Get contents of render target into CPU readable memory

Question

I'm struggling to get the contents of my framebuffer into memory for use by the CPU. Has anyone else managed to do this?

Here is my code so far - every call succeeds etc. but the final result is a black + zero alpha texture, which makes me believe I am doing something very wrong.

 ID3D11Texture2D* pSurface;
 HRESULT hr = m_swapChain->GetBuffer( 0, __uuidof( ID3D11Texture2D ), reinterpret_cast< void** >( &pSurface ) );
 if( pSurface )
 {
  const int width = static_cast<int>(m_window->Bounds.Width * m_dpi / 96.0f);
  const int height = static_cast<int>(m_window->Bounds.Height * m_dpi / 96.0f);
  unsigned int size = width * height;
  if( m_captureData )
  {
   freeFramebufferData( m_captureData );
  }
  m_captureData = new unsigned char[ width * height * 4 ];

  ID3D11Texture2D* pNewTexture = NULL;

  D3D11_TEXTURE2D_DESC description =
  {
   width, height, 1, 1, DXGI_FORMAT_R8G8B8A8_UNORM,
   { 1, 0 }, // DXGI_SAMPLE_DESC
   D3D11_USAGE_STAGING,
   0, D3D11_CPU_ACCESS_READ | D3D11_CPU_ACCESS_WRITE, 0
  };

  HRESULT hr = m_d3dDevice->CreateTexture2D( &description, NULL, &pNewTexture );
  if( pNewTexture )
  {
   m_d3dContext->CopyResource( pNewTexture, pSurface );
   D3D11_MAPPED_SUBRESOURCE resource;
   unsigned int subresource = D3D11CalcSubresource( 0, 0, 0 );
   HRESULT hr = m_d3dContext->Map( pNewTexture, subresource, D3D11_MAP_READ_WRITE, 0, &resource );
   //resource.pData; // TEXTURE DATA IS HERE

   const int pitch = width << 2;
   const unsigned char* source = static_cast< const unsigned char* >( resource.pData );
   unsigned char* dest = m_captureData;
   for( int i = 0; i < height; ++i )
   {
    memcpy( dest, source, width * 4 );
    source += pitch;
    dest += pitch;
   }

   m_captureSize = size;
   m_captureWidth = width;
   m_captureHeight = height;

   return;
  }
      
  freeFramebufferData( m_captureData );
 }

Answer 1

Fixed with the following:

ID3D11Texture2D* pSurface;

HRESULThr = m_swapChain->GetBuffer( 0, __uuidof( ID3D11Texture2D), reinterpret_cast< void** >( &pSurface ) );

if( pSurface )

{

constintwidth = static_cast<int>(m_window->Bounds.Width * m_dpi / 96.0f);

constintheight = static_cast<int>(m_window->Bounds.Height * m_dpi / 96.0f);

unsignedintsize = width * height;

if( m_captureData )

{

freeFramebufferData( m_captureData );

}

m_captureData =

newunsignedchar[ width * height * 4 ];

ID3D11Texture2D* pNewTexture = NULL;

D3D11_TEXTURE2D_DESCdescription;

pSurface->GetDesc( &description );

description.BindFlags = 0;

description.CPUAccessFlags =

D3D11_CPU_ACCESS_READ| D3D11_CPU_ACCESS_WRITE;

description.Usage =

D3D11_USAGE_STAGING;

HRESULThr = m_d3dDevice->CreateTexture2D( &description, NULL, &pNewTexture );

if( pNewTexture )

{

m_d3dContext->CopyResource( pNewTexture, pSurface );

D3D11_MAPPED_SUBRESOURCEresource;

unsignedintsubresource = D3D11CalcSubresource( 0, 0, 0 );

HRESULThr = m_d3dContext->Map( pNewTexture, subresource, D3D11_MAP_READ_WRITE, 0, &resource );

//resource.pData; // TEXTURE DATA IS HERE

constintpitch = width << 2;

constunsignedchar* source = static_cast< constunsignedchar* >( resource.pData );

unsignedchar* dest = m_captureData;

for( inti = 0; i < height; ++i )

{

memcpy( dest, source, width * 4 );

source += pitch;

dest += pitch;

}

m_captureSize = size;

m_captureWidth = width;

m_captureHeight = height;

return;

}

      

freeFramebufferData( m_captureData );

}

Answer 2

 

Thanks for sharing the solution.

 

Best regards,

Jesse

---

Jesse Jiang [MSFT]
MSDN Community Support | Feedback to us