DirectDraw Tutorial 2

DirectDraw Tutorial 2

Alpha 混合

 

一、              基本alpha混合

 

Alhpa混合有多种算法，基本的计算公式如下：

result = ALPHA * srcPixel + ( 1 - ALPHA ) * destPixel

其中ALHPA的范围从0.0到1.0，result为混合后的颜色，srcPixel为前景色，destPixel为背景色。也就是说较大的ALPHA值意味着更多得源颜色。

上述公式中出现了两个乘法，可以通过合并来优化它：

       result = ALPHA * ( srcPixel - destPixel ) + destPixel

现在ALPHA是一个浮点数，因此可以将它转化为整数：

result = ( ALPHA * ( srcPixel - destPixel ) ) / 256 + destPixel

现在ALPHA 的取值范围为0 到256 ，对于除法可以通过右移来代替：

       result = ( ALPHA * ( srcPixel - destPixel ) ) >>8+ destPixel

根据上面的公式，可以得到ALHPHA 混合的代码：

//DESC:Blt a surface pdds to (x,y) of backbuffer

//This function just can be used to blt a 32bit bitmap,if you want to blt a 8bit,16bit or 24bit

//you should rewrite this function, and int this version the parameter RECT* prc is not be used

//you should pass a NULL as value, alpha from 0 to 255, pdds is the source surface

HRESULT CDisplay::AlphaBlt( DWORD x, DWORD y, LPDIRECTDRAWSURFACE7 pdds,

                           RECT* prc, USHORT alpha )

{

    HRESULT hr;

 

    //Get the desc

    DDSURFACEDESC2 ddsdSrc;

    DDSURFACEDESC2 ddsdDest;

    ZeroMemory( &ddsdSrc, sizeof( ddsdSrc ) );

    ddsdSrc.dwSize = sizeof(ddsdSrc);

    ZeroMemory( &ddsdDest, sizeof( ddsdDest ) );

    ddsdDest.dwSize = sizeof(ddsdDest);

 

    //Get the mask info

    DDPIXELFORMAT ddpf;

    ZeroMemory( &ddpf, sizeof(ddpf) );

    ddpf.dwSize = sizeof(ddpf);

    pdds->GetPixelFormat(&ddpf);

    DWORD rMask = ddpf.dwRBitMask;

    DWORD bMask = ddpf.dwBBitMask;

    DWORD gMask = ddpf.dwGBitMask;

 

    if( FAILED( hr = pdds->Lock( NULL, &ddsdSrc, DDLOCK_WAIT, NULL )))

        return hr;

 

    //the target area

    RECT rect = {x, y, x+ddsdSrc.dwWidth, y+ddsdSrc.dwHeight} ;

 

    if( FAILED( hr = m_pddsBackBuffer->Lock( &rect, &ddsdDest, DDLOCK_WAIT, NULL )))

        return hr;

   

    BYTE* srcPt = (BYTE*)ddsdSrc.lpSurface;

    DWORD srcPitch = ddsdSrc.lPitch;

 

    BYTE* destPt = (BYTE*)ddsdDest.lpSurface;

    DWORD destPitch = ddsdDest.lPitch;

   

    //the width and height of area need to process

    int iWidth = ddsdSrc.dwWidth;

    int iHeight = ddsdSrc.dwHeight;

 

    DWORD dwDestPad = ddsdDest.lPitch - ( iWidth * 4 );

    DWORD dwSrcPad    = ddsdSrc.lPitch - ( iWidth * 4 );

 

    for( int i = 0; i < ddsdSrc.dwHeight; i++ )

    {

        for( int j = 0; j < ddsdSrc.dwWidth; j++ )

        {

           DWORD dwSourceTemp = *((DWORD*)srcPt);

 

           //if the source pixel is not black

           if(( dwSourceTemp & 0xffffff ) != 0 )

           {

              DWORD dwTargetTemp = *((DWORD*)destPt);

              // Extract the red channels.

              DWORD dwTgtRed = dwTargetTemp & rMask;

              DWORD dwSrcRed = dwSourceTemp & rMask;

 

              // Extract the green channels.

              DWORD dwTgtGreen = dwTargetTemp & gMask;

              DWORD dwSrcGreen = dwSourceTemp & gMask;

 

              // Extract the blue channel.

              DWORD dwTgtBlue = dwTargetTemp & bMask;

              DWORD dwSrcBlue = dwSourceTemp & bMask;

 

              // Calculate the destination pixel.

              dwTargetTemp =

                  ( ( ( ( alpha * ( dwSrcRed - dwTgtRed ) >> 8 ) + dwTgtRed ) & rMask ) |

                  ( ( ( alpha * ( dwSrcGreen - dwTgtGreen ) >> 8 ) + dwTgtGreen ) & gMask ) |

                  ( ( alpha * ( dwSrcBlue - dwTgtBlue ) >> 8 ) + dwTgtBlue ) );

              *((DWORD*)destPt) = dwTargetTemp;

           }

           //move to next pixel

           srcPt += 4;

           destPt += 4;

 

        }

       //move to next line

        srcPt += dwSrcPad;

        destPt += dwDestPad;

    }

 

    pdds->Unlock(NULL);

    m_pddsBackBuffer->Unlock(&rect);

 

    return hr;

}

 

二、              代码优化

可以使用MMX、SSE来优化上述代码，下面给出一些网上关于alpha混合的主题

http://www.gamedev.net/reference/articles/article1487.asp

http://www.gamedev.net/reference/programming/features/mmxblend/default.asp

http://francis.dupont.free.fr/coindev/english/dd.htm

http://dev.gameres.com/Program/Visual/2D/curtain.htm

http://softwarecommunity.intel.com/articles/eng/1713.htm