Windows Media Foundation 音視訊採集

smilestone322發表於2017-11-15

Windows

轉自：https://www.cnblogs.com/rhzhang/p/5185686.html

寫在前面

我是個講文明的人…… 不過有的時候實在忍不住了也要吐槽幾句：

1. 我真是跟不上時代，到現在了還在研究 Windows 應用開發…… 咳；

2. DirectShow 是傻X！我只是想要獲取 Camera 裸資料，尼瑪卻要讓我學習神馬各種 .ax, filter, graph... 相關資料少、又晦澀；

3. 在此祝願 Windows XP 及其之前的版本早點退出歷史舞臺，這樣 DirectShow 就不是必須的了！

音視訊採集

Windows 7 以後， Media Foundation 支援音視訊採集。視訊採集裝置必須支援 UVC 1.1 驅動。音訊採集裝置通過 Windows Audio Session API (WASAPI) 來支援。

採集裝置在 Media Foundation 中表現為 Media Source 物件，暴露出 IMFMediaSource 介面。但是一般情況下應用程式不會直接用這個介面，而會使用更高層的API，比如 Source Reader IMFSourceReader 來控制採集裝置。

列舉裝置並設定裝置

1. 呼叫 MFCreateAttributes 函式建立屬性庫；

2. 設定 MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE 屬性（MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_GUID 或 MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID）；

3. 呼叫 MFEnumDeviceSources 函式就會得到一組 IMFActivate 指標。每個指標指向一個裝置；

4. 呼叫 GetAllocatedString 函式可以獲取裝置名稱，其中，

4.1. 用 GetAllocatedString 函式的 MF_DEVSOURCE_ATTRIBUTE_FRIENDLY_NAME 引數可以獲取顯示名稱

4.2. 用 GetAllocatedString 函式的 MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_SYMBOLIC_LINK 引數可以獲取視訊採集裝置的物理名稱，名稱是唯一的，但不好閱讀

4.3. 用 GetAllocatedString 函式的 MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_ENDPOINT_ID 引數可以獲取音訊採集裝置的物理名稱，名稱是唯一的，但不好閱讀

5. 呼叫 IMFActivate::ActivateObject 或 MFCreateDeviceSource 方法來建立一個可用的 Media Source 物件例項。對於具體的一個採集裝置（建立好了的 Media Source），獲取裝置所支援的格式，並設定採集格式：

5.1. 呼叫 IMFMediaSource::CreatePresentationDescriptor 獲取媒體源（Media Source）的表現描述符（presentation descriptor）

5.2. 呼叫 IMFPresentationDescriptor::GetStreamDescriptorByIndex 獲取視訊流的流描述符（Stream Descriptor）

5.3. 呼叫 IMFStreamDescriptor::GetMediaTypeHandler 獲取流描述符的媒體型別控制程式碼（IMFMediaTypeHandler）

5.4. 呼叫 IMFMediaTypeHandler:: 的 GetMediaTypeCount 和 GetMediaTypeByIndex 來獲取每一條目的格式（IMFMediaType，每一條目的格式包含許多屬性，如寬高、取樣格式、幀率），另外，教程還有一些實用函式，用於列印 IMFAttributes 。

6. 最後， IMFMediaTypeHandler::SetCurrentMediaType 設定指定條目的格式（IMFMediaType）。如果要修改 IMFMediaType 的某個屬性（比如 Frame Rate）的話，就呼叫 IMFMediaType::SetItem 來設定具體的屬性（比如 MF_MT_FRAME_RATE）。具體要看程式碼來理解。

通過 Source Reader 讀取媒體

當獲取到裝置以後，一般用 Source Reader 來獲取媒體

1. 呼叫 MFCreateSourceReaderFromMediaSource 關聯 IMFMediaSource 和 IMFSourceReader;

2. 呼叫 IMFSourceReader::GetNativeMediaType 獲取各種 Source 提供的媒體型別；

3. 呼叫 IMFSourceReader::SetCurrentMediaType 設定想要的媒體型別；

4. 呼叫 IMFSourceReader::ReadSample 來獲取音訊取樣或視訊取樣（視訊幀）。如果想要使用非同步方式來 ReadSample，則需要設定一下 MF_SOURCE_READER_ASYNC_CALLBACK 。

其他功能

音視訊採集 + 編碼（Transcode）： https://msdn.microsoft.com/en-us/library/ff485863.aspx

播放媒體檔案：https://msdn.microsoft.com/en-us/library/ms703190.aspx

程式碼地址

在 Windows SDK 的 Samples\multimedia\mediafoundation 目錄中

Enumerating Video Capture Devices

This topic describes how to enumerate the video capture devices on the user's system, and how create an instance of a device.

To enumerate the video capture devices on the system, do the following:

Call MFCreateAttributes to create an attribute store. This function receives anIMFAttributes pointer.
Call IMFAttributes::SetGUID to set the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE attribute. Set the attribute value to MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID.
Call MFEnumDeviceSources. This function receives an array ofIMFActivate pointers and the array size. Each pointer represents a distinct video capture device.

To create an instance of a capture device:

Call IMFActivate::ActivateObject to get a pointer to theIMFMediaSource interface.

