On Thu, 22 Nov 2001, Kevin Brosius wrote:
> Michel D�nzer wrote:
> >
> >
> > On Thu, 2001-11-22 at 13:48, Peter Surda wrote:
> > > On Thu, Nov 22, 2001 at 11:50:19AM +0100, Michel D�nzer wrote:
> > > > > Or am I missing something here (as usually <g>)?
> > > > Yes, only one image can be displayed at a time per port so why bother.
> > > How "displayed"? If the image is already on the visible part of the screen,
> > > does it magically disappear when you tell the card to do
> > > scaling/conversion/drawing into another window? If it did, that would be "a
> > > really stupid thing" (TM). Are you sure this "magical disappearance" is caused
> > > by card's hardware and not some X obscurity?
> >
> > Yes, that's hiw an overlay works. May be different for Xv adaptors using
> > the texture engine or similar, but I don't think that changes the
> > semantics of an Xv port.
> >
>
> Xv currently supports one port which is tied to the hardware as you
> mention. At the moment, one window is programmed for the overlay, and
> the implementation only supports one window on hardware which only has
> one set of overlay registers. Thinking about it a little, I suspect it
> may be possible to do multiple windows, since overlay usually uses color
> keying to determine which portions overlay. You'd need to merge the
> window regions, plus handle overlap of the display data correctly prior
> to copying, and then only color key for the displayable regions. Seems
> doable on the S3 Virge series, for example, although we'd have memory
> problems if the regions are to big. There may be other reasons why it's
> undesirable...
This isn't a workable scenario. Scale factors may be different
for each window.
Mark.
_______________________________________________
Xpert mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xpert
