SkeyePlayer(Windows)中录像采纳GPAC的MP4Box库来封装MP4,上面我将简略介绍MP4的封装调用流程和须要留神的点;
一、GPAC库的编译,GPAC是跨平台的库,windows和linux都能很不便多编译,再次不做过多赘述,大家可去GPAC官网或者Github上下载;
二、创立MP4
bool SkeyeMP4Writer::CreateMP4File(char*filename,int flag)
{
SaveFile();
m_audiostartimestamp=-1;
m_videostartimestamp=-1;
if(filename==NULL)
{
char filename2[256]={0};
sprintf(filename2,"%d-gpac%d.mp4",time(NULL),rand());
p_file=gf_isom_open(filename2,GF_ISOM_OPEN_WRITE,NULL);//关上文件
}else
p_file=gf_isom_open(filename,GF_ISOM_OPEN_WRITE,NULL);//关上文件
if (p_file==NULL)
{
return false;
}
gf_isom_set_brand_info(p_file,GF_ISOM_BRAND_MP42,0);
//if(flag&ZOUTFILE_FLAG_VIDEO)
//{
// m_videtrackid=gf_isom_new_track(p_file,0,GF_ISOM_MEDIA_VISUAL,1000);
// gf_isom_set_track_enabled(p_file,m_videtrackid,1);
//}
//if(flag&ZOUTFILE_FLAG_AUDIO)
//{
// m_audiotrackid=gf_isom_new_track(p_file,0,GF_ISOM_MEDIA_AUDIO,1000);
// gf_isom_set_track_enabled(p_file,m_audiotrackid,1);
//}
m_nCreateFileFlag = flag;
return true;
}
创立MP4很简略,调用gf_isom_open函数就能轻松搞定,gf_isom_set_brand_info函数设置以后写MP4的版本为MP4V2;值得注意的中央是:
1>. 创立文件之前须要对所有的参数进行初始化,以及如果文件正在写入则须要将其敞开,这个操作次要是32位程序写的MP4文件大于4G可能呈现不能播放的问题,为了不便写MP4文件进行分片,这个将在系列文章后续中进行解说;
2>. 大家能够看到上段代码有屏蔽了局部代码flag&ZOUTFILE_FLAG_VIDEO和flag&ZOUTFILE_FLAG_AUDIO的判断,这两段代码是用来在MP4文件中创立音频轨和视频轨(默认各只创立一个),请留神:如果这里曾经创立了音频和视频轨,然而后续的写入过程中如果只写音频或者视频的话,某些播放器可能是播不进去的(比方windows自带的播放器),所以,如果只写音频的话只须要创立音频轨就能够了,视频同理。
三、写入视频H264的SPS和PPS头信息
bool SkeyeMP4Writer::WriteH264SPSandPPS(unsigned char*sps,int spslen,unsigned char*pps,int ppslen,int width,int height)
{
if (m_nCreateFileFlag&ZOUTFILE_FLAG_VIDEO)
{
m_videtrackid = gf_isom_new_track(p_file, 0, GF_ISOM_MEDIA_VISUAL, 1000);
gf_isom_set_track_enabled(p_file, m_videtrackid, 1);
}
else
{
return false;
}
p_videosample=gf_isom_sample_new();
p_videosample->data=(char*)malloc(1024*1024);
p_config=gf_odf_avc_cfg_new();
gf_isom_avc_config_new(p_file,m_videtrackid,p_config,NULL,NULL,&i_videodescidx);
gf_isom_set_visual_info(p_file,m_videtrackid,i_videodescidx,width,height);
GF_AVCConfigSlot m_slotsps={0};
GF_AVCConfigSlot m_slotpps={0};
p_config->configurationVersion = 1;
p_config->AVCProfileIndication = sps[1];
p_config->profile_compatibility = sps[2];
p_config->AVCLevelIndication = sps[3];
m_slotsps.size=spslen;
m_slotsps.data=(char*)malloc(spslen);
memcpy(m_slotsps.data,sps,spslen);
gf_list_add(p_config->sequenceParameterSets,&m_slotsps);
m_slotpps.size=ppslen;
m_slotpps.data=(char*)malloc(ppslen);
memcpy(m_slotpps.data,pps,ppslen);
gf_list_add(p_config->pictureParameterSets,&m_slotpps);
gf_isom_avc_config_update(p_file,m_videtrackid,1,p_config);
free(m_slotsps.data);
free(m_slotpps.data);
return true;
}
首先,通过gf_odf_avc_cfg_new()创立一个设置AVC信息的配置构造p_config,而后对构造中指定的信息,如:长,宽,SPS和PPS等要害参数写入配置构造,调用gf_isom_avc_config_update函数写入参数信息;当然这里只是H264格局的参数设置,像其余的格局比方H265的设置也相似,这将在后续系列中进行解说;
