The Other Side of Open GL

[Sockratease]

OK, We've all seen it.

The ever familiar grid that IS Open GL.

We make our images and rotate them, change the angle of viewing, and make our spline and smoothing adjustments and what ever other parameters our program of choice offers.

But what about Open GL's Seedy Underbelly?

Am I the only one who enjoys turning these things over and seeing what lies beneath?

Here's a good example -

I made this 2D render in Mutatorkammer:



It looked like this in Open GL (rotated):



But... If one rotates the grid a full 180 degrees around the z axis, you can see the Bottom of the fractal!




I have gotten some very nice results and interesting comparisons by doing this.  Here is another pair of "Top and Bottom" views:





I mirrored the bottom image of the above pair (the only image editing in these images!) to demonstrate the symmetry (looking at the bottom of these naturally produces mirror images of the other side).


And some more -








One last comparison, just because it's so weird...


A 3D Terrain type fractal




And the underside:

[cKleinhuis]

ROFL; it is a feature
 

no, the problem lies in the normals, and the front/backside rendering of triangles,

you know, you can achieve a similar effect in mutatorkammer by adjusting the heightmap values ( the grey bars in the palette window ) to the opposite, so that high values become low values and vice versa...

and btw, this one is cool:
http://i106.photobucket.com/albums/m278/sockratease/y2.png

[Duncan C]

ROFL; it is a feature
 

no, the problem lies in the normals, and the front/backside rendering of triangles,

you know, you can achieve a similar effect in mutatorkammer by adjusting the heightmap values ( the grey bars in the palette window ) to the opposite, so that high values become low values and vice versa...

I've played around with looking at the underside of my 3D plots in FractalWorks before, but have never been satisfied with the lighting because the normals are set up for top-side viewing.

Today I decided to add a "flip plot" option, and I'm pretty pleased with the results.

Here's a sample image:


You could actually argue that this plot is right side up, since it uses distance estimate values for the height, and distance estimates are smallest near the Mandelbrot set. My program normally inverts the values so Mandelbrot points are at the peak and points far from the set are at the bottom.

[Cyclops]

What exactly is open GL?

[cKleinhuis]

What exactly is open GL?

opengl is a "API"  - Application Programmers Interface Specification - to acess 3d hardware, which encapsulates 3d math, and gives functionality for managing 3d objects ( made out of triangles )
similar api is directx, which offers a bit more multimedia features, but also the 3d things from opengl ...

 

http://en.wikipedia.org/wiki/OpenGL

[Cyclops]

Oh right so basically its a library of code used by Incendia,Apophysis,Bryce, Terragen etc

[cKleinhuis]

yes, a library of code implemented the same way on many different systems, e.g. you can use the opengl api the same way you would use it under windows or linux ....

 

[lycium]

as far as i know, of those programs only incendia and possibly bryce use opengl, and only for previews (the high quality rendering being done with their own engines, since that's the whole point of these programs!).

in fact, i would argue that triangle rasterisation (which is what opengl and directx use) is completely unsuited to fractal rendering. then again, half the programs you listed aren't fractal renderers

[HPDZ]

I just love you all's 3D images...Please keep them coming!

I hope this question isn't getting too far off-topic, but I've been wondering how you get these. Are you using a fractal rendering program that also has a built-in 3D rendering component, or do you generate the fractal data with one program (e.g. UF or Apophysis) and then use commercial or home-grown 3D rendering rendering software?

I've been considering adding an OpenGL or DirectX 3D component to my deep-zoom animation software (home grown) but if there's a already commercial software that can import the fractal height data and render it in 3D that would save me a huge amount of reinventing of the wheel.

Thanks.

[lycium]

it's surprising to me that you've written "home grown" and "reinventing the wheel", two diametrically opposite views of the same thing, in the same sentence!

we've all written our own renderers, ducanc and trifox leverage the power of opengl whereas the other programs you mentioned have purely software-based rendering engines.

[HPDZ]

it's surprising to me that you've written "home grown" and "reinventing the wheel", two diametrically opposite views of the same thing, in the same sentence!

Good point! Probably a better way to say it would have been "...save me a huge amount of FURTHER reinventing of the wheel".

we've all written our own renderers, ducanc and trifox leverage the power of opengl whereas the other programs you mentioned have purely software-based rendering engines.

Good to know. The math for all this is well within my ability, but it looks like a lot of work to get it all coded into functional, stable software. Maybe someday...

Meanwhile, is there anything that can import a series of files with fractal count values and make a series of 3D images? I've done it with Mathematica once or twice for still images, but it's really tedious and I'm not sure it's practical if I were to try doing this with an animation.

[HPDZ]

it's surprising to me that you've written "home grown" and "reinventing the wheel", two diametrically opposite views of the same thing, in the same sentence!

Good point! Probably a better way to say it would have been "...save me a huge amount of FURTHER reinventing of the wheel".

Actually, sometimes, it feels more like reinventing the automatic transmission... 

[cKleinhuis]

so far, i think opengl is a very good base to implement a decent 3d renderer, and you know the main difference ?

opengl is used for REAL-TIME display of gfx, those kind of grafix can never (really) compete with advanced renders,
but,the 3d realtime functionality and fractal formulas can form a wonderfull marriage, but, writing something that
works like google earth on 3d objects ( meaning refining the structure on every zoom level ) is extremly hard to
program ( level of detail on non-static data ), my dream is ( if i had more time ) to implement a hardware renderer
of 3d objects ....

but believe me, it is a very deep field ...

lets say you could calculate a reasonable 3dimensional field ( e.g. 1024x1024x1024 ) then you only have some
values, if you know would like to display it as a 3d mesh, you have to find countours in the values.....
and then we have the things like calculating the normals of those meshes, for nice (simple direct ) light calculations

i have implemented an iso surface renderer of those fractal formula fields in mutatorkammer ( unreleased version )
but you need a hell of detail ( remember, fractals are infinite, and 3d fractals at low iterations look not as good
as the same 2d mandelbrot for example ... )


enough talking .... it is hard, but i would like to participate ....


writing a height map renderer in opengl is somewhat like an afternoon of work ...
greets
ck