Hello, 

I have download and build the  Linux version of IntelMediaSDK from the belew GitHubLink

https://github.com/Intel-Media-SDK/MediaSDK

Since we intend to use SDK for decoding H.264 stream I have started with the simple_decode sample. Our application will be running on Linux system that might not have graphics cards, so I change the acceleration type from MFX_IMPL_AUTO_ANY to MFX_IMPL_SOFTWARE.

But this results in MFXVideoSession::Init() giving error mfx_err_unsupported. Does Linux IntelSDK library supports only hardware processing ? Even I noticed in the source-code comments also

    // OS specific notes
    // - On Windows both SW and HW libraries may present
    // - On Linux only HW library only is available

Is the only way to use software acceleration is through Windows SDK ?

Thanks

Manish Narang

I've been successfully using the Intel Media SDK GStreamer plugins on Windows for a while now (specifically this branch), but I've encountered an error and I'm wondering if someone can help me out. 

What I'm trying to do is generate BGRA output from either the h264 or h265 encoders, but some negotiation with downstream elements in the pipeline is failing. NV12 works just fine, and so I know I've got things hooked up right. The failure occurs when wanting the encoder to generate any other colorspace but NV12. 

A bit more detail

I've got an RTSP stream that I'm reading, and I'm wanting to draw it onto some Windows Presentation Foundation control, which only allows BGRA/RGBA/RGB/etc. I'm using the plugin in code, but I can replicate the problem using just gst-launch command: 

gst-launch-1.0.exe rtspsrc location=rtsp://address/live ! decodebin ! 'video/x-raw(memory:MFXSurface),format=BGRA' ! autovideosink

This will fail. The decodebin does use the Intel Media plugins when building the pipeline. If you enable debugging through GST_DEBUG=4 or higher, you will get the following error/warning: 

WARN                GST_PADS gstpad.c:4230:gst_pad_peer_query:<mfxdec_h264-0:src> could not send sticky events

The net effect is the pipeline never fully transitions to the play state. If I watch the GPU and CPU, I can see it's chewing, but downstream elements don't get any data. 

Flipping BGRA to NV12, the stream will work: the pipeline transitions to PLAY and downstream elements get data and I can view my stream:

gst-launch-1.0.exe rtspsrc location=rtsp://address/live ! decodebin ! 'video/x-raw(memory:MFXSurface),format=NV12' ! autovideosink

When I inspect the MFX GStreamer elements, they all advertise they can handle BGRA. For example, here are the pad outputs generated by gst-inspect when examining the H265 encoder: 
 

Pad Templates:
  SINK template: 'sink'
    Availability: Always
    Capabilities:
      video/x-h265
              alignment: au
                profile: { (string)main, (string)main-10 }
          stream-format: byte-stream

  SRC template: 'src'
    Availability: Always
    Capabilities:
      video/x-raw(memory:MFXSurface)
                 format: { (string)NV12, (string)BGRA, (string)P010_10LE, (string)YUY2, (string)ENCODED }
                  width: [ 1, 2147483647 ]
                 height: [ 1, 2147483647 ]
              framerate: [ 0/1, 2147483647/1 ]
      video/x-raw
                 format: { (string)NV12, (string)BGRA, (string)P010_10LE, (string)YUY2 }
                  width: [ 1, 2147483647 ]
                 height: [ 1, 2147483647 ]
              framerate: [ 0/1, 2147483647/1

I'm running on Windows 10. My CPU is a i7-6700K @ 4GHZ. 

Any insight or thoughts would be appreciated. I'm digging through the Media SDK's GStreamer plugin code right now to find the source of the issue, but I'm hoping someone can offer a pointer or two. Thanks. 

Edit: 

I have found that the problem at least manifests itself at this point in the code. The call to gst_video_decoder_finish_frame() returns GST_FLOW_NOT_NEGOTIATED if BGRA is specified, GST_FLOW_OK if NV12 is specified. 

