On Mon, 13 Aug 2012, Lolitha Ratnayake wrote:
Thanks! During last three weeks I was doing this. I used python & Quickly for development along with pygst. 60% is done. What I'm doing is open source video mixer over LAN. Main software have three nodes makers. Camera node, mixer node and display node. Camera node is the node that capture and transmit data to network (currently using v4l2 in gstreamer). mixer is node that works as video mixing software to mix inputs from camera nodes and Display is the node connect to output screen.
Very interesting. I have started to do a SMIL implementation in ClutterGst using Python and Introspection. On SD resolutions it worked as a charm, but HD did require a very specific setup. Live input was on my list as well. You do know there exist dvswitch?
And btw I need to know what are the pros and cons of using clutter for video processing acceleration. Since clutter doesn't come with many default Linux distributions would it be too much negative to use this?
The point is that at some point the CPU acceleration of the screen just gives up, giving tearing images. Probably the problem is two-fold: decoding in software, and display on a surface. Now when I started my queste there was being work done on ClutterGst with libva accelerated decoding, since surfaces should already be accelerated in Clutter when using the correct backend.
One of the other key issues still need to be addressed are 'vector text'. Now it is obvious that the first step of fontrendering in Cluster is vector based. But when this surface is ever scaled, or transformed bitmap scaling is applied. This might not be an issue for scaling down, but scaling up eventuell could get very ugly.
Stefan _______________________________________________ clutter-app-devel-list mailing list clutter-app-devel-list@clutter-project.org http://lists.clutter-project.org/listinfo/clutter-app-devel-list
