I have 40 years in computer programming and graphics, 25 in 3D animation and yet virtually every day I see truly jaw dropping work appearing here! What we need is more of these research papers making it into products like Blender
How hard do you think it is for a normal person to take whats in these papers and put it into blender? You think its possible for people for you? Or normal people? Or people who have some knowledge or just the person who made the paper? Sorry for the over load of questions I’m just curious what you think
Yeah, I really want to see a simulation like this and sound simulation like that other video in blender, although the biggest issue I have with blender is it doesn't handle large dimensions very well, things break after not many kilometers in the viewport.
@@MrLordofgames it should be possible for every person who can read the mathematical formulas good enough to be able to transfer them into a programming language like Python or C. I'm pretty sure I can write the code, but I doubt I'll be able to read the formulas correctly. Edit: After checking the paper, which is suprisingly short, I think this is more easy than I thought. :o It's pretty well described.
Thank you so much for making Two Minute Papers! Makes it easier for us gamedevs to keep track of the next gen tech that will be comming! :) Can be difficult to keep track of Siggraph papers!
Dr. Zholnai-Feher skimmed over it, but the car was run just 7.5% slower than real time. Real-time interactive physics simulation is only a few papers away!
@@bravebee1272 Well you could run it on a PC with two AMD Epyc 7742 CPUs. That would be a 128 core PC, but it doesn't matter because this is probably running mostly on a GPU.
Not really since I'm guessing that they both use similar simulation techniques and BeamNG already is pretty realistic and has better performance than this simulation (< 0.1 ms average physics calculation time per frame for one car in Gridmap). If BeamNG wanted to make even more realistic deformation physics, all they would have to do is to make the cars out of more nodes and beams, although that would come at the cost of a slower simulation. But I think BeamNG's main goal is balancing performance and quality of the simulation.
Beamng goes beyond that (with much simplified physics engine), it takes into account clutch, breaks and tires temperature, breaks fading, pneumatic pistons in the suspension, etc. It will take much more effort to make a general physics engine that won't take shortcuts for all these things along with curved body interactions
@@kerimikkelson7746 To be fair simulations are math and math are simulations of reality so you do indeed have exams on simulations, like simulating what train will arrive first if one starts at x and the other at y at an acceleration of z and a.
This could have amazing applications in the virtual reality space, allowing for more accurate real-time physical interactions between objects and players
@@yudoball i wanna have a sword fight in vr but the problem is you can't really physically fight them,swords would just stay in place if 2 people hit and stuff,vr is far from being real enough but it's still fun
@@fitmotheyap It's really lacking forcefeedback, if you could change the weights and make controllers hard to turn it would be a drastic improvement over todays.
@@incription "Real-time" isn't a hard and fast line, or even really a useful descriptor. Any simulation could be executed in "real-time" if the thing being simulated is simple enough. The compute times for the different simulations in this paper span two orders of magnitude. If it was only the fastest of those simulations which can run at real-time, that would mean that the car simulation would run at ~100 seconds per frame. Don't be a dick.
"We dont even need a coffee break, we dont even need to wait at all" oh my god the rendering scene is about to change forever, and gonna get overworked to death lol
@@carnap355 there is real soft body physics in beam ng. Beam ng does simulate the physical properties of the mesh ( metal in this case ) and how it should react when reforming. There are stronger parts and weaker parts, attachment points and all the like. You can tell this through the fact both simulations have the same tyre mechanics. The rubber is simulated to deform differently like metal, to plastic, to glass etc.
@@carnap355 From what I've read, the simulation in this paper is done by constraining primitives and the mesh is bound to them. And in BeamNG, constraints are beams, primitives are triangles and visible mesh are bound to them. There is no major difference in that point.
i will miss those "classic" sounds that you hear over and over again in games and movies.. there are so many movies where people draw a pistol and i get flashbacks to trophy hunter...
That's what people thought about physics a few decades ago, and here we are today with half the assets in games still being hand animated. That's the thing about videogames, they're not pure engineering products. A lot of artistry go into them which is why some things will keep being hand-made for a long, long while. When audio designers want something to sound exactly right, it's often easier to record audio than to coerce an algorithmic engine into producing the right stuff. Procedural audio will be a useful tool for sure, but I wouldn't expect it to replace traditional foley any time soon.
