Media (Intel® Video Processing Library, Intel Media SDK)
Access community support with transcoding, decoding, and encoding in applications using media tools like Intel® oneAPI Video Processing Library and Intel® Media SDK
Announcements
The Intel Media SDK project is no longer active. For continued support and access to new features, Intel Media SDK users are encouraged to read the transition guide on upgrading from Intel® Media SDK to Intel® Video Processing Library (VPL), and to move to VPL as soon as possible.
For more information, see the VPL website.

Output Size of H264 encode is strange

song_l_
Beginner
306 Views

Hi,

I used one image data to compress into  a AVC video.the output buffer size is strange.

GopPicSize =1;  AsyncDepth=1;

for (int i = 0; i < 100; ++i){
    DWORD dt = GetTickCount();

    memset(&bitStream, 0, sizeof(mfxBitstream));
    bitStream.MaxLength = bufferSize;
    bitStream.Data = buffer;
    for (;;)
    {
        status = MFXVideoENCODE_EncodeFrameAsync(session, NULL, &surface, &bitStream, &syncp);
        //....some common routing.... 
    }
    MFXVideoCORE_SyncOperation(session, syncp, INFINITE);
    printf("%ld, %ld\r\n", bitStream.DataLength, GetTickCount() - dt);
}

 The output is:

49747, 15
43905, 16
38935, 0
35141, 0
29245, 0
25950, 16
23541, 0
18919, 0
16195, 0
16195, 16
16195, 0
16195, 0
16195, 0
16195, 0
16195, 16
16195, 0
16195, 0
16195, 0
16195, 0
16195, 0
16195, 0
16195, 16
16195, 0
16195, 0
16195, 0

 The problem is :

1)The output DataLength should be the same when GopPicSize =1(I Frame only).

Maybe the first frame will be a little bigger because of some header information. 

2) The encode time is not the same. if there is a cache, It should be 16, 0,0,0,0......

Is there anyone who can explain the output data?

0 Kudos
2 Replies
Jeffrey_M_Intel1
Employee
306 Views

Many things could affect frame size -- especially the input stream.  To understand expected behavior you can start with the tutorials.  The simple_3_encode tutorial is designed to be simple so you can quickly add your own modifications to GOP parameters, BRC, etc.  I suspect you will see different behavior (perhaps more like you expect?) if you switch to CQP instead of CBR/VBR.

For encode time, since the implementation is asynchronous the time/latency to encode is not expected to be constant.  The low latency tutorials can be used as a starting point/reference for tracking encode time by frame.

Regards, Jeff

 

0 Kudos
song_l_
Beginner
306 Views

TO:Jeffrey Mcallister (Intel)
    Thanks for you answer。

   

0 Kudos
Reply