4 replies to this topic

[Enerjak]

Ok, for SOME REASON, my creation of the texture isn't working as it says i specified a wrong format for it;

here is the create resources function

bool VideoDemo::createResources()
{
HRESULT hr;
hr = m_Direct3D->CheckDeviceFormat(0,D3DDEVTYPE_HAL,m_DispMode.Format,D3DUSAGE_DYNAMIC,D3DRTYPE_TEXTURE,D3DFMT_X8B8G8R8);
if(FAILED(hr))
{
  const char* err = DXGetErrorDescriptionA(hr);
  MessageBoxA(NULL,(LPCSTR)err,NULL,MB_OK);
}
hr = m_Direct3DDevice->CreateTexture(512,512,1,D3DUSAGE_DYNAMIC,D3DFMT_A8B8G8R8,D3DPOOL_DEFAULT,&m_pTexture,NULL);
if(FAILED(hr))
{
  const char* err = DXGetErrorDescriptionA(hr);
  MessageBoxA(NULL,(LPCSTR)err,NULL,MB_OK);
}
hr = m_pTexture->LockRect(0,&this->m_Rect,NULL,D3DLOCK_DISCARD);
if(FAILED(hr))
{
  const char* err = DXGetErrorDescriptionA(hr);
  MessageBoxA(NULL,(LPCSTR)err,"",MB_OK);
}
for(int x = 0; x < this->m_pTexture->Width; x++)
{
  for(int y = 0; y < this->m_pTexture->Height; y++)
  {
   DWORD* imageData = reinterpret_cast<DWORD*>(&m_Rect.pBits);
   int index = x * m_Rect.Pitch / 4 * y;
   imageData[index++] = 0xFF000000;
   imageData[index++ / 2] = 0xFF00FF00;
  }
}
m_pTexture->UnlockRect(0);
hr = D3DXSaveTextureToFileA("Image.png",D3DXIFF_PNG,m_pTexture,NULL);
if(FAILED(hr))
{
  const char* err = DXGetErrorDescriptionA(hr);
  MessageBoxA(NULL,(LPCSTR)err,"",MB_OK);
}
return true;
}

not really sure WHY it does this.

[kubera]

Maybe you would consider calling:

• D3DXFillTextureTX
• D3DXFillTexture

[mhagain]

Try D3DFMT_A8R8B8B8 instead of D3DFMT_A8B8G8R8 (i.e. ARGB instead of ABGR).

[Enerjak]

well, that worked (switching from A8B and all) i didn't even notice i had that :-D isn't that A8B8G8R8 for openGL? just wondering

[mhagain]

Not really. Internally the GPU will store textures in the same format, irrespecive of what you specify in the API (for this purposes of this discussion it's assumed that "format" refers to 32-bit 4-component textures, of which the fourth component may be alpha or unused). So far as this internal GPU storage is concerned there is no difference between D3D or OpenGL - it's just a chunk of data that's accessed by the hardware. Where API differences do come in is with respect to how you specify the texture format CPU-side, which will affect transfers of texture data from system RAM to video RAM. OpenGL allows for fairly flexible specification of how data is stored in system RAM, but the tradeoff is that transfers to GPU RAM may need to go through an additional software stage to convert this to the GPU-side format and layout. With D3D the system RAM format and layout must match the GPU-side format and layout. This of course will only be relevant for initial specification of all textures and for subsequent updates of dynamic textures. So far as all other usages are concerned it's not something that you need to be concerned about.