Welcome to Fractal Forums

Here: http://www.cs.uaf.edu/~olawlor/2011/gpuifs/

Awesome! This is quite a significant leap in speed over previous implementations. I get about 10 FPS on my laptop here.

I've been wondering how to do this. Yey!

It's great but the interface is ... horrible :dink: - almost everything is missing to make the program usable

well it's open source - so what are you waiting for ;)

That's just a kind of prototype to illustrate the method. It's an academic work. Too bad it can't (well in theory it's possible but ...) to use it as preview for existing flame fractal softwares. The main drawback of this method is that the blurring is not avoidable for large fractals and/or when zooming in. Moreover past 8kx8k images one have apparently to use doubles. Nevertheless, It is impressive, clever and the article is surprisingly eazy to read (for an academic work  ;D).

Interesting paper, looks like they're trying to solve the inverse problem for a very few (just 3?) xforms, to avoid doing the random access. Not a good idea IMO.

Also, they didn't compare against probably the best GPU fractal renderer out there at the moment, cuburn: https://github.com/stevenrobertson/cuburn

Have you folks seen this short video from this fellow?
I wonder if there's something in what he's done that could be applied to flames to get them more 3D then currently they are.  I don't have enough of a technical grasp of flames or ifs to even guess.

http://www.youtube.com/watch?v=ZKDGgOxX9HU

Hi guys, this is my first post here.

I've spent the last 9 months on and off re-writing the implementation from Professor Orion Lawlor. I started a Google code project page here:

http://code.google.com/p/orion-flames/

I took a special interest in his project because it was the only one I knew of that rendered in real-time. Also, it did so in a very clever way by using the inverses of the variation functions.

After getting about midway through it, I realized there were others that rendered in real-time by doing a few tricks. Only recently did I discover Cuburn, which I haven't ran yet.

I also realized that it will never be a complete replacement for the original flam3 algorithm. Still, I think this implementation is worth keeping around and improving because it can render a limited number of scenarios very fast at reasonable quality.

The GUI on my project is not much more than was present on the original, however I plan to change that to make it easier to play with xforms/variations just like all of the other major fractal flames editors.

One note about mine is that the flame rendering engine is completely separate from the GUI, meaning it can be used in other projects. I spent a ton of time untangling the original from the GUI with that exact goal in mind. The reason is that most flame implementations are tightly coupled to the GUI they run in, so they can't be used elsewhere. My hope is that by fully decoupling them, it can eventually get used in places like music visualizers and video games.

Again, the project is in it's infancy so please don't laugh or get upset if it's not the nicest. There is a Windows executable available for download at:

http://code.google.com/p/orion-flames/downloads/list

I probably won't be doing much work on the project for at least a couple of months. I will be spending my time trying to understand the original flam3 code better, as well as several of the other renderers out there.

Feedback is welcome.

Do you have any pictures? So we can decide whether to download or not.

Nice!!! I can save the result as image sequence?!

program does not work for me ... :( by clicking on start ifs it hangs, i have an amd gpu

Hey guys, thanks for the replies. Here are some answers:

No pictures, it's a small enough download. Just get it and play with it.

No capability for saving at the moment.

Sorry to hear it doesn't work on your AMD card. That really sucks. I'm not sure I can be of much help because I only have nVidia cards both at home and at work, so there's no way I can debug to see what the problem is. (I had a nightmare scenario once with an AMD card being faulty right out of the box and swore I'd never buy one again). I don't believe I have anything nVidia specific in there, just regular OpenGL stuff. Perhaps someone who has  an AMD card can try building and running the source on their system and let us know what happens?


A further note regarding saving pictures/video:

The point of this program is to help re-think how we think about fractal flames. For the past 10+ years, everyone's mind has been calibrated to follow these steps:

1) Come up with a bunch of parameters.
2) Hit render button.
3) Wait a few hours, or maybe a few days.
4) Out pops a very visually stunning image.

My aim is to move from that mode, to a mode of real-time visualization, similar to Milkdrop or Magnetosphere. Historically, flames have been way, way too computationally intensive to even think about rendering in real-time. However, there have been enough software and hardware improvements which now make it possible to render a few scenarios in real-time at perhaps less quality. It's a trade-off that I think is well worth it. Again, this is just a prototype with plenty of obvious rendering bugs that I'm sure you've seen, such as the screen blanking out randomly. I hope to make it better in the future.

Thanks for checking it out!

You guys might be interested to know that there is a Web version using this
fractal flames algorithm: The link to the online version can be found in this youtube video:  http://youtu.be/lFPJD7tDWN4

Wow, very cool! Good job! So glad to see someone pick this up.

This looks really promising.

Are there any updates on this method ?

mfeemster : your executable crashes on Win 10 when starting IFS.. :(

I only worked on that project very briefly as my intro to fractal flames. I quickly switched gears and made a program using the standard method of rendering fractal flames, but on the GPU. You can see it here: www.fractorium.com. I also post update notices here: http://mfeemster.deviantart.com/

I currently have builds for Windows and Linux, but will be releasing a new build soon that has new features, bug fixes, and an installer for Mac.

If you are still interested in the inverse method for fractal flames, please see the earlier posts on this thread where someone implemented it in a web browser. That will be a much better experience than whatever I wrote 5 years ago. Sadly, that method never really seemed to take off.