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->picture_timing.present = 0;
49 h->buffering_period.present = 0;
50 h->frame_packing.present = 0;
51 h->display_orientation.present = 0;
54 h->a53_caption.a53_caption_size = 0;
55 av_freep(&h->a53_caption.a53_caption);
58 static int decode_picture_timing(H264SEIPictureTiming *h, GetBitContext *gb,
59 const H264ParamSets *ps, void *logctx)
62 const SPS *sps = ps->sps;
64 for (i = 0; i<MAX_SPS_COUNT; i++)
65 if ((!sps || !sps->log2_max_frame_num) && ps->sps_list[i])
66 sps = (const SPS *)ps->sps_list[i]->data;
69 av_log(logctx, AV_LOG_ERROR, "SPS unavailable in decode_picture_timing\n");
70 return AVERROR_PS_NOT_FOUND;
73 if (sps->nal_hrd_parameters_present_flag ||
74 sps->vcl_hrd_parameters_present_flag) {
75 h->cpb_removal_delay = get_bits_long(gb, sps->cpb_removal_delay_length);
76 h->dpb_output_delay = get_bits_long(gb, sps->dpb_output_delay_length);
78 if (sps->pic_struct_present_flag) {
79 unsigned int i, num_clock_ts;
81 h->pic_struct = get_bits(gb, 4);
84 if (h->pic_struct > H264_SEI_PIC_STRUCT_FRAME_TRIPLING)
85 return AVERROR_INVALIDDATA;
87 num_clock_ts = sei_num_clock_ts_table[h->pic_struct];
89 for (i = 0; i < num_clock_ts; i++) {
90 if (get_bits(gb, 1)) { /* clock_timestamp_flag */
91 unsigned int full_timestamp_flag;
93 h->ct_type |= 1 << get_bits(gb, 2);
94 skip_bits(gb, 1); /* nuit_field_based_flag */
95 skip_bits(gb, 5); /* counting_type */
96 full_timestamp_flag = get_bits(gb, 1);
97 skip_bits(gb, 1); /* discontinuity_flag */
98 skip_bits(gb, 1); /* cnt_dropped_flag */
99 skip_bits(gb, 8); /* n_frames */
100 if (full_timestamp_flag) {
101 skip_bits(gb, 6); /* seconds_value 0..59 */
102 skip_bits(gb, 6); /* minutes_value 0..59 */
103 skip_bits(gb, 5); /* hours_value 0..23 */
105 if (get_bits(gb, 1)) { /* seconds_flag */
106 skip_bits(gb, 6); /* seconds_value range 0..59 */
107 if (get_bits(gb, 1)) { /* minutes_flag */
108 skip_bits(gb, 6); /* minutes_value 0..59 */
109 if (get_bits(gb, 1)) /* hours_flag */
110 skip_bits(gb, 5); /* hours_value 0..23 */
114 if (sps->time_offset_length > 0)
116 sps->time_offset_length); /* time_offset */
120 av_log(logctx, AV_LOG_DEBUG, "ct_type:%X pic_struct:%d\n",
121 h->ct_type, h->pic_struct);
128 static int decode_registered_user_data_afd(H264SEIAFD *h, GetBitContext *gb, int size)
133 return AVERROR_INVALIDDATA;
134 skip_bits(gb, 1); // 0
135 flag = get_bits(gb, 1); // active_format_flag
136 skip_bits(gb, 6); // reserved
140 return AVERROR_INVALIDDATA;
141 skip_bits(gb, 4); // reserved
142 h->active_format_description = get_bits(gb, 4);
149 static int decode_registered_user_data_closed_caption(H264SEIA53Caption *h,
150 GetBitContext *gb, void *logctx,
154 int user_data_type_code;
158 return AVERROR(EINVAL);
160 user_data_type_code = get_bits(gb, 8);
161 if (user_data_type_code == 0x3) {
162 skip_bits(gb, 1); // reserved
164 flag = get_bits(gb, 1); // process_cc_data_flag
166 skip_bits(gb, 1); // zero bit
167 cc_count = get_bits(gb, 5);
168 skip_bits(gb, 8); // reserved
171 if (cc_count && size >= cc_count * 3) {
172 const uint64_t new_size = (h->a53_caption_size + cc_count
176 if (new_size > INT_MAX)
177 return AVERROR(EINVAL);
179 /* Allow merging of the cc data from two fields. */
