_TOO_SMALL; |
358 | goto fail; |
359 | } |
360 | dvd_encode_rle(&q, vrect.data[0], vrect.w * 2, |
361 | vrect.w, (vrect.h + 1) >> 1, cmap); |
362 | offset2 = q - outbuf; |
363 | dvd_encode_rle(&q, vrect.data[0] + vrect.w, vrect.w * 2, |
364 | vrect.w, vrect.h >> 1, cmap); |
365 | |
366 | if (dvdc->even_rows_fix && (vrect.h & 1)) { |
367 | // Work-around for some players that want the height to be even. |
368 | vrect.h++; |
369 | *q++ = 0x00; // 0x00 0x00 == empty row, i.e. fully transparent |
370 | *q++ = 0x00; |
371 | } |
372 | |
373 | // set data packet size |
374 | qq = outbuf + 2; |
375 | bytestream_put_be16(&qq, q - outbuf); |
376 | |
377 | // send start display command |
378 | bytestream_put_be16(&q, (h->start_display_time*90) >> 10); |
379 | bytestream_put_be16(&q, (q - outbuf) /*- 2 */ + 8 + 12 + 2); |
380 | *q++ = 0x03; // palette - 4 nibbles |
381 | *q++ = (out_palette[3] << 4) | out_palette[2]; |
382 | *q++ = (out_palette[1] << 4) | out_palette[0]; |
383 | *q++ = 0x04; // alpha - 4 nibbles |
384 | *q++ = (out_alpha[3] & 0xF0) | (out_alpha[2] >> 4); |
385 | *q++ = (out_alpha[1] & 0xF0) | (out_alpha[0] >> 4); |
386 | |
387 | // 12 bytes per rect |
388 | x2 = vrect.x + vrect.w - 1; |
389 | y2 = vrect.y + vrect.h - 1; |
390 | |
391 | *q++ = 0x05; |
392 | // x1 x2 -> 6 nibbles |
393 | *q++ = vrect.x >> 4; |
394 | *q++ = (vrect.x << 4) | ((x2 >> 8) & 0xf); |
395 | *q++ = x2; |
396 | // y1 y2 -> 6 nibbles |
397 | *q++ = vrect.y >> 4; |
398 | *q++ = (vrect.y << 4) | ((y2 >> 8) & 0xf); |
399 | *q++ = y2; |
400 | |
401 | *q++ = 0x06; |
402 | // offset1, offset2 |
403 | bytestream_put_be16(&q, offset1); |
404 | bytestream_put_be16(&q, offset2); |
405 | |
406 | *q++ = forced ? 0x00 : 0x01; // start command |
407 | *q++ = 0xff; // terminating command |
408 | |
409 | // send stop display command last |
410 | bytestream_put_be16(&q, (h->end_display_time*90) >> 10); |
411 | bytestream_put_be16(&q, (q - outbuf) - 2 /*+ 4*/); |
412 | *q++ = 0x02; // set end |
413 | *q++ = 0xff; // terminating command |
414 | |
415 | qq = outbuf; |
416 | bytestream_put_be16(&qq, q - outbuf); |
417 | |
418 | av_log(NULL, AV_LOG_DEBUG, "subtitle_packet size=%"PTRDIFF_SPECIFIER"\n", q - outbuf); |
419 | ret = q - outbuf; |
420 | |
421 | fail: |
422 | av_free(vrect_data); |
423 | return ret; |
424 | } |
425 | |
426 | static int dvdsub_init(AVCodecContext *avctx) |
427 | { |
428 | DVDSubtitleContext *dvdc = avctx->priv_data; |
429 | static const uint32_t default_palette[16] = { |
430 | 0x000000, 0x0000FF, 0x00FF00, 0xFF0000, |
431 | 0xFFFF00, 0xFF00FF, 0x00FFFF, 0xFFFFFF, |
432 | 0x808000, 0x8080FF, 0x800080, 0x80FF80, |
433 | 0x008080, 0xFF8080, 0x555555, 0xAAAAAA, |
434 | }; |
435 | AVBPrint extradata; |
436 | int i, ret; |
437 | |
438 | av_assert0(sizeof(dvdc->global_palette) == sizeof(default_palette)); |
439 | memcpy(dvdc->global_palette, default_palette, sizeof(dvdc->global_palette)); |
440 | |
441 | av_bprint_init(&extradata, 0, 1); |
442 | if (avctx->width && avctx->height) |
443 | av_bprintf(&extradata, "size: %dx%d\n", avctx->width, avctx->height); |
444 | av_bprintf(&extradata, "palette:"); |
445 | for (i = 0; i < 16; i++) |
446 | av_bprintf(&extradata, " %06"PRIx32"%c", |
447 | dvdc->global_palette[i] & 0xFFFFFF, i < 15 ? ',' : '\n'); |
448 | |
449 | ret = avpriv_bprint_to_extradata(avctx, &extradata); |
450 | if (ret < 0) |
451 | return ret; |
452 | |
453 | return 0; |
454 | } |
455 | |
456 | static int dvdsub_encode(AVCodecContext *avctx, |
457 | unsigned char *buf, int buf_size, |
458 | const AVSubtitle *sub) |
459 | { |
460 | //DVDSubtitleContext *s = avctx->priv_data; |
461 | int ret; |
462 | |
463 | ret = encode_dvd_subtitles(avctx, buf, buf_size, sub); |
464 | return ret; |
465 | } |
466 | |
467 | #define OFFSET(x) offsetof(DVDSubtitleContext, x) |
468 | #define SE AV_OPT_FLAG_SUBTITLE_PARAM | AV_OPT_FLAG_ENCODING_PARAM |
469 | static const AVOption options[] = { |
470 | {"even_rows_fix", "Make number of rows even (workaround for some players)", OFFSET(even_rows_fix), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, SE}, |
471 | { NULL }, |
472 | }; |
473 | |
474 | static const AVClass dvdsubenc_class = { |
475 | .class_name = "VOBSUB subtitle encoder", |
476 | .item_name = av_default_item_name, |
477 | .option = options, |
478 | .version = LIBAVUTIL_VERSION_INT, |
479 | }; |
480 | |
481 | AVCodec ff_dvdsub_encoder = { |
482 | .name = "dvdsub", |
483 | .long_name = NULL_IF_CONFIG_SMALL("DVD subtitles"), |
484 | .type = AVMEDIA_TYPE_SUBTITLE, |
485 | .id = AV_CODEC_ID_DVD_SUBTITLE, |
486 | .init = dvdsub_init, |
487 | .encode_sub = dvdsub_encode, |
488 | .priv_class = &dvdsubenc_class, |
489 | .priv_data_size = sizeof(DVDSubtitleContext), |
490 | }; |
491 |