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 av_buffer_unref(&h->a53_caption.buf_ref);
57 int ff_h264_sei_process_picture_timing(H264SEIPictureTiming *h, const SPS *sps,
62 init_get_bits(&gb, h->payload, h->payload_size_bits);
64 if (sps->nal_hrd_parameters_present_flag ||
65 sps->vcl_hrd_parameters_present_flag) {
66 h->cpb_removal_delay = get_bits_long(&gb, sps->cpb_removal_delay_length);
67 h->dpb_output_delay = get_bits_long(&gb, sps->dpb_output_delay_length);
69 if (sps->pic_struct_present_flag) {
70 unsigned int i, num_clock_ts;
72 h->pic_struct = get_bits(&gb, 4);
75 if (h->pic_struct > H264_SEI_PIC_STRUCT_FRAME_TRIPLING)
76 return AVERROR_INVALIDDATA;
78 num_clock_ts = sei_num_clock_ts_table[h->pic_struct];
80 for (i = 0; i < num_clock_ts; i++) {
81 if (get_bits(&gb, 1)) { /* clock_timestamp_flag */
82 H264SEITimeCode *tc = &h->timecode[h->timecode_cnt++];
83 unsigned int full_timestamp_flag;
84 unsigned int counting_type, cnt_dropped_flag;
85 h->ct_type |= 1 << get_bits(&gb, 2);
86 skip_bits(&gb, 1); /* nuit_field_based_flag */
87 counting_type = get_bits(&gb, 5); /* counting_type */
88 full_timestamp_flag = get_bits(&gb, 1);
89 skip_bits(&gb, 1); /* discontinuity_flag */
90 cnt_dropped_flag = get_bits(&gb, 1); /* cnt_dropped_flag */
91 if (cnt_dropped_flag && counting_type > 1 && counting_type < 7)
93 tc->frame = get_bits(&gb, 8); /* n_frames */
94 if (full_timestamp_flag) {
96 tc->seconds = get_bits(&gb, 6); /* seconds_value 0..59 */
97 tc->minutes = get_bits(&gb, 6); /* minutes_value 0..59 */
98 tc->hours = get_bits(&gb, 5); /* hours_value 0..23 */
100 tc->seconds = tc->minutes = tc->hours = tc->full = 0;
101 if (get_bits(&gb, 1)) { /* seconds_flag */
102 tc->seconds = get_bits(&gb, 6);
103 if (get_bits(&gb, 1)) { /* minutes_flag */
104 tc->minutes = get_bits(&gb, 6);
105 if (get_bits(&gb, 1)) /* hours_flag */
106 tc->hours = get_bits(&gb, 5);
111 if (sps->time_offset_length > 0)
113 sps->time_offset_length); /* time_offset */
117 av_log(logctx, AV_LOG_DEBUG, "ct_type:%X pic_struct:%d\n",
118 h->ct_type, h->pic_struct);
124 static int decode_picture_timing(H264SEIPictureTiming *h, GetBitContext *gb,
127 int index = get_bits_count(gb);
128 int size_bits = get_bits_left(gb);
129 int size = (size_bits + 7) / 8;
132 av_log(logctx, AV_LOG_ERROR, "Unaligned SEI payload\n");
133 return AVERROR_INVALIDDATA;
135 if (size > sizeof(h->payload)) {
136 av_log(logctx, AV_LOG_ERROR, "Picture timing SEI payload too large\n");
137 return AVERROR_INVALIDDATA;
139 memcpy(h->payload, gb->buffer + index / 8, size);
141 h->payload_size_bits = size_bits;
147 static int decode_registered_user_data_afd(H264SEIAFD *h, GetBitContext *gb, int size)
152 return AVERROR_INVALIDDATA;
153 skip_bits(gb, 1); // 0
154 flag = get_bits(gb, 1); // active_format_flag
155 skip_bits(gb, 6); // reserved
159 return AVERROR_INVALIDDATA;
160 skip_bits(gb, 4); // reserved
161 h->active_format_description = get_bits(gb, 4);
168 static int decode_registered_user_data_closed_caption(H264SEIA53Caption *h,