Honestly, Dirt Rally 1+2 are so insanely close, simulating it at this stage is just a fun exercise more than anything else. I'd rather we'd have some kind of vestibular stimulation device allowing us to completely feel like we're driving a car with all the forces happening.
The RC-car simultion is unreal. I've had many hobbygrade cars in the past years, and this simulation is incredibly close to a real life 1:10 of 1:12 scale buggy (although without much or thin oil in the shock absorbers). Absolutely mindblowing
They do it already, just with slower algorithms on supercomputers. The cost of a few dozen crashed cars can buy you a decent amount of computing power.
Car manufacturers simulate using algorithmic methods that aim to model real structural physics and fluid dynamics. These "simulations" are geared towards getting close to physics, without needing a lot of compute. If you tried to use them to predict real physical systems, they would fail badly. There is a reason that fluid dynamics has not been replaced by smooth-particle-hydrodynamics and structural simulations still are not run on GPU's. And that is that when you want to do real physics, you can't parallelize it.
Looney I have, and it’s pretty great, but this seems more general purpose and potentially more robust, and I just want this kind of simulation in more games, not just niche titles. But also more niche titles. Whenever I see tech demos like this, I just want awesome physics simulation sandbox type games where I can just be a god of tearing food apart or punching physically accurate holes through objects and all that fun stuff XD
Wow! I'm speechless! I hope to one day be smart and learned enough to understand these papers - maybe implement and play around. And, maybe, contribute in some shape or form! Such a wonderful thing for the research to be made available for free, and the author of this video to share this with us! I have got be able to able to write my own papers to hold on to them! What a time to be alive!
DANG the recent advances in this area are insane. The other day you did that paper about super detailed liquid/solid-coupled simulation, and now this! Maybe one day we really will get a real time simulation of all of everyday life physics interacting in arbitrary ways!
Not this kind of deformation. BeamNG's is rigid deformation, this has soft deformation. Edit: My original comment was incorrect. BeamNG does have softbody deformation physics.
Imagine an RPG like Bethesda games with realistic physics such as this, rather than twitching ghost corpses, dinner settings that vibrate, bookshelves which explode if you look at them wrong and animals which clip through the ground.
Folks, if you were there, Do you remember waiting for those soft body simulations in 3dsmax? it took ages XD Coffee, coffe, more coffee . omg the bad old days
0:54 well there is a game called beamng.drive that can do that in real time, its amazing the things you can do in the game like deformations when you crash( sorry for my shiti english ) regards from Catalonia and stay safe :)
I think the exact same thing for 70% of the stuff on this channel. Though it's not an easy task for anyone to implement things into software, let alone some of the more advanced findings. I wish I was smart enough to understand just 30% of it, or else I'd try to do it.
They got so involved in programming deformation, they forgot to program the oil filled shocks, instead of just springs by themselves. Cars aren’t that bouncy guys.
@@skrimper no its not beamng is a soft body physics engine made in 2013!! SEVEN YEARS BEFORE THIS, and now everybodys so stupid and think beam is "JuSt LikE AnY OThER CAr GaME" and its stupid beam has done this, way better, with higher fps, and better physics LONG before this stupid shit hes acting like hes never seen a tier model before, idiots.
I believe this is just a soft body physics simulator just like BeamNG. And BeamNG probably uses the same simulation techniques that this simulation uses, but BeamNG actually has better performance than this simulation (< 0.1 ms average physics calculation time per frame for one car in Gridmap). Although that could be due to BeamNG using fewer nodes and beams to simulate the cars than this simulation does but I'm not sure about the key differences between the two simulators.
@@sgbench actually they are superior,assuming this is done on a computer that runs beam at even 30 fps,why am i saying this you may ask? beamng literally simulates every single part of a car and a good enough rig can run it at 105 fps on the most intensive map.all while simulating almost every aspect of a car on max graphics.
@@Lavender34124 as if running into a wall at 500kmh wouldnt make stuff clip into eachother.come on you are making me laugh,if it went fast enough i wouldnt doubt it'd clip the car and break it
Holy crap, I'm almost crying. This is unbelievable and I need to get this to Unity asap. Also, now is DEFINITELY the time for a new Lego Racers game; I've waiting forever, someone should make it happen!
As an automotive engineer I can not even imagine how useful this algorithms can be in my field for vehicle dynamic simulation. It usually takes up to minutes to perform even a simple steady state simulation. This on the other hand...Oh my god I'm getting goosebumps
Amazing! I come from a content creation background and have a love and hate relationship with dynamic simulations... This looks really promising... Love your channel btw!