The following code shows these steps:

C++

Copy

HRESULT CreateVideoDeviceSource(IMFMediaSource **ppSource)
{
    *ppSource = NULL;

    IMFMediaSource *pSource = NULL;
    IMFAttributes *pAttributes = NULL;
    IMFActivate **ppDevices = NULL;

    // Create an attribute store to specify the enumeration parameters.
    HRESULT hr = MFCreateAttributes(&pAttributes, 1);
    if (FAILED(hr))
    {
        goto done;
    }

    // Source type: video capture devices
    hr = pAttributes->SetGUID(
        MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE, 
        MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID
        );
    if (FAILED(hr))
    {
        goto done;
    }

    // Enumerate devices.
    UINT32 count;
    hr = MFEnumDeviceSources(pAttributes, &ppDevices, &count);
    if (FAILED(hr))
    {
        goto done;
    }

    if (count == 0)
    {
        hr = E_FAIL;
        goto done;
    }

    // Create the media source object.
    hr = ppDevices[0]->ActivateObject(IID_PPV_ARGS(&pSource));
    if (FAILED(hr))
    {
        goto done;
    }

    *ppSource = pSource;
    (*ppSource)->AddRef();

done:
    SafeRelease(&pAttributes);

    for (DWORD i = 0; i < count; i++)
    {
        SafeRelease(&ppDevices[i]);
    }
    CoTaskMemFree(ppDevices);
    SafeRelease(&pSource);
    return hr;
}

After you create media source, release the interface pointers and free the memory for the array:

C++

Copy

    SafeRelease(&pAttributes);

    for (DWORD i = 0; i < count; i++)
    {
        SafeRelease(&ppDevices[i]);
    }
    CoTaskMemFree(ppDevices);

Audio/Video Capture in Media Foundation

Microsoft Media Foundation supports audio and video capture. Video capture devices are supported through the UVC class driver and must be compatible with UVC 1.1. Audio capture devices are supported through Windows Audio Session API (WASAPI).

A capture device is represented in Media Foundation by a media source object, which exposes the IMFMediaSource interface. In most cases, the application will not use this interface directly, but will use a higher-level API such as the Source Reader to control the capture device.

Enumerate Capture Devices

To enumerate the capture devices on the system, perform the following steps:

Call the MFCreateAttributes function to create an attribute store.
Set the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE attribute to one of the following values:

Value Description
MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_GUID Enumerate audio capture devices.
MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID Enumerate video capture devices.
Call the MFEnumDeviceSources function. This function allocates an array of IMFActivate pointers. Each pointer represents an activation object for one device on the system.
Call the IMFActivate::ActivateObject method to create an instance of the media source from one of the activation objects.

Value	Description
MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_GUID	Enumerate audio capture devices.
MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID	Enumerate video capture devices.

The following example creates a media source for the first video capture device in the enumeration list:

C++

Copy

HRESULT CreateVideoCaptureDevice(IMFMediaSource **ppSource)
{
    *ppSource = NULL;

    UINT32 count = 0;

    IMFAttributes *pConfig = NULL;
    IMFActivate **ppDevices = NULL;

    // Create an attribute store to hold the search criteria.
    HRESULT hr = MFCreateAttributes(&pConfig, 1);

    // Request video capture devices.
    if (SUCCEEDED(hr))
    {
        hr = pConfig->SetGUID(
            MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE, 
            MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID
            );
    }

    // Enumerate the devices,
    if (SUCCEEDED(hr))
    {
        hr = MFEnumDeviceSources(pConfig, &ppDevices, &count);
    }

    // Create a media source for the first device in the list.
    if (SUCCEEDED(hr))
    {
        if (count > 0)
        {
            hr = ppDevices[0]->ActivateObject(IID_PPV_ARGS(ppSource));
        }
        else
        {
            hr = MF_E_NOT_FOUND;
        }
    }

    for (DWORD i = 0; i < count; i++)
    {
        ppDevices[i]->Release();
    }
    CoTaskMemFree(ppDevices);
    return hr;
}

You can query the activation objects for various attributes, including the following:

The MF_DEVSOURCE_ATTRIBUTE_FRIENDLY_NAME attribute contains the display name of the device. The display name is suitable for showing to the user, but might not be unique.
For video devices, the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_SYMBOLIC_LINK attribute contains the symbolic link to the device. The symbolic link uniquely identifies the device on the system, but is not a readable string.
For audio devices, the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_ENDPOINT_ID attribute contains the audio endpoint ID of the device. The audio endpoint ID is similar to a symbolic link. It uniquely identifies the device on the system, but is not a readable string.

The following example takes an array of IMFActivate pointers and prints the display name of each device to the debug window:

C++

Copy

void DebugShowDeviceNames(IMFActivate **ppDevices, UINT count)
{
    for (DWORD i = 0; i < count; i++)
    {
        HRESULT hr = S_OK;
        WCHAR *szFriendlyName = NULL;
    
        // Try to get the display name.
        UINT32 cchName;
        hr = ppDevices[i]->GetAllocatedString(
            MF_DEVSOURCE_ATTRIBUTE_FRIENDLY_NAME,
            &szFriendlyName, &cchName);

        if (SUCCEEDED(hr))
        {
            OutputDebugString(szFriendlyName);
            OutputDebugString(L"\n");
        }
        CoTaskMemFree(szFriendlyName);
    }
}

If you already know the symbolic link for a video device, there is another way to create the media source for the device:

Call MFCreateAttributes to create an attribute store.
Set the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE attribute to MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID.
Set the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_SYMBOLIC_LINK attribute to the symbolic link.
Call either the MFCreateDeviceSource or MFCreateDeviceSourceActivate function. The former returns an IMFMediaSource pointer. The latter returns an IMFActivate pointer to an activation object. You can use the activation object to create the source. (An activation object can be marshaled to another process, so it is useful if you want to create the source in another process. For more information, see Activation Objects.)

