dirkdeftly

9th June 2002 21:05 UTC

Holy sh*t
This guy might just put us all out of business. Check out Queen of Hearts, Gemstone, and Sierpinski (that's right, and it's named correctly, too).

http://www.winamp.com/plugins/detail...onentId=107040

NoSage

9th June 2002 22:05 UTC

There are some nice presets and I have never seen anything like queen of hearts, But, puttin me out of the biz? maybe, but UnConeD?

UnConeD

9th June 2002 22:26 UTC

Actually the slow-rendering high-quality fractals idea has been done a lot. Just do a search on Deviant-Art for 'Julia' in the AVS section. The fastest way I could get a Julia on screen was by interlacing ( http://www.deviantart.com/deviation.php?id=86655 ).
And El-vis has done some nice IFS (iterated fractal systems) too, I believe it was called 'Dragon'. The Sierpinski triangle is a rather boring IFS if you ask me :).

I've been experimenting with recreating the fern IFS in AVS too, but never got around to turning it in a nice preset.

Don't get me wrong though, Focus II is a great pack!

dirkdeftly

10th June 2002 00:52 UTC

Well as long as we're on the subject...How can you do a reasonable level of iterations without a) creating an array of objects or b) reproducing the code about 100 times?

And then there's Gemstone: a Julia fractal of a random order.

UnConeD

10th June 2002 02:21 UTC

You don't... if you look at Queen of Hearts you will see the same code copied multiple times :).

And powers of complex numbers become rather easy if you use their polar form... multiply angle by the desired power, and raise the distance to the desired power.

UnConeD

10th June 2002 02:45 UTC

Iterative SSC
Actually I've just implemented a new kind of fractal superscope renderer... it uses the iterative nature of the superscope (lots of points) to calculate a fractal. It keeps working on the same point until it hits the maximum iterations, or until it satisfies the escape condition. So it renders simple fractals a lot faster than filled ones.

This is of course just a little proof-of-concept... it's definately faster for simple fractals (it doesn't waste time calculating unnecessary steps) and it allows for an arbitrary number of maximum interations (the 'mi' value).

dirkdeftly

10th June 2002 03:16 UTC

yeah, I checked that out
I've created one that renders a fractal every frame, except it's really, really slow and it only goes up to 25 iterations.

Nic01

12th June 2002 02:50 UTC

That's Focus 2... AVS-King already released Focus 4...

Just try to find Focus 4...

For a little preview : In that pack, he made metaballs WITHOUT an APE, another fractal, and lotza other cools stuffs...

The closest AVS pack to make it in my Deviantart Favorites list...

UnConeD

12th June 2002 04:27 UTC

avs-king's metaballs are actually metacircles though. They're completely 2D... doing the same in 3D is tons more difficult. Almost all the time is spent calculating the 3D isosurface of values that define the balls.

I don't mean to sound cocky, but tons of people have been nagging about the APE's, that they're not authentic and such. An APE might look like an everything-in-one effect, but under the hood it's incredibly more complex than a preset (metaballs is about 1500 lines of code).

dirkdeftly

12th June 2002 07:18 UTC

Meh, I've checked out the code for Sierpinski, and it's really not a "real" IFS; the order of triangle in which a point is placed is randomized for every point. (I don't think that's the way it's done...is it?)

BTW, I've made one of my own, it goes up to some 30 iterations. I found out why mine was so much slower than UnConeD's and Hansen's (is that AVS-King?) - mine goes into a lot more detail than yours ;) I was able to exploit the origin-symmetry, though, so it only takes half as long as expected; it also semi-responds to the music: when a new fractal is started, it calculates the volume level, and the difference of color between two levels of iteration is related to the volume. And I got another couple of cool effects in with it...
I'm not going to post it here, but I will when I'm done with my pack (I'm getting closer!)

UnConeD

12th June 2002 07:31 UTC

Well I think the randomized method of drawing a Sierpinski triangle is the most common way. In any case, how would you draw a shape which has no surface? :rolleyes:

As for the Julia's, usually you can get away with a low amount of iterations. It only gets a lot twists and details when the constant is close to the edge of the Mandelbrot set. The origin symmetry is a nice property for square Julias too... indeed using a movement "x=if(below(y,0),x,-x);y=-abs(y)" you only need to draw half of the image.