2:20 I do not understand this at all? Are these simulations being done in blender? I've done a similiar one and never had that 'old technique' or whatever',s problem, I don't understand.
Holy grail! Just calmed down since the series about a liquid solver, that respects medium boundaries, thus being able to simulate air effects of all kinds... How great could it improve PhoenixFD, for example! And now the physics solver comes into play, that could not just give a huge kick in speed and accuracy of TyFlow simulations, but turn MadCar into a realtime BeamNG experience, but many times more physically correct! What a time to have all of my physical sandboxes experience!
So awesome! Can't believe it's that fast with that much geometry! They could make the tire pop or something; racing games will never be the same after this!
4:12 Being a person who has minimal knowledge of this, hearing happy noises from Károly makes me know that this is something to marvel at. Really amazing!
Now imagine games using such an davanced physics in VR games using an haptic glove...closest thing to being a real thing yet non existent object simulated in VR.
What _a _*_time_* to _be _*_alive_* & *_able_* to _see _*_reality_* if we are _first _*_willing_*_ enough_ to & _ask _*_hard questions_* that *_challenge_* our _current _*_beliefs_* as _often as _*_necessarily possible_* !
![The moment we stopped understanding AI [AlexNet]](http://i.ytimg.com/vi/UZDiGooFs54/mqdefault.jpg)
I wasn't holding on to my papers tight enough and they went everywhere.
Hmmm
I know right, it happend to me too
I wasn't holding on to my papers and I ended up flying away.
I blasted my papers all over my pants.
Yeah when mine hit the wall they crinkled and deformed quite realistically
I have 40 years in computer programming and graphics, 25 in 3D animation and yet virtually every day I see truly jaw dropping work appearing here! What we need is more of these research papers making it into products like Blender
Yes
How hard do you think it is for a normal person to take whats in these papers and put it into blender? You think its possible for people for you? Or normal people? Or people who have some knowledge or just the person who made the paper? Sorry for the over load of questions I’m just curious what you think
Yeah, I really want to see a simulation like this and sound simulation like that other video in blender, although the biggest issue I have with blender is it doesn't handle large dimensions very well, things break after not many kilometers in the viewport.
I was going to say that , blender is a pretty fertile environment to try these out
@@MrLordofgames it should be possible for every person who can read the mathematical formulas good enough to be able to transfer them into a programming language like Python or C. I'm pretty sure I can write the code, but I doubt I'll be able to read the formulas correctly. Edit: After checking the paper, which is suprisingly short, I think this is more easy than I thought. :o It's pretty well described.
who is this "et al" and why is he so smart?
This ET guy is really smart, his biopìc movie doesn't make him justice, and best lets not talk about their videogame....
He's so smart that he has made academical papers in every branch of science. What a madlad!
made me giggle lol
@@martiddy ikr
He seems like he's in every single paper I read!
this deforms my brain in real time
in 60fps?
@@holywizard6945 120 FPS, My brain deforms in high refresh rate
this deforms my brain in real time at nearly two thousand FPS
this simulated my brain turning into jelly in real time
@@aljon5947 and throw armadillo at it
Jeez! The tires compressing is crazy! I was not expecting that part.
Is there a discord community or something that you're in where you got access to this early? Or is it a patreon or something?
@@aydensmith8859 yeah you press the "join" button by the "subscribe" button for many perks including early access
@@aydensmith8859 two minute papers discord community is for all. Its link is available in the channel.
BeamNG Drive anyone?
@@me.unpredictable280 where is this discord link ?
This is amazing. I love seeing the papers you share. I'd have no idea about these amazing things otherwise.
I really love these clips. So cool!
"at the end of the video, I tell you how much time it will take to simulate this"
*Looks at video title*
well..
Yeah, the title is kinda spoiler. It reminds me of the episodes titles from Dragon Ball haha
Well, he did only tell us the actual time at the end - the title only gave away the order of magnitude.
NO SPOILERS
Not only did the title have a spoiler, the thumbnail had one too! Specifically, the spoiler on the back of the car.
Thank you so much for making Two Minute Papers! Makes it easier for us gamedevs to keep track of the next gen tech that will be comming! :) Can be difficult to keep track of Siggraph papers!
You are very kind, thank you! There is no such thing as too many papers, especially when it comes to SIGGRAPH! 👌
@@TwoMinutePapers You are most definitely right about that!
