1 /*****************************************************************************
3 *****************************************************************************
4 * Copyright (C) 2003-2011 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.
23 * This program is also available under a commercial proprietary license.
24 * For more information, contact us at licensing@x264.com.
25 *****************************************************************************/
28 #include <gpac/isomedia.h>
33 #define malloc gf_malloc
40 GF_AVCConfig *p_config;
41 GF_ISOSample *p_sample;
51 int i_dts_compress_multiplier;
54 static void recompute_bitrate_mp4( GF_ISOFile *p_file, int i_track )
56 u32 count, di, timescale, time_wnd, rate;
61 esd = gf_isom_get_esd( p_file, i_track, 1 );
65 esd->decoderConfig->avgBitrate = 0;
66 esd->decoderConfig->maxBitrate = 0;
69 timescale = gf_isom_get_media_timescale( p_file, i_track );
70 count = gf_isom_get_sample_count( p_file, i_track );
71 for( u32 i = 0; i < count; i++ )
73 GF_ISOSample *samp = gf_isom_get_sample_info( p_file, i_track, i+1, &di, &offset );
76 x264_cli_log( "mp4", X264_LOG_ERROR, "failure reading back frame %u\n", i );
80 if( esd->decoderConfig->bufferSizeDB < samp->dataLength )
81 esd->decoderConfig->bufferSizeDB = samp->dataLength;
83 esd->decoderConfig->avgBitrate += samp->dataLength;
84 rate += samp->dataLength;
85 if( samp->DTS > time_wnd + timescale )
87 if( rate > esd->decoderConfig->maxBitrate )
88 esd->decoderConfig->maxBitrate = rate;
93 gf_isom_sample_del( &samp );
96 br = (Double)(s64)gf_isom_get_media_duration( p_file, i_track );
98 esd->decoderConfig->avgBitrate = (u32)(esd->decoderConfig->avgBitrate / br);
100 esd->decoderConfig->avgBitrate *= 8;
101 esd->decoderConfig->maxBitrate *= 8;
103 gf_isom_change_mpeg4_description( p_file, i_track, 1, esd );
104 gf_odf_desc_del( (GF_Descriptor*)esd );
107 static int close_file( hnd_t handle, int64_t largest_pts, int64_t second_largest_pts )
109 mp4_hnd_t *p_mp4 = handle;
114 if( p_mp4->p_config )
115 gf_odf_avc_cfg_del( p_mp4->p_config );
117 if( p_mp4->p_sample )
119 if( p_mp4->p_sample->data )
120 free( p_mp4->p_sample->data );
122 p_mp4->p_sample->dataLength = 0;
123 gf_isom_sample_del( &p_mp4->p_sample );
128 /* The mdhd duration is defined as CTS[final] - CTS[0] + duration of last frame.
129 * The mdhd duration (in seconds) should be able to be longer than the tkhd duration since the track is managed by edts.
130 * So, if mdhd duration is equal to the last DTS or less, we give the last composition time delta to the last sample duration.
131 * And then, the mdhd duration is updated, but it time-wise doesn't give the actual duration.
132 * The tkhd duration is the actual track duration. */
133 uint64_t mdhd_duration = (2 * largest_pts - second_largest_pts) * p_mp4->i_time_inc;
134 if( mdhd_duration != gf_isom_get_media_duration( p_mp4->p_file, p_mp4->i_track ) )
136 uint64_t last_dts = gf_isom_get_sample_dts( p_mp4->p_file, p_mp4->i_track, p_mp4->i_numframe );
137 uint32_t last_duration = (uint32_t)( mdhd_duration > last_dts ? mdhd_duration - last_dts : (largest_pts - second_largest_pts) * p_mp4->i_time_inc );
138 gf_isom_set_last_sample_duration( p_mp4->p_file, p_mp4->i_track, last_duration );
141 /* Write an Edit Box if the first CTS offset is positive.
142 * A media_time is given by not the mvhd timescale but rather the mdhd timescale.
143 * The reason is that an Edit Box maps the presentation time-line to the media time-line.
