Announcement

Collapse
No announcement yet.

Wayland's Weston Gets Output Scaling Support

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #11
    Originally posted by stqn View Post
    I?ve never understood the point of Wayland, but if it lets us do that, then YES it?s suddenly interesting .
    its not X. For a better answer... literally go to my profile and search for all my Wayland threads. I typically have to explain it at least once per thread
    All opinions are my own not those of my employer if you know who they are.

    Comment


    • #12
      Does the scaling allows for scaling up and down?
      And with what precision, limits?

      Comment


      • #13
        Originally posted by plonoma View Post
        Does the scaling allows for scaling up and down?
        And with what precision, limits?
        This, and also: what scaling algorithms are available?

        Comment


        • #14
          Originally posted by dee. View Post
          This, and also: what scaling algorithms are available?
          It looks like it's just using Cairo to do all the scaling (cairo_scale), although i didn't spot what algorithm it had set.

          Comment


          • #15
            Originally posted by smitty3268 View Post
            It looks like it's just using Cairo to do all the scaling (cairo_scale), although i didn't spot what algorithm it had set.
            No, that's only for the clients, when they want to render at the full native resolution to _avoid_ scaling. If any scaling does actually occur, it'll happen in the render, with whatever scaling GLES or Pixman (or, in RPi's case, DispManX) applies.

            Comment


            • #16
              Originally posted by dee. View Post
              This, and also: what scaling algorithms are available?
              It's only meant up-scaling. While down-scaling might allow you to fit a fixed-size application window on a small screen, it would look ugly and probably unreadable.

              The scaling algorithm is not specified. If we did specify a certain algorithm, we would immediate exclude half of the hardware support. Specify a simple algorithm: most hardware will probably support it, but it will not allow capable hardware to make any better even if they could. Specify a fancy algorithm: lots of hardware simply may not be able to do it. This is also Wayland core protocol, so it is not only about PCs or desktops. Therefore the exact algorithm is left as an implementation detail.

              Comment


              • #17
                @pq__

                it would look ugly and probably unreadable.
                Not necessarily.
                Scaling video to fit on a smaller screen could look good while doing high-quality scaling on the client device.
                Not all applications are about reading.

                Things like miniature previews of windows can also use the scale down support, the content does not have to be readable.


                There might be other use cases where scaling down is useful and readability is not important.

                I hope there is a way for wayland devices to poll and negociate rescaling algorithms available?
                This would allow for the most flexible way to adapt to hardware while avoiding support and compatibility problems.

                Comment


                • #18
                  Originally posted by plonoma View Post
                  @pq__

                  Not necessarily.
                  Scaling video to fit on a smaller screen could look good while doing high-quality scaling on the client device.
                  Not all applications are about reading.

                  Things like miniature previews of windows can also use the scale down support, the content does not have to be readable.
                  [IMG---/IMG]

                  There might be other use cases where scaling down is useful and readability is not important.

                  I hope there is a way for wayland devices to poll and negociate rescaling algorithms available?
                  This would allow for the most flexible way to adapt to hardware while avoiding support and compatibility problems.
                  This is especially for high density monitors, not for per-window scaling, fullscreening, or anything else.

                  That means that everything on that output is *always* scaled by the specified factor, either by the compositor by default, or the clients rendering at the proper resolution as opt-in. There is nothing to negotiate of algorithms, a compositor uses what it has available.

                  There is a separate protocol extension being sketched for client-controlled per-surface scaling, especially for hardware scaling of videos but usable for anything.

                  Comment


                  • #19
                    @pg__

                    You gotta implement these things for being able to efficient do them and allow for hardware acceleration.
                    Being able to specify scaling in a unified way makes it easier to implement hardware acceleration.

                    Comment


                    • #20
                      That's very nice. The current, X way of scaling things up (changing DPI) doesn't work well, usually only changes text size. Having an ability to upscale everything would be really nice, even on today's monitors, given that full HD already gives a lot more space than many programs in the past considered possible to achieve.

                      Comment

                      Working...
                      X