static void premultiplyAlpha(wxBitmap& bitmap)
{
const int width = bitmap.GetWidth();
const int height = bitmap.GetHeight();
wxAlphaPixelData pixData(bitmap, wxPoint(0,0), wxSize(width, height));
wxAlphaPixelData::Iterator p(pixData);
for (int y=0; y<height; y++) {
wxAlphaPixelData::Iterator rowStart = p;
for (int x=0; x<width; x++) {
if (p.Red() || p.Green() || p.Blue() || p.Alpha()) {
byte a = p.Alpha();
p.Red() = PREMULTIPLY(p.Red(), a);
p.Green() = PREMULTIPLY(p.Green(), a);
p.Blue() = PREMULTIPLY(p.Blue(), a);
}
++p;
}
p = rowStart;
p.OffsetY(pixData, 1);
}
}
static void ApplyAlphaMultiply(uint8_t* rgba, int alpha_first,
int w, int h, int stride) {
const __m128i zero = _mm_setzero_si128();
const int kSpan = 2;
const int w2 = w & ~(kSpan - 1);
while (h-- > 0) {
uint32_t* const rgbx = (uint32_t*)rgba;
int i;
if (!alpha_first) {
const __m128i kMask = _mm_set_epi16(0xff, 0, 0, 0, 0xff, 0, 0, 0);
const __m128i kMult =
_mm_set_epi16(0, 0x8081, 0x8081, 0x8081, 0, 0x8081, 0x8081, 0x8081);
for (i = 0; i < w2; i += kSpan) {
APPLY_ALPHA(rgbx[i], _MM_SHUFFLE(0, 3, 3, 3), kMask, kMult);
}
} else {
const __m128i kMask = _mm_set_epi16(0, 0, 0, 0xff, 0, 0, 0, 0xff);
const __m128i kMult =
_mm_set_epi16(0x8081, 0x8081, 0x8081, 0, 0x8081, 0x8081, 0x8081, 0);
for (i = 0; i < w2; i += kSpan) {
APPLY_ALPHA(rgbx[i], _MM_SHUFFLE(0, 0, 0, 3), kMask, kMult);
}
}
// Finish with left-overs.
for (; i < w; ++i) {
uint8_t* const rgb = rgba + (alpha_first ? 1 : 0);
const uint8_t* const alpha = rgba + (alpha_first ? 0 : 3);
const uint32_t a = alpha[4 * i];
if (a != 0xff) {
const uint32_t mult = MULTIPLIER(a);
rgb[4 * i + 0] = PREMULTIPLY(rgb[4 * i + 0], mult);
rgb[4 * i + 1] = PREMULTIPLY(rgb[4 * i + 1], mult);
rgb[4 * i + 2] = PREMULTIPLY(rgb[4 * i + 2], mult);
}
}
rgba += stride;
}
}
