Rendering a Buddhabrot at 4K and Other Bad Ideas
benedikt-bitterli.meThat video is the coolest thing I've seen for a long time. I feel like I have experienced seeing in 4 dimensions. The bit where there are hundreds of spheres then it converges to a 2d fractal is my favorite bit.
I'm always absolutely amazed by people who have both the considerable technical skill and the deep sense of aesthetics required to produce things like this.
The beautiful thing about fractals, is that everyone can get a start with very basic coding skills and with very minimal maths as well. A basic Mandelbrot or Julia fractal can be done with a handful of lines of code (cue the entrance of the golfers?). It's something pretty much anyone can start playing with.
Yet if you want to, you can spend the rest of your life figuring out shortcuts and hacks and tweaks to make your rendering faster (to let you zoom further or increase resolution) or to produce things like this video.
It's a lot of fun to play with because you get aesthetic results very easily, yet you can continue to add improve on your results as long as you have patience for.
Here's [1] a page where you are only allowed to use as much code as would fit in a Twitter message. There are fractals being created and other cool stuff (third one by "icecuber", ctrl+f). So, not really much code needed for some awesome effects.
I almost read the title as about rendering it in a 4KB binary - realtime fractal rendering is a pretty common effect in the demoscene and definitely doable in that size category. Here are some examples:
https://www.youtube.com/watch?v=0w_xEUoK79o
This guy is very talented, he has several other cool projects on his homepage.
The fluid simulations, like [1], could and should enable so many cool games.
I can't wait to see a revival of Lemmings, or Worms, with this kind of physics.
Wetrix [1] comes to mind. One of my favorite games back in the time.
Then look at his cv and what he has accomplished in so little time.
Amazing, and what a talented and tenacious person. Well worth his efforts from my view!
Browsing the rest of the website is also recommended. And also very depressing.
Why depressing?
Presumably because Benedikt is extremely talented which can create a sense of inferiority and hopelessness for others.
When first watching the video I thought the frames were some kind of combination of Mandelbrot Set images with the Hubble telescope image of the Eagle nebula pillars. From wikipedia - https://encrypted-tbn1.gstatic.com/images?q=tbn:ANd9GcTkrY20...
I thought that was pretty cool but after reading the article the actual way it was done was much more interesting.
If nothing else whatever field this is from has very fancy sounding jargon.
From another post on the site "I researched various different approaches to this problem and implemented the algorithms Conjugate Projected Gradient, Projected Steepest Descent, Projected Gauss-Seidel, Subspace Minimization with Conjugate Gradient, Minimum Residual and Symmetric LQ solvers, as well as Projected Gauss-Seidel with line search using the Armijo rule."
These are almost all algorithms I've either heard about or actually implemented in a graduate level numerical methods course given in a mechanical engineering department. Nothing extremely esoteric but it depends on your background. I've no clue about most classic CS algorithms (dijkstra, algo for tree structures, etc).
I ran Subspace Minimization with Conjugate Gradient on your post and find it satisfactory.
They're all various optimization technique, variations on gradient descent.
> I am not convinced it was worth it, but I'm happy to finally put the project to rest :)
So say we all.