1 /*****************************************************************************
2 * mp4.c: x264 mp4 output module
3 *****************************************************************************
4 * Copyright (C) 2003-2009 x264 project
6 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
7 * Loren Merritt <lorenm@u.washington.edu>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
22 *****************************************************************************/
25 #include <gpac/isomedia.h>
30 GF_AVCConfig *p_config;
31 GF_ISOSample *p_sample;
40 int64_t i_prev_timestamps[2];
44 static void recompute_bitrate_mp4( GF_ISOFile *p_file, int i_track )
46 u32 i, count, di, timescale, time_wnd, rate;
51 esd = gf_isom_get_esd( p_file, i_track, 1 );
55 esd->decoderConfig->avgBitrate = 0;
56 esd->decoderConfig->maxBitrate = 0;
59 timescale = gf_isom_get_media_timescale( p_file, i_track );
60 count = gf_isom_get_sample_count( p_file, i_track );
61 for( i = 0; i < count; i++ )
63 GF_ISOSample *samp = gf_isom_get_sample_info( p_file, i_track, i+1, &di, &offset );
66 fprintf( stderr, "mp4 [error]: failure reading back frame %u\n", i );
70 if( esd->decoderConfig->bufferSizeDB < samp->dataLength )
71 esd->decoderConfig->bufferSizeDB = samp->dataLength;
73 esd->decoderConfig->avgBitrate += samp->dataLength;
74 rate += samp->dataLength;
75 if( samp->DTS > time_wnd + timescale )
77 if( rate > esd->decoderConfig->maxBitrate )
78 esd->decoderConfig->maxBitrate = rate;
83 gf_isom_sample_del( &samp );
86 br = (Double)(s64)gf_isom_get_media_duration( p_file, i_track );
88 esd->decoderConfig->avgBitrate = (u32)(esd->decoderConfig->avgBitrate / br);
90 esd->decoderConfig->avgBitrate *= 8;
91 esd->decoderConfig->maxBitrate *= 8;
93 gf_isom_change_mpeg4_description( p_file, i_track, 1, esd );
94 gf_odf_desc_del( (GF_Descriptor*)esd );
97 static int close_file( hnd_t handle, int64_t largest_pts, int64_t second_largest_pts )
99 mp4_hnd_t *p_mp4 = handle;
100 uint64_t total_duration = 0;
105 if( p_mp4->p_config )
106 gf_odf_avc_cfg_del( p_mp4->p_config );
108 if( p_mp4->p_sample )
110 if( p_mp4->p_sample->data )
111 free( p_mp4->p_sample->data );
113 gf_isom_sample_del( &p_mp4->p_sample );
118 /* The mdhd duration is defined as CTS[final] - CTS[0] + duration of last frame.
119 * The mdhd duration (in seconds) should be able to be longer than the tkhd duration since the track is managed by edts.
120 * So, if mdhd duration is equal to the last DTS or less, we give the last composition time delta to the last sample duration.
121 * And then, the mdhd duration is updated, but it time-wise doesn't give the actual duration.
122 * The tkhd duration is the actual track duration. */
123 uint64_t mdhd_duration = (2 * largest_pts - second_largest_pts - p_mp4->i_delay_time) * p_mp4->i_time_inc;
124 total_duration = gf_isom_get_media_duration( p_mp4->p_file, p_mp4->i_track );
125 if( mdhd_duration != total_duration )
127 uint64_t last_dts = gf_isom_get_sample_dts( p_mp4->p_file, p_mp4->i_track, p_mp4->i_numframe );
128 uint32_t last_duration = (uint32_t)( mdhd_duration > last_dts ? mdhd_duration - last_dts : (largest_pts - second_largest_pts) * p_mp4->i_time_inc );
129 gf_isom_set_last_sample_duration( p_mp4->p_file, p_mp4->i_track, last_duration );
130 total_duration = gf_isom_get_media_duration( p_mp4->p_file, p_mp4->i_track );
133 /* Write an Edit Box if the first CTS offset is positive.
134 * A media_time is given by not the mvhd timescale but rather the mdhd timescale.
135 * The reason is that an Edit Box maps the presentation time-line to the media time-line.
