I'm reading O'Reilly's <a href="http://www.amazon.de/Beautiful-Code-Theory-Practice-OReilly/dp/0596510047/ref=pd_bbs_sr_1?ie=UTF8&s=books-intl-de&qid=1196633416&sr=8-1";>Beautiful Code</a>. One essay is talking about creating code at runtime in the Windows 95 BitBlit function. BitBlit copies rectangular bitmap areas. Pixels can be copied in 256 different modes in which pixels are combined. The modes are defined by three different parameters. Instead of using conditionals to determine the mode (which is the same for every pixel), BitBlit renders the parameters to assembler code once, and executes this code for all the pixels. This gave a massive performance boost.
I'm not sure how the current Flash 3D libraries work. Maybe they don't use per pixel rendering. Maybe they already use the same technology as the one explained above. Just in case they don't ... Cheers Ralf. _______________________________________________ osflash mailing list [email protected] http://osflash.org/mailman/listinfo/osflash_osflash.org
