NVIDIA’s Tech: Impossible Water Simulation!

❤️ Check out Lambda here and sign up for their GPU Cloud:

📝 The papers are available here:

📝 My paper on simulations that look almost like reality is available for free here:

Or this is the orig. Nature Physics link with clickable citations:

🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:
Alex Balfanz, Alex Haro, B Shang, Benji Rabhan, Gaston Ingaramo, Gordon Child, John Le, Juan Benet, Kyle Davis, Loyal Alchemist, Lukas Biewald, Martin, Michael Albrecht, Michael Tedder, Owen Skarpness, Richard Sundvall, Taras Bobrovytsky,, Thomas Krcmar, Tybie Fitzhugh, Ueli Gallizzi.
If you wish to appear here or pick up other perks, click here:

My research:
X/Twitter:
Thumbnail design: Felícia Zsolnai-Fehér –

#nvidia

Joe Lilli
 

  • @IAm-n7d says:

    “Thank the papers” 🙏

  • @Mrdastapleton says:

    What a time to be alive!

  • @cocamaster0 says:

    what a time to be alive

  • @vozkolisk63 says:

    Playing minecraft with this water physics would be hella fucking cool

  • @Keatwonobe says:

    That’s insane. I love combining two things and getting a third way better thing!

  • @randomworld-ik7tk says:

    … Depends on if we’re able to ocntrol the viscosity also.

  • @urkururear says:

    What a time to be alive!!

  • @DdZ-lv3jq says:

    I always wonder. Is the paper detailed enough to recreate this demo? Or is it really vague with a nice video attached to it?

    • @itscaptainterry says:

      Leaving a comment so I see if someone replies to this! Curious as well! I feel like, even if you understand the maths, coding up a simulation that uses them correctly must still be a task of its own.

    • @24Carrrot says:

      after a quick glance i think this paper went pretty in depth on the processes and algorithms used to make it work, and they even managed to render it at 40fps on a 2080 mobile GPU, which is crazy imo. definitely worth a full read!

    • @syth-1 says:

      Would be great if they open sourced it if they are willing to share the papers on how they did it,

      A working example means others can implement it into their own projects, even if they don’t entirely understand the technical ‘how’ it works part

    • @teekanne15 says:

      One of the main purposes of publishing papers is the reproductions by other to challenge or enhance your findings.

    • @YOEL_44 says:

      ​@@teekanne15 For nVidia is just a matter of looking good for investors, as in: “last year we published 17 industry leading papers in the fields of computer graphics and simulation”

  • @f4r3l0 says:

    Simulated water always seem like it has no surface tension, too high fluidity

    • @sebastianjost says:

      I assume most experiments shown are on a larger scale, where surface tension is almost negligible

    • @AD34534 says:

      I was thinking the same thing.

    • @MarkHennessyBarrett says:

      More like ethanol than water.

    • @asdfadfafsdfa says:

      you have to build the rules up before you get to something close to reality.

    • @Anthonyg5005 says:

      it depends on the simulation settings and engine. for example I’ve only used blender mantaflow, blender flipped fluids addon, and houdini flip solver. all of them have their own settings for all of those, houdini’s flip being the best looking/fastest I’ve tried so far, which could be because faster = more time to play with settings

  • @TheRedRanger123 says:

    Well to be honest, if the ground is not sealed it can soak up some water, so disappearing water actually looks more realistic in some circumstances.

    • @Anthonyg5005 says:

      it depends on what you want in your scene but overall it’s not a good thing. If you really wanted that, you do have features like particle lifetime to have the particles disappear when you actually want it to

  • @Quast says:

    Man, if I’d get to see that in the next 5 years in any AAA game with a beach and boats… I’d be out words!

  • @GregoryCarnegie says:

    James Cameron just woke up in a cold sweat

  • @BS-gr3bi says:

    What a time to be a paper!

  • @AE_Sub says:

    If they manage to optimize this to run real-time on consumer grade hardware, I can foresee an era of new simulation based games that will inspire the next generation of engineers, designers, and scholars alike!

  • @panzerofthelake4460 says:

    it’s so great. If it’s really done in real time, this will be great for games

  • @napilopez says:

    Navier thought I’d see it happen. Stokes my interest in future papers!

  • @garyhamilton2104 says:

    aRE YOU KIDDING ME THIS IS SICK

  • @mirroredchaos says:

    bringing complex blender animations like realistic water physics over to video game engines in real time would create an entire new era of video games.

    • @shnydercom says:

      2.5D fluid simulations with height maps already work in realtime. There are plug-ins for both unreal and unity making that possible. And there’s a Blender 2.5D fluid Sim that’s entirely made with geometry nodes. I can imagine one or the other game dev is already experimenting with this. Although full 3D simulations are a whole different beast. Exiting times =)

  • @timothywcrane says:

    I love how the algo were not just layered but merged… thanks. It may seem simple to some, but there is a lot of optimization that would come from abstractions over more and more detail in the end.

  • @Insomn says:

    With this technology I would start filling all of my NPCs with real liquid so that when they squish or puncture, the resulting splashes would be numerically perfect simulations of theoretical fluid dynamics.

  • >