Hi, maybe this is not the best place to ask this question, but maybe someone can help me....

After finishing the process of registering Intel Media Server Community Edition, I get the message: Thank you for activating your Intel® Media Server Studio - Community Edition product. But the link to start the download does not appear, however when i tried the same with  Intel® Media Server Studio Professional Edition it is displayed to download. Does anyone know why this happens?

Thank's you

the latest intel media sdk only support win10.

but our parogram is running on win7, i've found that intel medis sdk 2016 supoort win7,but i cann't get the sample with 2016.
 

i need help ,thanks.

(Environment)
Windows10 Pro
Visual Studio 2010 Professional
openCV V4.0.0
CMake V3.14.5
Intel Media SDK

(Problem)
After install Intel Media SDK, I tryed to include "WITH_MFX" to openCV with CMake, but CMake-GUI does not show "WITH_MFX" on the list. That's why I can not inclue it.
Please tell me how to include "WITH_MFX" to openCV. My procedure is as follow.

1. Install Intel Media SDK
2. Create "build" directory
3. Open CMake-GUI
4. Set source code path and build the binaries path on CMake-GUI
5. Execute "Configure"
6. After Configure, turn on  "WITH_MFX" on the list. *But I can not turn on it because it is no shown on the list.
 

Hi, I have a question.

writer[0].open( cFileList[GnIndex], cv::CAP_INTEL_MFX, GnFourccCode, GnFrameRate, cvSize(GnFrameWidth, GnFrameHeight), true);

I can open this in the following CPU.

CPU:Intel(R) Core(TM)i7-8650U CPU@1.90GHz 2.11GHz

But I can not open this in the following CPU.

CPU:Intel(R) Atom(TM) CPU E3825@1.33GHz 1.33GHz

Please tell me why can not open it.

Hi, 

In the release notes I find this list:

Intel® Celeron®and Pentuim® processorscodenamed "Apollo Lake I"
Intel® Xeon® E3-1200 v4 Family with C226 chipset
Intel® Xeon® E3-1200 and E3-1500 v5 Family with C236 chipset
5th Generation Intel® Core™
6th Generation Intel® Core™
7th Generation Intel® Core™
8th Generation Intel® Core™
9th Generation Intel® Core™
10th Generation Intel® Core™
 

I've got some questions about:

• The only Xeon (enterprise server processor) are the one listed and so the latest Intel Xeon E3 v6 are not supported ?
• For instance the Xeon E3-1585 v5 comes with a Iris Pro P580. I guess is the Iris Pro that does the hardware processing. There will be further products like the Iris Pro in the future  ? Are they going to come with a new Xeon ?
• Can you indicate the End of Sales of the E3-1585 v5 and E3-1585L v5 processors ?
• The release notes states that a lot of gen of core processors are supported. Do they need to have a grafic processor like the HD 4600 to accellerate encoding ? I mean 9th gen Core i3-9300T will work as well as Core i3-9100F of the same gen ? the latter is missing the UHD 630.

I'm sorry of bothering with so much question that implies some times to answer, but we are starting a whole transcoding farm based on E3-1585 v5 and we want to be sure that the hardware is not end of sales or the will be Enterprise alternatives in the future.

BR

Francesco

Hi, 

I'm working on linux with the MediaServerStudioEssential.

There, it seems pretty clear that if you need to accelerate HEVC you need a commercial license, correct me if I'm wrong please.
Instead, if you are just using the hardware acceleration on h264 the is no need of a commercial license.

Please keep in mind that we use a custom build of ffmpeg for transcoding.

So it seems that we just need the media SDK and not the Media Studio Server.
The Media SDK 2019 Q1 on git hub come with a nice MediaStack.

We are guessing what kind of license do we need to use ffmpeg with hardware acceleration for h264,MPEG2,HEVC using the MediaStack on github.
I read the License that comes with the media stack but we'd like to be really sure.
Please consider that transcoded videos are for commercial use.

