Welcome to Fractal Forums

Deepest Mandelbrot set zoom to date, but soon to be surpassed

http://www.youtube.com/v/Rzmv2nBCIjw?hd=1

Comparison of scale:
The universe has a volume of 3,5*10^80 m^3 while a planck space is 1,62*10^-35 m^3. So the number of space quantums that fit inside the observable universe is already very large, but now consider the number of spacetime quantums in the observable universe and its entire history too. The universe is 13,8 billion years old, while a plancktime is 5,39*10^-44 sec. That's 1,75*10^174 spacetime quantums. Take this number to the power of 8, then you're getting somewhere close to the final magnification of this zoom video.

You can also download the original video file from either 180upload or rapidshare which is 60 frames per second (youtube allows only 30). You won't miss many details, but it's much smoother and there are no compression artifacts (just the calculation inaccuracy artifacts).
http://180upload.com/540qxtc0wq0y
http://rapidshare.com/files/3126003377/5000_zooms.mp4
Download instructions: download as free user, you want to wait, don't use 180upload's downloader program.

Final magnification:
2^5000
E1500

Music tracklisting:
1. A guy called Gerald - Life Unfolds His Mystery
2. A guy called Gerald - Voodoo Rage

Watch on YouTube:
http://youtu.be/Rzmv2nBCIjw

---

And also finished on the same day: Golden Mandelbrot zoom

http://www.youtube.com/v/49By5R-yprE?hd=1

Golden Mandelbrot zoom

Final magnification:
2^350
2.2934986159900715116108208953021 E105

Music:
Nitrostirumpa - Matrix IV (with minor edit)

Watch on YouTube:
http://youtu.be/49By5R-yprE

great animation!!!!

thank you

Excellet. What is the resolution of the first movie? WHO will ever surpass this...?

The frames were rendered at 800×500 and upscaled to 1600×1000 because youtube gives better bitrate for higher resolution.

But you could easily surpass this with your program that uses the SuperFractalThing rendering method.

Is it with Fractal Extreme, so that the resolution of the result movie is 400x250?
You mentioned somewhere that the time to render this was 2 weeks - and that is just amazing even for a 400x250 movie!
Shouldn't you be able to make a HD movie to the same position in 4 weeks?

Hmm... need to think more about the SFT scaling so that I can beat your record... ;)

Wooooooohaaaaaaaaa I have just experienced warp-speed to the power of a googolplex! :dance: :dance: :dance: :dance:
Makes you think...  O0

Fractal extreme doesn't decrease the resolution. It's just 800×500. A HD render of this is doable, but I have too much else to render.

Both Vids Are Amaizing , 2^5000 i didn't imagine that it was doable at this time ... :o

Golden Mandelbrot zoom, is Wonderfull , each step is a nice painting on itself, that's sad that you don't get rid of color stepping, smooth gradient would be much more nice to see.

I believe fractal extreme make keyframes with double resolution of the result movie.
Anyway, if 800x500 is not considered HD, what is then?

HD is considered to be 1920×1080, but if you take the words "high definition" literally, it doesn't have to be a particular resolution.

 :D
HD or HD ready, 1280 x 720 (720p - progressive or 720i - interlaced)
Full HD, 1920 x 1080 (1080i or 1080p)
4K2K, 4096 x 2160 (2160p)

From what I have read and noticed about fx is that when you create a 1280x720 movie fx creates keyframes with that resolution for every doubled zoom size, then the resulting raw movie is half size, 640x360, and then it is doubled to 1280x720 again. When I create a 1280x720 movie I render the frames in 640x360 and then double it to get the same resolution and quality. Att least I thought I'd get the same...?

I don't know exactly how fractal extreme works, but I'm currently rendering a video at 1280×800. It really doesn't look like the frames it renders are being upscaled.
http://i538.photobucket.com/albums/ff342/formule/1280800.png~original

When you play a fractal extreme 'movie' using the movie player press F5 to 'show polys'.
It illustrates the behaviour better than I can describe.

When you are at an integer zoom level (ie no decimals to number of zooms) the image is shown at native resolution.
As is passes the .00 zoom level the next zoom is faded in at 1/2 resolution in the centre of the frame.
The outer edges are now at a 'digital zoom', as the zoom approaches the next integer (.90) the edges show
more blur.

