2 * H.26L/H.264/AVC/JVT/14496-10/... SEI decoding
3 * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
5 * This file is part of FFmpeg.
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.
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.
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
24 * H.264 / AVC / MPEG-4 part10 SEI decoding.
25 * @author Michael Niedermayer <michaelni@gmx.at>
35 #define AVERROR_PS_NOT_FOUND FFERRTAG(0xF8,'?','P','S')
37 static const uint8_t sei_num_clock_ts_table[9] = {
38 1, 1, 1, 2, 2, 3, 3, 2, 3
41 void ff_h264_sei_uninit(H264SEIContext *h)
43 h->recovery_point.recovery_frame_cnt = -1;
45 h->picture_timing.dpb_output_delay = 0;
46 h->picture_timing.cpb_removal_delay = -1;
48 h->buffering_period.present = 0;
49 h->frame_packing.present = 0;
50 h->display_orientation.present = 0;
53 h->a53_caption.a53_caption_size = 0;
54 av_freep(&h->a53_caption.a53_caption);
57 static int decode_picture_timing(H264SEIPictureTiming *h, GetBitContext *gb,
58 const H264ParamSets *ps, void *logctx)
61 const SPS *sps = ps->sps;
63 for (i = 0; i<MAX_SPS_COUNT; i++)
64 if ((!sps || !sps->log2_max_frame_num) && ps->sps_list[i])
65 sps = (const SPS *)ps->sps_list[i]->data;
68 av_log(logctx, AV_LOG_ERROR, "SPS unavailable in decode_picture_timing\n");
69 return AVERROR_PS_NOT_FOUND;
72 if (sps->nal_hrd_parameters_present_flag ||
73 sps->vcl_hrd_parameters_present_flag) {
74 h->cpb_removal_delay = get_bits_long(gb, sps->cpb_removal_delay_length);
75 h->dpb_output_delay = get_bits_long(gb, sps->dpb_output_delay_length);
77 if (sps->pic_struct_present_flag) {
78 unsigned int i, num_clock_ts;
80 h->pic_struct = get_bits(gb, 4);
83 if (h->pic_struct > SEI_PIC_STRUCT_FRAME_TRIPLING)
84 return AVERROR_INVALIDDATA;
86 num_clock_ts = sei_num_clock_ts_table[h->pic_struct];
88 for (i = 0; i < num_clock_ts; i++) {
89 if (get_bits(gb, 1)) { /* clock_timestamp_flag */
90 unsigned int full_timestamp_flag;
92 h->ct_type |= 1 << get_bits(gb, 2);
93 skip_bits(gb, 1); /* nuit_field_based_flag */
94 skip_bits(gb, 5); /* counting_type */
95 full_timestamp_flag = get_bits(gb, 1);
96 skip_bits(gb, 1); /* discontinuity_flag */
97 skip_bits(gb, 1); /* cnt_dropped_flag */
98 skip_bits(gb, 8); /* n_frames */
99 if (full_timestamp_flag) {
100 skip_bits(gb, 6); /* seconds_value 0..59 */
101 skip_bits(gb, 6); /* minutes_value 0..59 */
102 skip_bits(gb, 5); /* hours_value 0..23 */
104 if (get_bits(gb, 1)) { /* seconds_flag */
105 skip_bits(gb, 6); /* seconds_value range 0..59 */
106 if (get_bits(gb, 1)) { /* minutes_flag */
107 skip_bits(gb, 6); /* minutes_value 0..59 */
108 if (get_bits(gb, 1)) /* hours_flag */
109 skip_bits(gb, 5); /* hours_value 0..23 */
113 if (sps->time_offset_length > 0)
115 sps->time_offset_length); /* time_offset */
119 av_log(logctx, AV_LOG_DEBUG, "ct_type:%X pic_struct:%d\n",
120 h->ct_type, h->pic_struct);
125 static int decode_registered_user_data_afd(H264SEIAFD *h, GetBitContext *gb, int size)
130 return AVERROR_INVALIDDATA;
131 skip_bits(gb, 1); // 0
132 flag = get_bits(gb, 1); // active_format_flag
133 skip_bits(gb, 6); // reserved
137 return AVERROR_INVALIDDATA;
138 skip_bits(gb, 4); // reserved
139 h->active_format_description = get_bits(gb, 4);
146 static int decode_registered_user_data_closed_caption(H264SEIA53Caption *h,
147 GetBitContext *gb, void *logctx,
151 int user_data_type_code;
155 return AVERROR(EINVAL);
157 user_data_type_code = get_bits(gb, 8);
158 if (user_data_type_code == 0x3) {
159 skip_bits(gb, 1); // reserved
161 flag = get_bits(gb, 1); // process_cc_data_flag
163 skip_bits(gb, 1); // zero bit
164 cc_count = get_bits(gb, 5);
165 skip_bits(gb, 8); // reserved
168 if (cc_count && size >= cc_count * 3) {
169 const uint64_t new_size = (h->a53_caption_size + cc_count
173 if (new_size > INT_MAX)
174 return AVERROR(EINVAL);
176 /* Allow merging of the cc data from two fields. */
177 ret = av_reallocp(&h->a53_caption, new_size);
