[poppler] [RFC] Add option to keep alpha channel to Qt frontends

[Albert Astals Cid]

El Divendres, 3 de juliol de 2015, a les 00:43:40, Adam Reichold va escriure:
> Hello again,
> 
> Am 02.07.2015 um 23:01 schrieb Albert Astals Cid:
> > Another thing i wanted to try for a while but didn't get to was killing
> > the
> > Splash XRGB8 mode (which is an extension i created) and using the plain
> > Splash RGB8 mode since now QImage is smart enough to know how to use
> > that, would you mind testing if that yields any benefit?
> 
> It does improve rendering that includes the background composition by
> more than two percent with structurally simple documents (i.e. some
> mathematical texts which do not contain any images themselves).
> 
> Not doing the alpha blending (but exporting the alpha channel) is still
> faster, i.e. also slightly above two percent in the same benchmarks.
> (Meaning XBGR8 with paper colour is 2.2 % slower than RGB8 with paper
> colour which itself is 2.2 % slower than XBGR8 without paper colour.)

What about BGR8 without paper colour?

Cheers,
  Albert

> 
> An extended patch is attached. (I am not totally sure about the
> endianess conversion there.)
> 
> Also note that while conversion from QImage to QPixmap did show no
> difference between Qt's handlign of ARGB32 and RGB888, painting the
> resulting QPixmap is 40 % faster for ARGB32, i.e. XBGR8 in Splash
> parlance. But since painting those images is usually very fast, this
> will very probably not offset the 2.2 % gain for rendering.
> 
> Best regards, Adam.