Two Minute Papers - 2016-06-05
Today we will talk about Eulerian and Lagrangian smoke and fluid simulations and how this excellent technique can incorporate a variant of Schrödinger's equation and make an excellent fluid simulator out of it. :) _________________ The paper "Schrödinger's Smoke" and its implementation is available here: http://multires.caltech.edu/pubs/SchrodingersSmoke.pdf http://multires.caltech.edu/pubs/SchrodingersSmokeCode.zip The publisher's version is expected to show up here soon: http://www.multires.caltech.edu/pubs/pubs.htm The Short Science website is available here: http://www.shortscience.org/ WE WOULD LIKE TO THANK OUR GENEROUS SUPPORTERS WHO MAKE TWO MINUTE PAPERS POSSIBLE: David Jaenisch, Sunil Kim, Julian Josephs. https://www.patreon.com/TwoMinutePapers Subscribe if you would like to see more of these! - http://www.youtube.com/subscription_center?add_user=keeroyz Splash screen/thumbnail design: Felícia Fehér - http://felicia.hu Károly Zsolnai-Fehér's links: Facebook → https://www.facebook.com/TwoMinutePapers/ Twitter → https://twitter.com/karoly_zsolnai Web → https://cg.tuwien.ac.at/~zsolnai/
Your explanations are fantastic:D
Thank you, happy to hear that you're enjoying the series! :)
no offense but they distract
.....it's a great idea but please keep it simple
watch the Utah presentations for SIGGRAPH
One thing I really like about this work is that it very intuitively shows what the Planck's constant does. It's, roughly, a measure of the amount of detail you can get. Smaller constant -> more detail.
Yeah, I found it absolutely crazy that you can actually modify it. At first, it felt like "insert the desired value of pi here". :)
@Károly Zsolnai-Fehér you can, in principle, change all the fundamental constants of nature how ever you please. It's probably the best way to figure out what the constant even does :)
(Of course you can only do that in simulation... Unless you have reality altering super powers :o) )
Wow, that was an absolutely fantastic paper, and video quality and narration were great too! 10/10 huge props to the guys who behind the paper, really phenomenal stuff!
I love your work! Im a multi interest person and love to learn new computational techniques but had no idea how to stay updated. your videos summarizes everything greatly and also provides link to the papers so I can learn in more detail.
Excellent, very happy to hear you're enjoying the series! It's always important to keep an open mind, that's what it's all about! :)
Isn't literally everybody a 'multi-interest' person?
I found your channel today and I am so amazed by it. You explain the content so well and it's easy to hear your own excitement which makes it even more enjoyable.
I study Computational Engineering Science and specialize in fluid mechanics. So this is exactly my field of interest!
Keep up the great work!
Thanks for the kind words, very happy to hear you're enjoying the channel. Welcome to our growing club of Fellow Scholars! :)
This is so obvious. I'm flabbergasted. Incredible. Thank you for posting this, and all of your videos.
Best out there regarding short informative videos. Thank you
Hah I always like it when I've seen stuff before you covered it :D (But I also like your covers of them)
@Mohd Maqbool Alam I'm pretty sure Károly is mostly just browsing for what he finds interesting (plus/minus what ever the Patreon crowd voted on). That's very different from reading papers for the purpose or research: He doesn't have to find very specific obscure papers to get there.
It's easy to get lost in papers. Unfortunately, many are written in quite an impenetrable style. I'd suggest:
- try finding better-written papers (hard and unlikely)
- tackling papers basically one paragraph / proof / subconcept at a time. It's implausible to "get" any given paper the first time through, especially if the covered concepts are fairly new to you. They tend to be super dense in information.
If you are just skimming, trying to figure out what to quote as "related works" or something like that, that's a different subject though. Full understanding is not necessary then. Just check for relevance, basically.
+Kram1032 thanks
It is difficult indeed. However, don't worry too much about it, the more time spend with reading, the more experienced you will get, and before you know it, you'll get the gist of a scientific work in a matter of minutes after opening the paper. Details still take time, though! :) Good luck!
@Károly Zsolnai-Fehér did you have videos where to find interesting research papers to read. what are most popular website for research papers.
thanks
Read the reddits for the topics you're interested in, and maybe try this Short Science website and see if it works for you! :)
Mind blown!
Though I didn't see any mention of the performance gain if any?