OH MY GOD THAT COMPUTE TIME THOUGH- this is crazy!
Dr. Zholnai-Feher skimmed over it, but the car was run just 7.5% slower than real time. Real-time interactive physics simulation is only a few papers away!
@@Zeuskabob1 Those results depend alot on the hardware they are using
what if they used 128 core cpu.wkwk
@@bravebee1272 they are running it on a gpu most likely as nvidia is involved in the project
@@bravebee1272 Well you could run it on a PC with two AMD Epyc 7742 CPUs. That would be a 128 core PC, but it doesn't matter because this is probably running mostly on a GPU.
Beamng drive is gonna improve so much after this
unlikely they'll implement it
@@descai10 yeah, and the performance is already shit
Not really since I'm guessing that they both use similar simulation techniques and BeamNG already is pretty realistic and has better performance than this simulation (< 0.1 ms average physics calculation time per frame for one car in Gridmap). If BeamNG wanted to make even more realistic deformation physics, all they would have to do is to make the cars out of more nodes and beams, although that would come at the cost of a slower simulation. But I think BeamNG's main goal is balancing performance and quality of the simulation.
Beamng goes beyond that (with much simplified physics engine), it takes into account clutch, breaks and tires temperature, breaks fading, pneumatic pistons in the suspension, etc. It will take much more effort to make a general physics engine that won't take shortcuts for all these things along with curved body interactions
this does not look to groundbreaking here. beam has all of that and more
"Dear fellow scholars"
Me who's watching this when I'm supposed to be in school: Uh... yes..
E
I mean as long as you pass the exam ... no need to waste more time than needed
@@lucienledune1077 I wish I had exams on simulations. I only have exams on plants and math and the middle ages
@@Permafrost107 EE
@@kerimikkelson7746 To be fair simulations are math and math are simulations of reality so you do indeed have exams on simulations, like simulating what train will arrive first if one starts at x and the other at y at an acceleration of z and a.
This could have amazing applications in the virtual reality space, allowing for more accurate real-time physical interactions between objects and players
nice observation
Great idea. Maybe its just me but i want to 1v1 someone in VR mma style
@@yudoball i wanna have a sword fight in vr but the problem is you can't really physically fight them,swords would just stay in place if 2 people hit and stuff,vr is far from being real enough but it's still fun
@@fitmotheyap It's really lacking forcefeedback, if you could change the weights and make controllers hard to turn it would be a drastic improvement over todays.
I guessed that the compute time would be around 5-10 seconds a frame, and I thought I was being greedy. Just milliseconds!? That's crazy!
what do you think real time means?? lol
@@incription "Real-time" isn't a hard and fast line, or even really a useful descriptor. Any simulation could be executed in "real-time" if the thing being simulated is simple enough. The compute times for the different simulations in this paper span two orders of magnitude. If it was only the fastest of those simulations which can run at real-time, that would mean that the car simulation would run at ~100 seconds per frame.
Don't be a dick.
"We dont even need a coffee break, we dont even need to wait at all" oh my god the rendering scene is about to change forever, and gonna get overworked to death lol
Paper: shows some bouncy marbles
Two minute Papers: Wow! What a time to be alive!
i really love the excitement of u at 4:12
love that sound, Woohoo!
btw amazing research paper and amazing video!
"Holy Mother of Papers" :D
He could've just said "Holy Trees".
@@imveryangryitsnotbutter that would’ve been very smart
BeamNG Drive: *am I a joke to you?*
Muller et al.: Yes.
heheheh got ya
@@carnap355 there is real soft body physics in beam ng. Beam ng does simulate the physical properties of the mesh ( metal in this case ) and how it should react when reforming. There are stronger parts and weaker parts, attachment points and all the like.
You can tell this through the fact both simulations have the same tyre mechanics. The rubber is simulated to deform differently like metal, to plastic, to glass etc.
@@carnap355
From what I've read, the simulation in this paper is done by constraining primitives and the mesh is bound to them.
And in BeamNG, constraints are beams, primitives are triangles and visible mesh are bound to them.
There is no major difference in that point.
And all other racing game with advance tyre deformation, temperature factor, compound and wear xD
the algorithmic sound will be the future of computer games! now we can make worlds with materials and dont even have to design sounds anymore!
