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The Intel Media SDK project is no longer active.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.
3066 Discussions

Composition of different resolution streams using VPP

Mohnish_J_
Beginner
381 Views

Hi,

I am developing a sample application for overlaying four NV12 streams of different resolution. For this I am modifying simple_4_vpp_resize_denoise application present in mediasdk-tutorials-0.0.3 and I have followed the procedure as specified in https://software.intel.com/en-us/articles/video-composition-using-intel-media-sdk. I am using Intel(R) Core(TM) i7-4770R CPU @ 3.20GHz, which has Iris Pro Graphics 5200, Linux OS 3.10.0-229.1.2.39163.MSSr4.el7.centos.x86_64,  Media Server Studio Essentials 2015 R6 API version 1.16 for my development.

The application works fine if the input streams are of same resolution i.e. 1920x1080 with the below parameters:

#define NUM_STREAMS 4

#define STREAM_FILE_ONE     "test1.yuv"
#define IN_HEIGHT_ONE       1080
#define IN_WIDTH_ONE        1920
#define CROPH_ONE           1080
#define CROPW_ONE           1920
#define CROPX_ONE           0
#define CROPY_ONE           0
#define DESTX_ONE           0
#define DESTY_ONE           0
#define DESTH_ONE           720
#define DESTW_ONE           1280
#define ALPHA_ONE           255

#define STREAM_FILE_TWO     "test2.yuv"
#define IN_HEIGHT_TWO       1080
#define IN_WIDTH_TWO        1920
#define CROPH_TWO           1080
#define CROPW_TWO           1920
#define CROPX_TWO           0
#define CROPY_TWO           0
#define DESTX_TWO           0
#define DESTY_TWO           0
#define DESTH_TWO           270
#define DESTW_TWO           480
#define ALPHA_TWO           128

#define STREAM_FILE_THREE   "test3.yuv"
#define IN_HEIGHT_THREE     1080
#define IN_WIDTH_THREE      1920
#define CROPH_THREE         1080
#define CROPW_THREE         1920
#define CROPX_THREE         0
#define CROPY_THREE         0
#define DESTX_THREE         500
#define DESTY_THREE         0
#define DESTH_THREE         270
#define DESTW_THREE         480
#define ALPHA_THREE         200

#define STREAM_FILE_FOUR    "test4.yuv"
#define IN_HEIGHT_FOUR      1080
#define IN_WIDTH_FOUR       1920
#define CROPH_FOUR          1080
#define CROPW_FOUR          1920
#define CROPX_FOUR          0
#define CROPY_FOUR          0
#define DESTX_FOUR          100
#define DESTY_FOUR          300
#define DESTH_FOUR          270
#define DESTW_FOUR          480
#define ALPHA_FOUR          256

#define STREAM_FILE_OUT "out.yuv"
#define VPP_OUT_HEIGHT          720
#define VPP_OUT_WIDTH           1280

#define VPP_IN_HEIGHT           1080
#define VPP_IN_WIDTH            1920

and I get the below output:

proper_overlay_out.PNG

Now I change stream3 to 960x540 and accordingly I modify my parameters also as below:

#define NUM_STREAMS 4

#define STREAM_FILE_ONE     "test1.yuv"
#define IN_HEIGHT_ONE       1080
#define IN_WIDTH_ONE        1920
#define CROPH_ONE           1080
#define CROPW_ONE           1920
#define CROPX_ONE           0
#define CROPY_ONE           0
#define DESTX_ONE           0
#define DESTY_ONE           0
#define DESTH_ONE           720
#define DESTW_ONE           1280
#define ALPHA_ONE           255

#define STREAM_FILE_TWO     "test2.yuv"
#define IN_HEIGHT_TWO       1080
#define IN_WIDTH_TWO        1920
#define CROPH_TWO           1080
#define CROPW_TWO           1920
#define CROPX_TWO           0
#define CROPY_TWO           0
#define DESTX_TWO           0
#define DESTY_TWO           0
#define DESTH_TWO           270
#define DESTW_TWO           480
#define ALPHA_TWO           128

#define STREAM_FILE_THREE   "test3.yuv"
#define IN_HEIGHT_THREE     540
#define IN_WIDTH_THREE      960
#define CROPH_THREE         540
#define CROPW_THREE         960
#define CROPX_THREE         0
#define CROPY_THREE         0
#define DESTX_THREE         500
#define DESTY_THREE         0
#define DESTH_THREE         270
#define DESTW_THREE         480
#define ALPHA_THREE         200

#define STREAM_FILE_FOUR    "test4.yuv"
#define IN_HEIGHT_FOUR      1080
#define IN_WIDTH_FOUR       1920
#define CROPH_FOUR          1080
#define CROPW_FOUR          1920
#define CROPX_FOUR          0
#define CROPY_FOUR          0
#define DESTX_FOUR          100
#define DESTY_FOUR          300
#define DESTH_FOUR          270
#define DESTW_FOUR          480
#define ALPHA_FOUR          255

