> > which means: to get really great fps one should think about a > > different way to render those 8.5mio vectors to the screen, eg. to > > render only a subset (LODs comes to mind, carmack's megatexture > > approach is interesting as well) > > LODs need some structure to it, so I'm not sure that he can create them > easily.
yep, i wasnt really thinking of using LODs as we know them but to give a hint about certain techniques; eg. rendering huge images (32k x 32k) to a 400x400 resolution output device with a naive approach will end up freezing the machine by all this scaling of the original data while only a fraction of the original pixel will end up on the output device. > Ronan, do you have an example image of what your data will look like? that would be most interesting. > Doing that efficiently for a vector soup is not trivial, though. well, by just throwing geometry to the hardware and hoping for good results works pretty fine when one expects moores law to produce fast enough hardware in the near future. so, in some months the problem is gone and he has .. 5fps :) cheers, mathias -- [uid] mathias gumz [mail] akira at fluxbox dot org [pgp] 1024D/F6F6B18C [www] http://www.darkshed.net/ [irc] ak|ra (#fluxbox at freenode.org) ------------------------------------------------------------------------- This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone _______________________________________________ Opensg-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/opensg-users
