2D High-resolution Rayleigh-Taylor Instability
category: gfx [glöplog]
Gotta love the chaotic evolution from a seemingly simple initial state:
http://www.youtube.com/watch?v=P8Cs2tO5w74&feature=youtube_gdata_player
Now, go make a demo about it!
http://www.youtube.com/watch?v=P8Cs2tO5w74&feature=youtube_gdata_player
Now, go make a demo about it!
And for people like me who have no idea what said concept is: http://en.wikipedia.org/wiki/Rayleigh%E2%80%93Taylor_instability.
Looks yummy, prolly would make a good 4k :D
Looks yummy, prolly would make a good 4k :D
yeh... looks cool. kinda like milk in the coffee or "relaxed" cigarette smoke?
It is also an opportunity to show off "code-fu" tricks... from the author comments on the video:
Realtime challenge at that resolution?
Quote:
I used my own software and unpublished algorithm. I don't have the actual performance numbers handy, but I'd guess it took about a day on a quad-core 2.5GHz Phenom. It was probably 4096x4096, and took roughly 1 minute per frame.
Realtime challenge at that resolution?
A challenge eyy? ;) Interesting and what coincidence! Re-registered today after a decade of absence from the demo scene, and the first topic I stumble upon is this one! And it's this type of simulations happens to be my all-time favorite "can't be done" that in my mind just have to be transformed to "done!" :P
This can certainly be done in real-time, I had plans to implement something similar back in 1998, extending upon the algorithm/technique (with its "code-fu" tricks :)) I used in our last demo "Enforcing the Law" where it ran real-time smoke on the Amiga.
Seems difficult to do it in 4k*4k resolution on that platform though hehe ;)
Anyway.. On what hardware? With the specs quoted, you should be able to achieve something very similar at 1920x1080 in real-time. Would be a really cool screen saver indeed! The thing is that you have screen resolution, simulation resolution(s), precision and accurancy - and they're related but may differ. In other words: If you sacrifice the simulation part in favor of high screen res, you get a smooth high resolution result but lacking all the fine details as seen in the video.
..."resolution" is a tricky term when doing this kind of stuff! :)
This can certainly be done in real-time, I had plans to implement something similar back in 1998, extending upon the algorithm/technique (with its "code-fu" tricks :)) I used in our last demo "Enforcing the Law" where it ran real-time smoke on the Amiga.
Seems difficult to do it in 4k*4k resolution on that platform though hehe ;)
Anyway.. On what hardware? With the specs quoted, you should be able to achieve something very similar at 1920x1080 in real-time. Would be a really cool screen saver indeed! The thing is that you have screen resolution, simulation resolution(s), precision and accurancy - and they're related but may differ. In other words: If you sacrifice the simulation part in favor of high screen res, you get a smooth high resolution result but lacking all the fine details as seen in the video.
..."resolution" is a tricky term when doing this kind of stuff! :)
