Transferring Real Honey Into A Simulation 🍯

Transferring Real Honey Into A Simulation 🍯


Dear Fellow Scholars, this is Two Minute Papers
with Dr. Károly Zsolnai-Fehér. It’s time for some fluid simulations again!
Writing fluid simulations is one of the most fun things we can do within computer graphics,
because we can create a virtual scene, add the laws of physics for fluid motion, and
create photorealistic footage with an absolutely incredible amount of detail and realism. Note
that we can do this ourselves, so much so, that for this scene, I ran the fluid and light
simulation myself here at the Two Minute Papers studio, and, on consumer hardware. However, despite this amazing looking footage,
we are not nearly done yet! There is still so much to explore! For instance, a big challenge
these days is trying to simulate fluid-solid interactions. This means that the sand is
allowed to have an effect on the fluid, but at the same time, as the fluid sloshes around,
it also moves the sand particles within. This is what we refer to as two-way coupling. We also know that there are different kinds
of two-way coupling, and only the more advanced ones can correctly simulate how real honey
supports the dipper and there is barely any movement. This may be about the only place
on the internet where we are super happy that nothing at all is happening. However, many of you astute Fellow Scholars
immediately ask, okay, but what kind of honey are we talking about? We can buy tens, if
not hundreds of different kinds of honey at the market. If we don’t know what kind of
honey we are using, how do we know if this simulation is too viscous, or not viscous
enough? Great question! Just to make sure we don’t get lost, viscosity
means the amount of resistance against deformation, therefore, as we go up, you can witness this
kind of resistance increasing. And now, hold on to your papers because this new technique
comes from the same authors as the previous one with the honey dipper, and enables us
to import our real-world honey into our simulation. That sounds like science fiction. Importing
real-world materials into a computer simulation? How is that even possible? Well, with this
solution, all we need to do is point a consumer smartphone camera at the phenomenon and record
it. The proposed technique does all the heavy lifting by first, extracting the silhouette
of the footage, and then, creating a simulation that tries to reproduce this behavior. The
closer it is, the better. However, at first, of course, we don’t know the exact parameters
that would result in this, however, now we have an objective we can work towards. The
goal is to re-run this simulation with different parameter sets, in a way to minimize the difference
between the simulation and reality. This is not just working by trial and error but through
a technique that we refer to as mathematical optimization. As you see, later, the technique was able
to successfully identify the appropriate viscosity parameter. And when evaluating these results,
note that this work does not deal with how things look – for instance, whether the honey
has the proper color or translucency is not the point here. What we are trying to reproduce
is not how it looks, but how it moves. It works on a variety of different fluid types.
I have slowed down some of these videos to make sure we can appreciate together how amazingly
good these estimations are. And we’re not even done yet! If we wish
to, we can even set up a similar scene as the real-world one with our simulation as
a proxy for the real honey or caramel flow. After that, we can perform anything we want
with this virtual piece of fluid, even including putting it into novel scenarios, like this
scene which would otherwise be difficult to control, and quite wasteful, or even creating
the perfect honey dipper experiment. Look at how perfect the symmetry is there down
below! Yum! Normally, in a real-world environment, we cannot pour the honey and apply forces
this accurately, but in a simulation, we can do anything we want! And now, we can also
import the exact kind of materials from our real world repertoire. If you can buy it,
you can simulate it. What a time to be alive! This episode has been supported by Linode.
Linode is the world’s largest independent cloud computing provider. Unlike entry-level
hosting services, Linode gives you full backend access to your server, which is your step
up to powerful, fast, fully configurable cloud computing. Linode also has One-Click Apps
that streamline your ability to deploy websites, personal VPNs, game servers, and more. If
you need something as small as a personal online portfolio, Linode has your back, and
if you need to manage tons of client’s websites and reliably serve them to millions of visitors,
Linode can do that too. What’s more, they offer affordable GPU instances featuring the
Quadro RTX 6000 which is tailor-made for AI, scientific computing and computer graphics
projects. If only I had access to a tool like this while I was working on my last few papers!
To receive $20 in credit on your new Linode account, visit linode.com/papers or click
the link in the description and give it a try today! Our thanks to Linode for supporting
the series and helping us make better videos for you. Thanks for watching and for your generous
support, and I’ll see you next time!