180 ret = av_reallocp(&h->a53_caption, new_size);
184 for (i = 0; i < cc_count; i++) {
185 h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
186 h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
187 h->a53_caption[h->a53_caption_size++] = get_bits(gb, 8);
190 skip_bits(gb, 8); // marker_bits
195 for (i = 0; i < size - 1; i++)
202 static int decode_registered_user_data(H264SEIContext *h, GetBitContext *gb,
203 void *logctx, int size)
205 uint32_t country_code;
206 uint32_t user_identifier;
209 return AVERROR_INVALIDDATA;
212 country_code = get_bits(gb, 8); // itu_t_t35_country_code
213 if (country_code == 0xFF) {
214 skip_bits(gb, 8); // itu_t_t35_country_code_extension_byte
218 /* itu_t_t35_payload_byte follows */
219 skip_bits(gb, 8); // terminal provider code
220 skip_bits(gb, 8); // terminal provider oriented code
221 user_identifier = get_bits_long(gb, 32);
223 switch (user_identifier) {
224 case MKBETAG('D', 'T', 'G', '1'): // afd_data
225 return decode_registered_user_data_afd(&h->afd, gb, size);
226 case MKBETAG('G', 'A', '9', '4'): // closed captions
227 return decode_registered_user_data_closed_caption(&h->a53_caption, gb,
230 skip_bits(gb, size * 8);
237 static int decode_unregistered_user_data(H264SEIUnregistered *h, GetBitContext *gb,
238 void *logctx, int size)
243 if (size < 16 || size >= INT_MAX - 16)
244 return AVERROR_INVALIDDATA;
246 user_data = av_malloc(16 + size + 1);
248 return AVERROR(ENOMEM);
250 for (i = 0; i < size + 16; i++)
251 user_data[i] = get_bits(gb, 8);
254 e = sscanf(user_data + 16, "x264 - core %d", &build);
255 if (e == 1 && build > 0)
256 h->x264_build = build;
257 if (e == 1 && build == 1 && !strncmp(user_data+16, "x264 - core 0000", 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->arrangement_id = get_ue_golomb_long(gb);
317 h->arrangement_cancel_flag = get_bits1(gb);
318 h->present = !h->arrangement_cancel_flag;
321 h->arrangement_type = get_bits(gb, 7);
322 h->quincunx_sampling_flag = get_bits1(gb);
323 h->content_interpretation_type = get_bits(gb, 6);
325 // spatial_flipping_flag, frame0_flipped_flag, field_views_flag
327 h->current_frame_is_frame0_flag = get_bits1(gb);
328 // frame0_self_contained_flag, frame1_self_contained_flag
331 if (!h->quincunx_sampling_flag && h->arrangement_type != 5)
332 skip_bits(gb, 16); // frame[01]_grid_position_[xy]
333 skip_bits(gb, 8); // frame_packing_arrangement_reserved_byte
334 h->arrangement_repetition_period = get_ue_golomb_long(gb);
336 skip_bits1(gb); // frame_packing_arrangement_extension_flag
341 static int decode_display_orientation(H264SEIDisplayOrientation *h,
344 h->present = !get_bits1(gb);
347 h->hflip = get_bits1(gb); // hor_flip
348 h->vflip = get_bits1(gb); // ver_flip
350 h->anticlockwise_rotation = get_bits(gb, 16);
351 get_ue_golomb_long(gb); // display_orientation_repetition_period
352 skip_bits1(gb); // display_orientation_extension_flag
358 static int decode_green_metadata(H264SEIGreenMetaData *h, GetBitContext *gb)
360 h->green_metadata_type = get_bits(gb, 8);
362 if (h->green_metadata_type == 0) {
363 h->period_type = get_bits(gb, 8);
365 if (h->period_type == 2)
366 h->num_seconds = get_bits(gb, 16);
367 else if (h->period_type == 3)
368 h->num_pictures = get_bits(gb, 16);