Yeah it‘s astonishing. No more recycled sounds. Alll organically created sounds. I can‘t wait.
i will miss those "classic" sounds that you hear over and over again in games and movies..
there are so many movies where people draw a pistol and i get flashbacks to trophy hunter...
The experience of doing something by yourself can increase your creativity instead of being limited to what someone else create for you
@anurag this! virtual instruments! even when not directly playing something in VR, this adds a whole new dimension of sound creation for synthesizing!
That's what people thought about physics a few decades ago, and here we are today with half the assets in games still being hand animated. That's the thing about videogames, they're not pure engineering products. A lot of artistry go into them which is why some things will keep being hand-made for a long, long while. When audio designers want something to sound exactly right, it's often easier to record audio than to coerce an algorithmic engine into producing the right stuff. Procedural audio will be a useful tool for sure, but I wouldn't expect it to replace traditional foley any time soon.
This is freaking amazing. Have been following since few weeks and love your videos. Keep up the awesome work !!!!
I couldn't remember what this channel was called, so I just looked up "dear fellow scholars" and got there ;)
Can't wait to see this implemented in 3d software like c4d or Maya. The amount of time it would save
Turns out my papers were just a simulation.
Wintergatam: *heavy breathing*
I swear I could hear some cranks being thrown out of the window :)
Holy shit! Imagine car games in the future
Probably all games in the future will use simulations instead of animations for just about everything.
Chech out: Beam ng drive!!!!
Honestly, Dirt Rally 1+2 are so insanely close, simulating it at this stage is just a fun exercise more than anything else. I'd rather we'd have some kind of vestibular stimulation device allowing us to completely feel like we're driving a car with all the forces happening.
Imagine a new Lego Racers game with this tech.
Cars in the future: Holy Shit! Imagine humans in the past
What a time to be alive, indeed! Love your work! Thanks for sharing.
BeamNG: *laughing hysterically*
The RC-car simultion is unreal. I've had many hobbygrade cars in the past years, and this simulation is incredibly close to a real life 1:10 of 1:12 scale buggy (although without much or thin oil in the shock absorbers). Absolutely mindblowing
Okay addition: The car rendering in 55 frames per second sounds like the most amazing RC-game ever. Re-volt 2 Remastered anyone?
car manufacturers: hmm, maybe we can use this and spend less on crashing actual ones
They don't do that, it's another company, the manufacturers earn money instead
They do it already, just with slower algorithms on supercomputers. The cost of a few dozen crashed cars can buy you a decent amount of computing power.
They've been doing it for years, if not decades
Car manufacturers simulate using algorithmic methods that aim to model real structural physics and fluid dynamics. These "simulations" are geared towards getting close to physics, without needing a lot of compute. If you tried to use them to predict real physical systems, they would fail badly. There is a reason that fluid dynamics has not been replaced by smooth-particle-hydrodynamics and structural simulations still are not run on GPU's. And that is that when you want to do real physics, you can't parallelize it.
@@avialexander nature is real, physics is partially real, maths is not real, algorithms are only aproximations in this case, whether parallel or not.
ABSOLUTELY LOVING IT as always! Thanks for keeping us up to date with our wonderful time to be alive!
A two minute paper vid aaannd an accursed farms video! What a time to be alive
I am in total awe and love with what i see in this video ! I'll get my hands on this paper !!
"Holy mother of papers"
Now that's a fellow man of culture
Thanks for vid, you doing great!
This channel always reminds me that progress is now. Thank you for the amazing videos
i've never clicked like so fast in my life, even before watching, I need this in my games like now
Try BeamNG
Looney I have, and it’s pretty great, but this seems more general purpose and potentially more robust, and I just want this kind of simulation in more games, not just niche titles. But also more niche titles. Whenever I see tech demos like this, I just want awesome physics simulation sandbox type games where I can just be a god of tearing food apart or punching physically accurate holes through objects and all that fun stuff XD
Wow! I'm speechless! I hope to one day be smart and learned enough to understand these papers - maybe implement and play around. And, maybe, contribute in some shape or form! Such a wonderful thing for the research to be made available for free, and the author of this video to share this with us! I have got be able to able to write my own papers to hold on to them! What a time to be alive!
I'm not a scholar or anything but I love these videos and his voice is so calming
Your personality takes what would otherwise be mildly interesting and makes it very entertaining.
I’m done Dr. K, I have no papers left, I held them as hard as I could but it was useless, I have nothing to loose now.
Yes, this was a proper paper, so that's understandable. Excused!