181 for (i = 0; i < cc_count; i++) {
182 h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
183 h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
184 h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
187 skip_bits(gb, 8); // marker_bits
192 for (i = 0; i < size - 1; i++)
199 static int decode_registered_user_data(H264SEIContext *h, GetBitContext *gb,
200 void *logctx, int size)
202 uint32_t country_code;
203 uint32_t user_identifier;
206 return AVERROR_INVALIDDATA;
209 country_code = get_bits(gb, 8); // itu_t_t35_country_code
210 if (country_code == 0xFF) {
211 skip_bits(gb, 8); // itu_t_t35_country_code_extension_byte
215 /* itu_t_t35_payload_byte follows */
216 skip_bits(gb, 8); // terminal provider code
217 skip_bits(gb, 8); // terminal provider oriented code
218 user_identifier = get_bits_long(gb, 32);
220 switch (user_identifier) {
221 case MKBETAG('D', 'T', 'G', '1'): // afd_data
222 return decode_registered_user_data_afd(&h->afd, gb, size);
223 case MKBETAG('G', 'A', '9', '4'): // closed captions
224 return decode_registered_user_data_closed_caption(&h->a53_caption, gb,
227 skip_bits(gb, size * 8);
234 static int decode_unregistered_user_data(H264SEIUnregistered *h, GetBitContext *gb,
235 void *logctx, int size)
240 if (size < 16 || size >= INT_MAX - 16)
241 return AVERROR_INVALIDDATA;
243 user_data = av_malloc(16 + size + 1);
245 return AVERROR(ENOMEM);
247 for (i = 0; i < size + 16; i++)
248 user_data[i] = get_bits(gb, 8);
251 e = sscanf(user_data + 16, "x264 - core %d", &build);
252 if (e == 1 && build > 0)
253 h->x264_build = build;
254 if (e == 1 && build == 1 && !strncmp(user_data+16, "x264 - core 0000", 16))
257 if (strlen(user_data + 16) > 0)
258 av_log(logctx, AV_LOG_DEBUG, "user data:\"%s\"\n", user_data + 16);
264 static int decode_recovery_point(H264SEIRecoveryPoint *h, GetBitContext *gb)
266 h->recovery_frame_cnt = get_ue_golomb_long(gb);
268 /* 1b exact_match_flag,
269 * 1b broken_link_flag,
270 * 2b changing_slice_group_idc */
276 static int decode_buffering_period(H264SEIBufferingPeriod *h, GetBitContext *gb,
277 const H264ParamSets *ps, void *logctx)
283 sps_id = get_ue_golomb_31(gb);
284 if (sps_id > 31 || !ps->sps_list[sps_id]) {
285 av_log(logctx, AV_LOG_ERROR,
286 "non-existing SPS %d referenced in buffering period\n", sps_id);
287 return sps_id > 31 ? AVERROR_INVALIDDATA : AVERROR_PS_NOT_FOUND;
289 sps = (const SPS*)ps->sps_list[sps_id]->data;
291 // NOTE: This is really so duplicated in the standard... See H.264, D.1.1
292 if (sps->nal_hrd_parameters_present_flag) {
293 for (sched_sel_idx = 0; sched_sel_idx < sps->cpb_cnt; sched_sel_idx++) {
294 h->initial_cpb_removal_delay[sched_sel_idx] =
295 get_bits_long(gb, sps->initial_cpb_removal_delay_length);
296 // initial_cpb_removal_delay_offset
297 skip_bits(gb, sps->initial_cpb_removal_delay_length);
300 if (sps->vcl_hrd_parameters_present_flag) {
301 for (sched_sel_idx = 0; sched_sel_idx < sps->cpb_cnt; sched_sel_idx++) {
302 h->initial_cpb_removal_delay[sched_sel_idx] =
303 get_bits_long(gb, sps->initial_cpb_removal_delay_length);
304 // initial_cpb_removal_delay_offset
305 skip_bits(gb, sps->initial_cpb_removal_delay_length);
313 static int decode_frame_packing_arrangement(H264SEIFramePacking *h,
316 h->frame_packing_arrangement_id = get_ue_golomb_long(gb);
317 h->frame_packing_arrangement_cancel_flag = get_bits1(gb);
318 h->present = !h->frame_packing_arrangement_cancel_flag;
321 h->frame_packing_arrangement_type = get_bits(gb, 7);
322 h->quincunx_sampling_flag = get_bits1(gb);
323 h->content_interpretation_type = get_bits(gb, 6);
325 // the following skips: spatial_flipping_flag, frame0_flipped_flag,
326 // field_views_flag, current_frame_is_frame0_flag,
327 // frame0_self_contained_flag, frame1_self_contained_flag
330 if (!h->quincunx_sampling_flag && h->frame_packing_arrangement_type != 5)
331 skip_bits(gb, 16); // frame[01]_grid_position_[xy]