The following example takes the symbolic link of a video device and creates a media source.

C++

Copy

HRESULT CreateVideoCaptureDevice(PCWSTR *pszSymbolicLink, IMFMediaSource **ppSource)
{
    *ppSource = NULL;
    
    IMFAttributes *pAttributes = NULL;
    IMFMediaSource *pSource = NULL;

    HRESULT hr = MFCreateAttributes(&pAttributes, 2);

    // Set the device type to video.
    if (SUCCEEDED(hr))
    {
        hr = pAttributes->SetGUID(
            MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE,
            MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_GUID
            );
    }


    // Set the symbolic link.
    if (SUCCEEDED(hr))
    {
        hr = pAttributes->SetString(
            MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_SYMBOLIC_LINK,
            (LPCWSTR)pszSymbolicLink
            );            
    }

    if (SUCCEEDED(hr))
    {
        hr = MFCreateDeviceSource(pAttributes, ppSource);
    }

    SafeRelease(&pAttributes);
    return hr;    
}

There is an equivalent way to create an audio device from the audio endpoint ID:

Call MFCreateAttributes to create an attribute store.
Set the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE attribute to MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_GUID.
Set the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_ENDPOINT_ID attribute to the endpoint ID.
Call either the MFCreateDeviceSource or MFCreateDeviceSourceActivate function.

The following example takes an audio endpoint ID and creates a media source.

C++

Copy

HRESULT CreateAudioCaptureDevice(PCWSTR *pszEndPointID, IMFMediaSource **ppSource)
{
    *ppSource = NULL;
    
    IMFAttributes *pAttributes = NULL;
    IMFMediaSource *pSource = NULL;

    HRESULT hr = MFCreateAttributes(&pAttributes, 2);

    // Set the device type to audio.
    if (SUCCEEDED(hr))
    {
        hr = pAttributes->SetGUID(
            MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE,
            MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_GUID
            );
    }

    // Set the endpoint ID.
    if (SUCCEEDED(hr))
    {
        hr = pAttributes->SetString(
            MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_AUDCAP_ENDPOINT_ID,
            (LPCWSTR)pszEndPointID
            ); 
    }

    if (SUCCEEDED(hr))
    {
        hr = MFCreateDeviceSource(pAttributes, ppSource);
    }

    SafeRelease(&pAttributes);
    return hr;    
}

Use a capture device

After you create the media source for a capture device, use the Source Reader to get data from the device. The Source Reader delivers media samples that contain the capture audio data or video frames. The next step depends on your application scenario:

Video preview: Use Microsoft Direct3D or Direct2D to display the video.
File capture: Use the Sink Writer to encode the file.
Audio preview: Use WASAPI.

If you want to combine audio capture with video capture, use the aggregate media source. The aggregate media source contains a collection of media sources and combines all of their streams into a single media source object. To create an instance of the aggregate media source, call the MFCreateAggregateSource function.

Shut down the capture device

When the capture device is no longer needed, you must shut down the device by calling Shutdown on the IMFMediaSource object you obtained by calling MFCreateDeviceSource or IMFActivate::ActivateObject. Failure to call Shutdown can result in memory links because the system may keep a reference to IMFMediaSource resources until Shutdown is called.

C++

Copy

if (g_pSource)
{
    g_pSource->Shutdown();
    g_pSource->Release();
    g_pSource = NULL;
}

If you allocated a string containing the symbolic link to a capture device, you should release this object as well.

C++

Copy

    CoTaskMemFree(g_pwszSymbolicLink);
    g_pwszSymbolicLink = NULL;

    g_cchSymbolicLink = 0;

Handling Video Device Loss

This topic describes how to detect device loss when using a video capture device. It contains the following sections:

Register For Device Notification

Before you start capturing from the device, call the RegisterDeviceNotification function to register for device notifications. Register for the KSCATEGORY_CAPTURE device class, as shown in the following code.

C++

Copy

#include <Dbt.h>
#include <ks.h>
#include <ksmedia.h>

HDEVNOTIFY  g_hdevnotify = NULL;

BOOL RegisterForDeviceNotification(HWND hwnd)
{
    DEV_BROADCAST_DEVICEINTERFACE di = { 0 };
    di.dbcc_size = sizeof(di);
    di.dbcc_devicetype  = DBT_DEVTYP_DEVICEINTERFACE;
    di.dbcc_classguid  = KSCATEGORY_CAPTURE; 

    g_hdevnotify = RegisterDeviceNotification(
        hwnd,
        &di,
        DEVICE_NOTIFY_WINDOW_HANDLE
        );

    if (g_hdevnotify == NULL)
    {
        return FALSE;
    }

    return TRUE;
}

Get the Symbolic Link of the Device

Enumerate the video devices on the system, as described in Enumerating Video Capture Devices. Choose a device from the list, and then query the activation object for the MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_SYMBOLIC_LINK attribute, as shown in the following code.