DANG the recent advances in this area are insane. The other day you did that paper about super detailed liquid/solid-coupled simulation, and now this!
Maybe one day we really will get a real time simulation of all of everyday life physics interacting in arbitrary ways!
I actually said “excuse me !?” out loud 😂
Ouuuu this is so excellent for future video games and so much more. Beautiful visuals and explanation as always. Thank you!
Isn't this already present in BeamNG? At least the deformation simulations.
Not this kind of deformation. BeamNG's is rigid deformation, this has soft deformation.
Edit: My original comment was incorrect. BeamNG does have softbody deformation physics.
@@TornaitSuperBird What's the difference? I thought BeamNG was all about soft bodies.
@@TornaitSuperBird Beamng does has soft body physics
this is an improvement on it. older algorithms tend to go full kraken. this one is apparently smooth as japanese dog fur.
@@TornaitSuperBird my guy the whole game is soft body physics XD
'Holy mother of papers" your vids just keep getting better!
Everyone: I wanna play a game like that!!!
BeamNG Drive: Whimpers in a corner
Imagine an RPG like Bethesda games with realistic physics such as this, rather than twitching ghost corpses, dinner settings that vibrate, bookshelves which explode if you look at them wrong and animals which clip through the ground.
I love your enthusiasm, it's contagious :)
Combine this with BeamNG Drive's softbody physics for body deformation and I think we simulated real life
It's not that easy. They would need to make a brand new game
@@FSXgta I never said it was easy I just said the outcome would be significant
@@SpeedySpee true
you sound quieter and smoother károly. it is nice on my ears. like ASMR but computer science. you are the Bob Ross of research.
Folks, if you were there, Do you remember waiting for those soft body simulations in 3dsmax?
it took ages XD Coffee, coffe, more coffee . omg the bad old days
no i dont remember cuz i do blender lol
@@Lrelni
It would be the same in blender though...
Fantastic stuff. I love your very clear explanations into ai based graphics simulations.
0:54 well there is a game called beamng.drive that can do that in real time, its amazing the things you can do in the game like deformations when you crash( sorry for my shiti english ) regards from Catalonia and stay safe :)
It would be amazing if this was used in the next racing simulator.
5:42 Oh god that's like my headphones
My earphones :-(
This research is absolutely awesome, I am speachless! Great video as usual, thanks for sharing :)
how can i get this in blender? I need this in blender
I think the exact same thing for 70% of the stuff on this channel. Though it's not an easy task for anyone to implement things into software, let alone some of the more advanced findings. I wish I was smart enough to understand just 30% of it, or else I'd try to do it.
Hopefully by 2025 there will be a ton of tutorials on youtube showing us how
lmao
True, blender for destruction is just npt quite there yet, the only thing we have ist cell fracture addon, which works well like 20% of the times
They got so involved in programming deformation, they forgot to program the oil filled shocks, instead of just springs by themselves. Cars aren’t that bouncy guys.
Hasn't Beam.NG been able to do these things for a while now?
No, it's done differently
@@skrimper no its not beamng is a soft body physics engine made in 2013!! SEVEN YEARS BEFORE THIS, and now everybodys so stupid and think beam is "JuSt LikE AnY OThER CAr GaME" and its stupid beam has done this, way better, with higher fps, and better physics LONG before this stupid shit hes acting like hes never seen a tier model before, idiots.
@@skrimper and done better.
What a time to be alive indeed, this is so far one of my favorite papers, this is absolutely amazing!
laughs in Beamng
@@carnap355 that's super cool, thanks for the comparison and insight!
@@robinrai4973 who is this guy and why does he keep deleting his comments, what is he saying?!?!?
Wow, I'm amazed by the string at 5:50 !!!
I believe this is just a soft body physics simulator just like BeamNG. And BeamNG probably uses the same simulation techniques that this simulation uses, but BeamNG actually has better performance than this simulation (< 0.1 ms average physics calculation time per frame for one car in Gridmap). Although that could be due to BeamNG using fewer nodes and beams to simulate the cars than this simulation does but I'm not sure about the key differences between the two simulators.
This is cutting-edge research. Whatever techniques BeamNG uses are inferior. (Or just plain different.)
@@sgbench actually they are superior,assuming this is done on a computer that runs beam at even 30 fps,why am i saying this you may ask?
beamng literally simulates every single part of a car and a good enough rig can run it at 105 fps on the most intensive map.all while simulating almost every aspect of a car on max graphics.