169 GetBitContext *gb, void *logctx,
173 int user_data_type_code;
177 return AVERROR(EINVAL);
179 user_data_type_code = get_bits(gb, 8);
180 if (user_data_type_code == 0x3) {
181 skip_bits(gb, 1); // reserved
183 flag = get_bits(gb, 1); // process_cc_data_flag
185 skip_bits(gb, 1); // zero bit
186 cc_count = get_bits(gb, 5);
187 skip_bits(gb, 8); // reserved
190 if (cc_count && size >= cc_count * 3) {
191 int old_size = h->buf_ref ? h->buf_ref->size : 0;
192 const uint64_t new_size = (old_size + cc_count
196 if (new_size > INT_MAX)
197 return AVERROR(EINVAL);
199 /* Allow merging of the cc data from two fields. */
200 ret = av_buffer_realloc(&h->buf_ref, new_size);
204 /* Use of av_buffer_realloc assumes buffer is writeable */
205 for (i = 0; i < cc_count; i++) {
206 h->buf_ref->data[old_size++] = get_bits(gb, 8);
207 h->buf_ref->data[old_size++] = get_bits(gb, 8);
208 h->buf_ref->data[old_size++] = get_bits(gb, 8);
211 skip_bits(gb, 8); // marker_bits
216 for (i = 0; i < size - 1; i++)
223 static int decode_registered_user_data(H264SEIContext *h, GetBitContext *gb,
224 void *logctx, int size)
226 uint32_t country_code;
227 uint32_t user_identifier;
230 return AVERROR_INVALIDDATA;
233 country_code = get_bits(gb, 8); // itu_t_t35_country_code
234 if (country_code == 0xFF) {
235 skip_bits(gb, 8); // itu_t_t35_country_code_extension_byte
239 /* itu_t_t35_payload_byte follows */
240 skip_bits(gb, 8); // terminal provider code
241 skip_bits(gb, 8); // terminal provider oriented code
242 user_identifier = get_bits_long(gb, 32);
244 switch (user_identifier) {
245 case MKBETAG('D', 'T', 'G', '1'): // afd_data
246 return decode_registered_user_data_afd(&h->afd, gb, size);
247 case MKBETAG('G', 'A', '9', '4'): // closed captions
248 return decode_registered_user_data_closed_caption(&h->a53_caption, gb,
251 skip_bits(gb, size * 8);
258 static int decode_unregistered_user_data(H264SEIUnregistered *h, GetBitContext *gb,
259 void *logctx, int size)
264 if (size < 16 || size >= INT_MAX - 1)
265 return AVERROR_INVALIDDATA;
267 user_data = av_malloc(size + 1);
269 return AVERROR(ENOMEM);
271 for (i = 0; i < size; i++)
272 user_data[i] = get_bits(gb, 8);
275 e = sscanf(user_data + 16, "x264 - core %d", &build);
276 if (e == 1 && build > 0)
277 h->x264_build = build;
278 if (e == 1 && build == 1 && !strncmp(user_data+16, "x264 - core 0000", 16))
285 static int decode_recovery_point(H264SEIRecoveryPoint *h, GetBitContext *gb, void *logctx)
287 unsigned recovery_frame_cnt = get_ue_golomb_long(gb);
289 if (recovery_frame_cnt >= (1<<MAX_LOG2_MAX_FRAME_NUM)) {
290 av_log(logctx, AV_LOG_ERROR, "recovery_frame_cnt %u is out of range\n", recovery_frame_cnt);
291 return AVERROR_INVALIDDATA;
294 h->recovery_frame_cnt = recovery_frame_cnt;
295 /* 1b exact_match_flag,
296 * 1b broken_link_flag,
297 * 2b changing_slice_group_idc */
303 static int decode_buffering_period(H264SEIBufferingPeriod *h, GetBitContext *gb,
304 const H264ParamSets *ps, void *logctx)
310 sps_id = get_ue_golomb_31(gb);
311 if (sps_id > 31 || !ps->sps_list[sps_id]) {