144 * Any demuxers should follow the Edit Box if it exists. */
145 GF_ISOSample *sample = gf_isom_get_sample_info( p_mp4->p_file, p_mp4->i_track, 1, NULL, NULL );
146 if( sample && sample->CTS_Offset > 0 )
148 uint32_t mvhd_timescale = gf_isom_get_timescale( p_mp4->p_file );
149 uint64_t tkhd_duration = (uint64_t)( mdhd_duration * ( (double)mvhd_timescale / p_mp4->i_time_res ) );
150 gf_isom_append_edit_segment( p_mp4->p_file, p_mp4->i_track, tkhd_duration, sample->CTS_Offset, GF_ISOM_EDIT_NORMAL );
152 gf_isom_sample_del( &sample );
154 recompute_bitrate_mp4( p_mp4->p_file, p_mp4->i_track );
155 gf_isom_set_pl_indication( p_mp4->p_file, GF_ISOM_PL_VISUAL, 0x15 );
156 gf_isom_set_storage_mode( p_mp4->p_file, GF_ISOM_STORE_FLAT );
157 gf_isom_close( p_mp4->p_file );
165 static int open_file( char *psz_filename, hnd_t *p_handle, cli_output_opt_t *opt )
170 FILE *fh = fopen( psz_filename, "w" );
173 FAIL_IF_ERR( !x264_is_regular_file( fh ), "mp4", "MP4 output is incompatible with non-regular file `%s'\n", psz_filename )
176 if( !(p_mp4 = malloc( sizeof(mp4_hnd_t) )) )
179 memset( p_mp4, 0, sizeof(mp4_hnd_t) );
180 p_mp4->p_file = gf_isom_open( psz_filename, GF_ISOM_OPEN_WRITE, NULL );
182 p_mp4->b_dts_compress = opt->use_dts_compress;
184 if( !(p_mp4->p_sample = gf_isom_sample_new()) )
186 close_file( p_mp4, 0, 0 );
190 gf_isom_set_brand_info( p_mp4->p_file, GF_ISOM_BRAND_AVC1, 0 );
197 static int set_param( hnd_t handle, x264_param_t *p_param )
199 mp4_hnd_t *p_mp4 = handle;
201 p_mp4->i_delay_frames = p_param->i_bframe ? (p_param->i_bframe_pyramid ? 2 : 1) : 0;
202 p_mp4->i_dts_compress_multiplier = p_mp4->b_dts_compress * p_mp4->i_delay_frames + 1;
204 p_mp4->i_time_res = p_param->i_timebase_den * p_mp4->i_dts_compress_multiplier;
205 p_mp4->i_time_inc = p_param->i_timebase_num * p_mp4->i_dts_compress_multiplier;
206 FAIL_IF_ERR( p_mp4->i_time_res > UINT32_MAX, "mp4", "MP4 media timescale %"PRIu64" exceeds maximum\n", p_mp4->i_time_res )
208 p_mp4->i_track = gf_isom_new_track( p_mp4->p_file, 0, GF_ISOM_MEDIA_VISUAL,
211 p_mp4->p_config = gf_odf_avc_cfg_new();
212 gf_isom_avc_config_new( p_mp4->p_file, p_mp4->i_track, p_mp4->p_config,
213 NULL, NULL, &p_mp4->i_descidx );
215 gf_isom_set_track_enabled( p_mp4->p_file, p_mp4->i_track, 1 );
217 gf_isom_set_visual_info( p_mp4->p_file, p_mp4->i_track, p_mp4->i_descidx,
218 p_param->i_width, p_param->i_height );
220 if( p_param->vui.i_sar_width && p_param->vui.i_sar_height )
222 uint64_t dw = p_param->i_width << 16;
223 uint64_t dh = p_param->i_height << 16;
224 double sar = (double)p_param->vui.i_sar_width / p_param->vui.i_sar_height;
229 gf_isom_set_pixel_aspect_ratio( p_mp4->p_file, p_mp4->i_track, p_mp4->i_descidx, p_param->vui.i_sar_width, p_param->vui.i_sar_height );
