3D Heightfield rendering

[David Makin]

Hi all, I just realised that my alternative to the standard distance estimator could be adapted for use in rendering heightfield fractals.

First results:



If no image above:

http://makinmagic.deviantart.com/art/Mandelbrot-Heightfield-128316417

And:



If no image above:

http://makinmagic.deviantart.com/art/Minibrot-Heightfield-128316816

Rendered using pure ray-casting i.e. no triangles or polygons.
In both the height is atan(smooth iter - 3)/atan(max. iter - 3) or 0 if iter<3
max iter was 60
bailout was 1024
On my 3GHz P4HT @640*480 the full Mandy took 30secs and the minibrot zoom took 2mins 40secs (I cranked the detail up on that such that the ray-stepping was reduced by a factor of 2.5).

[David Makin]

Hi, I tinkered with the way I was calculating the step distances - new results:

bailout 1024
max. iter. 60
"Inside" height 2.0



If no image above:

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

The above took just 11.78 seconds @640*480, the heights based on atan(smooth iteration).



If no image above:

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

The above took just 11.28 secs.
Height based on standard Distance estimator colouring.

[David Makin]

Hi all,



If no image above then:

http://makinmagic.deviantart.com/art/Orbit-Trap-Heightfield-128850151

This is more like what I'm aiming at with rendering heightfields.
Height based on "Orbit Trap" "closest".
Nice image but unfortunately to render this well using my current algorithm took nearly 15 mins @640*480.
I'm workiing on the speed, hopefully getting the quality of rendering this image as good in under 1 minute - if I can do that then the algorithm will basically be able to render heightfields based on any colouring algorithm.