C++

Copy

WCHAR      *g_pwszSymbolicLink = NULL;
UINT32     g_cchSymbolicLink = 0;

HRESULT GetSymbolicLink(IMFActivate *pActivate)
{
    return pActivate->GetAllocatedString(
        MF_DEVSOURCE_ATTRIBUTE_SOURCE_TYPE_VIDCAP_SYMBOLIC_LINK,
        &g_pwszSymbolicLink,
        &g_cchSymbolicLink
        );
}

Handle WM_DEVICECHANGE

In your message loop, listen for WM_DEVICECHANGE messages. The lParam message parameter is a pointer to a DEV_BROADCAST_HDR structure.

C++

Copy

    case WM_DEVICECHANGE:
        if (lParam != 0)
        {
            HRESULT hr = S_OK;
            BOOL bDeviceLost = FALSE;

            hr = CheckDeviceLost((PDEV_BROADCAST_HDR)lParam, &bDeviceLost);

            if (FAILED(hr) || bDeviceLost)
            {
                CloseDevice();

                MessageBox(hwnd, L"Lost the capture device.", NULL, MB_OK);
            }
        }
        return TRUE;

Next, compare the device notification message against the symbolic link of your device, as follows:

Check the dbch_devicetype member of the DEV_BROADCAST_HDR structure. If the value is DBT_DEVTYP_DEVICEINTERFACE, cast the structure pointer to a DEV_BROADCAST_DEVICEINTERFACE structure.
Compare the dbcc_name member of this structure to the symbolic link of the device.

C++

Copy

HRESULT CheckDeviceLost(DEV_BROADCAST_HDR *pHdr, BOOL *pbDeviceLost)
{
    DEV_BROADCAST_DEVICEINTERFACE *pDi = NULL;

    if (pbDeviceLost == NULL)
    {
        return E_POINTER;
    }

    *pbDeviceLost = FALSE;
    
    if (g_pSource == NULL)
    {
        return S_OK;
    }
    if (pHdr == NULL)
    {
        return S_OK;
    }
    if (pHdr->dbch_devicetype != DBT_DEVTYP_DEVICEINTERFACE)
    {
        return S_OK;
    }

    // Compare the device name with the symbolic link.

    pDi = (DEV_BROADCAST_DEVICEINTERFACE*)pHdr;

    if (g_pwszSymbolicLink)
    {
        if (_wcsicmp(g_pwszSymbolicLink, pDi->dbcc_name) == 0)
        {
            *pbDeviceLost = TRUE;
        }
    }

    return S_OK;
}

Unregister For Notification

Before the application exits, call UnregisterDeviceNotification to unregister for device notifications/

C++

Copy

void OnClose(HWND /*hwnd*/)
{
    if (g_hdevnotify)
    {
        UnregisterDeviceNotification(g_hdevnotify);
    }

    PostQuitMessage(0);
}

Media Type Debugging Code

You can use the following code to view the contents of a media type while debugging.

C++

Copy

// The following code enables you to view the contents of a media type while 
// debugging.

#include <strsafe.h>

LPCWSTR GetGUIDNameConst(const GUID& guid);
HRESULT GetGUIDName(const GUID& guid, WCHAR **ppwsz);

HRESULT LogAttributeValueByIndex(IMFAttributes *pAttr, DWORD index);
HRESULT SpecialCaseAttributeValue(GUID guid, const PROPVARIANT& var);

void DBGMSG(PCWSTR format, ...);

HRESULT LogMediaType(IMFMediaType *pType)
{
    UINT32 count = 0;

    HRESULT hr = pType->GetCount(&count);
    if (FAILED(hr))
    {
        return hr;
    }

    if (count == 0)
    {
        DBGMSG(L"Empty media type.\n");
    }

    for (UINT32 i = 0; i < count; i++)
    {
        hr = LogAttributeValueByIndex(pType, i);
        if (FAILED(hr))
        {
            break;
        }
    }
    return hr;
}

HRESULT LogAttributeValueByIndex(IMFAttributes *pAttr, DWORD index)
{
    WCHAR *pGuidName = NULL;
    WCHAR *pGuidValName = NULL;

    GUID guid = { 0 };

    PROPVARIANT var;
    PropVariantInit(&var);

    HRESULT hr = pAttr->GetItemByIndex(index, &guid, &var);
    if (FAILED(hr))
    {
        goto done;
    }

    hr = GetGUIDName(guid, &pGuidName);
    if (FAILED(hr))
    {
        goto done;
    }

    DBGMSG(L"\t%s\t", pGuidName);

    hr = SpecialCaseAttributeValue(guid, var);
    if (FAILED(hr))
    {
        goto done;
    }
    if (hr == S_FALSE)
    {
        switch (var.vt)
        {
        case VT_UI4:
            DBGMSG(L"%d", var.ulVal);
            break;

        case VT_UI8:
            DBGMSG(L"%I64d", var.uhVal);
            break;

        case VT_R8:
            DBGMSG(L"%f", var.dblVal);
            break;

        case VT_CLSID:
            hr = GetGUIDName(*var.puuid, &pGuidValName);
            if (SUCCEEDED(hr))
            {
                DBGMSG(pGuidValName);
            }
            break;

        case VT_LPWSTR:
            DBGMSG(var.pwszVal);
            break;

        case VT_VECTOR | VT_UI1:
            DBGMSG(L"<<byte array>>");
            break;

        case VT_UNKNOWN:
            DBGMSG(L"IUnknown");
            break;

        default:
            DBGMSG(L"Unexpected attribute type (vt = %d)", var.vt);
            break;
        }
    }

done:
    DBGMSG(L"\n");
    CoTaskMemFree(pGuidName);
    CoTaskMemFree(pGuidValName);
    PropVariantClear(&var);
    return hr;
}

