@@alext2046 i mean, trump IS pretty fucking stupid, and personally he doesn't make for a good president. the fact he denies global warming only proves that
@@michaeladams3762 only in good simulations... there are lots of terrible simulation techniques out there... and a lot of software uses them by default...
Amazing! Hard to believe that at the first SIGGRAPH they were "oohing" and "awing" at rotating, non-shadowed or reflected company logos. My, how far we've come...
It's not a question of technical limitations, it's a question of how long you're willing to wait for the simulation to run. If you set the resolution high enough and get the parameters right it's long been possible to create fluid simulations that are entirely indistinguishable from reality.
@@Bob3D2000 We can real time ray trace fluid Sims!, just not anywhere near cloth like this, that would be insane. Give is a couple years. AI is going to fix all of this.
Super impressive stuff... I guess the next step is to simulate the varying levels of saturation throughout the towel after you ring it to get those dry/wet creases we see in real life...
Willy J. I think if how much the cloth can retain liquids can be adjusted, that would be fixed. Then again, this is a huge ball of liquid so maybe it's normal for them to be dripping wet for so long.
I feel the same, and after water is absorbed it also stais on the surface as liquid making the cloth really reflective for way too long, its not sucked into the cloth well
From 4:39 on, why does the yellow cloth suddenly sag _before_ the falling water even makes contact with it? You can see it clearly with the slow-motion replay at 4:44.
We've had all this shit with mind-blowing demos with Relief Mapping, Digital Molecular Matter, Euphoria, Wiimote tracking, Geometrical Shaders, etc. It didn't get anywhere and was not used in any game (except like 5 ten year old games) and this one will die the same way.
@@fyun0acm OK, thanks. With the focus on the liquid I assumed the cloth would be stretched tight at rest and didn't consider that it would need to settle as well.
it looks amazing and seems accurate, but for what you gonna use it? for wet t-shirt contest simulations?, exept that you can not do that because you never showed how two fabrics in contact transfer the liquid to each other (you know, the t-shirt and the bra) aside from that, for commercials?
For CG character coming out of the water. They'll wear clothes. Have you watched Alita: Battle Angel and notice there's a second when Alita walked out of water?
Yeah this is cool but i have one problem an that is that cloth when wet will stay on top of the water not go through is, that is an improovement yu can add
Would anyone be able to explain the process of how simulations like these are made? Is there certain software being used to create and render the animations? Are they coded from scratch, including the models themselves? Are the models created in 3D modeling software then imported? If so, how are they imported? Are the physics engines and rendering engines coded from scratch? I apologize if I’m asking bad questions. I’ve always wanted to get into this kind of stuff but have never figured out where to even start.
can we start doing simulations with a real world physics as control? Setup a real world test that simulation is trying to mimic and show us how closely the simulation is matching with what actually happens in real world
You may find that info in our paper: www.cs.columbia.edu/cg/wetcloth/main.pdf Generally the sim will take several seconds for a time step, on a 32-core workstation.
This seems like such an internal and proprietary software, with the channel likely existing for sales and demonstration purposes... which begs the question, why the hell are comments turned on? This has to be counter intuitive to whatever you hope to achieve here. I'd love to know though, what validates any of this beyond "you watched a video where it was similar in real life"... and how industry a term exactly is "fuzzy towel"?
It depends on the size and length of the sim. For each example shown in the video it tooks in average 2~3 days for simulation on a 32-core workstation, and 1 day for the rendering on a small renderfarm with ~280 cores.
But if it's a simulation, how does the cloth relax if the vertices don't move, which they clearly don't in the video? It appears that the cloth is at rest and equilibrium under a constant gravitational force. Then before the water hits, without any external stimuli or the vertices moving, it dips before contact with the body of liquid.
4:50 is a perfect example of what is wrong with the simulation. The fluid is weird. No liquid when dropped in the shape with that volume would ever react that way unless it was some super viscous liquid. But then when it travels through the yarn you can tell it not super viscous skin the answer has to be. Your fluids dont behave enough like fluids. If you want more realistic fluid cloth interactions you need to start with realistic fluid
Нет. Мы имитируем наш собственный код (см. Ссылку выше). В будущем физический движок в режиме реального времени, такой как PhysX, может его поддерживать.
![[SIGGRAPH 2017] A Multi-Scale Model for Simulating Liquid-Hair Interactions](http://i.ytimg.com/vi/9Oo0TlprwAQ/mqdefault.jpg)
![[SIGGRAPH 2017] A Multi-Scale Model for Simulating Liquid-Hair Interactions](/img/tr.png)
![[SIGGRAPH 2018] Animating Fluid Sediment Mixture in Particle-Laden Flows](http://i.ytimg.com/vi/kt_TlUsv6rU/mqdefault.jpg)
![[SIGGRAPH 2017] Anisotropic Elastoplasticity for Cloth, Knit and Hair Frictional Contact](/img/n.gif)
Now this is how you show your research. It doesn't need to be a lifeless video. 10/10 soundtrack. Thanks
Very well done!
i can already smell the GPU burnin
@@chrisguy6301 maybe PhysX accelerated
lemme rephrase that, hardware accelerated**
You can run it long enough to take a single cool screenshot to show off, then your PC will rest in pieces
cuDNN
mb CPU? nope? -_-
I'm predicting this will first be used commercially in simulating a wet t-shirt contest.
