Title: Slowness of some formulas and parameters. reasons? Post by: Mrz00m on March 06, 2012, 06:11:33 AM Some parameters render very slowly and in a grainy way, even though the DE looks simple and uncluttered and the surfaces are smooth and shouldnt be grainy, i have to put the raystep multiplier right down to 0.02 to get a smooth pic and the render takes 3 hours for a high quality image.
What are the reasons for a mix of formulas going slowly? i.e. i know devide and exp are slow, can you explain practically what formulas etc will tend to be slow for those reasons? what other reasons? any tips and tricks to make slow formulas go faster? why do i have to put raystep multiplier to 0.025? i.e. here is a slow one, why is it slow? any ways of speeding up the rendering? Code: Mandelbulb3Dv17{Title: Re: Slowness of some formulas and parameters. reasons? Post by: DarkBeam on March 06, 2012, 11:38:15 AM Some parameters render very slowly and in a grainy way, even though the DE looks simple and uncluttered and the surfaces are smooth and shouldnt be grainy, i have to put the raystep multiplier right down to 0.02 to get a smooth pic and the render takes 3 hours for a high quality image. What are the reasons for a mix of formulas going slowly? i.e. i know devide and exp are slow, can you explain practically what formulas etc will tend to be slow for those reasons? what other reasons? any tips and tricks to make slow formulas go faster? why do i have to put raystep multiplier to 0.025? i.e. here is a slow one, why is it slow? any ways of speeding up the rendering? Hey everybody is wondering why... It's too complicated to explain btw! Moving this to the right section! :police: Title: Re: Slowness of some formulas and parameters. reasons? Post by: lenord on March 06, 2012, 04:10:21 PM Some parameters render very slowly and in a grainy way yep, they sure do, learn to live with it and which ones to stay away from. Title: Re: Slowness of some formulas and parameters. reasons? Post by: Jesse on March 06, 2012, 10:33:07 PM It is not that complicate to explain, basically it is because of bad distance estimates. For formulas that are non-analytic and hyrids with those, the local 3d gradient of the vector length after bailout is used to get a more or less accurate distance, the fractal border should be. Some formulas like the reciprocal ones makes this calculation fully inaccurate because flipping the real distance by the transformation, making the internal metric to the outside for example -> point zero becomes infinity. Maybe there are some cooler mathematic equations that can handle those formulas and hybrids better in a general way, if i had 3 brains and four lives i would spend one for this issue. :) |