Ferro Fluid software algorithm (Ferrolic clock in sowtware)
category: code [glöplog]
Hello,
I want to make a software version of Ferrolic clock http://www.youtube.com/watch?v=AisJ9PSe1-0 work on chinese wild hardware with 480x272 LCD. This hardware board without GPU, only with 350MHz MIPS CPU. Needed a "light" simple liquid algorithm for slow CPU, maybe from 90' era...
PS: If anyone interesting, my old fun with this board:
http://www.youtube.com/watch?v=kZM_so_r4Pw
http://www.youtube.com/watch?v=n4QLd8uGxgM
http://www.youtube.com/watch?v=yx89zQR92aM
I want to make a software version of Ferrolic clock http://www.youtube.com/watch?v=AisJ9PSe1-0 work on chinese wild hardware with 480x272 LCD. This hardware board without GPU, only with 350MHz MIPS CPU. Needed a "light" simple liquid algorithm for slow CPU, maybe from 90' era...
PS: If anyone interesting, my old fun with this board:
http://www.youtube.com/watch?v=kZM_so_r4Pw
http://www.youtube.com/watch?v=n4QLd8uGxgM
http://www.youtube.com/watch?v=yx89zQR92aM
i guess you can do it with 2d metaballs
Software is as close as most are going to get to that sort of clock. I found it and the price back when it went around the web before... a Nixie clock would be cheaper.
If it isn't fast enough to do it on the fly (real time), which I wouldn't know, you could always precalculate the transitions for numbers individually, on the assumption they're far enough apart they won't interact with each other. However that could get repetitive.
Also this isn't just a liquid simulation. Ferro fluid has some very interesting properties and the clock is influencing the fluid by way of magnetic attraction.
If it isn't fast enough to do it on the fly (real time), which I wouldn't know, you could always precalculate the transitions for numbers individually, on the assumption they're far enough apart they won't interact with each other. However that could get repetitive.
Also this isn't just a liquid simulation. Ferro fluid has some very interesting properties and the clock is influencing the fluid by way of magnetic attraction.
I think it should be possible to do in software if you're clever enough about it. Will be trickier than metaballs. I'm wondering how realistic you can get with just 350MHz.
The trick is too handle every particle locally. So you can try do it with cellular automata. Not sure about the amount of neighbourhoods you'll need, maybe Moore will work, but you can give it a try with Von-neumann first. Traditional 2D metaballs require too much calculation per point. So doing it with local neighbourhoods is alot faster. Maybe you can do it with more "colors", meaning that this point has more states and will actually move faster or slower depending on the curves of their respective neighbour colors.
Uhm. If you can do (at least fake) it in 2D with a 1MHz C64, a 7MHz Atari/Amiga, shouldn't you be able to do it easily with a 350MHz CPU?
Thanks for metaballs hint. First attempt: https://youtu.be/ic8yBpFKefU
how is that a clock?
fell, yes i know. i meant his balls dont really tell time :)
well, an idiot am I!
nice.. it looks a little unrealistic, however.. like the area of combined 'balls' is much too large compared to the individual 'ball' area..
