Intel Quick Sync Video Consumption by Applications

I wrote a few posts on hardware H.264 encoding (e.g. this and the latest one Applying Hardsubs to H.264 Video with GPU). A blog reader asked a question regarding availability of the mentioned Intel Quick Sync Video support with low end Intel x5-Z8300 CPU.

[…] Intel has advertised that the Cherry Trail CPUs support H264 encoding and / or QSV, but nowhere have I seen a demo of this being used […].
What did you use to encode the video? Is the QSV codec available in the x5-z8300 for possible 720p realtime encoding? I’d like to see this checked in regards to using software like FFmpeg with qsv_h264 -codec and OBS. […]

A picture below explains how applications are consuming Intel’s hardware video compression offering in Windows.

Intel QSV includes hardware implementation of the encoder and corresponding drivers which provide a frontend API to software. This includes a component which integrates the codec with Microsoft’s Media Foundation API. Applications are to choose between interfacing the codec using Windows API – this is the way stock Microsoft applications work, and this is the way I used for video encoding development mentioned on the blog. Other applications prefer to interface through Intel Media SDK, which is an alternate route ending up at the same hardware-backed services.

Intel x5-Z8300 system in question has H.264 video encoding support integrated into Windows API and the services can be consumed without additional Intel runtime and/or development kit. The codec, according to the benchmarks made earlier, is fast enough to handle real-time 720p video encoding nevertheless the device is a budget thing.

Bug in Media Foundation MPEG-4 File Source related to timestamping video frames of a fragmented MP4 file

Some recent update in Media Foundation platform introduced a new bug related to fragmented MP4 files and H.264 video. The bug shows up consistently with file versions:

  • mfplat.dll – 10.0.14393.351 (rs1_release_inmarket.161014-1755)    15-Oct-16 05:48
  • mfmp4srcsnk.dll – 10.0.14393.351 (rs1_release_inmarket.161014-1755)    15-Oct-16 05:45

The nature of the problem is that MPEG-4 File Source is incorrectly time stamping the data: frame time stamps are incorrect, they seems to be getting wrong durations and increments, then quickly jumps into future… and on playback this leads to unobvious playback freezes. As Media Foundation is used by Windows Media Player, Windows 10 Movies & TV Player, the bug is present there as well.

The original report is on MSDN Forums.

Presumably it is possible to roll certain Windows Update package back, or alternatively one has to wait for Microsoft to fix the problem and deliver a new update deploying the fix.

Applicability of Virtual DirectShow Sources

Virtual DirectShow  sources have been a long time synonym of software-only camera implementation exposed to applications along with physical cameras in a way that applications consume the sources without making a difference whether the camera is real or virtual. Vivek’s template was a starting point for many:

Capture Source Filter filter (version 0.1) 86 KB zipped, includes binaries.  A sample source filter that emulates a video capture device contributed by Vivek (rep movsd from the public newsgroups).  Thanks Vivek!  TMH has not tested this filter yet.  Ask questions about this on microsoft.public.win32.programmer.directx.video.

With API changes over years, the sample and the concept is still understood as the method of adding a virtual camera, however new scenarios exist where the concept no longer works. Typical problems:

  1. 64-bit applications cannot consume virtual 32-bit virtual sources
  2. Virtual sources are no visible and accessible to applications consuming video using Media Foundation API

The diagram below explains the applicability of virtual cameras:

Applicability of Virtual DirectShow Sources

Important is that virtual sources can only be consumed by the DirectShow-based applications of the same bitness.

If source developer needs to synchronize virtual source throughout multiple applications (e.g. video is synthesized by another application and needs to be deliverable to multiple clients), he needs to add interprocess synchronization on the backyard of virtual source.

If developer needs to support both 32- and 64-bit apps, he needs both variants of virtual sources registered, and possibly synchronization of the kind of the paragraph above.

The only virtual device which is visible to all video capture applications if implemented by kernel level driver (implementations are rare but exist).

See also:

KB3176938’s Frame Server update visually

  1. M-JPEG and H.264 media types are available again (good)
  2. Nevertheless connected, H.264 video is not processed correctly; new bug or old one? Not clear. Even though it sort of works, in DirectShow it looks broken in another new way (this and not just this), perhaps a collateral damage and maybe never ever fixed…
  3. There is no camera sharing between the applications even though it was the justification for the changes in first place. For now Frame Server is just useless overhead, which adds bad stuff, is polished a bit to do not so much harm, and maybe turns to be good some time later.
    • for the record, the camera works in Skype when it is not consumed elsewhere concurrently

BTW the hack that bypasses FrameServer survived the update and remains in good standing.

DirectShowCaptureCapabilities and MediaFoundationCaptureCapabilities: API version of EnableFrameServerMode state

Both tools now include exact version of the API and also include an export or registry key related to frame server.

Capture Capabilities: API Version and State

mfcore.dll version of 10.0.14393.105 corresponds to Cumulative Update for Windows 10 Version 1607: August 31, 2016 also known as KB3176938 with DirectShow and Media Foundation improvement for Windows 10 Anniversary Update that restores availability of compressed media types.

