If I understand your question right use this:
#define RGB16(r,g,b) ((r << 11) + (g << 5) + b)

I would recommend pink. it''s:
RGB(255,0,255) or RGB16(31,0,31)

Remember that RGB16''s max values are 31,63,31

/MindWipe


"If it doesn''t fit, force it; if it breaks, it needed replacement anyway."

Both of the above poster''s suggestions will break if run on a videocard with the "wrong" pixel format. If you use a 16-bit mode, your code must adapt to the current format.

I forget the specific function name, and it might have changed slightly in DX8, but in DX7 the IDirectDrawSurface interface had a function which would return a DDPIXELFORMAT structure which held info about the surface''s format. This included mask values which could be used in color conversion code.

...
shuffle shuffle
...

I just searched through my source code, and found the function that stores the format data in my game. The lo___bit and num___bits variables are globals which I''ve defined here so you know what they are.

	WORD loREDbit, numREDbits;	WORD loGREENbit, numGREENbits;	WORD loBLUEbit, numBLUEbits; 

In addition, the code uses these helper functions.

  WORD HighBitPos( DWORD dword ){	DWORD test=1;	test<<=31;	for (WORD i=0;i<32;i++)	{		if ( dword & test )			return (WORD)(31-i);		test>>=1;	}	return 0;}WORD LowBitPos( DWORD dword ){	DWORD test=1;	for (WORD i=0;i<32;i++)	{		if ( dword & test )			return i;		test<<=1;	}	return 0;}  

Here''s the function:

  BOOL StorePixelFormatData(){	DDPIXELFORMAT format;	ZeroMemory( &format, sizeof(format) );	format.dwSize=sizeof(format);	if (primsurf->GetPixelFormat( &format )!=DD_OK)	{		TRACE("StorePixelFormatData() failed\n");		return FALSE;	}	loREDbit = LowBitPos( format.dwRBitMask );	WORD hiREDbit = HighBitPos( format.dwRBitMask );	numREDbits=(WORD)(hiREDbit-loREDbit+1);	loGREENbit = LowBitPos( format.dwGBitMask );	WORD hiGREENbit = HighBitPos( format.dwGBitMask );	numGREENbits=(WORD)(hiGREENbit-loGREENbit+1);	loBLUEbit  = LowBitPos( format.dwBBitMask );	WORD hiBLUEbit  = HighBitPos( format.dwBBitMask );	numBLUEbits=(WORD)(hiBLUEbit-loBLUEbit+1);	return TRUE;}  

The StorePixelFormatData function is called after creating the flipping surfaces. To see how this data is used, here is my function that copies a bitmap into a 16-bit surface. (The bmpbuf parameter is a pointer to the bitmap''s bits.)

  BOOL Copy_Bmp24_Surface16( LPDIRECTDRAWSURFACE7 surf, 										 BYTE* bmpbuf, int w, int h ){	if (surf==0 || bmpbuf==0)		return FALSE;		DDSURFACEDESC2 desc;	ZeroMemory( &desc, sizeof(desc) );	desc.dwSize = sizeof(desc);	HRESULT r=surf->Lock( 0, &desc, DDLOCK_WAIT | DDLOCK_WRITEONLY, 0 );	if (r!=DD_OK)	{		TRACE("Copy_Bmp24_Surface16: Lock() failed\n");		return FALSE;	}	int bytesrequired=w*3;	int bytesgiven=(bytesrequired+3) & ~3;	BYTE* surfbits = (BYTE*)desc.lpSurface;	BYTE* imagebits = (BYTE*)(&bmpbuf[(h-1)*bytesgiven]);	float REDdiv=(float)256/(float)pow( 2, numREDbits );	float GREENdiv=(float)256/(float)pow( 2, numGREENbits );	float BLUEdiv=(float)256/(float)pow(2, numBLUEbits );	for( int i=0; i<h; i++ )	{		USHORT* pixptr=(unsigned short*)surfbits;		RGBTRIPLE* triple=(RGBTRIPLE*)imagebits;		for (int p=0;p<w;p++)		{			float rf=(float)triple->rgbtRed/REDdiv;			float gf=(float)triple->rgbtGreen/GREENdiv;			float bf=(float)triple->rgbtBlue/BLUEdiv;			WORD r=(WORD)((WORD)rf<<loREDbit);			WORD g=(WORD)((WORD)gf<<loGREENbit);			WORD b=(WORD)((WORD)bf<<loBLUEbit);			*pixptr = (WORD)(r|g|b);			triple++;			pixptr++;		}		surfbits += desc.lPitch;		imagebits -= bytesgiven;	}	surf->Unlock( 0 );	return TRUE;}  

One thing to note about this method is that it will work with any format that could ever possibly exist. Even if run on some non-standard videocard, it would still display colors properly.

Anyway, I hope this helps you.