100 thoughts on “Transferring Real Honey Into A Simulation 🍯”

  1. Two more papers down the line and they will be able to simulate the air bubbles too. Thanks for the upload.

  2. I'm pretty sure honey simulation like this will mostly be used for jizz in erotic videogames.
    You know that's where the future is, everybody can tell

  3. Something i noticed , and I might be wrong about it, it seems to generate a simulation with more surface tension interacting with itself than with the world. When it pours on the paper (or whatever that is that looks like paper from here) the real world one has borders/edges that are less rounded , since it is adhering to the paper and the simulation has very rounded edges, looking like its not interacting with the environment as much as real world. I still find this outstanding news, but i wanted to hear if you see the same thing i see and if you do, how far do you think we are to solving that problem?

  4. Okay this is getting REALLY intense. Wow!! Thank you so much for sharing this with us, Two Minute Papers. 🙂 This is honestly mind-blowing.

  5. Could this approach be used to take physical measurements at a distance, eg monitoring fluid viscosity in a continuous flow process?

  6. The next 10 years will be absolutely amazing when it comes to realistic graphics in games …

    If humanity still exists so long

  7. One of the differences I am seeing is the difference in the flow fronts of the fluids from measured to simulated. It seems like the viscosity is being modeled well, but maybe the surface tension of the two contacting bodies need to be added, or modeled. I am perhaps over thinking the simulation.

  8. wow amazing video. math teachers need to show this to highschool students when they ask them why do they need to learn math

  9. Do you think that if you allowed the bot second and third perspectives simultaneously it would be able to recognize that there is an external source of light projecting a reflection onto the liquid and perhaps speeding up how quickly it can render a correct simulation? What if eventually it controls the zoom/focus functions of the camera to better view the image youre asking it to duplicate‽

  10. What if scientists taught them how to simulate mirrors and we were confronted with a reflection of who they believe themselves to be?

  11. I've watched a lot of Two Minute Papers' videos and I can say…

    …that I can pronounce his name, almost

  12. your content is just so well done, at this point I wouldn't even be annoyed if I saw a raid shadow legends ad on this channel

  13. Doc, i have a question, it happens that i admire you a lot, and i am finishing school, so i need to pick a career, the point is that i want to know what you studied because i am interested

  14. One of the things I learned from your cool video: yum! ^^
    Thanks for your interesting video. 🙂 I like watching them even if I only know a few basics of programming.

  15. Seems like they need to work on fluid tension, a lot of the liquids look like they are heading up even in large amounts

  16. Viscosity is not enough, you also need surface tension modeling – one can clearly see that the modeled honey spreads out with proper speed but is too flat.
    BTW viscosity is related to relative sizes of particles that constitute the substance. Might be useful if the simulation uses model based on particles.

  17. The only thing I disliked in this video was the title, particularly the emoji in it. In other words: Great video!

  18. Hey, Dr Zsolani-Feher, I'd like to give a suggestion as an experimental fluid dynamicist. Though it's important to praise the work of computational fluid mechanicists, I'm sure you're aware of the limitations. It would be as good to see your criticism and your comments on the challenges faced by this bleeding edge simulations.

    As an experimental fluid dynamicist I think we are much farther behind than you make it appear in your videos (and I'm sure you know it). It's definitely exciting times that we can simulate those things and make it visually appealing, but we're not making useful engineering predictions in many cases. In my experience, the simulated flows look convincing and quite beautiful, but the underlying data is not accurate – to the point where designing a critical system is still not possible without using millions of supercomputer hours. In a sense, that's why we (experimentalists) still have a job – because simulations are quite far behind.

    In any case, I think many would appreciate a more balanced commentary on the papers, also showing the shortcomings of the techniques.

Leave a Reply

Your email address will not be published. Required fields are marked *