332 skip_bits(gb, 8); // frame_packing_arrangement_reserved_byte
333 h->frame_packing_arrangement_repetition_period = get_ue_golomb_long(gb);
335 skip_bits1(gb); // frame_packing_arrangement_extension_flag
340 static int decode_display_orientation(H264SEIDisplayOrientation *h,
343 h->present = !get_bits1(gb);
346 h->hflip = get_bits1(gb); // hor_flip
347 h->vflip = get_bits1(gb); // ver_flip
349 h->anticlockwise_rotation = get_bits(gb, 16);
350 get_ue_golomb_long(gb); // display_orientation_repetition_period
351 skip_bits1(gb); // display_orientation_extension_flag
357 static int decode_green_metadata(H264SEIGreenMetaData *h, GetBitContext *gb)
359 h->green_metadata_type = get_bits(gb, 8);
361 if (h->green_metadata_type == 0) {
362 h->period_type = get_bits(gb, 8);
364 if (h->period_type == 2)
365 h->num_seconds = get_bits(gb, 16);
366 else if (h->period_type == 3)
367 h->num_pictures = get_bits(gb, 16);
369 h->percent_non_zero_macroblocks = get_bits(gb, 8);
370 h->percent_intra_coded_macroblocks = get_bits(gb, 8);
371 h->percent_six_tap_filtering = get_bits(gb, 8);
372 h->percent_alpha_point_deblocking_instance = get_bits(gb, 8);
374 } else if (h->green_metadata_type == 1) {
375 h->xsd_metric_type = get_bits(gb, 8);
376 h->xsd_metric_value = get_bits(gb, 16);
382 int ff_h264_sei_decode(H264SEIContext *h, GetBitContext *gb,
383 const H264ParamSets *ps, void *logctx)
387 while (get_bits_left(gb) > 16 && show_bits(gb, 16)) {
394 if (get_bits_left(gb) < 8)
395 return AVERROR_INVALIDDATA;
396 type += show_bits(gb, 8);
397 } while (get_bits(gb, 8) == 255);
400 if (get_bits_left(gb) < 8)
401 return AVERROR_INVALIDDATA;
402 size += show_bits(gb, 8);
403 } while (get_bits(gb, 8) == 255);
405 if (size > get_bits_left(gb) / 8) {
406 av_log(logctx, AV_LOG_ERROR, "SEI type %d size %d truncated at %d\n",
407 type, 8*size, get_bits_left(gb));
408 return AVERROR_INVALIDDATA;
410 next = get_bits_count(gb) + 8 * size;
413 case SEI_TYPE_PIC_TIMING: // Picture timing SEI
414 ret = decode_picture_timing(&h->picture_timing, gb, ps, logctx);
416 case SEI_TYPE_USER_DATA_REGISTERED:
417 ret = decode_registered_user_data(h, gb, logctx, size);
419 case SEI_TYPE_USER_DATA_UNREGISTERED:
420 ret = decode_unregistered_user_data(&h->unregistered, gb, logctx, size);
422 case SEI_TYPE_RECOVERY_POINT:
423 ret = decode_recovery_point(&h->recovery_point, gb);
425 case SEI_TYPE_BUFFERING_PERIOD:
426 ret = decode_buffering_period(&h->buffering_period, gb, ps, logctx);
428 case SEI_TYPE_FRAME_PACKING:
429 ret = decode_frame_packing_arrangement(&h->frame_packing, gb);
431 case SEI_TYPE_DISPLAY_ORIENTATION:
432 ret = decode_display_orientation(&h->display_orientation, gb);
434 case SEI_TYPE_GREEN_METADATA:
435 ret = decode_green_metadata(&h->green_metadata, gb);
438 av_log(logctx, AV_LOG_DEBUG, "unknown SEI type %d\n", type);
440 if (ret < 0 && ret != AVERROR_PS_NOT_FOUND)
445 skip_bits_long(gb, next - get_bits_count(gb));
447 // FIXME check bits here
454 const char *ff_h264_sei_stereo_mode(const H264SEIFramePacking *h)
456 if (h->frame_packing_arrangement_cancel_flag == 0) {
457 switch (h->frame_packing_arrangement_type) {
458 case SEI_FPA_TYPE_CHECKERBOARD:
459 if (h->content_interpretation_type == 2)
460 return "checkerboard_rl";
462 return "checkerboard_lr";
463 case SEI_FPA_TYPE_INTERLEAVE_COLUMN:
464 if (h->content_interpretation_type == 2)
465 return "col_interleaved_rl";
467 return "col_interleaved_lr";
468 case SEI_FPA_TYPE_INTERLEAVE_ROW:
469 if (h->content_interpretation_type == 2)
470 return "row_interleaved_rl";
472 return "row_interleaved_lr";
473 case SEI_FPA_TYPE_SIDE_BY_SIDE:
474 if (h->content_interpretation_type == 2)
478 case SEI_FPA_TYPE_TOP_BOTTOM:
479 if (h->content_interpretation_type == 2)
483 case SEI_FPA_TYPE_INTERLEAVE_TEMPORAL:
484 if (h->content_interpretation_type == 2)
488 case SEI_FPA_TYPE_2D:
492 } else if (h->frame_packing_arrangement_cancel_flag == 1) {