Hi, I'm trying to optimize some graphics processing for GUI and found that things like copying images and alphablending are very fast. However what I need is alpha composition of 2 images, where the target's alpha is not important (that I think is possible), source's alpha is used, AND also a global alpha is used to multiply the source's alpha. So:
This reminds me - would it be possible to implement something like AlphaBlend in Windows API? I mean, IPP's implementation of simple alphablend is faster, but it would need the global alpha and possibility to resize the output, linear interpolation is just fine. I think this would be very handy for many people and save A LOT of CPU time...
well there's no code, since as I read it's not supposed to be featured.
Like the OP I needed alphablending of a premultiplied source onto a target, in which both the source's own alpha channel & a global alpha are used. This is normally the most basic use for alphablending, when you wanna draw a masked image onto another.
Weirdly everything I find on the net uses the source's alpha channel, or a global alpha, but rarely both (like the GDI's AlphaBlend does).