OK, challenge accepted :)
I have started an e2000 rendering now, we'll see when it is done, it is running on my old 32-bit dualcore laptop, with 640x360 frames (no pixel stretching as Fractal extreme ;)

When my move is done and published on youtube, I will make my program available (don't want you guys with super-hardware to precede me though :D ).
Some features of my program:
 - Free of charges - we are enthusiasts aren't we!?
 - Unlimited pertubation SFT rendering (I finally managed to go beyond the e616 limit of double)!!
     (ok, e10'000 is the limit, however that can be extended)
 - Lossless animations
 - Automatic find of Minibrot - Just select an interesting path half the depth that you want your animation, choose the option "Find Minibrot" and sit back and watch!

Sounds interesting??

That sounds very interesting, asdklfjdf!

Let's see how long it takes for someone to render a zoom to e10000.

What software did you use for this zoom? I heard there are some new improved algorithms?

edit: ok next time I should first read the thread before posting lol

asdklfjdf, what language are you writing your program in? Man this is so exciting, great news about that super fractal thing invention.

Why stop at E10000? So far my experiences with SFT have demonstrated that perturbation will speedily render just about anything that isn't too complex reguardless of depth or iteration count. While it may take weeks to render images of extremely complex deep minibrots, we could have very short rendering times of beautiful stacked shapes. Even if they go all the way down to E20000. Heck, get rid of the limit all together.

Are you aware how deep e10000 is? You will probably have RSI before you get there if you zoom non-stop. Of course no limit is always better, but extending the limit requires extra programming work if I'm not mistaken.

I remember from reading the SuperFractalThing threads that there were some issues with double precision past E308 and arbitrary precision past E616, but asdklfjdf says that he has managed to solve the problem and get arbitrary precision to work. He said that the limit is E10000, but I don't think that the precision stopped working at such a round number. I guess we'll wait until asdklfjdf gets back to this thread.

C++

The rendering with SFT is actually very dependent of the iteration count, and the depth also still affects, since it takes time to calculate the reference point which still requires arbitrary precision. The time processing arbitrary precision arithmetic grows exponentially, and above E2000 the calculation of the reference point takes more than half the time to render the whole image. For an animation, the reference point can be reused and then it only affects the first frame. However, a zoom to E2000 requires some 40000 frames if you don't want it to be too fast and just flickering, so it takes time...

No, the limit E10000 of my program is the limit of the arbitrary precision library, that is still used for the reference point. That is just a defined number specifying the size of the array of integers, so it can be extended to a much larger number.
If Dinkydau and others with super hardware wants to use my program, which I hope, and wants to go even deeper, I can just change the number and compile again :)

Wait, are you saying you are calculating each frame? I thought that zoom movies are made by calculating only a subset (every doubling in size or something like that) and calculating the rest of the frames by intrapolation?

No, I don't calculate every frame completely, but that is the number of frames in the movie.
These frames can be created from every doubled size, as Fractal Extreme does, which still requires many calculated frames, log(1e2000)/log(2)=6644.
However I create my moves as a backward zoom-out, for each frame I only calculate the new revealed pixels at the edge and squeeze the previous frame into the middle. By doing so the center of the images get smooth without use of anti-alias.

Edit: Also, by doing so no pixels are stretched. However the edges of the movies does not eliminate interference patterns, which is visible when getting close to a minibrot - but not too close since then it just shows random noise.

That method of zoom rendering sounds like a very good idea, anti-aliasing where you most need it without any extra render time.

Assuming no technical limit of extending that way, why don't you make it something like e500 000 000? That should be "practically unlimited".

How about you add a parameter in the program that allows the user to set the size of the array of integers, and just pass that value to the arbitrary precision library? That way we aren't using giant E100000 integer arrays for more shallow zooms, but can set it back up to E100000 if we plan to zoom that far. (I wish you luck if you plan to navigate all the way down there.)

My arbitrary precision library has the integer array in a fixed size. The array is on the stack, which makes the code much faster than if it would be allocated on the heap for all variables. There exist other libraries that are capable of billions of decimals, apfloat for example that is used to calculate pi decimals, but the calculation of the reference point would take very much longer time, even for shallow depths. And of course a library that can be used for fractals needs to be able to control the precision and not use more decimals than necessary for each depth.

That was my point, that we don't need huge arrays for calculation of shallower zooms. I see that you have already thought of that.

Best of luck on your E2000 dive!