// color1 and color2 are R4G4B4 12bit RGB color values, alpha is 0-255
uint16_t blend_12bit( uint16_t color1, uint16_t color2, uint8_t alpha ) {
	uint64_t c1 = (uint64_t) color1;
	uint64_t c2 = (uint64_t) color2;
	uint64_t a = (uint64_t)( alpha >> 4 );
	// bit magic to alpha blend R G B with single mul
	c1 = ( c1 | ( c1 << 12 ) ) & 0x0f0f0f;
	c2 = ( c2 | ( c2 << 12 ) ) & 0x0f0f0f;
	uint32_t o = ( ( ( ( c2 - c1 ) * a ) >> 4 ) + c1 ) & 0x0f0f0f;
	return (uint16_t) ( o | ( o >> 12 ) );
}

// color1 and color2 are R5G6B5 16bit RGB color values, alpha is 0-255
uint16_t blend_16bit( uint16_t color1, uint16_t color2, uint8_t alpha )	{
    uint32_t c1 = (uint32_t) color1;
    uint32_t c2 = (uint32_t) color2;
    uint32_t a = (uint32_t)( alpha >> 3 );
    // bit magic to alpha blend R G B with single mul
    c1 = ( c1 | ( c1 << 16 ) ) & 0x07e0f81fu;
    c2 = ( c2 | ( c2 << 16 ) ) & 0x07e0f81fu;
    uint32_t o = ( ( ( ( c2 - c1 ) * a ) >> 5 ) + c1 ) & 0x07e0f81fu; 
    return (uint16_t) ( o | ( o >> 16 ) );
}


// color1 and color2 are R8G8B8 24bit RGB color values, alpha is 0-255
uint32_t blend_24bit( uint32_t color1, uint32_t color2, uint8_t a ) {
	uint64_t c1 = (uint64_t) color1;
	uint64_t c2 = (uint64_t) color2;
	uint64_t a = (uint64_t) a;
	// bit magic to alpha blend R G B with single mul
	c1 = ( c1 | ( c1 << 24 ) ) & 0x00ff00ff00ffull;
	c2 = ( c2 | ( c2 << 24 ) ) & 0x00ff00ff00ffull;
	uint64_t o = ( ( ( ( c2 - c1 ) * a ) >> 8 ) + c1 ) & 0x00ff00ff00ffull;
	return (uint32_t) ( o | ( o >> 24 ) );
}