Flying around in a Mandelbox - 4 mins 58 secs - 720p

[lukesleftleg]

Think I'm beginning to get the hang of this now.
I still need to get my head around optimising for YouTube though.

<a href="http://www.youtube.com/v/5df7VqGXI3I&rel=1&fs=1&hd=1" target="_blank">http://www.youtube.com/v/5df7VqGXI3I&rel=1&fs=1&hd=1</a>

[mclarekin]

Yep, you surely gained  your fractal navigation spacecraft licence. 

[lukesleftleg]

Well, thank you very much indeed.
And coming from one with such an extensive and impressive presence in this gallery, I take it as even more of a compliment.

I think I still need to work on my cornering, and I haven't quite got the feel for braking distances just yet, but it's coming.

I guess I'd better get started on the next one then, although I'm just waiting for another project to finish rendering out.
Not an actual fractal flight this time, but still highly relevant to the Mandelbox.
More news as it breaks.

[vinz]

thanks for the visit , there are some cool moments  

[BadCRC]

Real nice flight.  Did you keep a static zoom through most of the video?

[lukesleftleg]

Real nice flight.  Did you keep a static zoom through most of the video?

Thank you very much. 

Yes, I did, or at least tried to. There's a few moments where I had to change speed though, and I'm not entirely happy with these. I still need to work out a decent numerical sequence for speed up/slow down events, but hey, it's all part of the learning process.

I should have used a better quality of Ambient Occlusion as well, but it took several weeks to render out as it was.

I need a renderfarm so badly.
Maybe I'll try and win the lottery. You only need to get the numbers right, how hard can it be?

[cKleinhuis]

look into log-zoom, somewhere in the help from ultrafractal it is described, you need to slow down the farther you zoom, just think of what does a scale 2 make with your original area, (it quadruples it ) and likewise later on, you need to adjust that scaling value, but i am afraid that i have no direct link for it ... if you do not do so it is getting faster and faster at the end!

[taurus]

log-zoom or not, I recommend to leave the zoom value untouched after the initial approach. zooming in mostly works fine, but controlling a zoom-out while moving forward is next to impossible in mandelbulber.

Acceleration curves are not so difficult to make. Simplified I draw a desired smooth accel curve in Blender (from 0,0 to 100,100 or 1000,1000), then I let blender intersect it along x in say 3,4 or 5 pieces. the y coordinate of the intersection points are usable as percentage values.
take them to a spreadsheet to automatically calculate coordinates - usually I take a starting- and an end point or a start point and a first step to calculate values for my absolute movement. Unfortunately I postboned my latest animation, so I can't show an example, but it worked quite nice so far.
Creating the spreadsheet is a bit of work, but needs only few math skills and gives quick and reliable results.

[mclarekin]

I do similar to Taurus66 but not as skilfully, and more of a rough approximation backwards method.

I produce a reduced resolution draft of the animation, and note the areas that require fixing for speed, frame rate or positional problems.

At times I lazily  navigate  mainly  in two planes (with relatively small movement in the third plane)  to make the graph analysis easier.

I tediously cut-paste  the keyframe  co-ordinates from these problem areas, into Excel,    I then adjust the graphed co-ordinates by eye to approximate a transition spiral between the sections of keyframes that don't need fixing i.e. logarithmic looking curves.

I also  have needed to do similar for some smooth flowing parameter changes.

[lukesleftleg]

Thanks very much for the hint, cKleinhuis. I didn't really pay much attention at school, and only realised I was actually capable of doing maths relatively recently, but I'll certainly be taking a look at the log function.

I've been playing around with some easing functions, which I found on Robert Penner's site here, mostly for my next project (which I think is almost ready to announce), but I used a simple Sine wave function at the end of this one, just for the tween near the end where the Mandelbox changes scale (oh, and right at the end, on the camera lens).

But spreadsheets! Of course! I don't know why I didn't think of that. There I was, loading a couple of hundred files into a text editor and trying to do complex search and replace with macros and stuff, and trying to plan out this Java program which was going to do a whole ton of stuff, and it didn't even cross my mind to load up Excel. Doh!

So thanks very much mclarekin and taurus.

In fact, isn't it possible to program Excel? I know it can run macros at the very least. I wonder if those easing functions can be applied somehow?
Hmm, it's too late in the day now. If I start asking questions like that, I'll never sleep.
I'll look into it in the morning.

Thanks again for the replies and food for thought guys.

[cKleinhuis]

the thing with the log zoom is called exponential interpolation, here is a copy of the text from the ultrafractal help file,
i think it explains very good what the problem is, which is the reason i use it here, its a pitty that the actual formula
for it is not included ;(

For floating-point parameters, Ultra Fractal offers an additional interpolation setting called exponential interpolation. This can be set independently of the normal interpolation methods in the Timeline tool window.

Exponential interpolation should be used for parameters that are exponential in nature. This means that in order to experience the same apparent increase of the parameter, you need to double it each time, instead of adding something.

A perfect example is the Magnification parameter on the Location tab. It starts at 1. If you add 1 to make it 2, the fractal is magnified by a factor of two. If you add 1 again, the fractal is magnified by only a factor of 1.5. If you keep adding 1, the apparent difference gets less and less. At a magnification of 1000, you will probably not notice it if you change the magnification to 1001. In contrast, if you keep multiplying the magnification with two, you will each time experience the same zoom effect. Clearly, the Magnification parameter is exponential in nature. In contrast, the Rotation Angle parameter is linear in nature.

This is important when interpolating animations, because exponential parameters must also be interpolated exponentially. For example, if you create an animation of 99 frames where Magnification ranges from 1 at frame 1 to 16 at frame 99, it must be 4 at frame 50 to give the effect of a gradually increasing zooming level. With normal interpolation, it will be 8.5. Fortunately, the Magnification parameter uses exponential interpolation by default.

the easing you have done with the sine function has to stay intact, otherwise you would have an abrupt ending,
you can as well search for the "smoothstep" which is the commonly used method in computer graphics, but the sine
function works well too
http://en.wikipedia.org/wiki/Smoothstep

[lukesleftleg]

Aha! Yep, that makes sense, but I'm going to have to let it sink in for a bit. See what my brain comes back with after a day or two of mulling it over.

And I hadn't come across the smoothstep algorithm before. Very handy indeed.

Thanks very much CK.

[Fracturbator]

Nice animation - well done