[Scummvm-devel] blendPixelPtr()

[Marcus Comstedt]

"Marcus Comstedt" <marcus at mc.pp.se> writes:

> Although I must confess I'm slightly puzzled by this result; even
> thought the blend function was wrong, I would have expected the
> resulting overlay pixels to have alpha=1.0, in which case there should
> be no difference between the two blend functions (becase SRC_ALPHA ==
> ONE).  Isn't the background gradient fully opaque?

Ah, it seems there is some bug in the code for drawing the _non_ round
parts of the box, which decreases the alpha.  Maybe that is where the
non-opaque pixels came from?

http://mc.pp.se/Screenshot_2014-01-10-17-53-29.jpg

(I'm pretty sure Sam's nose is not supposed to be visible through the
 bottom of the Load button... :)


  // Marcus