HRESULT GetGUIDName(const GUID& guid, WCHAR **ppwsz)
{
    HRESULT hr = S_OK;
    WCHAR *pName = NULL;

    LPCWSTR pcwsz = GetGUIDNameConst(guid);
    if (pcwsz)
    {
        size_t cchLength = 0;
    
        hr = StringCchLength(pcwsz, STRSAFE_MAX_CCH, &cchLength);
        if (FAILED(hr))
        {
            goto done;
        }
        
        pName = (WCHAR*)CoTaskMemAlloc((cchLength + 1) * sizeof(WCHAR));

        if (pName == NULL)
        {
            hr = E_OUTOFMEMORY;
            goto done;
        }

        hr = StringCchCopy(pName, cchLength + 1, pcwsz);
        if (FAILED(hr))
        {
            goto done;
        }
    }
    else
    {
        hr = StringFromCLSID(guid, &pName);
    }

done:
    if (FAILED(hr))
    {
        *ppwsz = NULL;
        CoTaskMemFree(pName);
    }
    else
    {
        *ppwsz = pName;
    }
    return hr;
}

void LogUINT32AsUINT64(const PROPVARIANT& var)
{
    UINT32 uHigh = 0, uLow = 0;
    Unpack2UINT32AsUINT64(var.uhVal.QuadPart, &uHigh, &uLow);
    DBGMSG(L"%d x %d", uHigh, uLow);
}

float OffsetToFloat(const MFOffset& offset)
{
    return offset.value + (static_cast<float>(offset.fract) / 65536.0f);
}

HRESULT LogVideoArea(const PROPVARIANT& var)
{
    if (var.caub.cElems < sizeof(MFVideoArea))
    {
        return MF_E_BUFFERTOOSMALL;
    }

    MFVideoArea *pArea = (MFVideoArea*)var.caub.pElems;

    DBGMSG(L"(%f,%f) (%d,%d)", OffsetToFloat(pArea->OffsetX), OffsetToFloat(pArea->OffsetY), 
        pArea->Area.cx, pArea->Area.cy);
    return S_OK;
}

// Handle certain known special cases.
HRESULT SpecialCaseAttributeValue(GUID guid, const PROPVARIANT& var)
{
    if ((guid == MF_MT_FRAME_RATE) || (guid == MF_MT_FRAME_RATE_RANGE_MAX) ||
        (guid == MF_MT_FRAME_RATE_RANGE_MIN) || (guid == MF_MT_FRAME_SIZE) ||
        (guid == MF_MT_PIXEL_ASPECT_RATIO))
    {
        // Attributes that contain two packed 32-bit values.
        LogUINT32AsUINT64(var);
    }
    else if ((guid == MF_MT_GEOMETRIC_APERTURE) || 
             (guid == MF_MT_MINIMUM_DISPLAY_APERTURE) || 
             (guid == MF_MT_PAN_SCAN_APERTURE))
    {
        // Attributes that an MFVideoArea structure.
        return LogVideoArea(var);
    }
    else
    {
        return S_FALSE;
    }
    return S_OK;
}

void DBGMSG(PCWSTR format, ...)
{
    va_list args;
    va_start(args, format);

    WCHAR msg[MAX_PATH];

    if (SUCCEEDED(StringCbVPrintf(msg, sizeof(msg), format, args)))
    {
        OutputDebugString(msg);
    }
}

#ifndef IF_EQUAL_RETURN
#define IF_EQUAL_RETURN(param, val) if(val == param) return L#val
#endif

