Title: DE vs no-DE Post by: ker2x on May 12, 2010, 09:07:11 PM i can't decide which one is better.
I use mandelbulber 0.5. The 1st one (cacao) have a low resolution DE : i like the tiny holes everywhere, but the very sharp shapes aren't nice. The 2nd one (coffee) use "no DE" (much slower) : finer, nicer, but less holes :( what do you think ? (http://fractals.s3.amazonaws.com/cacao-fractal1.jpg) (http://fractals.s3.amazonaws.com/coffee-fractal.jpg) Title: Re: DE vs no-DE Post by: ker2x on May 12, 2010, 10:02:44 PM i changed the url of the image so they are now hosted on amazon S3 (the pics are not recompressed/touched/whatever and i can hotlink as much as i want)
Hope it works for everyone, this is the 1st time i use Amazon S3 Title: Re: DE vs no-DE Post by: Bent-Winged Angel on May 12, 2010, 10:26:23 PM #1 :)
Title: Re: DE vs no-DE Post by: Buddhi on May 12, 2010, 10:44:03 PM For me first one is more interesting, because there is visible more details. But I think you used to high value of "DE step factor". Standard value is 1.0 but for higher quality you should try value around 0.5. Then ray-matching will be more accurate.
"No DE shading mode" didn't use distance estimation for calculating normal vectors (necessary for calculation angle of incidence of the light). It "analyse" fractal shape around calculated pixel by calculating pseudo gradient (of binary potential function) using around 200 points. It is very slow but works very good also when calculation of distance estimation is impossible or not accurate. Title: Re: DE vs no-DE Post by: ker2x on May 12, 2010, 11:00:02 PM For me first one is more interesting, because there is visible more details. But I think you used to high value of "DE step factor". Standard value is 1.0 but for higher quality you should try value around 0.5. Then ray-matching will be more accurate. "No DE shading mode" didn't use distance estimation for calculating normal vectors (necessary for calculation angle of incidence of the light). It "analyse" fractal shape around calculated pixel by calculating pseudo gradient (of binary potential function) using around 200 points. It is very slow but works very good also when calculation of distance estimation is impossible or not accurate. Yes, i used 1.0 DE step factor. i'm trying with lower value. Thx for the tips, i'll upload the result :) Title: Re: DE vs no-DE Post by: ker2x on May 13, 2010, 02:48:21 AM ~12h of cpu time later (3200x3200 with 0.1 "DE Step Factor" and 4 "Resolution") :
A 800x800 version : (http://fractals.s3.amazonaws.com/cacao-fractal2-800x800.png) you can download the 12MByte, 3200x3200, original version here : http://fractals.s3.amazonaws.com/cacao-fractal2-original-high.jpg (http://fractals.s3.amazonaws.com/cacao-fractal2-original-high.jpg) Title: Re: DE vs no-DE Post by: ker2x on May 13, 2010, 02:57:48 AM i had to reduce *a lot* the glow factor when i reduced the DE step factor to 0.1, but it's still too much when compared to the earlier version. here is a version with less glow effect : (http://fractals.s3.amazonaws.com/cacao-fractal2-less-glow-800x800.png) and the 3200x3200 version : http://fractals.s3.amazonaws.com/cacao-fractal2-less-glow-3200x3200.jpg (http://fractals.s3.amazonaws.com/cacao-fractal2-less-glow-3200x3200.jpg) |