四、写入音频AAC头信息
//写入AAC信息
bool SkeyeMP4Writer::WriteAACInfo(unsigned char*info,int len, int nSampleRate, int nChannel, int nBitsPerSample)
{
if (m_nCreateFileFlag&ZOUTFILE_FLAG_AUDIO)
{
m_audiotrackid = gf_isom_new_track(p_file, 0, GF_ISOM_MEDIA_AUDIO, 1000);
gf_isom_set_track_enabled(p_file, m_audiotrackid, 1);
}
else
{
return false;
}
p_audiosample=gf_isom_sample_new();
p_audiosample->data=(char*)malloc(1024*10);
GF_ESD*esd= gf_odf_desc_esd_new(0);
esd->ESID=gf_isom_get_track_id(p_file,m_audiotrackid);
esd->OCRESID=gf_isom_get_track_id(p_file,m_audiotrackid);
esd->decoderConfig->streamType=0x05;
esd->decoderConfig->objectTypeIndication=0x40;//0x40;
esd->slConfig->timestampResolution=1000;//1000;//工夫单元
esd->decoderConfig->decoderSpecificInfo=(GF_DefaultDescriptor*)gf_odf_desc_new(GF_ODF_DSI_TAG);
esd->decoderConfig->decoderSpecificInfo->data=(char*)malloc(len);
memcpy(esd->decoderConfig->decoderSpecificInfo->data,info,len);
esd->decoderConfig->decoderSpecificInfo->dataLength=len;
GF_Err gferr=gf_isom_new_mpeg4_description(p_file, m_audiotrackid, esd, NULL, NULL, &i_audiodescidx);
if (gferr!=0)
{
// TRACE("mpeg4_description:%d\n",gferr);
}
gferr=gf_isom_set_audio_info(p_file,m_audiotrackid,i_audiodescidx, nSampleRate,nChannel, nBitsPerSample);//44100 2 16
if (gferr!=0)
{
// TRACE("gf_isom_set_audio:%d\n",gferr);
}
free(esd->decoderConfig->decoderSpecificInfo->data);
return true;
}
调几个 API就搞定了,判若两人的简略–!,这里说一下一些要害参数的配置:
1> esd->decoderConfig->streamType=0x05,这里的0x05标示为AAC,当然还指出其余的类型,如MP3,AC3等等,具体可查问MP4BOX相干文档获取;
2> 函数出入的头两个参数大家看起来有点费解,这里示意的是音频解码参数组合的一个串,具体格局解析如下:(这个原本想独自开一篇博客来专门论述的,然而鉴于没多少内容就在这里一并表述进去)
看上面代码段:
// 前五位为 AAC object types LOW 2
// 接着4位为 码率index 16000 8
// 采样标记规范:
// static unsigned long tnsSupportedSamplingRates[13] = //音频采样率规范(标记),下标为写入标记
// { 96000,88200,64000,48000,44100,32000,24000,22050,16000,12000,11025,8000,0 };
// 接着4位为 channels 个数 2
// 最初3位用0补齐
// 应打印出的正确2进制模式为 00010 | 1000 | 0010 | 000
// 2 8 2
// BYTE ubDecInfoBuff[] = {0x12,0x10};//00010 0100 0010 000
//音频采样率规范(标记),下标为写入标记
unsigned long tnsSupportedSamplingRates[13] = { 96000,88200,64000,48000,44100,32000,24000,22050,16000,12000,11025,8000,0 };
int nI = 0;
for ( nI = 0; nI<13; nI++)
{
if (tnsSupportedSamplingRates[nI] == sample_rate )
{
break;
}
}
unsigned char ucDecInfoBuff[2] = {0x12,0x10};//
unsigned short nDecInfo = (1<<12) | (nI << 7) | (channels<<3);
int nSize = sizeof(unsigned short);
memcpy(ucDecInfoBuff, &nDecInfo, nSize);
SWAP(ucDecInfoBuff[0], ucDecInfoBuff[1]);
int unBuffSize = sizeof(ucDecInfoBuff)*sizeof(unsigned char);
大家看懂了吧,比方当初有个示意解码信息的串为 00010 | 0100 | 0010 | 000 ,那么它则示意为AAC-LC 44100采样率 双声道音频,是不是很好了解呢!!!
