Hi Alessandro, Sharing threads would be a nightmare to code and maintain, is certainly not an avenue that you'd want to pursue.
With my work on Present3D I added to code to start and stop movies so that only a few are active at any one time, with big presentations there might be several dozen movies all loaded and we haven't seen major problems with this approach. Robert. On 7/2/07, alessandro terenzi <[EMAIL PROTECTED]> wrote:
I wonder if there can be a workaround to avoid to start a new thread for each movie... in my application I need to open many movies but, actually, just one or two will be played at the same time, what could I do to use the same thread for different movies? ... I'm not sure, maybe I could reuse the same ImageStream? will this avoid new thread creation? or pheraps, replacing a movie's subgraph with another subgraph for the newly requested movie?...maybe in this case there will be always just one thread alive each time? or there is another way you could suggest? Thank you. Best regards. Alessandro. On 6/27/07, Robert Osfield <[EMAIL PROTECTED]> wrote: > The movie plugins will start a background thread to read the movies - > one per movie. You can start and stop movies via the ImageStream > interface, this won't switch off the threading though. > > On 6/27/07, alessandro terenzi <[EMAIL PROTECTED]> wrote: > > Thank you very much for the quick answer... > > > > ...overall the size of the models I load (ive files) is 200MB + 180MB for > > some videos (avi, mpg, mov). > > ...actually I realized that performances problems arise just with the videos > > and not with the usual geometry... I'm trying to understand the reason... > > maybe the stream is always running from the moment I load the video file? > > and maybe I have to stop each stream... > > > > ..by the way my machine's ram is 2GB > > > > Regards. > > Alessandro > > > > > > On 6/27/07, Robert Osfield <[EMAIL PROTECTED]> wrote: > > > Hi Alessandro, > > > > > > If the NodeMask is 0 then the intersection and cull traversals should > > > all ignore those subgraphs and shouldn't affect performance on their > > > own. > > > > > > Could it be that you are running out of memory? How big is each model? > > > Check the overal memory consumption as the models are loaded. > > > > > > W.r.t reducing the time it takes to introduce a new subgraph in for > > > the first time, the time cost is usually down to the compilation of > > > display lists and texture objects. Running the > > > osgUtil::GLObjectsVisitor on these new subgraphs before viewing can be > > > done for each graphics context to ensure that everything is ready when > > > you need it. This will bump memory usuage though... > > > > > > Robert. > > > > > > On 6/27/07, alessandro terenzi <[EMAIL PROTECTED]> wrote: > > > > I'm working on OSG+osgART and I'm experiencing performances problems > > when I > > > > load many (50) 3D models in the scene. > > > > > > > > Actually, I load every single model when the application first starts, I > > > > attach them to the scenegraph but each model is displayed only when a > > > > fiducial marker is recognized inside a videostream that is the scene > > > > background. But, actually, it looks like that every model is taken into > > > > account for many computations (for example for mouse picking...) even > > though > > > > it is not visible in the scene. To figure out what is happening I dumped > > the > > > > scene in a file and I realize that nodes corresponding to not-visible > > models > > > > have the nodeMask set to 0x0 and cullingActive to true. > > > > > > > > Can you imagine why I'm experiencig such bad perfomances? If I just load > > one > > > > or few models, everything is fine... > > > > When I say bad performances I mean, of course, low fps and, if there is > > a 3d > > > > video plane, also no smooth video playback...but the videostream from > > the > > > > background is always smooth. > > > > > > > > I aslo have another problem, at startup I load each model and I thought > > that > > > > they were put in some kind of cache, but I realize that actually, maybe, > > > > they are not really loaded in ram until they are displayed, because, at > > > > runtime, when a marker is recognized, it takes some time to display the > > > > corresponding model (I guess because it is loaded in that moment and not > > > > actually when the application started...) ... so my question is: is > > there a > > > > way to pre-load object in a cache? > > > > > > > > Thank you. > > > > Best regards. > > > > Alessandro > > > > _______________________________________________ > > > > osg-users mailing list > > > > osg-users@openscenegraph.net > > > > http://openscenegraph.net/mailman/listinfo/osg-users > > > > http://www.openscenegraph.org/ > > > > > > > _______________________________________________ > > > osg-users mailing list > > > osg-users@openscenegraph.net > > > http://openscenegraph.net/mailman/listinfo/osg-users > > > http://www.openscenegraph.org/ > > > > > > > > > _______________________________________________ > > osg-users mailing list > > osg-users@openscenegraph.net > > http://openscenegraph.net/mailman/listinfo/osg-users > > http://www.openscenegraph.org/ > > > _______________________________________________ > osg-users mailing list > osg-users@openscenegraph.net > http://openscenegraph.net/mailman/listinfo/osg-users > http://www.openscenegraph.org/ > _______________________________________________ osg-users mailing list osg-users@openscenegraph.net http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
_______________________________________________ osg-users mailing list osg-users@openscenegraph.net http://openscenegraph.net/mailman/listinfo/osg-users http://www.openscenegraph.org/