#define STREAM_FILE_OUT "out.yuv"
#define VPP_OUT_HEIGHT          720
#define VPP_OUT_WIDTH           1280

#define VPP_IN_HEIGHT           1920
#define VPP_IN_WIDTH            1080

With this I get the output as:

improper_overlay_0.PNG

I have done these changes in my application:

1. VPP Parameters change for the largest stream:
    // Input data
    VPPParams.vpp.In.FourCC = MFX_FOURCC_NV12;
    VPPParams.vpp.In.ChromaFormat = MFX_CHROMAFORMAT_YUV420;
    VPPParams.vpp.In.CropX = 0;
    VPPParams.vpp.In.CropY = 0;
    VPPParams.vpp.In.CropW = VPP_IN_WIDTH;
    VPPParams.vpp.In.CropH = VPP_IN_HEIGHT;
    VPPParams.vpp.In.PicStruct = MFX_PICSTRUCT_PROGRESSIVE;
    VPPParams.vpp.In.FrameRateExtN = 30;
    VPPParams.vpp.In.FrameRateExtD = 1;
    // width must be a multiple of 16
    // height must be a multiple of 16
    // in case of frame picture and a multiple
    // of 32 in case of field picture
    VPPParams.vpp.In.Width = MSDK_ALIGN16(VPP_IN_WIDTH);
    VPPParams.vpp.In.Height =
        (MFX_PICSTRUCT_PROGRESSIVE == VPPParams.vpp.In.PicStruct) ?
        MSDK_ALIGN16(VPP_IN_HEIGHT) :
        MSDK_ALIGN32(VPP_IN_HEIGHT);

 

    // Output data
    VPPParams.vpp.Out.FourCC = MFX_FOURCC_NV12;
    VPPParams.vpp.Out.ChromaFormat = MFX_CHROMAFORMAT_YUV420;
    VPPParams.vpp.Out.CropX = 0;
    VPPParams.vpp.Out.CropY = 0;
    VPPParams.vpp.Out.CropW = VPP_OUT_WIDTH;
    VPPParams.vpp.Out.CropH = VPP_OUT_HEIGHT;
    VPPParams.vpp.Out.PicStruct = MFX_PICSTRUCT_PROGRESSIVE;
    VPPParams.vpp.Out.FrameRateExtN = 30;
    VPPParams.vpp.Out.FrameRateExtD = 1;
    // width must be a multiple of 16
    // height must be a multiple of 16 in case of frame
    // picture and a multiple of 32 in case of field picture
    VPPParams.vpp.Out.Width = MSDK_ALIGN16(VPP_OUT_WIDTH);
    VPPParams.vpp.Out.Height =
        (MFX_PICSTRUCT_PROGRESSIVE == VPPParams.vpp.Out.PicStruct) ?
        MSDK_ALIGN16(VPP_OUT_HEIGHT) :
        MSDK_ALIGN32(VPP_OUT_HEIGHT);

2. mfxFrameInfo structure for input streams:

 mfxFrameInfo inputStreams[NUM_STREAMS];
    /** Fill frame params in mFrameInfo structures
     * with the above parameters **/
    for (mfxU16 i = 0; i < NUM_STREAMS; i++){
        memcpy(&inputStreams,&(VPPParams.vpp.In), sizeof(mfxFrameInfo));
    }

    inputStreams[0].Width = IN_WIDTH_ONE;
    inputStreams[0].Height = IN_HEIGHT_ONE;
    inputStreams[0].CropX = CROPX_ONE;
    inputStreams[0].CropY = CROPY_ONE;
    inputStreams[0].CropW = CROPW_ONE;
    inputStreams[0].CropH = CROPH_ONE;

    inputStreams[1].Width = IN_WIDTH_TWO;
    inputStreams[1].Height = IN_HEIGHT_TWO;
    inputStreams[1].CropX = CROPX_TWO;
    inputStreams[1].CropY = CROPY_TWO;
    inputStreams[1].CropW = CROPW_TWO;
    inputStreams[1].CropH = CROPH_TWO;

    inputStreams[2].Width = IN_WIDTH_THREE;
    inputStreams[2].Height = IN_HEIGHT_THREE;
    inputStreams[2].CropX = CROPX_THREE;
    inputStreams[2].CropY = CROPY_THREE;
    inputStreams[2].CropW = DESTW_THREE;
    inputStreams[2].CropH = DESTH_THREE;

    inputStreams[3].Width = IN_WIDTH_FOUR;
    inputStreams[3].Height = IN_HEIGHT_FOUR;
    inputStreams[3].CropX = CROPX_FOUR;
    inputStreams[3].CropY = CROPY_FOUR;
    inputStreams[3].CropW = CROPW_FOUR;
    inputStreams[3].CropH = CROPH_FOUR;