230 gf_isom_set_track_layout_info( p_mp4->p_file, p_mp4->i_track, dw, dh, 0, 0, 0 );
233 p_mp4->p_sample->data = malloc( p_param->i_width * p_param->i_height * 3 / 2 );
234 if( !p_mp4->p_sample->data )
240 static int write_headers( hnd_t handle, x264_nal_t *p_nal )
242 mp4_hnd_t *p_mp4 = handle;
243 GF_AVCConfigSlot *p_slot;
245 int sps_size = p_nal[0].i_payload - 4;
246 int pps_size = p_nal[1].i_payload - 4;
247 int sei_size = p_nal[2].i_payload;
249 uint8_t *sps = p_nal[0].p_payload + 4;
250 uint8_t *pps = p_nal[1].p_payload + 4;
251 uint8_t *sei = p_nal[2].p_payload;
255 p_mp4->p_config->configurationVersion = 1;
256 p_mp4->p_config->AVCProfileIndication = sps[1];
257 p_mp4->p_config->profile_compatibility = sps[2];
258 p_mp4->p_config->AVCLevelIndication = sps[3];
259 p_slot = malloc( sizeof(GF_AVCConfigSlot) );
262 p_slot->size = sps_size;
263 p_slot->data = malloc( p_slot->size );
266 memcpy( p_slot->data, sps, sps_size );
267 gf_list_add( p_mp4->p_config->sequenceParameterSets, p_slot );
271 p_slot = malloc( sizeof(GF_AVCConfigSlot) );
274 p_slot->size = pps_size;
275 p_slot->data = malloc( p_slot->size );
278 memcpy( p_slot->data, pps, pps_size );
279 gf_list_add( p_mp4->p_config->pictureParameterSets, p_slot );
280 gf_isom_avc_config_update( p_mp4->p_file, p_mp4->i_track, 1, p_mp4->p_config );
284 memcpy( p_mp4->p_sample->data + p_mp4->p_sample->dataLength, sei, sei_size );
285 p_mp4->p_sample->dataLength += sei_size;
287 return sei_size + sps_size + pps_size;
290 static int write_frame( hnd_t handle, uint8_t *p_nalu, int i_size, x264_picture_t *p_picture )
292 mp4_hnd_t *p_mp4 = handle;
296 memcpy( p_mp4->p_sample->data + p_mp4->p_sample->dataLength, p_nalu, i_size );
297 p_mp4->p_sample->dataLength += i_size;
299 if( !p_mp4->i_numframe )
300 p_mp4->i_delay_time = p_picture->i_dts * -1;
302 if( p_mp4->b_dts_compress )
304 if( p_mp4->i_numframe == 1 )
305 p_mp4->i_init_delta = (p_picture->i_dts + p_mp4->i_delay_time) * p_mp4->i_time_inc;
306 dts = p_mp4->i_numframe > p_mp4->i_delay_frames
307 ? p_picture->i_dts * p_mp4->i_time_inc
308 : p_mp4->i_numframe * (p_mp4->i_init_delta / p_mp4->i_dts_compress_multiplier);
309 cts = p_picture->i_pts * p_mp4->i_time_inc;
313 dts = (p_picture->i_dts + p_mp4->i_delay_time) * p_mp4->i_time_inc;
314 cts = (p_picture->i_pts + p_mp4->i_delay_time) * p_mp4->i_time_inc;
317 p_mp4->p_sample->IsRAP = p_picture->b_keyframe;
318 p_mp4->p_sample->DTS = dts;
319 p_mp4->p_sample->CTS_Offset = (uint32_t)(cts - dts);
320 gf_isom_add_sample( p_mp4->p_file, p_mp4->i_track, p_mp4->i_descidx, p_mp4->p_sample );
322 p_mp4->p_sample->dataLength = 0;
328 const cli_output_t mp4_output = { open_file, set_param, write_headers, write_frame, close_file };