Ulrich Sibiller wrote on Wed, 29 Jan 2020 10:28:38 +0100:
That's too bad, because performance is excellent even on weak systems and slow lines.
Yes, normal performance is very good. But when it comes to 3d graphics or video it is not.
I actually meant that even video performance is excellent.
I've tried an other video output driver, that plays video on the client, but is extremely limited compared to XVideo.
I am not really sure what you mean by that.
I told the player to use XImage/Shm instead of XVideo. Videos play fine and smoothly, but the XImage/Shm output driver is somewhat limited (like when changing from windowed playback to fullscreen).
Playing video on the client means playing it locally using the capabilities of the local graphics hardware.
The client hardware should be fine.
As far as I understand, the VcXsrv's display just don't answers X11's Xv requests. It reports XvQueryExtension() okay, but XvQueryAdaptors() returns 0 adapters, which means that either the call is "blocked" or VcXsrv has no (full?) XVideo support.
We _could_ pass on X-Video requests to the X-Video extension on the client side. But I doubt that would bring significant improvements because the data would still have to travel the wire back and forth.
It's not about performance, but presumably about capabilities. VcXsrv with (hopefully) build-in XVideo support should do the rest.
Since I always compile everything for the X2Go server by myself, I'd be happy to test if XVideo passing could easily be added and if you could provide either XVideo passing sources or a patch to me.
There's a technology called "Telekinesis" you might want to try.
I'll have a look at it, but would prefer X2Go.
Ingo