Welcome to Fractal Forums

Here is a Mandelbrot heightfield fractal using meshes. The heights are calculated using a Vepstas/Harkonen algorithm, which is essentially an iteration smoothing algorithm that works for almost all fractal types. The mesh uses phong shading.

Hi Ron, adapting DE to work for heightfields may be a better option - I did some trials and got these on a Julia using a version of delta DE:

Solid on atan(smooth iter) @640*480 in 11.78 secs, maxiter 60, bailout 1024:

(http://fc08.deviantart.net/fs48/f/2009/189/0/2/Julia_Iter_Heightfield_by_MakinMagic.png)

If no image above then look here:

http://makinmagic.deviantart.com/art/Julia-Iter-Heightfield-128717839 (http://makinmagic.deviantart.com/art/Julia-Iter-Heightfield-128717839)

Solid on DE @640*480 in 11.28 secs, max iter 60, bailout 1024:

(http://fc06.deviantart.net/fs46/f/2009/189/4/e/Julia_DE_Heightfield_by_MakinMagic.png)

If no image above then go here:

http://makinmagic.deviantart.com/art/Julia-DE-Heightfield-128718029 (http://makinmagic.deviantart.com/art/Julia-DE-Heightfield-128718029)

I'm pretty sure an analytical DE could be developed for both solid on smooth iteration height and solid on DE height that would be more optimum than using delta DE.

Heightfields done using DE are definitely possible on other colourings too, but perhaps a little more difficult to impliment both fast and accurately.

To my eye, there is a lot of detail missing.

I used a fairly high "solid threshold" on those and I've since improved the algorithm, here's some renders of a Mandy:

Atan(smooth iter):

http://www.fractalforums.com/gallery/?sa=view;id=1092 (http://www.fractalforums.com/gallery/?sa=view;id=1092)

DE:

http://www.fractalforums.com/gallery/?sa=view;id=1093 (http://www.fractalforums.com/gallery/?sa=view;id=1093)

DE zoom (higher detail):

http://www.fractalforums.com/gallery/?sa=view;id=1094 (http://www.fractalforums.com/gallery/?sa=view;id=1094)

David,

I can't get those links to work. They take me to top level pages here on Fractal Forums. Can you see if the links work for you?


Duncan C

Like David, I have had good luck adapting DE values to create height maps. I then render them as triangle meshes using OpenGL.

I've created a lot of images this way. some examples:

A fairly conventional view of the whole Mandelbrot set in 3D:
(http://www.fractalforums.com/gallery/1/190_19_02_10_5_01_40.jpeg)

An image colored using integer iteration values:
(http://www.pbase.com/duncanc/image/111851328/original.jpg)


A 3D version of a plot from "The Beauty of Fractals": (map 49 from that book)
(http://www.pbase.com/duncanc/image/96827384/original.jpg)

Another Mandelbrot set from a different perspective (with some clouds added after the fact)
(http://www.pbase.com/duncanc/image/106063460/original.jpg)

And a Julia set rendered with much more dramatic height values, from a very close camera angle that makes it look like a flower:
(http://www.pbase.com/duncanc/image/101662299/original.jpg)


Regards,

Duncan C

Beautiful!

(http://www.pbase.com/duncanc/image/96827384/original.jpg)

I think these images could be much better if you invert heights and assign water material to flat area (and also some rocky material instead of phong to everything else)

Very strange ! Let's try this instead:

Atan(smooth iter):

(http://www.fractalforums.com/gallery/1/141_23_11_09_9_10_00_0.jpg)

DE:

(http://www.fractalforums.com/gallery/1/141_23_11_09_9_10_00_1.jpg)


DE zoom:

(http://www.fractalforums.com/gallery/1/141_23_11_09_9_13_48.jpg)

btw what about using normal mapping? here's a rough test.

it could create nice effects, if used for lower resolution meshes :D
especially when a little finer grade is used for attentuation or small effects ( using distance estimators )
:D:D

but never tried it really!