That's reassuring. Thanks.
Since I found this thread, instead of finding github, I'll stick this here for any that want it:
void Image::convertToPremultipliedAlpha()
{
if (!m_pixels.empty())
{
Uint8* source = &m_pixels[0];
for (unsigned int y = 0; y < m_size.y; ++y)
{
for (unsigned int x = 0; x < m_size.x; ++x)
{
Uint32 alpha = *(source + 3);
*(source ) = Uint8((Uint32(*(source )) * alpha) / 255);
*(source + 1) = Uint8((Uint32(*(source + 1)) * alpha) / 255);
*(source + 2) = Uint8((Uint32(*(source + 2)) * alpha) / 255);
source += 4;
}
}
}
}
optimizations exist for this I'm sure, but this works, and I don't have a profiler to hand.