LPCWSTR GetGUIDNameConst(const GUID& guid)
{
    IF_EQUAL_RETURN(guid, MF_MT_MAJOR_TYPE);
    IF_EQUAL_RETURN(guid, MF_MT_MAJOR_TYPE);
    IF_EQUAL_RETURN(guid, MF_MT_SUBTYPE);
    IF_EQUAL_RETURN(guid, MF_MT_ALL_SAMPLES_INDEPENDENT);
    IF_EQUAL_RETURN(guid, MF_MT_FIXED_SIZE_SAMPLES);
    IF_EQUAL_RETURN(guid, MF_MT_COMPRESSED);
    IF_EQUAL_RETURN(guid, MF_MT_SAMPLE_SIZE);
    IF_EQUAL_RETURN(guid, MF_MT_WRAPPED_TYPE);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_NUM_CHANNELS);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_SAMPLES_PER_SECOND);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_FLOAT_SAMPLES_PER_SECOND);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_AVG_BYTES_PER_SECOND);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_BLOCK_ALIGNMENT);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_BITS_PER_SAMPLE);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_VALID_BITS_PER_SAMPLE);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_SAMPLES_PER_BLOCK);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_CHANNEL_MASK);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_FOLDDOWN_MATRIX);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_WMADRC_PEAKREF);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_WMADRC_PEAKTARGET);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_WMADRC_AVGREF);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_WMADRC_AVGTARGET);
    IF_EQUAL_RETURN(guid, MF_MT_AUDIO_PREFER_WAVEFORMATEX);
    IF_EQUAL_RETURN(guid, MF_MT_AAC_PAYLOAD_TYPE);
    IF_EQUAL_RETURN(guid, MF_MT_AAC_AUDIO_PROFILE_LEVEL_INDICATION);
    IF_EQUAL_RETURN(guid, MF_MT_FRAME_SIZE);
    IF_EQUAL_RETURN(guid, MF_MT_FRAME_RATE);
    IF_EQUAL_RETURN(guid, MF_MT_FRAME_RATE_RANGE_MAX);
    IF_EQUAL_RETURN(guid, MF_MT_FRAME_RATE_RANGE_MIN);
    IF_EQUAL_RETURN(guid, MF_MT_PIXEL_ASPECT_RATIO);
    IF_EQUAL_RETURN(guid, MF_MT_DRM_FLAGS);
    IF_EQUAL_RETURN(guid, MF_MT_PAD_CONTROL_FLAGS);
    IF_EQUAL_RETURN(guid, MF_MT_SOURCE_CONTENT_HINT);
    IF_EQUAL_RETURN(guid, MF_MT_VIDEO_CHROMA_SITING);
    IF_EQUAL_RETURN(guid, MF_MT_INTERLACE_MODE);
    IF_EQUAL_RETURN(guid, MF_MT_TRANSFER_FUNCTION);
    IF_EQUAL_RETURN(guid, MF_MT_VIDEO_PRIMARIES);
    IF_EQUAL_RETURN(guid, MF_MT_CUSTOM_VIDEO_PRIMARIES);
    IF_EQUAL_RETURN(guid, MF_MT_YUV_MATRIX);
    IF_EQUAL_RETURN(guid, MF_MT_VIDEO_LIGHTING);
    IF_EQUAL_RETURN(guid, MF_MT_VIDEO_NOMINAL_RANGE);
    IF_EQUAL_RETURN(guid, MF_MT_GEOMETRIC_APERTURE);
    IF_EQUAL_RETURN(guid, MF_MT_MINIMUM_DISPLAY_APERTURE);
    IF_EQUAL_RETURN(guid, MF_MT_PAN_SCAN_APERTURE);
    IF_EQUAL_RETURN(guid, MF_MT_PAN_SCAN_ENABLED);
    IF_EQUAL_RETURN(guid, MF_MT_AVG_BITRATE);
    IF_EQUAL_RETURN(guid, MF_MT_AVG_BIT_ERROR_RATE);
    IF_EQUAL_RETURN(guid, MF_MT_MAX_KEYFRAME_SPACING);
    IF_EQUAL_RETURN(guid, MF_MT_DEFAULT_STRIDE);
    IF_EQUAL_RETURN(guid, MF_MT_PALETTE);
    IF_EQUAL_RETURN(guid, MF_MT_USER_DATA);
    IF_EQUAL_RETURN(guid, MF_MT_AM_FORMAT_TYPE);
    IF_EQUAL_RETURN(guid, MF_MT_MPEG_START_TIME_CODE);
    IF_EQUAL_RETURN(guid, MF_MT_MPEG2_PROFILE);
    IF_EQUAL_RETURN(guid, MF_MT_MPEG2_LEVEL);
    IF_EQUAL_RETURN(guid, MF_MT_MPEG2_FLAGS);
    IF_EQUAL_RETURN(guid, MF_MT_MPEG_SEQUENCE_HEADER);
    IF_EQUAL_RETURN(guid, MF_MT_DV_AAUX_SRC_PACK_0);
    IF_EQUAL_RETURN(guid, MF_MT_DV_AAUX_CTRL_PACK_0);
    IF_EQUAL_RETURN(guid, MF_MT_DV_AAUX_SRC_PACK_1);
    IF_EQUAL_RETURN(guid, MF_MT_DV_AAUX_CTRL_PACK_1);
    IF_EQUAL_RETURN(guid, MF_MT_DV_VAUX_SRC_PACK);
    IF_EQUAL_RETURN(guid, MF_MT_DV_VAUX_CTRL_PACK);
    IF_EQUAL_RETURN(guid, MF_MT_ARBITRARY_HEADER);
    IF_EQUAL_RETURN(guid, MF_MT_ARBITRARY_FORMAT);
    IF_EQUAL_RETURN(guid, MF_MT_IMAGE_LOSS_TOLERANT); 
    IF_EQUAL_RETURN(guid, MF_MT_MPEG4_SAMPLE_DESCRIPTION);
    IF_EQUAL_RETURN(guid, MF_MT_MPEG4_CURRENT_SAMPLE_ENTRY);
    IF_EQUAL_RETURN(guid, MF_MT_ORIGINAL_4CC); 
    IF_EQUAL_RETURN(guid, MF_MT_ORIGINAL_WAVE_FORMAT_TAG);
    
    // Media types

    IF_EQUAL_RETURN(guid, MFMediaType_Audio);
    IF_EQUAL_RETURN(guid, MFMediaType_Video);
    IF_EQUAL_RETURN(guid, MFMediaType_Protected);
    IF_EQUAL_RETURN(guid, MFMediaType_SAMI);
    IF_EQUAL_RETURN(guid, MFMediaType_Script);
    IF_EQUAL_RETURN(guid, MFMediaType_Image);
    IF_EQUAL_RETURN(guid, MFMediaType_HTML);
    IF_EQUAL_RETURN(guid, MFMediaType_Binary);
    IF_EQUAL_RETURN(guid, MFMediaType_FileTransfer);

