Aurora_Attractor 3 has finally been started!

[aurora]

OK, so the first thing you are asking yourself is where/when did I miss Aurora_Attractor 2 (AA2)? Well you did miss it, completely. It has been the sole playground for me alone as it was written in Matlab and c++, thus never entering LW or any other renderer other then my own OGL rendering engine, the c++ part of the project. Now thanks to certain advances taking place AA3 is starting to shape up and will be available well hmm, that part is not up to me as it depends on other issues.

So what is AA3? Its a beast unlike anything you have seen before. AA1 was just a small set of purely artistic manipulations of a smaller set of strange attractor equations. Its sole purpose make semi-almost-pretty pictures. AA2 was designed to aid me in exploring the dynamics of non-linear equations. Its a mathematical scientific toolset, which happens to produced nicer semi-pretty pictures and a whole heck of alot of mathematical information. AA3 is everything AA1 and AA2 was with even more mathematical tools to study non-linear dynamics. If you want to understand what stability, instability, chaos, period doubling and a ton of other mathematical tools used to study non-liner dynamics, AA3 will provide those options. The mathematical side is for the causal study for those with only a simple math background to the professional post doc researcher in applied mathematics.

So what will it bring for the artist? Really nice and pretty pictures (so I'm promised I'll have the ability to produce one way or another). It will include all the attractors you normally are used to seeing and many, many, many,... more. It will include all kinds of new transformers and introduces remappings. Remappings? No not the ability to shift from one attractor to another in an animation but more mathematical in nature. As in creating Poincare sections and other sections. Meaning there will be a huge host of different graphing options made available. So finally bifurcation maps will be seen in a glory never before seen and oh so much more!!!

So does that mean you still can't shift from one attractor to another, nope that's what I mean that's a whole different type of thing and guess what, yes that will be available too!

Cool, so you are wondering, will there be other cool new animation dynamics? Yep, there sure will be. Winds, collisions, and other dynamic features will be possible. Plus individual points will be allowed to have properties such as gravity, charge, ect. 'Whoa, wait are you tell me there will be n-body dynamics involved?' you are asking yourself. Yep! In fact you want to recreate a spiral galaxy from an elliptical galaxy, or hey what about galactic collisions, black holes sucking up a Wolf-Rayet star through its accretion disk into its infinite mull, or just a solar system. Yep that is now part of AA3 it used to be another toolset of mine which only I have used but is part of the plan for AA3.x.

What else can there possibly be. Ohhh little cricket there is MUCH, MUCH more. So stay tuned as I'll keep you all updated on the progress of its development. Sadly there's little else I can say about the specifics of how or where it will work. I'm sure many of you can take a pretty accurate guess why. In other words you probably won't be playing with it till next year. But trust me the wait will be worth it!


[EDIT]I just answered an Email asking about non-Windoze32 builds for AA_1. Well I have no plans on doing any new builds of AA_1. AA_2 is always being worked on, almost daily for a couple years now thanks to school, but you guys never see AA_2. So what about AA_3? It will come in many flavors. Including, if a friend delivers as promised, a build for you people that have yet to buy a 'real mans machine' quoting two profs of mine .

[Exper]

Great news my friend!

Well, I feel that it'll be compatible with LightWave 9.x only!

[aurora]

Well, I feel that it'll be compatible with LightWave 9.x only!

Ah, uhmmm, think X.x only.
I'm doing a mock up model of the designed setup right now, granted its in Matlab but it will test the over all plan for AA3 as well as segregate the many different components for the CG version.

[aurora]

I've received a few more questions which I can answer. As a note to that if you have tried Emailing me directly the past 3 weeks and I have not responded please PMail me. For some beepen reason the server my Email and site is on has been DEAD, not that I'm bitter, oh wait I AM! Anyways.

And please if you have questions about anything PLEASE PMail me as you a have and will when looking at the rest of this post, see is I'm more then happy to answer them. Sometimes they are private and have to be answered in private. But I'm more then happy to answer them. And those that have asked what AA-3 is being built for. If I have not answered it publicly, assume I can't in private either save for the few that already know the answer. This in and of itself should also answer that question though.

Q - Will AA-3 be free?
A - YES!!! I'm a major believer that tools designed to teach and learn from should be freely available and that one should spend as much time as possible teaching what they can to people that want to learn. Sadly I'm in the learn mode more then teach mode thanks to school but everything in time and cycles.

Q - Will Mandelbrot, Julian, IFS and L-systems finally be added?
A - Yes on the first three and a sort of yes to L-systems. L-systems in CG are used for a huge variety of different things like making trees and other really cool branching artworks. Thus L-systems deserve a treatment all there own. Does that mean I won't work on L-systems? NOOOOO. I surely will just not to the full extent they deserve till a later version as I have a cut off point for V3.0 designed in and a 3.x already part of the 3.0 design phase. I'll cover those special 3.x additions down below.