312 av_log(logctx, AV_LOG_ERROR,
313 "non-existing SPS %d referenced in buffering period\n", sps_id);
314 return sps_id > 31 ? AVERROR_INVALIDDATA : AVERROR_PS_NOT_FOUND;
316 sps = (const SPS*)ps->sps_list[sps_id]->data;
318 // NOTE: This is really so duplicated in the standard... See H.264, D.1.1
319 if (sps->nal_hrd_parameters_present_flag) {
320 for (sched_sel_idx = 0; sched_sel_idx < sps->cpb_cnt; sched_sel_idx++) {
321 h->initial_cpb_removal_delay[sched_sel_idx] =
322 get_bits_long(gb, sps->initial_cpb_removal_delay_length);
323 // initial_cpb_removal_delay_offset
324 skip_bits(gb, sps->initial_cpb_removal_delay_length);
327 if (sps->vcl_hrd_parameters_present_flag) {
328 for (sched_sel_idx = 0; sched_sel_idx < sps->cpb_cnt; sched_sel_idx++) {
329 h->initial_cpb_removal_delay[sched_sel_idx] =
330 get_bits_long(gb, sps->initial_cpb_removal_delay_length);
331 // initial_cpb_removal_delay_offset
332 skip_bits(gb, sps->initial_cpb_removal_delay_length);
340 static int decode_frame_packing_arrangement(H264SEIFramePacking *h,
343 h->arrangement_id = get_ue_golomb_long(gb);
344 h->arrangement_cancel_flag = get_bits1(gb);
345 h->present = !h->arrangement_cancel_flag;
348 h->arrangement_type = get_bits(gb, 7);
349 h->quincunx_sampling_flag = get_bits1(gb);
350 h->content_interpretation_type = get_bits(gb, 6);
352 // spatial_flipping_flag, frame0_flipped_flag, field_views_flag
354 h->current_frame_is_frame0_flag = get_bits1(gb);
355 // frame0_self_contained_flag, frame1_self_contained_flag
358 if (!h->quincunx_sampling_flag && h->arrangement_type != 5)
359 skip_bits(gb, 16); // frame[01]_grid_position_[xy]
360 skip_bits(gb, 8); // frame_packing_arrangement_reserved_byte
361 h->arrangement_repetition_period = get_ue_golomb_long(gb);
363 skip_bits1(gb); // frame_packing_arrangement_extension_flag
368 static int decode_display_orientation(H264SEIDisplayOrientation *h,
371 h->present = !get_bits1(gb);
374 h->hflip = get_bits1(gb); // hor_flip
375 h->vflip = get_bits1(gb); // ver_flip
377 h->anticlockwise_rotation = get_bits(gb, 16);
378 get_ue_golomb_long(gb); // display_orientation_repetition_period
379 skip_bits1(gb); // display_orientation_extension_flag
385 static int decode_green_metadata(H264SEIGreenMetaData *h, GetBitContext *gb)
387 h->green_metadata_type = get_bits(gb, 8);
389 if (h->green_metadata_type == 0) {
390 h->period_type = get_bits(gb, 8);
392 if (h->period_type == 2)
393 h->num_seconds = get_bits(gb, 16);
394 else if (h->period_type == 3)
395 h->num_pictures = get_bits(gb, 16);
397 h->percent_non_zero_macroblocks = get_bits(gb, 8);
398 h->percent_intra_coded_macroblocks = get_bits(gb, 8);
399 h->percent_six_tap_filtering = get_bits(gb, 8);
400 h->percent_alpha_point_deblocking_instance = get_bits(gb, 8);
402 } else if (h->green_metadata_type == 1) {
403 h->xsd_metric_type = get_bits(gb, 8);
404 h->xsd_metric_value = get_bits(gb, 16);
410 static int decode_alternative_transfer(H264SEIAlternativeTransfer *h,
414 h->preferred_transfer_characteristics = get_bits(gb, 8);
