Welcome to Fractal Forums

Hi everyone,

I have just started working on a new 3D IFS program. At this point it is still proof of concept, and there is no ray tracing (there will be eventually). The current version uses the Painter's Algorithm for hidden surface removal. Coloring a very simply the scaled Z distance. Overall it is very fast. The Menger Sponge example below rendered in about 8 seconds.

Transforms can be easily applied to the image to warp the shape. Here is the Warped Menger Sponge - still very fast to render.

The formula is progressing. In addition to coloring by distance, raytrace/diffuse lighting is now working, along with coloring by by hit density (this is IFS  ;D) and by the map number. The next example is a five layer UF5 image which uses hit density, map coloring and raytrace/diffuse lighting. Rendering time is still quite fast - under 1 minute for a high resolution IFS. Still to come (soon I hope) - shadows and specular lighting.

I call the image Menger Building Blocks.

I believe I like the capabilities of XenoDream (http://www.XenoDream.com/) better for creating IFS type 3-D renderings.  And something like this can be done in a very short time period as well.  Plus, I can dynamically try out and apply various colorings, shadings, lightings, etc...  But the best part, is that it actually creates a 3-D object than can be exported as a mesh for other 3-D rendering applications.

(http://www.Nahee.com/Fractals/XenoDream/Images/PNL_MengerSponge_Stretch2-b.jpg)

XD definitely has the advantage if you want to export meshes, but rendering this type of fractal using the escape-time method in UF or ChaosPro etc. is still relatively in its infancy and even now if you want print-sized resolution (e.g.8000*6000) then the escape-time method is faster for plain affine IFS - myself and Ron just need to add more lighting and positioning options ;)

But they still will not be able to beat XenoDream's ability to apply virtually unlimited light sources and colors after the image has been rendered.  (And by the way, that image was twice the size when rendered and it took only a matter of seconds on my machine.)
 

Here is a 3D rendering of the 3D Cantor set.

Methinks the UF users throw the "3-D" word around way too much !!!

And if this was really 3-D from the UF program, where is the object file??  Where is the mesh to import into real 3-D applications??

Oh wait....  UF can not create real 3-D, it only simulates such things.    :dink:
 

Methinks you understand very little about UF5. It is possible to write a full featured Raytracer in UF5 (camera, multiple lights, specular and diffuse lighting, any level of reflection, transparency with choice of refractive index, perspective, rotations, etc.). Its too bad you really haven't explored the possibilites or tried to understand the code.

John Hart, who had done much for raytracing quaternions and 3D IFS fractals, has some IFS code in his thesis. I took his Elm Tree and Grass setting and put them into my UltraFractal 3D IFS program. Here is what I got:

Hi Ron - mind if I ask for Hart's "settings" for these ? I've never come across that source.

Hi Dave,
Its a PDF file I downloaded from the internet quite a while ago. I renamed the PDF as the orignal was a bunch of numbers ald letters, and I don't have the orginal name. Here is what is on the cover page. a search should find it.

COMPUTER DISPLAY OF
LINEAR FRACTAL SURFACES

JOHN C. HART
B.S., Aurora University, 1987
M.S., University of Illinois at Chicago, 1989

THESIS
Submitted as partial fulllment of the requirements
for the degree of Doctor of Philosophy in Electrical Engineering
and Computer Science
in the Graduate College of the
University of Illinois at Chicago, 1991

I forgot to mention, the thesis has the transforms for all the Platonic solids (Menger Sponge for the cube), and a cool extruded Sierpinski

Here is the extruded Sierpinski image:

Here is the Elm Tree using Hart's parameters rendered in XenoDream.

It look like prehistoric tree. Probably  most often IFS looks like prehistoric plants cause they are more basic.

Go on, export us a mesh of the 3D Cantor set :) Let's see how well suited triangular meshes are to rendering fractals... ;)

Prediction: the mesh will be absolutely massive, very little surface detail, take up a ton of disk space and render at least 100x more slowly than a direct intersection algorithm.

Well... I see both Aexion and Garth Thornton online now, so that's 3 people who've implemented mesh export... maybe we can get some more opinions?

Of course a triangular mesh of a fractal dust would be huge at any useful resolution. In practice meshes are always a compromise, especially for fractals with uniformly rough surfaces because there's not much scope for mesh simplification. Point cloud export would be more efficient for the 3D Cantor set, and the size is not constrained by memory. However, 100 million points (the current limit in XD point cloud export) is still only 464^3. You can do really cool animations with such point clouds in 3DS Max with Krakatoa or Realflow. However, for high resolution, export is a long way from competing with direct rendering (except for 3D printing, where it's currently ahead.)