Title: Masquerade Post by: Kali on January 27, 2013, 03:20:59 AM Masquerade
(http://nocache-nocookies.digitalgott.com/gallery/13/3869_27_01_13_3_20_59.jpeg) http://www.fractalforums.com/index.php?action=gallery;sa=view;id=13278 Amazing Surface for Fragmentarium get it here: http://www.fractalforums.com/index.php?action=downloads;sa=view;down=44 DE has some problems with some parameters and needs to be improved. Mandelbrot modes and non-rotated or slight rotated params could be problematic. At the moment you can use fudgefactor on the raytracer tab to get rid of the problem most of the times, but it will slow down the render :sad1: Any help to improve it will be much appreciated Title: Re: Masquerade Post by: Syntopia on January 27, 2013, 11:37:17 PM Great work, Kali. I'll add your frag to the distribution!
Title: Re: Masquerade Post by: Kali on January 28, 2013, 01:56:13 AM Ok, many thanks! I wish I could get rid of the DE problems, maybe your dual number approach?
Title: Re: Masquerade Post by: eiffie on January 31, 2013, 05:30:34 PM Great script Kali! I put the +1 back in the running derivative like this...
vec4 c=p;//p.w==1 and it helped quite a bit with the DE. For speed (and sanity) you may want to change the scaling to... vec4 scale=vec4(scale...,abs(scale));//remove the /MinRad2 then.. p*=scale/clamp(r2,MinRad2,1.0);//I believe this works out to be the same - check it ;) Thanks for posting - it will keep me busy exploring for quite awhile. Title: Re: Masquerade Post by: Kali on January 31, 2013, 07:48:26 PM Thanks eiffie! It works much better now, I updated the .frag in the download section.
Still some problems with non-rotated Julias, but it requires less raystep scaling now ;) Title: Re: Masquerade Post by: knighty on January 31, 2013, 09:36:54 PM Great work! I put the +1 back in the running derivative like this... You nailed it eiffie! ;)vec4 c=p;//p.w==1 and it helped quite a bit with the DE. Actually, you only need that in mandelbrot mode where it is necessary to obtain well behaved DE. Keep in mind that, just like the mandelbox and it's julias, it's usually necessary to scale down the DE in order to avoid overstepping. For speed (and sanity) you may want to change the scaling to... Nice trick, thanks a lot! :).vec4 scale=vec4(scale...,abs(scale));//remove the /MinRad2 then.. p*=scale/clamp(r2,MinRad2,1.0);//I believe this works out to be the same - check it ;) |