There are ways to achieve server-side scaling with the video encoders,
but that's meant to be used for saving bandwidth on video content and is
mostly done automagically.
This is not what is happening. I want to capture in 4K from the
virtual X fb server in the server while rendering in FullHD (1080) in
the client's machine for interaction purpose.
With 0.16, you can do this using "desktop scaling":
xpra attach --desktop-scaling=0.5
BUT this is the opposite of what this feature was designed for, and so
this will waste a lot of bandwidth and CPU.
IF your server is going to be sending mostly video content, which should
get encoded using one of the video encoders, you may want to start the
server with:
SCALING_HARDCODED=1/2 xpra start ...
This should ensure that we downscale the pixel data by half in each
dimension before the encoding step.
The xpra_Xdummy command above displays following in the console:
http://hastebin.com/tudewavuxe.avrasm
The relevant log is at:
http://hastebin.com/miqaxajusa.vhdl
I've just tried it in an Ubuntu Trusty virtual machine and got the
dreaded message:
"xf86OpenConsole: cannot open /dev/tty0"
Different error, but the same result: as I expected, Xdummy does not
work on Ubuntu.
That means Xdummy is ruled out in the Ubuntu's case. So rely on xvfb, Right?
Yes, and Xvfb is more limited: no RandR, no OpenGL, etc...
Cheers
Antoine
_______________________________________________
shifter-users mailing list
[email protected]
http://lists.devloop.org.uk/mailman/listinfo/shifter-users