    IF_EQUAL_RETURN(guid, MFVideoFormat_AI44); //     FCC('AI44')
    IF_EQUAL_RETURN(guid, MFVideoFormat_ARGB32); //   D3DFMT_A8R8G8B8 
    IF_EQUAL_RETURN(guid, MFVideoFormat_AYUV); //     FCC('AYUV')
    IF_EQUAL_RETURN(guid, MFVideoFormat_DV25); //     FCC('dv25')
    IF_EQUAL_RETURN(guid, MFVideoFormat_DV50); //     FCC('dv50')
    IF_EQUAL_RETURN(guid, MFVideoFormat_DVH1); //     FCC('dvh1')
    IF_EQUAL_RETURN(guid, MFVideoFormat_DVSD); //     FCC('dvsd')
    IF_EQUAL_RETURN(guid, MFVideoFormat_DVSL); //     FCC('dvsl')
    IF_EQUAL_RETURN(guid, MFVideoFormat_H264); //     FCC('H264')
    IF_EQUAL_RETURN(guid, MFVideoFormat_I420); //     FCC('I420')
    IF_EQUAL_RETURN(guid, MFVideoFormat_IYUV); //     FCC('IYUV')
    IF_EQUAL_RETURN(guid, MFVideoFormat_M4S2); //     FCC('M4S2')
    IF_EQUAL_RETURN(guid, MFVideoFormat_MJPG);
    IF_EQUAL_RETURN(guid, MFVideoFormat_MP43); //     FCC('MP43')
    IF_EQUAL_RETURN(guid, MFVideoFormat_MP4S); //     FCC('MP4S')
    IF_EQUAL_RETURN(guid, MFVideoFormat_MP4V); //     FCC('MP4V')
    IF_EQUAL_RETURN(guid, MFVideoFormat_MPG1); //     FCC('MPG1')
    IF_EQUAL_RETURN(guid, MFVideoFormat_MSS1); //     FCC('MSS1')
    IF_EQUAL_RETURN(guid, MFVideoFormat_MSS2); //     FCC('MSS2')
    IF_EQUAL_RETURN(guid, MFVideoFormat_NV11); //     FCC('NV11')
    IF_EQUAL_RETURN(guid, MFVideoFormat_NV12); //     FCC('NV12')
    IF_EQUAL_RETURN(guid, MFVideoFormat_P010); //     FCC('P010')
    IF_EQUAL_RETURN(guid, MFVideoFormat_P016); //     FCC('P016')
    IF_EQUAL_RETURN(guid, MFVideoFormat_P210); //     FCC('P210')
    IF_EQUAL_RETURN(guid, MFVideoFormat_P216); //     FCC('P216')
    IF_EQUAL_RETURN(guid, MFVideoFormat_RGB24); //    D3DFMT_R8G8B8 
    IF_EQUAL_RETURN(guid, MFVideoFormat_RGB32); //    D3DFMT_X8R8G8B8 
    IF_EQUAL_RETURN(guid, MFVideoFormat_RGB555); //   D3DFMT_X1R5G5B5 
    IF_EQUAL_RETURN(guid, MFVideoFormat_RGB565); //   D3DFMT_R5G6B5 
    IF_EQUAL_RETURN(guid, MFVideoFormat_RGB8);
    IF_EQUAL_RETURN(guid, MFVideoFormat_UYVY); //     FCC('UYVY')
    IF_EQUAL_RETURN(guid, MFVideoFormat_v210); //     FCC('v210')
    IF_EQUAL_RETURN(guid, MFVideoFormat_v410); //     FCC('v410')
    IF_EQUAL_RETURN(guid, MFVideoFormat_WMV1); //     FCC('WMV1')
    IF_EQUAL_RETURN(guid, MFVideoFormat_WMV2); //     FCC('WMV2')
    IF_EQUAL_RETURN(guid, MFVideoFormat_WMV3); //     FCC('WMV3')
    IF_EQUAL_RETURN(guid, MFVideoFormat_WVC1); //     FCC('WVC1')
    IF_EQUAL_RETURN(guid, MFVideoFormat_Y210); //     FCC('Y210')
    IF_EQUAL_RETURN(guid, MFVideoFormat_Y216); //     FCC('Y216')
    IF_EQUAL_RETURN(guid, MFVideoFormat_Y410); //     FCC('Y410')
    IF_EQUAL_RETURN(guid, MFVideoFormat_Y416); //     FCC('Y416')
    IF_EQUAL_RETURN(guid, MFVideoFormat_Y41P);
    IF_EQUAL_RETURN(guid, MFVideoFormat_Y41T);
    IF_EQUAL_RETURN(guid, MFVideoFormat_YUY2); //     FCC('YUY2')
    IF_EQUAL_RETURN(guid, MFVideoFormat_YV12); //     FCC('YV12')
    IF_EQUAL_RETURN(guid, MFVideoFormat_YVYU);