This is the most satisfying simulation footage I've seen in a while. Excellent video as always =)
beamng: *looks away
in beamng drive this clip through objects and stuff but with this it wouldn't
@@Lavender34124 as if running into a wall at 500kmh wouldnt make stuff clip into eachother.come on you are making me laugh,if it went fast enough i wouldnt doubt it'd clip the car and break it
Wow, this one seriously blew me away. Really excited to see it's applications
Woah -- NVIDIA? Copenhagen?
Sounds like the folks up at Blender have been busy!
Holy crap, I'm almost crying. This is unbelievable and I need to get this to Unity asap.
Also, now is DEFINITELY the time for a new Lego Racers game; I've waiting forever, someone should make it happen!
BeamNG 👀
As an automotive engineer I can not even imagine how useful this algorithms can be in my field for vehicle dynamic simulation. It usually takes up to minutes to perform even a simple steady state simulation. This on the other hand...Oh my god I'm getting goosebumps
Check out Beamng.Drive! They've been doing pretty accurate vehicle physics for years now, including deformation. Loooots of deformation ;)
ikr how do people not know about beam? and if they do they say "oH wElL iTS nOt SofT iTs RiGID" and it makes me very mad
@@Ripsaw460 yea its softbody.if it was rigid it'd be similar to wreckfest/flatout 2
I’ve been watching all of these videos thinking that you were at least 50 years old. You are wise beyond your years, sir.
15ms? Yeah, on what computer?
It's always funny seeing these simulations when they don't tell what hardware they used.
Amazing! I come from a content creation background and have a love and hate relationship with dynamic simulations... This looks really promising... Love your channel btw!
I mean.. Beam.ng drive already has this :shrug:
I'm all about the phrase "holy mother of papers".
One singular word. Beamng...
Its so Natural and Perfect it creates satisfaction and joy just by being hypnotized by the simulations
We've had this with voxels for years, check out Teardown
I am happy to be able to say that I was here when the original car-simulation paper was published (:
* completely ignores the existance of beamng.drive *
Some of these papers are so amazing it makes you wonder how they come up with such ideas.
I came for cool simulations... I stayed for the happy marbles! :)
2:20 I do not understand this at all? Are these simulations being done in blender? I've done a similiar one and never had that 'old technique' or whatever',s problem, I don't understand.
lol why would they be in blender
@@spaceman-pe5je wym
The future is looking bright. I love stuff like this.
5:45 now i can get my headphones tangled up inside my computer! what a time to be alive!
I don't understand what you are talking about, but you sound so happy, so i think it's a good news! Lets GOOOOOOOOOOOOOO
Holy grail!
Just calmed down since the series about a liquid solver, that respects medium boundaries, thus being able to simulate air effects of all kinds... How great could it improve PhoenixFD, for example!
And now the physics solver comes into play, that could not just give a huge kick in speed and accuracy of TyFlow simulations, but turn MadCar into a realtime BeamNG experience, but many times more physically correct!
What a time to have all of my physical sandboxes experience!
So awesome! Can't believe it's that fast with that much geometry! They could make the tire pop or something; racing games will never be the same after this!
Oh my God where are we going with this...thanks to the geniuses who are making it happen....I have always loved realistic simulation....
4:12 Being a person who has minimal knowledge of this, hearing happy noises from Károly makes me know that this is something to marvel at. Really amazing!
man!, I were able to recognise the different between cat's faces more than human's faces 😂😂 6:29
I like free things, thanks Nvidea!
I don't know why this was recommended to me but I'm glad it did
I do not know anything about programming, I rarely understand anything, but this is just watchable.
Now imagine games using such an davanced physics in VR games using an haptic glove...closest thing to being a real thing yet non existent object simulated in VR.
Smiling marble should be a new emoji.
Holy crap, this is some mindblowing stuff.
What _a _*_time_* to _be _*_alive_* & *_able_* to _see _*_reality_* if we are _first _*_willing_*_ enough_ to & _ask _*_hard questions_* that *_challenge_* our _current _*_beliefs_* as _often as _*_necessarily possible_* !
Haha these frame times are extremely impressive, nice!
Dont know about others but it just Feels good to hear “dear, fellow scholars” Every morning. Just makes your day much brighter.
oh my gosh this is so cool. dude i love seeing the advancements in simulation technology! :)