On Monday 07 February 2005 18:23, John Tsiombikas (Nuclear / the Lab) wrote: > On Mon, Feb 07, 2005 at 01:35:44PM +0100, Lourens Veen wrote: > > Good call. I've just been trying out the model with my new and shiny > > float25 class, and I actually got it to work. It appears that going from > > float32 to float25 does in fact cost some precision (1-2%) but my > > approximate reciprocal does not reduce quality any further (ie yields the > > exact same output on the textured image). > > Ah, I almost forgot (sorry about the multiple replies, I'm a bit off > right now), after applying your patch the antialiased line test failed, > choughing up some "ERROR"s. Are you aware of it? > I didn't try to find the problem since I'm not really familiar with the > internals of floating point arithmetic.
Oh, good point. Yes, I've noticed them as well, but forgot to mention. It may be a problem in my float25 class, or there is some rounding problem that now pops up due to the reduced precision. I haven't yet looked at the model code close enough to know what the problem might be. I'll see if I can figure it out later tonight. It had been a while since I'd done anything with floating point, but I think I've reacquainted myself by now :-). Lourens _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