418 int ff_h264_sei_decode(H264SEIContext *h, GetBitContext *gb,
419 const H264ParamSets *ps, void *logctx)
423 while (get_bits_left(gb) > 16 && show_bits(gb, 16)) {
424 GetBitContext gb_payload;
430 if (get_bits_left(gb) < 8)
431 return AVERROR_INVALIDDATA;
432 type += show_bits(gb, 8);
433 } while (get_bits(gb, 8) == 255);
436 if (get_bits_left(gb) < 8)
437 return AVERROR_INVALIDDATA;
438 size += show_bits(gb, 8);
439 } while (get_bits(gb, 8) == 255);
441 if (size > get_bits_left(gb) / 8) {
442 av_log(logctx, AV_LOG_ERROR, "SEI type %d size %d truncated at %d\n",
443 type, 8*size, get_bits_left(gb));
444 return AVERROR_INVALIDDATA;
447 ret = init_get_bits8(&gb_payload, gb->buffer + get_bits_count(gb) / 8, size);
452 case H264_SEI_TYPE_PIC_TIMING: // Picture timing SEI
453 ret = decode_picture_timing(&h->picture_timing, &gb_payload, logctx);
455 case H264_SEI_TYPE_USER_DATA_REGISTERED:
456 ret = decode_registered_user_data(h, &gb_payload, logctx, size);
458 case H264_SEI_TYPE_USER_DATA_UNREGISTERED:
459 ret = decode_unregistered_user_data(&h->unregistered, &gb_payload, logctx, size);
461 case H264_SEI_TYPE_RECOVERY_POINT:
462 ret = decode_recovery_point(&h->recovery_point, &gb_payload, logctx);
464 case H264_SEI_TYPE_BUFFERING_PERIOD:
465 ret = decode_buffering_period(&h->buffering_period, &gb_payload, ps, logctx);
467 case H264_SEI_TYPE_FRAME_PACKING:
468 ret = decode_frame_packing_arrangement(&h->frame_packing, &gb_payload);
470 case H264_SEI_TYPE_DISPLAY_ORIENTATION:
471 ret = decode_display_orientation(&h->display_orientation, &gb_payload);
473 case H264_SEI_TYPE_GREEN_METADATA:
474 ret = decode_green_metadata(&h->green_metadata, &gb_payload);
476 case H264_SEI_TYPE_ALTERNATIVE_TRANSFER:
477 ret = decode_alternative_transfer(&h->alternative_transfer, &gb_payload);
480 av_log(logctx, AV_LOG_DEBUG, "unknown SEI type %d\n", type);
482 if (ret < 0 && ret != AVERROR_PS_NOT_FOUND)
487 if (get_bits_left(&gb_payload) < 0) {
488 av_log(logctx, AV_LOG_WARNING, "SEI type %d overread by %d bits\n",
489 type, -get_bits_left(&gb_payload));
492 skip_bits_long(gb, 8 * size);
498 const char *ff_h264_sei_stereo_mode(const H264SEIFramePacking *h)
500 if (h->arrangement_cancel_flag == 0) {
501 switch (h->arrangement_type) {
502 case H264_SEI_FPA_TYPE_CHECKERBOARD:
503 if (h->content_interpretation_type == 2)
504 return "checkerboard_rl";
506 return "checkerboard_lr";
507 case H264_SEI_FPA_TYPE_INTERLEAVE_COLUMN:
508 if (h->content_interpretation_type == 2)
509 return "col_interleaved_rl";
511 return "col_interleaved_lr";
512 case H264_SEI_FPA_TYPE_INTERLEAVE_ROW:
513 if (h->content_interpretation_type == 2)
514 return "row_interleaved_rl";
516 return "row_interleaved_lr";
517 case H264_SEI_FPA_TYPE_SIDE_BY_SIDE:
518 if (h->content_interpretation_type == 2)
522 case H264_SEI_FPA_TYPE_TOP_BOTTOM:
523 if (h->content_interpretation_type == 2)
527 case H264_SEI_FPA_TYPE_INTERLEAVE_TEMPORAL:
528 if (h->content_interpretation_type == 2)
532 case H264_SEI_FPA_TYPE_2D:
536 } else if (h->arrangement_cancel_flag == 1) {