See:

Enumeration of DirectShow Capture Capabilities (Video and Audio)

Media Foundation Video/Audio Capture Capabilities

Number of streams served by IMFSourceReader interface

It looks confusing that IMFSourceReader interface does not offer a dedicated method to find out the number of streams behind it. There is a IMFMediaSource instance behind the reader, and its streams are available through IMFMediaSource::CreatePresentationDescriptor method and IMFPresentationDescriptor::GetStreamDescriptorCount method call.

I am under impression that source reader’s method just has to be there even though I am not seeing it looking at the list of methods. Okay, there are other methods, namely IMFSourceReader::GetStreamSelection method, which takes either ordinal stream index or an alias as the first argument, then returns MF_E_INVALIDSTREAMNUMBER if you run out of streams. However the problem is that this is associated with an internal exception, and I consider exceptions as exceptional conditions the code should not normally hit. I would expect to have a legal exception-free way to find out the number of streams. I am using debugger that breaks on exception or at least pollutes output log for no reason, I use other tools that intercept and log exceptions as something that needs attention – getting number of streams is nowhere near there.

Internal MF_E_INVALIDSTREAMNUMBER Exception

Even though it is not a real drawback of the API since it is still possible to get the data and the API acts as documented, I still think someone overlooked this and API like this should have have a normal method or argument to request number of streams explicitly. Or I am just not seeing it even though I am trying thoroughly.

Anniversary Webcam Saga: It’s clear who’s guilty, now what to do? (Updated)

As new and new people discover the Windows 10 Anniversary Update breaking changes (expectedly running mad), let’s reiterate the possible solutions:

  1. You don’t like the idea that video sharing service adds latency, and adds man-in-the-middle access/spying over a video feed.
    See #6 below.
  2. You are consuming raw video from camera using one of the uncompressed modes within USB 2.0 bandwidth.
    You are likely to be not affected by the changes.
  3. You are consuming raw video from camera but resolution/rate combination makes it unable to capture raw video, so you captured M-JPEG instead and decoded that, via DirectShow API.
    It is no longer possible, but you can use Media Foundation API instead. Or someone will develop a wrapper that re-exposes Media Foundation captured video via DirectShow.
  4. Same as #3 above but via Media Foundation API.
    You have the option to consume already decoded video, new subsystem will automatically capture M-JPEG and decode into NV12.
  5. You take advantage of compressed format of video captured (DirectShow or Media Foundation) so that you don’t need compress it for storage or network transmission purposes.
    Compressed captured video is no longer available, see #6 below.
  6. You take advantage of H.264 video capture offered by UVC 1.5 device, including fine tuning of hardware H.264 compression.
    Just as in #1 and #5 above, you are in trouble. Windows Camera Frame Server no longer offers access to such video feed. You need a hack (yes, it’s confirmed to be possible) that restores original behavior of video capture hardware.

These and other reasons related to the fact that applications no longer talk to real capture device, but rather a Frame Server Client that proxies a web camera, will possibly require that video capture applications are updated in order to work well in new version of the operating system.

It is unclear if and how Microsoft and Media Foundation team will respond to customer pain voices. First, it looked as a bug and one could expect a response and fix. But with the information from Windows Camera Team it looks completely different. No, they did not accidentally break it up – it was a planned change. Then they connected new behavior with new Microsoft Products – new products rely on new behavior. Then they did a few nasty things, not just one: added proxy service, killed UVC related compression control over the device, reduced range of operation modes for DirectShow they look for ways to deprecate, conceptually removed compressed video capture modes. I think there is no way back – Windows Camera Frame Server is new reality. The best to expect is that some of the mentioned problems are relaxed by offering greater flexibility by the platform. Maybe they will add some sort of exclusive modes for video capture or “professional” hardware which offers more through the API. In any event these changes are unlikely to appear soon, as they will go through the full cycle of development and take months to get delivered. Public pressure might force that to appear rather earlier than later, but I don’t think it is what is going to happen.

16 Aug update: Windows Camera Team reported that they see customer pain and will do something to ease it shortly. As I see it, they will address scenarios #3, #4, #5 above, for MJPG video, to allow compressed formats pass Frame Server so that users could consume them from their applications and Frame Server would be able to release frames not just after decoder, but also before the shared decoder. Also as use of H.264 is limited, it might be not included into hotfix at all, or will be included much later being given more serious consideration (which might end up as dropped support in DirectShow and something new introduced for Media Foundation).

19 Aug update: Someone took time to locate a registry value. User WithinRafael on MSDN Forums:

Try opening up

HKLM\SOFTWARE\Microsoft\Windows Media Foundation\Platform (32- and 64-bit OS)
HKLM\SOFTWARE\WOW6432Node\Microsoft\Windows Media Foundation\Platform (64-bit OS)

and add a DWORD value with name EnableFrameServerMode. Set its value to 0 and try again.

Put a sticky note on your monitor to revisit this if/when Microsoft issues a fix.

Also:

Untitled