DarkBeam
Global Moderator
Fractal Senior
Posts: 2512
Fragments of the fractal -like the tip of it
|
|
« Reply #30 on: December 20, 2016, 11:04:29 PM » |
|
I was literally unable to turn on pc last two weeks. Wait patiently I also need some asm help I am preparing great stuff. I also pinned the post so it will stay very visible to posterity like bulatov.frag To Johan - do three renders with the noise before or I won't give the new toy tou!
|
|
|
Logged
|
No sweat, guardian of wisdom!
|
|
|
mclarekin
|
|
« Reply #31 on: December 21, 2016, 02:46:43 AM » |
|
cool formula image rendered in Fragmentarium
|
|
|
Logged
|
|
|
|
|
DarkBeam
Global Moderator
Fractal Senior
Posts: 2512
Fragments of the fractal -like the tip of it
|
|
« Reply #33 on: December 21, 2016, 08:33:15 PM » |
|
|
|
|
Logged
|
No sweat, guardian of wisdom!
|
|
|
trafassel
Fractal Bachius
Posts: 531
|
|
« Reply #34 on: December 21, 2016, 08:38:17 PM » |
|
|
|
|
Logged
|
|
|
|
DarkBeam
Global Moderator
Fractal Senior
Posts: 2512
Fragments of the fractal -like the tip of it
|
|
« Reply #35 on: December 22, 2016, 09:51:45 AM » |
|
Here are the fragmentarium shaders: The first is from Jos Leys which don't work on my PC for unknown reason. I put it here especially for darkbeam in order to make a formula for MB3D and mclarekin to include it in mandelbulber. The second one was "rewritten" in order to look better as glsl code. It may look obfuscated or better . It is still work in progress. There are some ideas to try in order to make it faster / have better DE. The algorithm is very simple. The main idea is to use a "line of separation" in order to choose the transformation to apply. In those shaders, the line is approximated by a simple sinusoid. IMHO, it is related to "Schottky blobs" as described in this article (see for example figure 2). The drawback of this method is that in most cases the fractal will have cuts... but the cases where it works well is already huge. Further question. Just for speedup, may I replace sin() with a polynome accurate to 1e-5? And what may be the "exact" expression of the line? Tangent circle?...
|
|
|
Logged
|
No sweat, guardian of wisdom!
|
|
|
JosLeys
|
|
« Reply #36 on: December 22, 2016, 10:34:47 AM » |
|
The sinus as a separation line is just an approximation that works ok in most cases. You can replace it with anything else that works.
In fact, getting the real line is difficult. We can collect points on the separation line by iteratively transforming, in the nomenclature of the frag file, the point (box_size_x-b/2, a/2) until we get to an x value below -b/2, en then taking the symmetric line, as in the picture below.
However this is probably too expensive to do in Fragmentarium.
|
|
|
Logged
|
|
|
|
DarkBeam
Global Moderator
Fractal Senior
Posts: 2512
Fragments of the fractal -like the tip of it
|
|
« Reply #37 on: December 22, 2016, 11:40:35 AM » |
|
At this point we can try any curve with S shape like atan(x) (also slow) and fractional polynomes (faster). I will do some tests definitely. Also it is not necessary imho to find all points; there are methods that find an interpolating polynome given n points - polynomes work great for most anything and are CPU friendly . But idk how to find those points just now. Thanks so much for the share!
|
|
|
Logged
|
No sweat, guardian of wisdom!
|
|
|
knighty
Fractal Iambus
Posts: 819
|
|
« Reply #38 on: December 22, 2016, 01:39:15 PM » |
|
Update: new frag attached. Fixed: discontinuity. New option: fourGenerators check box. Default to 3 generators.
Now the "line of separation" uses an exponential function as suggested by Jos Leys... Still WIP.
|
|
|
Logged
|
|
|
|
mclarekin
|
|
« Reply #39 on: December 23, 2016, 09:22:18 AM » |
|
WOW, this may be the best Christmas ever!! Thanks guys
|
|
|
Logged
|
|
|
|
lycium
|
|
« Reply #40 on: December 23, 2016, 02:42:49 PM » |
|
Thanks very much for these Jos + Knighty! Best 3D fractal ever, by far
|
|
|
Logged
|
|
|
|
lycium
|
|
« Reply #41 on: December 24, 2016, 03:36:42 AM » |
|
Such an awesome DE! I had to set the fudge factor down a bit (e.g. 0.125) to stop it overstepping, and it looks pretty good with this setting Also, in the code I saw: float r=length(p); float r2=r*r;
which can be improved to: float r2=dot(p, p); float r=sqrt(r2);
Finally, my first test render (click for 5120x2880 resolution):
|
|
« Last Edit: December 24, 2016, 11:01:46 AM by lycium »
|
Logged
|
|
|
|
Crist-JRoger
|
|
« Reply #42 on: December 24, 2016, 06:24:54 AM » |
|
Beautiful fractal! Jos and Knighty thank you guys!
|
|
|
Logged
|
|
|
|
Sabine
|
|
« Reply #43 on: December 24, 2016, 11:26:10 AM » |
|
I am quite speechless... Thank you all!
|
|
|
Logged
|
sabine62.deviantart.com
|
|
|
Crist-JRoger
|
|
« Reply #44 on: December 24, 2016, 12:35:56 PM » |
|
DE-Kn render with reflections, volumetric light and bloom
|
|
|
Logged
|
|
|
|
|