五、解析H264帧写入MP4
限于篇幅,这里就不贴代码了(否则有靠代码凑字数的嫌疑,尽管我曾经贴了好多了 ,哈哈哈......),上面用文字描述,分三步走:
1> 解析H264 nal头,获取SPS和PPS, 因为咱们曾经通过设置函数设置了SPS和PPS等解码要害信息,所以咱们写入文件时,H264帧将转换为AVC格局,什么意思,就是说将以00000001以及000001结尾的NAL单元转换为以该NAL单元的长度来填满该四个字节(留神:所有的H264帧中的0x00000001和0x000001都要替换成NAL的长度,否则未替换的局部解码会花屏),默认三个字节的000001也用四个字节补齐,这次要是见于一帧多NAL的状况,这里有疑难我将在后续系列文章中解说;
2> 写入SPS和PPS头;
3> 写入以NAL长度为头四个字节的AVC帧,具体实现如下:
//写入一帧,前四字节为该帧NAL长度
bool SkeyeMP4Writer::WriteVideoFrame(unsigned char*data,int len,bool keyframe,long timestamp)
{
if (!p_videosample)
{
return false;
}
if (m_videostartimestamp==-1&&keyframe)
{
m_videostartimestamp=timestamp;
}
if (m_videostartimestamp!=-1)
{
p_videosample->IsRAP=keyframe;
p_videosample->dataLength=len;
memcpy(p_videosample->data,data,len);
p_videosample->DTS=timestamp-m_videostartimestamp;
p_videosample->CTS_Offset=0;
GF_Err gferr=gf_isom_add_sample(p_file,m_videtrackid,i_videodescidx,p_videosample);
if (gferr==-1)
{
p_videosample->DTS=timestamp-m_videostartimestamp+15;
gf_isom_add_sample(p_file,m_videtrackid,i_videodescidx,p_videosample);
}
}
return true;
}
六、AAC写入MP4(是否带ADTS头)
同写视频相似,写音频同样要先写如音频解码参数,上文曾经剖析过如何写解码参数,这里只需把解码参数信息组织成串,通过WriteAACInfo()函数写入即可。
写音频数据,实现和视频一样,调用gf_isom_add_sample函数即可;
须要留神:因为咱们曾经写入了音频解码信息,那么如果AAC数据中带有ADTS头,则须要去掉则7个字节的头,否则可能局部播放器不能失常播放,ADTS头以 0xFFF 开始;
七、写入MP4封装头,保留文件
保留文件,开释缓存和系统资源:
//保留文件
bool SkeyeMP4Writer::SaveFile()
{
if (m_psps)
{
delete m_psps;
m_psps = NULL;
}
if (m_ppps)
{
delete m_ppps;
m_ppps = NULL;
}
m_spslen=0;
m_ppslen=0;
if (m_pvps)
{
delete m_pvps;
m_pvps = NULL;
}
m_vpslen = 0;
m_audiostartimestamp=-1;
m_videostartimestamp=-1;
if (p_file)
{
gf_isom_close(p_file);
p_file=NULL;
}
if(p_config)
{
// delete p_config->pictureParameterSets;
p_config->pictureParameterSets=NULL;
// delete p_config->sequenceParameterSets;
p_config->sequenceParameterSets=NULL;
gf_odf_avc_cfg_del(p_config);
p_config=NULL;
}
if (p_hevc_config)
{
gf_odf_hevc_cfg_del(p_hevc_config);
p_hevc_config = NULL;
}
if( p_audiosample)
{
if( p_audiosample->data)
{
free(p_audiosample->data);
p_audiosample->data=NULL;
}
gf_isom_sample_del(&p_audiosample);
p_audiosample=NULL;
}
if( p_videosample)
{
if( p_videosample->data)
{
free(p_videosample->data);
p_videosample->data=NULL;
}
gf_isom_sample_del(&p_videosample);
p_audiosample=NULL;
}
m_bwriteaudioinfo = false;
m_bwritevideoinfo = false;
return true;
}
获取更多信息
邮件:support@Skeyedarwin.org
WEB:www.SkeyeDarwin.org
Copyright © SkeyeDarwin.org 2012-2016
[外链图片转存中…(img-M61RFUJj-1652408925267)]
发表回复