I was unable to find hard numbers in the paper. What I have found is spatial grid resolutions that range from 64^3 to 192^3, which are quite favorable numbers. If you pair this with a linearized version of the incompressible Schrödinger-equations, it sounds like a mighty efficient technique. :)
thanks :)
I love your videos!
Thank you, happy to have you around! :)
You are awesome. Actually im not a scientist, but i love science so much. Thank you for this video.
It is indeed a beautiful profession that is very difficult, and equally rewarding to do. :)
according to john taylor gatto, theire is nothing difficult in the world. ;) Theire is no reason to hurt yourself. Just love your passion and just do it :)
by seeing this I remember this tech demo by capcom for the panta rhei: https://www.youtube.com/watch?v=l3jG21MhpKw I really have no idea how it could work in real time
Wow, a lot of interesting videos about fluid simulation from you lately. Sounds to me like if you are going to simulate fluid on an Eulerian grid, you should use the Navier–Stokes equations or any variant of them (like the Euler equations), but I'm gonna give this one a try. I may become surprised. :)
🤟
Never stop doing what you do! Your video's are so inspiring.
I love playing with graphics and mathematics and algorithms as well, and i'm quite good at it. Right now i am super exited about distance fields and all its applications; lights, shadows, fog, clouds, physics, etc..
However i have a very hard time reading papers and obtaining new knowledge via papers. They are filled formulas that are gibberish to me - lots of non-descriptive variable names, and so many symbols i don't understand. I have a very good understating of mathematics though, i feel like it's my superpower - but its like when a person can play the violin well, and play almost anything by ear, but can't read sheets / notes. (Reading the formula as and algorithm in a programming language is much easier)
I obviously need some practice in reading this formula syntax, and i was wondering if you had any advice, or know of a good place to start? :)? Its a vague question i know. Sorry about that.
I understand where you are coming from. My understanding of mathematics, at least, how it translates to images in my mind tends to be quite different than other people's images. My way of trying to be more "compatible" with others is using their formalism, but translating the formalism into my own intuition. Conservative force fields are not integrals, but wind gusts that push a ball in the water around without rotating it. These intuitive translations helped me a great deal.
I am trying to promote this kind of learning in my university lectures where we systematically decompose difficult equations into small, intuitive ideas and suddenly everything becomes easy to understand. Have a look, hope it helps!
https://users.cg.tuwien.ac.at/zsolnai/gfx/rendering-course/
this is channel is gold !
Thanks for watching and happy to hear that you liked it! :)
So the smoke is treated as a wave function instead of multiple particles? Why doesnt this create problems like unwanted wave interference? I guess Id have to read the paper
So, would it be possible to link this to the idea of that we are part of a simulation inside another higher dimension? Just as waves don't behave like particles, the idea of this papers sounds similar to me. (I'm far away of deeply understanding the formulas on the paper)
Usually a more scientific way to approach the problem is first writing up observations that are surely true, and try to make a healthy conclusion building on them. Here, it seems that you have jumped to a conclusion immediately without any prerequisites. I personally do not see a strong parallel between the two, but our intuitions may be wildly different and many good observations come from a good hunch. Who knows! :)
Only thing that stopped me from liking is that at 1:40 you claimed Schödinger equation has absolutely nothing to do with large scale fluid simulation, but you are wrong. Don’t you know of the Madelung equations or its variations? The mass and momentum equations for fluids can be merged using complex numbers and we obtain Schödinger-like complex-valued equations.
So it's alittle like a procedural generation. You won't need to save everything on the creation, just need to load what the formula says it is.
I'm not sure why it should be Schrödinger equation. I guess Fokker-Planck equation (describing classical stochastic process) works better for this purpose. One advantage of using Schrödinger equation (that I can think of) is that the simulation could be easily implemented into quantum computers. But it seems that it is not the original purpose.
I think an important part here is that this is an incompressible Schrödinger-equation variant which makes the simulation stable (you don't have to do the Helmholtz-Hodge decomposition to make the pressure field divergence-free). I'll definitely read up a bit more on this. Loving the concept. :)
Oh my god. Convolutional Neural Network weather prediction.
what platform is the smoke simulated on? thanks!
Jorge C. M. - 2018-01-19
Your weed is burning and producing smoke and not doing it until you see it (and inhale it).
Schrödinger's Smoke