Julia Quaternion for VRay

[MasterBercon]

VRay/Max related code mostly direct copy of VRaySphere
Julia part is adaptation of Keenan Crane's Julia "Ray Tracing Quaternion Julia Sets on the GPU"

The thing doesn't have any practical use aside from being pretty.

You can freely scale, move and rotate it.

Animation: http://www.niksula.cs.hut.fi/~jylila...uliaAnim02.mov

Plugin (Max 2009 64bit, VRay sp2 only): http://www.niksula.cs.hut.fi/~jylila.../VrayJulia.dlo

Source code: http://www.niksula.cs.hut.fi/~jylila...e_8_6_2008.rar


The problem I have is that the normal calculation doesn't work properly. I'm not really sure how it should work in the first place. When calculating the normal the julia set escapes to infinity really fast making it impossible to calculate the normal. This happens when using value above 11-20 in Normal Iteration.

If somebody knows what might be the problem I'd be glad to hear it and finish the plugin.

Here is just the normal estimation part of the source code http://www.niksula.cs.hut.fi/~jylila...alEstimate.txt

- Jerry Ylilammi

[cecofuli]

MasterBercon,

usefull for modern statue. Any possibility for other version?

Thank's

cecofuli

[rajdarge]

julia

any chance of recompiling this for 2008, 32 bit?
Raj

[MasterBercon]

I want to solve the normal estimation problem first. Once I've figure it out I'll release 2008/2009 32/64bit versions.

[RayOfLight]

how's it going?

[dimo]

Hi Jerry,

Just a blind guess. You could try to measure the gradient from the intersection point to the point + deltas and not from point - delta to point + delta. Perhaps this gives more stable results. The size of delta will also have great impact.
This would make:
double4 gx1 = qP ;
double4 gx2 = qP + double4(delta, 0, 0, 0);
double4 gy1 = qP ;
double4 gy2 = qP + double4(0, delta, 0, 0);
double4 gz1 = qP ;
double4 gz2 = qP + double4(0, 0, delta, 0);

Best regards,
Dieter