Transforms a bitmap int RGB 8:8:8 24bpp into an equivalent YUV420p 4:2:0 12bpp

rgb2yuv.cc

void Bitmap2Yuv420p(boost::uint8_t *destination, boost::uint8_t *rgb, 
                    const int &width, const int &height) {
  const std::size_t image_size = width * height;
  boost::uint8_t *y = destination;
  boost::uint8_t *u = destination + image_size;
  boost::uint8_t *v = destination + image_size +  image_size / 4;
  
  boost::uint8_t *r = rgb;
  boost::uint8_t *g = rgb + 1;
  boost::uint8_t *b = rgb + 2;

  for (std::size_t line = 0; line < height; ++line) {
    if (!(line % 2)) {
      for (std::size_t x = 0; x < width; x += 2) {
        *(y++) = ((66 * (*r) + 129 * (*g) + 25 * (*b)) >> 8) + 16;
        *(u++) = ((-38 * (*r) - 74 * (*g) + 112 * (*b)) >> 8) + 128;
        *(v++) = ((112 * (*r) - 94 * (*g) - 18 * (*b)) >> 8) + 128;
        r += 3;
        g += 3;
        b += 3;
        *(y++) = ((66 * (*r) + 129 * (*g) + 25 * (*b)) >> 8) + 16;
        r += 3;
        g += 3;
        b += 3;
      }
    } else {
      for (std::size_t x = 0; x < width; ++x) {
        *(y++) = ((66 * (*r) + 129 * (*g) + 25 * (*b)) >> 8) + 16;
        r += 3;
        g += 3;
        b += 3;
      }
    }
  }
}

So it seems, multiplication is faster than lookup tables.