3. Initializing Composite extended buffer:

  mfxExtVPPComposite composite;
    memset(&composite, 0, sizeof(composite));
    composite.Header.BufferId = MFX_EXTBUFF_VPP_COMPOSITE;
    composite.Header.BufferSz = sizeof(mfxExtVPPComposite);
    composite.NumInputStream = NUM_STREAMS;
    composite.Y = 10;
    composite.U = 80;
    composite.V = 80;

    composite.InputStream = new mfxVPPCompInputStream[NUM_STREAMS];
    MSDK_CHECK_POINTER(composite.InputStream, MFX_ERR_MEMORY_ALLOC);
    memset(composite.InputStream, 0, sizeof(mfxVPPCompInputStream) * NUM_STREAMS);

    composite.InputStream[0].DstX = (mfxU32)DESTX_ONE;
    composite.InputStream[0].DstY = (mfxU32)DESTY_ONE;
    composite.InputStream[0].DstW = (mfxU32)DESTW_ONE;
    composite.InputStream[0].DstH = (mfxU32)DESTH_ONE;
    composite.InputStream[0].GlobalAlphaEnable = 1;
    composite.InputStream[0].GlobalAlpha = ALPHA_ONE;

    composite.InputStream[1].DstX = (mfxU32)DESTX_TWO;
    composite.InputStream[1].DstY = (mfxU32)DESTY_TWO;
    composite.InputStream[1].DstW = (mfxU32)DESTW_TWO;
    composite.InputStream[1].DstH = (mfxU32)DESTH_TWO;
    composite.InputStream[1].GlobalAlphaEnable = 1;
    composite.InputStream[1].GlobalAlpha = ALPHA_TWO;

    composite.InputStream[2].DstX = (mfxU32)DESTX_THREE;
    composite.InputStream[2].DstY = (mfxU32)DESTY_THREE;
    composite.InputStream[2].DstW = (mfxU32)DESTW_THREE;
    composite.InputStream[2].DstH = (mfxU32)DESTH_THREE;
    composite.InputStream[2].GlobalAlphaEnable = 1;
    composite.InputStream[2].GlobalAlpha = ALPHA_THREE;

    composite.InputStream[3].DstX = (mfxU32)DESTX_FOUR;
    composite.InputStream[3].DstY = (mfxU32)DESTY_FOUR;
    composite.InputStream[3].DstW = (mfxU32)DESTW_FOUR;
    composite.InputStream[3].DstH = (mfxU32)DESTH_FOUR;
    composite.InputStream[3].GlobalAlphaEnable = 1;
    composite.InputStream[3].GlobalAlpha = ALPHA_FOUR;

 

    // Add extended VPP buffers
    mfxExtBuffer* ExtBuffer[1];
    ExtBuffer[0] = (mfxExtBuffer*) &composite;
    VPPParams.NumExtParam = 1;
    VPPParams.ExtParam = (mfxExtBuffer**) &ExtBuffer[0];

4. In the processing loop, after getting surface index modifying surfaces Info and Data fields for 960x540 stream before I do LoadRawFrame call :

        memcpy(&(pVPPSurfacesIn[nSurfIdxIn]->Info), &(inputStreams[streamNum]), sizeof(mfxFrameInfo));
        //pVPPSurfacesIn->Data.Y = &surfaceBuffersIn[surfaceSize * i];

        pVPPSurfacesIn[nSurfIdxIn]->Data.U = pVPPSurfacesIn[nSurfIdxIn]->Data.Y + inputStreams[streamNum].Width*inputStreams[streamNum].Height;
        pVPPSurfacesIn[nSurfIdxIn]->Data.V = pVPPSurfacesIn[nSurfIdxIn]->Data.U + 1;
pVPPSurfacesIn[nSurfIdxIn]->Data.Pitch = inputStreams[streamNum].Width;

        // Load frame from file into surface
        sts = LoadRawFrame(pVPPSurfacesIn[nSurfIdxIn], fSource[streamNum]);
        streamNum++;

OR just modifying the frame info:

        memcpy(&(pVPPSurfacesIn[nSurfIdxIn]->Info), &(inputStreams[streamNum]), sizeof(mfxFrameInfo));
        //pVPPSurfacesIn->Data.Y = &surfaceBuffersIn[surfaceSize * i];

        // Load frame from file into surface
        sts = LoadRawFrame(pVPPSurfacesIn[nSurfIdxIn], fSource[streamNum]);
        streamNum++;

Please help me with this issue. Thanks in advance.

 

0 Kudos
2 Replies
Mohnish_J_
Beginner
381 Views

Hi,

I resolved the issue. Input 2 stream parameters were not proper:

    inputStreams[2].CropW = DESTW_THREE;
    inputStreams[2].CropH = DESTH_THREE;

was to be replaced with

    inputStreams[2].CropW = CROPW_THREE;
    inputStreams[2].CropH = CROPH_THREE;

0 Kudos
Sravanthi_K_Intel
381 Views

Thanks for the update Mohnish - glad you got the composition working. I am closing this thread.

0 Kudos
Reply