If you pass integers derived from floats (where the fractional parts are simply truncated) you can get a not-banker-rounded result by adding 0.5 to the float values (that will be truncated).
But I think your point is well taken -- in the context of computing pixel positions (that are inherently integral) you don't want banker's rounding to be used. Perhaps that's why many math libraries include more than one set of rounding methods... At 09:21 AM 4/20/2006, Peter Ritchie wrote >Except--unlike mathematical calculations--with drawing, I want >consistency; not statistical unbiasedness as my number of calculations >approach infinity. I certainly don't want all my even coordinate pixels >to be on one side of the line and all my odd coordinate pixels to be on >the other. >[snip] J. Merrill / Analytical Software Corp =================================== This list is hosted by DevelopMentor® http://www.develop.com View archives and manage your subscription(s) at http://discuss.develop.com