BR
Francesco

W/Adreno-ES20: <core_glUseProgram:1557>: GL_INVALID_VALUE
W/System.err: java.lang.RuntimeException: glGetAttribLocation uOrientationM: glError 1281
W/System.err:     at com.intel.inde.mp.android.graphics.EglUtil.checkEglError(EglUtil.java:215)
W/System.err:     at com.intel.inde.mp.android.graphics.ShaderProgram.getAttributeLocation(ShaderProgram.java:113)
W/System.err:     at com.intel.inde.mp.android.graphics.FullFrameTexture.draw(FullFrameTexture.java:93)
W/System.err:     at com.qzplugin.Capturing$EncodeThread.run(Capturing.java:90)

Hi,

I would like to use HW encoding(GPU) on the following environment.

(Environment)

OS : Windows 7

Processor : CPU:Intel(R) Atom(TM) CPU E3825@1.33GHz 1.33GHz

According to the following web page, Intel Media SDK does not support this processor.

https://software.intel.com/en-us/media-sdk/documentation/system-requirements

Are there any method to use HW encoding on this processor?

Would you please tell me how to solve the problem?

Is vfx good as career ?

We have been using Intel QSV for H264 Encode/Decode within our own video communications software application for quite some time without any problems.  Recently, I updated my PC to version 1903 of Windows 10.  MFXInit for both the H264 Encode/Decode now return MFX_ERR_UNSUPPORTED. 

This is the code we run, which used to work without any problems:

mfxVersion minHWversion = {0,1};

sts = MFXInit(MFX_IMPL_HARDWARE_ANY, &minHWversion, &Session);

As well as returning (MFX_ERR_UNSUPPORTED), the Visual Studio 2019 debugger also gives the following errors:

onecore\windows\directx\database\helperlibrary\lib\directxdatabasequeryimpl.cpp(53)\d3d9.dll!6FEA882A: (caller: 6FEA89DE) ReturnHr(106) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasequeryimpl.cpp(180)\d3d9.dll!6FEA89F7: (caller: 6FEA885D) ReturnHr(107) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasequeryimpl.cpp(163)\d3d9.dll!6FEA887F: (caller: 6FEA7949) ReturnHr(108) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasehelper.cpp(705)\d3d9.dll!6FEA79E3: (caller: 6FEA8116) ReturnHr(109) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasehelper.cpp(820)\d3d9.dll!6FEA8189: (caller: 6FE2A6A6) ReturnHr(110) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasehelper.cpp(515)\d3d9.dll!6FEA8264: (caller: 6FE0C673) ReturnHr(111) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasehelper.cpp(515)\d3d9.dll!6FEA8264: (caller: 6FE0C673) ReturnHr(112) tid(2f98) 80004002 No such interface supported

onecore\windows\directx\database\helperlibrary\lib\directxdatabasequeryimpl.cpp(180)\dxgi.dll!6B5AC110: (caller: 6B5ABF76) ReturnHr(16) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasequeryimpl.cpp(163)\dxgi.dll!6B5ABF98: (caller: 6B5AB3C9) ReturnHr(17) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasehelper.cpp(705)\dxgi.dll!6B5AB462: (caller: 6B5AB9D0) ReturnHr(18) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasehelper.cpp(820)\dxgi.dll!6B5ABA43: (caller: 6B57282E) ReturnHr(19) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasehelper.cpp(515)\dxgi.dll!6B5ABB0B: (caller: 6B594DF2) ReturnHr(20) tid(2f98) 80004002 No such interface supported
onecore\windows\directx\database\helperlibrary\lib\directxdatabasehelper.cpp(515)\dxgi.dll!6B5ABB0B: (caller: 6B594DF2) ReturnHr(21) tid(2f98) 80004002 No such interface supported

I am using a i7-7700K (Kaby Lake) with Intel HD Graphics 630.  At the time of the update I was using driver version (25.20.100.6374) which was working.  As the Windows 10 update broke the Intel QSV encode/decode, I upgraded the GPU driver to version (26.20.100.6912), which made no difference, it still doesn't work.

