On Friday 04 February 2005 12:11, Timothy Miller wrote: > > What if you're adding 0 (2^-127) and 2^-120? That won't be equal to > > 2^-120 since (I guess) you have more than 7 bits of precision. What if > > then, the result is multiplied? > > > > Well, maybe you should read this mail as: consider what the output is > > used for. If it doesn't go into a multiplier, you should be fine. > > Yes, that is a good point. The problem case is if you're multiplying > a really big number by a really small number. That could, > theoretically, put you back into range of whatever clamping happens > later and cause a problem.
It's a common case, just look at a polygon edge-on. Regards, Daniel _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
