Wow, A Simulation That Looks Like Reality! 🤯

Dear Fellow Scholars, this is Two Minute  Papers with Dr. Károly Zsolnai-Fehér. I am so happy to tell you that today, my  dream came true. I can’t believe it. So,   what happened? I am going to show you my new  paper that just came out. This is a short   Comment article published in Nature Physics,  where I had the opportunity to be a bit of an   ambassador for the computer graphics research  community. You see, this is a small field,   but it has so many absolutely amazing papers,  I am really honored for the opportunity to show   them to the entire world. So, let’s marvel  together today as you’ll see some amazing   papers throughout this video. For instance, what  about this one? What are we seeing here? Yes,   what you see here is a physics simulation from a  previous episode that is so detailed, it is almost   indistinguishable from reality. So, what you  are seeing here should be absolutely impossible.

Why is that? Well, to find out, we are going  to look at two different branches of fluid   simulations. Scientists who work in the area  of computational fluid dynamics have been able   to write such simulations for more than 60 years  now, and these are the really rigorous simulations   that can accurately tell how nature works in  hypothetical situations. This is super useful,   for instance, for wind tunnel tests for cars,  testing different aircraft and turbine designs,   and more. However, these simulations can  easily take from days to weeks to compute. Now, here is the other branch. Yes, that  is computer graphics. What is the goal for   computer graphics simulation research? Well,  the goal is exactly what you see here. Yes,   for us graphics people, a simulation does not need  to be perfect. The goal is to create a simulation   that is good enough to fool the human eye. These  are excellent for feature-length movies, virtual   worlds, quick visualizations, and more. And  here is the best part, in return, most computer   graphics techniques don’t run from days to weeks,  they run in a few minutes to a few hours at most. So, computational fluid dynamics, slow but  rigorous, computer graphics, fast but approximate.

These are two completely different branches. Now, of course, any self-respecting Scholar   would now ask, okay, but Károly, how do you  optimize an algorithm that took weeks to run,   to now run in a matter of minutes? Well,  here are two amazing ideas to perform that. One is spatial adaptivity. What does  that do? Well, it does this! Oh yes,   it allocates most of our computational  resources to regions where the real   action happens. You see these more turbulent  regions with dark blue. The orange or yellow   parts indicate calmer regions where we can  get away with less computation. You see,   this is an approximate model, but a really  good one which speeds up the computation   a great deal at a typically small cost. The  hallmark of a good computer graphics paper. Two, in my opinion, this is one of the most  beautiful papers in all of computer graphics,

Surface-Only Liquids. Legendary work. Here,  we lean on the observation that most of the   interesting detail in a fluid simulation  is visible on the surface of the liquids,   so why not concentrate on that. Now, of course,   this is easier said than done, and  this work really pulled it off. But, we are not stopping here. We can also use  this amazing paper to add realistic bubbles to   a simulation. And you will see in a moment that  it greatly enhances the realism of these videos.    Look at that. I absolutely love  it. If the video ends here,   it is because I cannot stop looking  at and running these simulations. Okay, it took a while, but I am back, I have  recovered. So, the key to this work is that   it does not do the full surface-tension  calculations that are typically required   for simulating bubbles, no-no. That would be  prohibitively expensive. It just says that look,   bubbles typically appear in regions where air  gets trapped by the fluid. And it also proposes   an algorithm to find these regions. And the  algorithm is so simple, I could not believe   my eyes. I thought that this was some kind of a  prank. It turns out, we only need to look at the   curvature of the fluid, and only nearby, which  is really cheap, simple and fast. And it really   works! I can’t believe it. Once again, this  is an approximate solution, not the real deal,   but it is extremely simple and fast, and it can  even be added as a post-processing step to a   simulation that is already finished without  the bubbles. How cool is that? Loving it! Now wait, we just said that surface-tension  computations are out of reach. That just costs   too much. Or, does it? Here is an incredible  graphics paper from last year that does just   that. So what is all this surface-tension thing  good for? Well, for instance, it can simulate   this paperclip floating on water. That is quite  remarkable because the density of the paperclip   is 8 times as much as the water itself, and  yet, it still sits on top of the water. We   can also drop a bunch of cherries into water and  milk and get these beautiful simulations. Yes,   these are not real videos, these are all simulated  on a computer. Can you tell the difference? I’d   love to know, let me know in the comments below!   And get this, for simpler scenes, we only need to   wait for a few seconds for each frame. That  is insanity. I told you that there are some   magical works within computer graphics. I am so  happy for the opportunity to share them with you! However, wait a second. Remember, we  said that these graphics works are fast   and approximate. But is this really true?   Can they really hold the candle to these   rigorous computational fluid dynamics  simulations that are super accurate,   but take so long? That is impossible,  right? A quick simulation on our home   computer cannot possibly tell us anything  new about aircraft designs, can it? Well,   hold on to your papers, because it can! And it  not only can, but you are already looking at it! This is a devilishly difficult test, a real  aerodynamic simulation in a wind tunnel. In   these cases, getting really accurate results is  critical. For instance, here we would like to   see that if we were to add a spoiler to this car,  how much of an aerodynamic advantage we would get   in return. Here are the results from the real wind  tunnel test, and now, let’s see how the new method   compares. Wow. Goodness! It is not perfect by any  means, but seems accurate enough that we can see   the wake flow of the car clearly enough so that  we can make an informed decision on that spoiler. Yes, this, and even more is possible with computer  graphics algorithms. These approximate solutions   became so accurate and so fast, that we are seeing  these two seemingly completely different branches,   computational fluid dynamics, and computer  graphics getting closer and closer to each   other. Even just a couple years ago, I did not  dare to think that this would ever be possible.    And yet, these fast and predictive  simulations are within arm’s reach,   and just a couple more papers down the line,  and we might be able to enter a world where   an engineer is able to test new ideas in aircraft  design every few minutes. What a time to be alive! So, all this came about because I am worried that  as the field of computer graphics is so small,

there are some true gems out there, and if we  don’t talk about these works, I am afraid that   almost no one will. And this is why Two Minute  Papers, and this paper came into existence. And   I am so happy to have a Comment paper accepted  to Nature Physics, and now, to be able to show   it to you. By the way, the paper is a quick  read, and you can read it for free through   the link in the video description. I think we are  getting so close to real life-like simulations,   everyone has to hear about it. So cool! If  you wish to help me spread the word about   these incredible works, please consider sharing  this with your friends and tweeting about it. I would also like to send a big thank you for  this amazing opportunity to write this paper,   and to Christopher Batty who sent super  useful comments. So, what do you think?    What would you use these techniques  for? Let me know in the comments below!

Thanks for watching and for your generous  support, and I'll see you next time!