I am using Intel Media SDK 2018 R2.  I've run the media SDK analyzer and it outputs the following (I've removed 1.00 to 1.24 for clarity):

        1.25    HW      Yes             X       X
        1.25    SW      Yes             X       X
        1.26    HW      Yes             X       X
        1.26    SW      Yes             X       X
        1.27    HW      Yes             X       X
        1.27    SW      Yes             X       X
        1.28    HW      Yes             X       X
        1.28    SW      No
        1.29    HW      Yes             X       X
        1.29    SW      No

Graphics Devices:
        Name                                         Version                    State
        AMD Radeon RX 5700                  26.20.13001.9005    Offline
        NVIDIA GeForce GTX 1660           26.21.14.3136          Offline
        Intel(R) HD Graphics 630              26.20.100.6912        Running / Full Power

System info:
        CPU:    Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz
        OS:     Microsoft Windows 10 Pro
        Arch:   64-bit

Installed Media SDK packages (be patient...processing takes some time):
        Intel(R) Media SDK 2018 R2 - HEVC GPU accelerated Encoder
        Intel(R) Media SDK 2018 R2 - Media Samples
        Intel(R) Media SDK 2018 R2 - Software Development Kit
        Intel(R) Media SDK 2018 R2 - Documentation for HEVC
        Intel(R) Media SDK 2018 R2 - HEVC SW Encoder
        Intel(R) INDE Media SDK for Windows 6.0.0.388
        Intel(R) Media SDK 2018 R2 - HEVC SW Decoder

Installed Media SDK DirectShow filters:

Installed Intel Media Foundation Transforms:
        Intel(R) Hardware M-JPEG Decoder MFT : {00C69F81-0524-48C0-A353-4DD9D54F9A6E}

I have re-installed GPU drivers and DirectX, while DxDiag reports no errors.

I am struggling to find the reason for the MFXInit failure, other than the Windows 10 (1903) update corrupting my PC system in some way.  Has anyone seen or knows anything about the errors I am getting?  Any suggestions about how to resolve this issue would be greatly appreciated.

Thank you.

1903 system application store where to report mistakes, can not install; Check again can not find the application, but check where the media reported missing documents, I hope to send or provide informed solutions, thank you very much!

I have trying out the noise reduction using media sdk which works fine but currently we are not in the position of using media sdk due to amount of work needed. The best option we have right now is using the vaapi and there when I try to use the noise reduction filter it does not seems to be doing anything. I am using the iHD driver with latest version of vaapi. My question is why is the noise reduction filter not working when I use the vaapi ? 

Hi,

I am trying to change codec profile to main for H264 stream from yuv using sample_encode,

What does -CodecProfile option takes as input to change the profile to main.

../sample_encode h264 -hw -vaapi -CodecProfile ??? -i old_town_cross_2160p50.y4m -o oldtownprofile.h264 -h 2160 -w 3840 -b 20000 -f 60
 

I want to use media sdk to perform denoising on linux. The problem I have is that I have VASurfaceID which i want to give to the media sdk to perform denoising. Is there any good way to convert this VASurfaceID to mfxFrameSurface1 or something similar so that I can use Media SDK?  

When I call MFXVideoDecode_Init() it fails with MFX_ERR_UNSUPPORTED. 

I'm using an annex-b .h264 file that I generated with FFMPEG. I get the same error when I use the .264 file from the samples; this is obvious ofcourse as VideoDECODEH264::Init() has a #ifdef that triggers the return of MFX_ERR_UNSUPPORTED, as you can see in the linked code

VideoDECODEH264::Init() returns MFX_ERR_UNSUPPORTED because MFX_VA_LINUX is defined in mfx_config.h:

 But I wasn't able to find where MFX_VA was defined.

How can I fix this?

