Incompatible DXGI format and alpha mode when creating Direct2D Image source

Question

The problem

I’m having trouble creating an Direct2D image source from a DXGI surface. I’m creating a Windows 10 UWP app in Visual Studio 2017, with the part that handles graphics written in Direct2D/3D in C++.

Basically, I get an array with luminance values from a YUV image that I for now want to draw as an black/white image. The values come once per frame in a stream, so I want to use a swapchain to efficiently render the data. With help from a Swapchain example i have managed to set up a swapchain that I could draw to when I tested with other data.

The code to create the image source is something like this:

void D2DPanel::DrawBackground(int32 dataPtr, int width, int height)
{
    D3D11_TEXTURE2D_DESC _d3d_texture_desc;
    _d3d_texture_desc.Width = width;
    _d3d_texture_desc.Height = height;
    _d3d_texture_desc.MipLevels = 1;
    _d3d_texture_desc.ArraySize = 1;
    _d3d_texture_desc.Format = DXGI_FORMAT::DXGI_FORMAT_R8_UNORM;    
    _d3d_texture_desc.Usage = D3D11_USAGE::D3D11_USAGE_DYNAMIC;
    _d3d_texture_desc.SampleDesc.Count = 1;
    _d3d_texture_desc.SampleDesc.Quality = 0;
    _d3d_texture_desc.BindFlags = 
    D3D11_BIND_FLAG::D3D11_BIND_SHADER_RESOURCE;
    _d3d_texture_desc.CPUAccessFlags = D3D11_CPU_ACCESS_FLAG::D3D11_CPU_ACCESS_WRITE;
    _d3d_texture_desc.MiscFlags = 0;

    D3D11_SUBRESOURCE_DATA _d3d_texture_data;
    _d3d_texture_data.pSysMem = (void*)IntPtr(dataPtr);
    _d3d_texture_data.SysMemPitch = width;

    ComPtr<ID3D11Texture2D> _d3d_texture;   
    DX::ThrowIfFailed(m_d3dDevice->CreateTexture2D(&_d3d_texture_desc, &_d3d_texture_data, &_d3d_texture)); 
    ComPtr<IDXGISurface> _dxgi_surface;
    DX::ThrowIfFailed(_d3d_texture.As(&_dxgi_surface));    
    IDXGISurface* surfaces[1] = { _dxgi_surface.Get() };                

    ComPtr<ID2D1ImageSource> d2d_image_source;

    //The following method call results in an exception
    //m_d2d_context2 is an ID2D1DeviceContext2
    DX::ThrowIfFailed(m_d2d_context2->CreateImageSourceFromDxgi(surfaces, 1,
        DXGI_COLOR_SPACE_TYPE::DXGI_COLOR_SPACE_RGB_FULL_G22_NONE_P709,
        D2D1_IMAGE_SOURCE_FROM_DXGI_OPTIONS::D2D1_IMAGE_SOURCE_FROM_DXGI_OPTIONS_LOW_QUALITY_PRIMARY_CONVERSION,
        &d2d_image_source));

    //….drawing, etc
}

When I call CreateImageSourceFromDXGI with the parameters above I get an exception with the following error message (error message description):

D2D DEBUG ERROR - The combination of alpha mode and DXGI format supplied
are not compatible with one another.

I can access the alpha mode connected to the ID2D1DeviceContext2 by calling for example auto foo = m_d2d_context2->GetPixelFormat(); I have tried setting the alpha mode for the render target bitmap created from the swapchain buffer to D2D1_ALPHA_MODE_PREMULTIPLIED (which it was in the example code my code is based on) and D2D1_ALPHA_MODE_IGNORE. D2D1_ALPHA_MODE_STRAIGHT resulted in an error when I tried to create the render target bitmap. I can see that the alpha mode really is changed through m_d2d_context2->GetPixelFormat(). Nothing has helped, I still get the same error message. When the swapchain is created alpha is set to DXGI_ALPHA_MODE_UNSPECIFIED and format is DXGI_FORMAT_B8G8R8A8_UNORM.

Before trying to create a black and white image, I successfully managed to create and render an image source from YUV data, where I used the DXGI-format DXGI_FORMAT_NV12 to create the D3D11_TEXTURE2D and the parameters DXGI_COLOR_SPACE_TYPE::DXGI_COLOR_SPACE_YCBCR_FULL_G22_NONE_P709_X601, D2D1_IMAGE_SOURCE_FROM_DXGI_OPTIONS::D2D1_IMAGE_SOURCE_FROM_DXGI_OPTIONS_LOW_QUALITY_PRIMARY_CONVERSION when calling CreateImageSourceFromDxgi.

My questions

• Is the problem that I am using an alpha mode that is incompatible with DXGI_FORMAT_R8_UNORM? If so, which are compatible and which are not?
• Does anyone have any idea what else could prevent me from creating the image source?
• Does anyone know of an example where someone is trying to render a single-channel image in Direct2D from an array of pixel data?

Edit:

I made a workaround and created a bitmap instead of an image source.

//create texture etc, see code from original question...

//access _dxgi_surface connected to texture
ComPtr<IDXGISurface> _dxgi_surface;
DX::ThrowIfFailed(_d3d_texture.As(&_dxgi_surface)); 

//create bitmap from texture with y channel values
D2D1_BITMAP_PROPERTIES1 YPlaneBitmapProp =
    BitmapProperties1(
        D2D1_BITMAP_OPTIONS_NONE,
        PixelFormat(DXGI_FORMAT_R8_UNORM, D2D1_ALPHA_MODE_IGNORE),
        width, height
    );
ComPtr<ID2D1Bitmap1> YPlaneBitmap;
DX::ThrowIfFailed(m_d2d_context2->CreateBitmapFromDxgiSurface(_dxgi_surface.Get(), &YPlaneBitmapProp, YPlaneBitmap.GetAddressOf()));    

//create empty rgb bitmap 
D2D1_BITMAP_PROPERTIES RGBBitmapProp =
    BitmapProperties(PixelFormat(DXGI_FORMAT_B8G8R8A8_UNORM, D2D1_ALPHA_MODE_IGNORE), 96.0f, 96.0f);
ComPtr<ID2D1Bitmap> RGBBitmap;
DX::ThrowIfFailed(m_d2dContext->CreateBitmap(D2D1::SizeU(width, height), RGBBitmapProp, RGBBitmap.GetAddressOf())); 

//Draw Y plane values from YPlaneBitmap on RGBBitmap using a custom effect/pixel shader
m_backgroundEffect->SetInput(0, RGBBitmap.Get());
m_backgroundEffect->SetInput(1, YPlaneBitmap.Get());    

//Scale up image to size of render target
m_scaleEffect->SetInputEffect(0, m_backgroundEffect.Get()); 
m_scaleEffect->SetValue(D2D1_SCALE_PROP_SCALE, D2D1::Vector2F(((float)m_renderTargetWidth / width), (float)m_renderTargetHeight / height));

//Draw image
m_d2dContext->DrawImage(m_scaleEffect.Get());

I still have no idea what the problem was with the code to create an image source.