    IF_EQUAL_RETURN(guid, MFAudioFormat_PCM); //              WAVE_FORMAT_PCM 
    IF_EQUAL_RETURN(guid, MFAudioFormat_Float); //            WAVE_FORMAT_IEEE_FLOAT 
    IF_EQUAL_RETURN(guid, MFAudioFormat_DTS); //              WAVE_FORMAT_DTS 
    IF_EQUAL_RETURN(guid, MFAudioFormat_Dolby_AC3_SPDIF); //  WAVE_FORMAT_DOLBY_AC3_SPDIF 
    IF_EQUAL_RETURN(guid, MFAudioFormat_DRM); //              WAVE_FORMAT_DRM 
    IF_EQUAL_RETURN(guid, MFAudioFormat_WMAudioV8); //        WAVE_FORMAT_WMAUDIO2 
    IF_EQUAL_RETURN(guid, MFAudioFormat_WMAudioV9); //        WAVE_FORMAT_WMAUDIO3 
    IF_EQUAL_RETURN(guid, MFAudioFormat_WMAudio_Lossless); // WAVE_FORMAT_WMAUDIO_LOSSLESS 
    IF_EQUAL_RETURN(guid, MFAudioFormat_WMASPDIF); //         WAVE_FORMAT_WMASPDIF 
    IF_EQUAL_RETURN(guid, MFAudioFormat_MSP1); //             WAVE_FORMAT_WMAVOICE9 
    IF_EQUAL_RETURN(guid, MFAudioFormat_MP3); //              WAVE_FORMAT_MPEGLAYER3 
    IF_EQUAL_RETURN(guid, MFAudioFormat_MPEG); //             WAVE_FORMAT_MPEG 
    IF_EQUAL_RETURN(guid, MFAudioFormat_AAC); //              WAVE_FORMAT_MPEG_HEAAC 
    IF_EQUAL_RETURN(guid, MFAudioFormat_ADTS); //             WAVE_FORMAT_MPEG_ADTS_AAC 

    return NULL;
}

音視訊系列之iOS: 音訊採集 AudioUnit
2017-12-14
iOS音訊
Android 音視訊採集那些事
2021-10-27
Android
音視訊入門之音訊採集、編碼、播放
2018-12-28
音訊
iOS採集錄製音視訊API選擇推薦
2019-05-11
iOSAPI
轉載:iOS音視訊實時採集硬體編碼
2018-06-04
iOS
Audio Unit採集音訊實戰
2019-05-11
音訊
Android音視訊(一) Camera2 API採集資料
2019-03-04
AndroidAPI
iOS視訊流採集概述(AVCaptureSession)
2019-04-13
iOSAPTSession
iOS視訊採集實戰(AVCaptureSession)
2019-04-13
iOSAPTSession
iOS - 直播系列一：視訊採集
2017-10-16
iOS
Linux下的音訊採集與回放(轉)
2007-08-11
Linux音訊
ffmpeg命令錄製windows音視訊
2018-03-20
Windows
利用爬蟲採集音訊資訊完整程式碼示例
2023-10-19
爬蟲音訊
JavaCV FFmpeg採集麥克風PCM音訊資料
2021-01-03
Java音訊
視訊直播技術詳解之採集
2016-08-30
（三）WebRTC手記之本地視訊採集
2015-09-20
Web
Audio Queue 採集音訊實戰(支援不同格式)
2019-05-11
音訊
iOS 實時音訊採集與播放Audio Unit使用
2018-08-29
iOS音訊
短視訊“音訊化”，音樂“視訊化”
2021-11-29
音訊
自定義相機採集及視訊編輯（1）-短視訊錄製
2018-02-26
音視訊--音訊入門
2018-12-18
音訊
音視訊–音訊入門
2018-12-18
音訊
ASP.NET視訊採集站核心技術分析
2009-07-31
ASP.NET
亞馬遜的主圖視訊/描述視訊有辦法同採集下來嗎？
2021-12-14
亞馬遜
音視訊--視訊入門
2018-12-25
Android音視訊之MediaPlayer音視訊播放
2016-06-15
Android
天貓商品採集軟體，怎麼一鍵批量採集主圖、評論圖以及視訊
2021-12-16
視訊採集：iOS平臺基於AVCaptureDevice的實現
2018-12-24
iOSAPTdev
亞馬遜裡的商品視訊是怎麼批量採集的
2020-10-03
亞馬遜
Android音視訊之MediaRecorder音視訊錄製
2016-06-14
Android
視音訊播放
2018-02-28
音訊
android 音訊採集、FLTP重取樣與AAC編碼推流
2017-11-27
Android音訊
Windows Presentation Foundation 安全沙箱
2005-12-16
Windows
Windows Presentation Foundation資源
2008-01-22
Windows
Windows Presentation Foundation慨述
2008-01-22
Windows
Android 音視訊 - MediaCodec 編解碼音視訊
2021-11-03
Android
能夠採集小紅書圖片、視訊的軟體，一鍵自動批量採集到電腦上
2021-11-26
IOS音視訊（二）AVFoundation視訊捕捉
2020-04-05
iOS

Windows Media Foundation 音視訊採集

寫在前面

音視訊採集

列舉裝置並設定裝置

通過 Source Reader 讀取媒體

其他功能

程式碼地址

Enumerating Video Capture Devices

Related topics

Audio/Video Capture in Media Foundation

Enumerate Capture Devices

Use a capture device

Shut down the capture device

Related topics

Handling Video Device Loss

Register For Device Notification

Get the Symbolic Link of the Device

Handle WM_DEVICECHANGE

Unregister For Notification

Related topics

Media Type Debugging Code

相關文章