CMAKE output

CPU info

Hi,

I am new to media SDK and any gfx development in general. I want use the sample 'simple_decode_d3d11' provided in tutorial to read the input from my webcam or USB camera connected to my system instead of a video file. I could not figure out how to specify camera as input in MSDK_FOPEN/fopen_s. I am using Intel(R) Media SDK 2018 R2 on wondows10. Appreciate any help regarding this.

-Surabhi

I am using GStreamer-vaapi on Intel Atom x5-Z8550. I cannot access the encoding feature called "variable bitrate" (VBR) which is available as a "rate-control" option for the GStreamer vaapih264enc plugin like this: "vaapih264enc rate-control=vbr"

Our questions:

1. Is variable bitrate supported on our processor?

2. Where can I find documentation that indicates support or the lack thereof for this feature?

I am building GStreamer-vaapi from source. I have been able to trace the problem into a VA-API library call: vaGetConfigAttributes().

For our processor, vaGetConfigAttributes() indicates support for only CBR 0x02 (constant-bitrate) and CQP 0x10 (constant-quality). vaGetConfigAttributes comes from the core VA-API library as documented here: http://intel.github.io/libva/group__api__core.html#gae51cad2e388d6cc63ce.... We need access to VBR 0x04 (variable-bitrate).

Environment:

Linux kernel: v4.9.115

vainfo
libva info: VA-API version 0.39.4
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/i965_drv_video.so
libva info: Found init function __vaDriverInit_0_39
libva info: va_openDriver() returns 0
vainfo: VA-API version: 0.39 (libva 1.7.3)
vainfo: Driver version: Intel i965 driver for Intel(R) CherryView - 1.7.3

Replication Steps:

This problem can be demonstrated by using the standard GStreamer command line application: "gst-launch-1.0".

I have tried similar pipelines in GStreamer 1.14.1, 1.14.5, and 1.16.0 with identical results.

This pipeline works with rate-control=constant bitrate (CBR):

gst-launch-1.0 -v videotestsrc num-buffers=1000 ! videoconvert ! vaapipostproc width=1280 height=720 ! vaapih264enc rate-control=cbr bitrate=8000 ! mp4mux ! filesink location=video.mp4

This pipeline fails with rate-control=variable bitrate (VBR):

gst-launch-1.0 -v videotestsrc num-buffers=1000 ! videoconvert ! vaapipostproc width=1280 height=720 ! vaapih264enc rate-control=vbr bitrate=8000 ! mp4mux ! filesink location=video.mp4

    ERROR: from element /GstPipeline:pipeline0/GstVideoTestSrc:videotestsrc0: Internal data stream error.
    streaming stopped, reason not-negotiated (-4)
    ERROR: pipeline doesn't want to preroll.

I am compiling the mfx_dispatch code that is distributed along with the Intel Media SDK 2019, and there is a bug when using it with 3rd generation processors that was not in the 2018 R2 sdk.

Line 213 of mfx_dispatcher.cpp checks whether it should call MFXInit or MFXInitEx by checking if the MFXInitEx entry in the MFX_DISP_HANDLE.callTable object is null. 

In 2019 R1, mfx_dispatcher.cpp was changed so that line 63 now reads:

    if (MFX_ERR_NONE != mfxRes)

instead of:

    if (MFX_ERR_NONE == mfxRes)

This means that when MFX_DISP_HANDLE::Close is called, the callTable variable is not reset, and so it is full of now invalid function pointers.  If the MFX_DISP_HANDLE::LoadSelectedDLL is called on a previously closed object, as main.cpp from the dispatcher does, you can end up with a callTable with invalid pointers in it.  This means the check on line 213 of mx_dipsatcher.cpp thinks it should call MFXInitEx when it shouldn't, and the program crashes.

An easy fix would be to reset the callTable object in the MFX_DISP_HANDLE::UnLoadSelectedDLL() method, but I wasn't sure if there would be any repercussions to doing that, that I wasn't aware of.