I think I might have read that one while researching - the one in which the argument is basically "Wolfenstein 3d must be possible because all the individual parts of the engine are simple" versus "scrolling the screen is even more simple, but the Sam doesn't seem to have the horsepower for that"?
At least this time I'm actually going to release some code. I've submitted to Colin a tiny little demo of the 'engine' up to the point of the most recent youtube video and it is my current understanding that it'll be on the disk with Sam Revival 22. Re: the precision issue, the location of each point ends up effectively being the sum of three numbers, each of them the product of at least three numbers. And each of my table based multiplications is likely to lose accuracy in at least the low bit, so the compounded loss of accuracy is almost certain to have a visible effect. I'm going to switch to a 2.14 fixed point scheme for matrix preparation. It'll cost a bit more per-object but no more per-vertex. And it's a much better idea if I want to implement proper local-axis rotations ala Elite rather than the hodge bodge of local and global axes that you get with Euler angles. On Fri, Jun 6, 2008 at 3:51 AM, Simon Cooke <[EMAIL PROTECTED]> wrote: > I'm amused... was just doing a search for some old friends on Google, and lo > and behold... we were talking about trying to get a good 3D renderer going > back in 1995. > > Some things never change ;-) > > Si > >
