MFXVideoDECODE_DecodeHeader and Windows Media Foundation

[chatelier__pierre]

[context]

I want to use oneAPI/vpl/MFXVideoDECODE for raw samples provided by an IMFSourceReader (part of Microsoft Windows Media Foundation). That way, I delegate the media reading to Media Foundation, and let H264 decoding occur through VPL. I need to to that explicitely, I don't want to use "plugins".

[what works]

I have a valid avi file containing valid H264 (AVC1) samples, that can be played by any movie player.

I can myself handle that file with the Media Foundation API.

[what does not work]

In order to call MFXVideoDECODE_Init() , I need a valid mfxVideoParam structure filled with information matching the movie.

Since it seems really difficult to fill correctly a mfxVideoParam by hand, and that there is no way to fill it from an IMFMediaType, I have to resort to MFXVideoDECODE_DecodeHeader().

However, whatever I send to MFXVideoDECODE_DecodeHeader(), it pretends to succeed but does not fill correctly the structure, making the next call to MFXVideoDECODE_Init() fail miserably

[some details]

I tried to feed the input bitstream of MFXVideoDECODE_DecodeHeader() with

-the complete raw file bytes

-the first IMFMediaSample from the IMFSourceReader

-the MF_MT_MPEG_SEQUENCE_HEADER  blob from the IMFMediaType (that contains the NALUs)

-as expected, I also manually set the mfx.codecId of the mfxVideoParam  as MFX_CODEC_AVC

in any case, MFXVideoDECODE_DecodeHeader() returns success, but no mfx.FrameInfo fields  are filled, and of course MFXVideoDECODE_Init() will fail

[the question]

What am I supposed to do to get a correct mfxVideoParam  from a sequence opened in Windows Media Foundation ?

[Pamela_H_Intel]

Pierre,

Have you begun using the oneVPL GPU implementation? If not, we will try to find a workaround for you.

However, know that the CPU implementation of oneVPL is a reference implementation and is not expected to be performant.

We have been working to improve our documentation to be more clear.

Pamela

[chatelier__pierre]

I don't have a supported GPU to test oneVPL GPU implementation.

I understand that CPU is not "a reference implementation and is not expected to be performant.", but once again it has nothing to do with the *bug* !
And that's the point : if it does not work at least with the CPU, I have zero confidence to go further with this technology.

Please reread this thread. It started 8 months ago and I have not been provided any useful hint or solution. From time to time, a new message was posted that just had nothing to do with addressing the problem.

This is one of the worst support I have ever seen.