On Wednesday 10 December 2003 16:03, Jason Wood wrote: > On Wednesday 10 December 2003 23:03, Rolf Dubitzky wrote: > > On Wednesday 10 December 2003 14:41, Jason Wood wrote: > > > Does piave treat rendering a scenelist to a file in the same way as > > > playing a scenelist to a video window? > > > > there is no difference. I would like there to be one, but this involves > > OpenGL / quality of effects and optimisations inlcuding hardware > > accelerated graphics. All these things are far away and don't affect the > > statement that the 'render' tag is superfluous, since output to a screen > > is also > > "rendering". it has nothing to do with file/screen/whatnot. > > Ok, so the differences that you are talking about are essentially > optimizations within piave, and do not (should not) affect the > communication between kdenlive and piave.
in principle you might want to configure the quality of the preview or the quality of a pre-render run, but I think we shouldn't worry about that now. I don't think it will be difficult to add what is needed for that at a later time, when we know what we want/need. > In this case, I was thinking of the "addOutStream" as a replacement for the > "createXWindow" tag that currently exists. Kdenlive would need some return > communication from piave to indicate success and any required attributes. > The only one currently being used is the X window ID when a video window is > created, but a very useful addition would be a unique identifier for each > outstream (future proofing for when we have multiple outstreams ;-)) Ah, ok. good that makes sense. > I agree it would be useful to be able to have multiple outstreams, so it is > best to consider a communication design where multiple outstreams are > possible. A nice feature would be an outstream to firewire (as a > replacement or in addition to the sdl stream); suddenly anyone with a > digital camcorder and a computer can use a separate tv monitor to watch the > video they are creating ;-) yes, that is not too difficult. I tried it when I tested the grabbing and in priciple it works, but that requires some additional thoughts about buffering and realtime requirements. Cheers, Rolf -- contacts: http://www.physi.uni-heidelberg.de/~dubitzky
