本文主要是介绍JM8.6中NALU（此处指VCL式的NALU)是如何写进码流的？，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！
在lencod.c的main函数中调用了encode_one_slice函数，在encode_one_slice函数中调用了frame_picture函数，调用了frame_picture函数后(frame_picture还会继续调用其他重要函数) 就实现了对第一帧的编码，这样全局变量frame_pic就发生了变化，得到了它想得到的码流，现在要写码流，怎么写呢？
在encode_one_frame中接着就调用到了writeout_picture函数，调用形式是：
writeout_picture (frame_pic);可见，把代码码流信息的frame_pic穿进去了，进入writeout_picture函数看看：
static int writeout_picture(Picture *pic)
{Bitstream *currStream;int partition, slice;Slice *currSlice;img->currentPicture=pic;// 遍历slice,通常，一个frame对应一个slicefor (slice=0; slice<pic->no_slices; slice++){currSlice = pic->slices[slice];// 目前都发现currSlice->max_part_nr为1for (partition=0; partition<currSlice->max_part_nr; partition++){currStream = (currSlice->partArr[partition]).bitstream;// currStream->bits_to_go相当于标记，必须为8assert (currStream->bits_to_go == 8); //! should always be the case, the //! byte alignment is done in terminate_slice// writeUnit函数的第一参数类型是Bitstream*writeUnit (currSlice->partArr[partition].bitstream,partition);} // partition loop} // slice loopreturn 0;
}可以看到，把一个picture对应的码流分成slice和数据分块来传给writeUnit的形参（一般认为：一个slice对应一个数据分块），进入到writeUnit函数看看：
static void writeUnit(Bitstream* currStream,int partition)
{NALU_t *nalu;// 实际上currStream->bits_to_go就是一个标记// 如果不为8, 则表明往下继续执行会有错误assert (currStream->bits_to_go == 8);// 分配堆空间， *4是为了安全起见nalu = AllocNALU(img->width*img->height*4);nalu->startcodeprefix_len = 2+(img->current_mb_nr == 0?ZEROBYTES_SHORTSTARTCODE+1:ZEROBYTES_SHORTSTARTCODE);
//printf ("nalu->startcodeprefix_len %d\n", nalu->startcodeprefix_len);nalu->len = currStream->byte_pos +1; // add one for the first byte of the NALU
//printf ("nalu->len %d\n", nalu->len);// 内存数据复制，这个是最重要的.memcpy (&nalu->buf[1], currStream->streamBuffer, nalu->len-1);// 下面对nalu的一些参数进行赋值// 其实就是对nalu header的8个比特赋值if (img->currentPicture->idr_flag){nalu->nal_unit_type = NALU_TYPE_IDR;nalu->nal_reference_idc = NALU_PRIORITY_HIGHEST;}else if (img->type == B_SLICE){//different nal header for different partitionsif(input->partition_mode == 0){nalu->nal_unit_type = NALU_TYPE_SLICE;}else{nalu->nal_unit_type = NALU_TYPE_DPA + partition;}if (img->nal_reference_idc !=0){nalu->nal_reference_idc = NALU_PRIORITY_HIGH;}else{nalu->nal_reference_idc = NALU_PRIORITY_DISPOSABLE;}}else // non-b frame, non IDR slice{//different nal header for different partitionsif(input->partition_mode == 0){nalu->nal_unit_type = NALU_TYPE_SLICE;}else{nalu->nal_unit_type = NALU_TYPE_DPA + partition;}if (img->nal_reference_idc !=0){nalu->nal_reference_idc = NALU_PRIORITY_HIGH;}else{nalu->nal_reference_idc = NALU_PRIORITY_DISPOSABLE;}}nalu->forbidden_bit = 0; // 必须为0// WriteNALU是指向函数的全局指针（提前赋值了）,// 所以要想用鼠标进入WriteNALU“函数”是不可能的// 真正调用到的函数是：WriteAnnexbNALU// 而WriteAnnexbNALU函数就是写码流函数stat->bit_ctr += WriteNALU (nalu); // 写码流FreeNALU(nalu);
}
不用再进入WriteAnnexbNALU函数了，因为，对这个函数太熟悉了.
这篇关于JM8.6中NALU（此处指VCL式的NALU)是如何写进码流的？的文章就介绍到这儿，希望我们推荐的文章对编程师们有所帮助！