136 * Any demuxers should follow the Edit Box if it exists. */
137 GF_ISOSample *sample = gf_isom_get_sample_info( p_mp4->p_file, p_mp4->i_track, 1, NULL, NULL );
138 if( sample->CTS_Offset > 0 )
140 uint32_t mvhd_timescale = gf_isom_get_timescale( p_mp4->p_file );
141 uint64_t tkhd_duration = (uint64_t)( mdhd_duration * ( (double)mvhd_timescale / p_mp4->i_time_res ) );
142 gf_isom_append_edit_segment( p_mp4->p_file, p_mp4->i_track, tkhd_duration, sample->CTS_Offset, GF_ISOM_EDIT_NORMAL );
144 gf_isom_sample_del( &sample );
146 recompute_bitrate_mp4( p_mp4->p_file, p_mp4->i_track );
147 gf_isom_set_pl_indication( p_mp4->p_file, GF_ISOM_PL_VISUAL, 0x15 );
148 gf_isom_set_storage_mode( p_mp4->p_file, GF_ISOM_STORE_FLAT );
149 gf_isom_close( p_mp4->p_file );
157 static int open_file( char *psz_filename, hnd_t *p_handle )
162 FILE *fh = fopen( psz_filename, "w" );
165 else if( !x264_is_regular_file( fh ) )
167 fprintf( stderr, "mp4 [error]: MP4 output is incompatible with non-regular file `%s'\n", psz_filename );
172 if( !(p_mp4 = malloc( sizeof(mp4_hnd_t) )) )
175 memset( p_mp4, 0, sizeof(mp4_hnd_t) );
176 p_mp4->p_file = gf_isom_open( psz_filename, GF_ISOM_OPEN_WRITE, NULL );
178 if( !(p_mp4->p_sample = gf_isom_sample_new()) )
180 close_file( p_mp4, 0, 0 );
184 gf_isom_set_brand_info( p_mp4->p_file, GF_ISOM_BRAND_AVC1, 0 );
191 static int set_param( hnd_t handle, x264_param_t *p_param )
193 mp4_hnd_t *p_mp4 = handle;
195 p_mp4->i_time_res = p_param->i_timebase_den;
196 p_mp4->i_time_inc = p_param->i_timebase_num;
198 p_mp4->i_init_delay = p_param->i_bframe ? (p_param->i_bframe_pyramid ? 2 : 1) : 0;
200 p_mp4->i_track = gf_isom_new_track( p_mp4->p_file, 0, GF_ISOM_MEDIA_VISUAL,
203 p_mp4->p_config = gf_odf_avc_cfg_new();
204 gf_isom_avc_config_new( p_mp4->p_file, p_mp4->i_track, p_mp4->p_config,
205 NULL, NULL, &p_mp4->i_descidx );
207 gf_isom_set_track_enabled( p_mp4->p_file, p_mp4->i_track, 1 );
209 gf_isom_set_visual_info( p_mp4->p_file, p_mp4->i_track, p_mp4->i_descidx,
210 p_param->i_width, p_param->i_height );
212 if( p_param->vui.i_sar_width && p_param->vui.i_sar_height )
214 uint64_t dw = p_param->i_width << 16;
215 uint64_t dh = p_param->i_height << 16;
216 double sar = (double)p_param->vui.i_sar_width / p_param->vui.i_sar_height;
221 gf_isom_set_track_layout_info( p_mp4->p_file, p_mp4->i_track, dw, dh, 0, 0, 0 );
224 p_mp4->p_sample->data = malloc( p_param->i_width * p_param->i_height * 3 / 2 );
225 if( !p_mp4->p_sample->data )
231 static int write_headers( hnd_t handle, x264_nal_t *p_nal )
233 mp4_hnd_t *p_mp4 = handle;
234 GF_AVCConfigSlot *p_slot;
236 int sei_size = p_nal[0].i_payload;
237 int sps_size = p_nal[1].i_payload - 4;
238 int pps_size = p_nal[2].i_payload - 4;
240 uint8_t *sei = p_nal[0].p_payload;
241 uint8_t *sps = p_nal[1].p_payload + 4;
242 uint8_t *pps = p_nal[2].p_payload + 4;
246 p_mp4->p_config->configurationVersion = 1;
247 p_mp4->p_config->AVCProfileIndication = sps[1];
248 p_mp4->p_config->profile_compatibility = sps[2];
249 p_mp4->p_config->AVCLevelIndication = sps[3];
250 p_slot = malloc( sizeof(GF_AVCConfigSlot) );
253 p_slot->size = sps_size;
254 p_slot->data = malloc( p_slot->size );
257 memcpy( p_slot->data, sps, sps_size );
258 gf_list_add( p_mp4->p_config->sequenceParameterSets, p_slot );
262 p_slot = malloc( sizeof(GF_AVCConfigSlot) );
265 p_slot->size = pps_size;
266 p_slot->data = malloc( p_slot->size );
269 memcpy( p_slot->data, pps, pps_size );
270 gf_list_add( p_mp4->p_config->pictureParameterSets, p_slot );
271 gf_isom_avc_config_update( p_mp4->p_file, p_mp4->i_track, 1, p_mp4->p_config );
275 memcpy( p_mp4->p_sample->data + p_mp4->p_sample->dataLength, sei, sei_size );
276 p_mp4->p_sample->dataLength += sei_size;
278 return sei_size + sps_size + pps_size;
280 static int write_frame( hnd_t handle, uint8_t *p_nalu, int i_size, x264_picture_t *p_picture )
282 mp4_hnd_t *p_mp4 = handle;
287 memcpy( p_mp4->p_sample->data + p_mp4->p_sample->dataLength, p_nalu, i_size );
288 p_mp4->p_sample->dataLength += i_size;
290 if( !p_mp4->i_numframe )
291 p_mp4->i_delay_time = p_picture->i_dts * -1;
293 if( !p_mp4->i_init_delay )
294 dts = cts = p_picture->i_pts * p_mp4->i_time_inc;
297 if( p_mp4->i_numframe <= p_mp4->i_init_delay )
298 dts = p_picture->i_dts + p_mp4->i_delay_time;
300 dts = p_mp4->i_prev_timestamps[ (p_mp4->i_numframe - p_mp4->i_init_delay) % p_mp4->i_init_delay ] + p_mp4->i_delay_time;
303 p_mp4->i_prev_timestamps[ p_mp4->i_numframe % p_mp4->i_init_delay ] = p_picture->i_dts + p_mp4->i_delay_time;
305 dts *= p_mp4->i_time_inc;
306 cts = (p_picture->i_pts + p_mp4->i_delay_time) * p_mp4->i_time_inc;
311 p_mp4->p_sample->IsRAP = p_picture->b_keyframe;
312 p_mp4->p_sample->DTS = dts;
313 p_mp4->p_sample->CTS_Offset = offset;
314 gf_isom_add_sample( p_mp4->p_file, p_mp4->i_track, p_mp4->i_descidx, p_mp4->p_sample );
316 p_mp4->p_sample->dataLength = 0;
322 cli_output_t mp4_output = { open_file, set_param, write_headers, write_frame, close_file };