twinbee


« Reply #195 on: September 10, 2009, 01:51:01 PM » 

Ace! I fear adding even shadowing (let alone global illumination) may choke on that!


« Last Edit: September 10, 2009, 01:59:49 PM by twinbee »

Logged




cKleinhuis


« Reply #196 on: September 10, 2009, 04:16:45 PM » 

a bit moreiterations would also be nice, diffuse direct lighting is fine !
i propose, for global illumination issues, a mesh export would be better, than to include a whole global illum renderer ....


« Last Edit: September 10, 2009, 04:18:32 PM by Trifox »

Logged


divide and conquer  iterate and rule  chaos is No random!




twinbee


« Reply #198 on: September 12, 2009, 01:57:42 AM » 

Cool ! How about a zoom in to 10,000x starting from 1x... (one where the camera moves in, rather than a camera type magnification (narrowing field of vision), which would only look like a 2D magnification thing, rather than an awesome 'fly through' with parallax). Naturally, the camera would need to slow down towards the end as it gets incredibly close to the object. Also, with a nice, large field of vision for maximum effect


« Last Edit: September 12, 2009, 02:08:59 AM by twinbee »

Logged




David Makin


« Reply #199 on: September 12, 2009, 05:06:33 AM » 




Logged




bugman


« Reply #200 on: September 12, 2009, 05:31:58 AM » 




Logged





cKleinhuis


« Reply #202 on: September 12, 2009, 04:24:50 PM » 

those orbit traps variation looks beautiful ! and for the 3d buddhabrot .... you fill a 3d array with hit counter values ?! and render it as a volumetric object ?!?



Logged


divide and conquer  iterate and rule  chaos is No random!



bugman


« Reply #203 on: September 12, 2009, 06:39:50 PM » 

Here's a comparison of two different ways of calculating higher order Mandelbrot sets for Twinbee's formula. One image uses the standard method of: {x,y,z}^8 = r^8{cos(8θ)cos(8φ),sin(8θ)cos(8φ),sin(8φ)}
The other image uses nesting: (({x,y,z}²)²)² where {x,y,z}² = r²{cos(2θ)cos(2φ),sin(2θ)cos(2φ),sin(2φ)}
In either case r=sqrt(x²+y²+z²), θ=atan(y/x), φ=atan(z/sqrt(x²+y²))


« Last Edit: September 12, 2009, 07:51:42 PM by bugman »

Logged




bugman


« Reply #204 on: September 14, 2009, 09:31:02 PM » 

those orbit traps variation looks beautiful ! and for the 3d buddhabrot .... you fill a 3d array with hit counter values ?! and render it as a volumetric object ?!? Yes, the 3D Buddhabrot is a volumetric rendering of a 3D array with hit counter values. I am going to try a completely different approach and see if I get better results.



Logged





bugman


« Reply #206 on: September 14, 2009, 09:33:22 PM » 

I like this one. What constant did you use for this Julia set?



Logged




David Makin


« Reply #207 on: September 14, 2009, 10:58:42 PM » 

I like this one. What constant did you use for this Julia set? It's a degree 4 using your version of Twinbee's formula  i.e. the "power" version rather than the repeated squaring  and the Julia constant is (0.8,0,0).



Logged





David Makin


« Reply #209 on: September 14, 2009, 11:36:17 PM » 

Actually for wellbehaved maths (like quats or "standard" hypercomplex) where there's an analytical distance estimator and for other fractals using the delta smooth iteration method then colouring the surface simply based on the smooth iteration value for the solid surface points should produce surface colour variation  using both methods the shape of the solid does not match the shape of the smooth iteration contours provided that the max. iteration used is large enough  in the delta case this is because "solid" is essentially based on the iteration density rather than the iteration value. Ways of colouring the surface is something that I will include in my new set of classes when I finally get around to some more fractal coding


« Last Edit: September 14, 2009, 11:38:46 PM by David Makin »

Logged