370 h->percent_non_zero_macroblocks = get_bits(gb, 8);
371 h->percent_intra_coded_macroblocks = get_bits(gb, 8);
372 h->percent_six_tap_filtering = get_bits(gb, 8);
373 h->percent_alpha_point_deblocking_instance = get_bits(gb, 8);
375 } else if (h->green_metadata_type == 1) {
376 h->xsd_metric_type = get_bits(gb, 8);
377 h->xsd_metric_value = get_bits(gb, 16);
383 static int decode_alternative_transfer(H264SEIAlternativeTransfer *h,
387 h->preferred_transfer_characteristics = get_bits(gb, 8);
391 int ff_h264_sei_decode(H264SEIContext *h, GetBitContext *gb,
392 const H264ParamSets *ps, void *logctx)
396 while (get_bits_left(gb) > 16 && show_bits(gb, 16)) {
403 if (get_bits_left(gb) < 8)
404 return AVERROR_INVALIDDATA;
405 type += show_bits(gb, 8);
406 } while (get_bits(gb, 8) == 255);
409 if (get_bits_left(gb) < 8)
410 return AVERROR_INVALIDDATA;
411 size += show_bits(gb, 8);
412 } while (get_bits(gb, 8) == 255);
414 if (size > get_bits_left(gb) / 8) {
415 av_log(logctx, AV_LOG_ERROR, "SEI type %d size %d truncated at %d\n",
416 type, 8*size, get_bits_left(gb));
417 return AVERROR_INVALIDDATA;
419 next = get_bits_count(gb) + 8 * size;
422 case H264_SEI_TYPE_PIC_TIMING: // Picture timing SEI
423 ret = decode_picture_timing(&h->picture_timing, gb, ps, logctx);
425 case H264_SEI_TYPE_USER_DATA_REGISTERED:
426 ret = decode_registered_user_data(h, gb, logctx, size);
428 case H264_SEI_TYPE_USER_DATA_UNREGISTERED:
429 ret = decode_unregistered_user_data(&h->unregistered, gb, logctx, size);
431 case H264_SEI_TYPE_RECOVERY_POINT:
432 ret = decode_recovery_point(&h->recovery_point, gb);
434 case H264_SEI_TYPE_BUFFERING_PERIOD:
435 ret = decode_buffering_period(&h->buffering_period, gb, ps, logctx);
437 case H264_SEI_TYPE_FRAME_PACKING:
438 ret = decode_frame_packing_arrangement(&h->frame_packing, gb);
440 case H264_SEI_TYPE_DISPLAY_ORIENTATION:
441 ret = decode_display_orientation(&h->display_orientation, gb);
443 case H264_SEI_TYPE_GREEN_METADATA:
444 ret = decode_green_metadata(&h->green_metadata, gb);
446 case H264_SEI_TYPE_ALTERNATIVE_TRANSFER:
447 ret = decode_alternative_transfer(&h->alternative_transfer, gb);
450 av_log(logctx, AV_LOG_DEBUG, "unknown SEI type %d\n", type);
452 if (ret < 0 && ret != AVERROR_PS_NOT_FOUND)
457 skip_bits_long(gb, next - get_bits_count(gb));
459 // FIXME check bits here
466 const char *ff_h264_sei_stereo_mode(const H264SEIFramePacking *h)
468 if (h->arrangement_cancel_flag == 0) {
469 switch (h->arrangement_type) {
470 case H264_SEI_FPA_TYPE_CHECKERBOARD:
471 if (h->content_interpretation_type == 2)
472 return "checkerboard_rl";
474 return "checkerboard_lr";
475 case H264_SEI_FPA_TYPE_INTERLEAVE_COLUMN:
476 if (h->content_interpretation_type == 2)
477 return "col_interleaved_rl";
479 return "col_interleaved_lr";
480 case H264_SEI_FPA_TYPE_INTERLEAVE_ROW:
481 if (h->content_interpretation_type == 2)
482 return "row_interleaved_rl";
484 return "row_interleaved_lr";
485 case H264_SEI_FPA_TYPE_SIDE_BY_SIDE:
486 if (h->content_interpretation_type == 2)
490 case H264_SEI_FPA_TYPE_TOP_BOTTOM:
491 if (h->content_interpretation_type == 2)
495 case H264_SEI_FPA_TYPE_INTERLEAVE_TEMPORAL:
496 if (h->content_interpretation_type == 2)
500 case H264_SEI_FPA_TYPE_2D:
504 } else if (h->arrangement_cancel_flag == 1) {