Q - Have you thought about doing something like Max's Thinking particles?
A - I had not really put much thought into this till I was chatting to a new Robotics prof at CU and I got to thinking that me swarming and other AI solutions I have created would be an extremely interesting tie in to AA-3. It will not be part of V3.0 but I have gone back to my design specs and added the ability for me to create an AI app which will work interactively with AA-3 and beyond. Heck I'll probably start that app before I try to move my fluid solution to something a CG user with normal machines can use.

Q - I've been asked this several times the past few weeks. Hows your fluid simulator coming?
A - Oto-Taca is incredible right now. Going back to school for it has been worth it a million fold beyond what I originally dreamed possible. But Oto-Taca is truly, purely a mathematics/scientific app requiring a hefty machine to run on. Something well beyond normal CG artists and since its not in Cuda running a Tesla will not help. I would love to try it on the new Octane by SGI though. That could prove real interesting indeed.
So to answer what you really want to hear, or actually what you don't want to hear, is that Oto-Taca will not be moved to a CG app anytime soon. In fact I guarantee I won't even think about such an endeavor before Christmas 2010 at the absolute possible soonest.
But no its not a vaporware concept either. Its just trying to figure out how to make the gap between Oto-Taca and CG like I'm doing with AA-2 to AA-3.
These following questions are from people that have seen/used my Matllab V2 version but apply to what V3 is all about.

Q - Out of the gazillion different equations and tools just how much can you move to a non-math based app?
A - As much as possible! Granted I have some non-linear systems that evolve into 20(and some many times that) system of 1st order ODEs. Which numerically solving on a 1000+ node cluster gives wall times of several minutes to 30+ hours. Obviously those systems will not show up in V3 as nobody will be able to calc them on there machines. Well at least for the next couple of years till the new hardware architectures become commercially available and no I don't mean Larabee.

Q - OK so will you still include as many systems as possible that can be ran inside a graphics app with a normal Joe Blow machine?
A - Yes and then some. For example I have over 20 variants of just the Henon and Rossler attractors and that many *2 for the Lorenz family and that's just a tithing of all the new attractors that will be available in V3.

Q - Will all the different mappings be available for each system of equations?
A - Yes and no. I'll provide all the mappings for each system that 'can' be provided. For example for 'true' Poincare mappings not all equations have Poincare functions. Yet if you take the stroboscopic approach to Poincare mappings then yes they can be made available. And of course this applies to other mappings as well.

Q - Then for something like Poincare mappings will the user be allowed to choose the slicing orientations for them?
A - I'm working on that issue right now. Its a tricky thing to do with higher dimensional systems especially if you try and derive the real P function for the system.

Q - So this means that higher dimensional systems will be part of AA-3?
A - Duh! Of course they will as well as tools to find all the different dimensions for a system.

Q - What about introduction of noise and rotations to the equations to create variants?
A - Ahhh, this is a tricky question actually as AA-3 is being designed for a 3D CG/animation system where the concept of noise and rotations are vastly different then those use applied mathematicians, physicists, ChemE's, EE's and ME's, think about so there will be switches to handle one the other and maybe both. Thereby making AA-3 a CG app for artists as well as scientists/engineers (my real vision for what I can provide to the various communities that I love!)

Q - Will we have the ability to add in our own systems?
A - Short answer yes. Will it be apart of release 3.0, I'm doubting it. That will probably be the main goal of a 3.x release as that will require so rather interesting setups to allow for all the different tools I plan on adding and to make it work for both scientific, educational and artistic facets. But is definitely a core design feature behind to the architecture I have been working on.

Q - Will you still have the standard Matlab plotting and the prettyPlotting available and will the speed of prettyPlotting still be so dang slow.
A NO and NO!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
A huge part of switching to a CG app as a base is to be able to create stunning images which have never been possible to the scientific community before. Sure there are some tools like Chaoscope and Apophysis but as we know they provide pretty pictures but not mathematically accurate. AA-3 will bridge the gap on both ends. And being part of a CG app being able to create an image with billions of iters will run as fast as my Matlab prettyPlotting routines with only 10,000 iters.

You just have to get past the calc times. And like AA-2 I will provide methods to save solution files for later rendering and/or continue calcing at later times! So say you are doing a DLA system and you know you can crank through the first 100,000 iters tonight but need your machine for other things in the morning. Sure stop the calcing as you can in V2 and then start back up again the next night or free period.

Q - Does this mean you will finally add a calc scheduler?
A - Its in my design specs to add that feature in as you know its something I have longed to add in. But I'm not sure if it will be in the 3.0 release or part of the custom creation release as it has some interesting issues that need worked out still. Hence why I have not added it to V2.