Games?
Matrix29bear what
Tuxfanturnip that's what "in"cel losers actually think. ;)
Matrix29bear found the virgin
Wow, you definitely have a problem with porn.
Can’t wait for games to be hyper realistic like this and probably run on VR. Ooooooh boy. Can’t wait till I’m 80.
i mean if we don't burn our entire planet before then
Nienke Timmermans orange man bad. Orange man make weather hot
@@alext2046 Are you a caveman?
@@titanicjames3146 woosh
@@alext2046 i mean, trump IS pretty fucking stupid, and personally he doesn't make for a good president. the fact he denies global warming only proves that
Born to late pioneer new lands, born to early to travel the galaxy, born just in time for wet towel simulation physics.
For some reason i was really amazed by that towel not clipping or weirdly stretching
thats literally the least impressive part of this video
cool opinion
nah but like if you've ever simulated cloth thats how it normally acts
@@michaeladams3762 only in good simulations... there are lots of terrible simulation techniques out there... and a lot of software uses them by default...
Robonator if impressive is defined as difficult to achieve, but achieved anyway, then it is not an opinion
This is way beyond anything I've seen so far. Just amazing. I can't even begin to imagine the work that went into this.
Amazing! Hard to believe that at the first SIGGRAPH they were "oohing" and "awing" at rotating, non-shadowed or reflected company logos. My, how far we've come...
It seems to drip too easily.
It is still very impressive though, probably just needs some tweaking and comparisons with real experiments to tune it.
Also water resolution is still noticeable, but that can ptobably be increased easily with more processing power..
Astounding
:0 Joseju!!
who knew that a TH-cam recommendation would help me find what i want to do for graduate studies
To me personally, animated water looks either way too viscous, or has too much surface tension.
Yeah, that's the problem with fluids in general, since they have a set resolution per particle of liquid they behave more like gel or sand.
@@Monkeynuts502 Damn you technological limitations!
It's not a question of technical limitations, it's a question of how long you're willing to wait for the simulation to run. If you set the resolution high enough and get the parameters right it's long been possible to create fluid simulations that are entirely indistinguishable from reality.
@@Bob3D2000
We can real time ray trace fluid Sims!, just not anywhere near cloth like this, that would be insane. Give is a couple years. AI is going to fix all of this.
@@index7787 i wanna see that normal pc that sims water in real time with shader and lighting with billions of particles in real time
ANyone else just getting cloth simulations in their recommended. I mean im not complaining tho
That's pretty impressive. Congratulations to everyone involved.
4:46 I don't understand why the simulation flexes downward when there's no force applied to it yet?
The plane was frozen in the air before the simulation and then when they hit play, gravity was enabled before the water hit it.
They should have run the simulation for a couple of seconds before dropping the ball of water, then just not shown the first couple of seconds.
For whatever reason this is recommended to me and I like it.
Super impressive stuff... I guess the next step is to simulate the varying levels of saturation throughout the towel after you ring it to get those dry/wet creases we see in real life...
The tighten the towel game looks like it's gonna be awesome. I'm gonna wring the hell out of that towel.
that water looks THICC
Σ5 never thought someone would say that
CG golden showers animators are so happy right now
It's cool but it doesn't feel quite right somehow. Isn't it dripping for way longer than it should ?
Willy J. I think if how much the cloth can retain liquids can be adjusted, that would be fixed. Then again, this is a huge ball of liquid so maybe it's normal for them to be dripping wet for so long.
@Terrance. It's normal for your mom... I'm just kidding dude. I'm an ass.
Look at the replay speeds. Most are 8 times slower than real time, some 4 times.
I feel the same, and after water is absorbed it also stais on the surface as liquid making the cloth really reflective for way too long, its not sucked into the cloth well
From 4:39 on, why does the yellow cloth suddenly sag _before_ the falling water even makes contact with it? You can see it clearly with the slow-motion replay at 4:44.
I think that is just gravity taking hold. At frame 0 the cloth is positioned completely flat, then it droops as the physics simulation is run.
Is the fluid in the demo simulated by Material Point Method? The rendering is also very cool.Using houdini?
Yes it's augmented Material Point Method, and yes it's using Houdini for the rendering.
The physics that went into this must be crazy!
That's impressive. However why does the water in the demo often look like clear liquid glue?
I guess because they're replayed in slow-mo.
Oh right good point
They are simulating water known as "Chubby Rain".
I am immensely stunned
Oh hey, Christopher Batty was my Numerical Computation professor. Pretty good prof.
Why does normal speed look like it's actually sped up???
so what your saying is expect to see this implemented in movie CG about next year
What musical composition is playing during the video? Interesting information and the video, good luck team!
Thanks! The soundtrack was done by SnowMusicStudio, and can be found on AudioJungle.
Didnt know you could use HTC Vive sensors to hang cloth to dry.
Really awesome stuff and very well presented! Congratulations! 👍👍👍 Any plans on working with hair and fur?
We have released our simulator for hair and fur last year (libwethair.info).
Raymond Yun Fei oh great, thank you 😊
This looks awesome.
I can't wait till commercial computers get good enough to run these kinds of simulations in real time for videogames!
At 4:57, why did the cloth sink in without the water ball colliding yet?
The cloth was not relaxed initially. It needs some time to get relaxed under gravity.
I see. Thanks!
Such an incredible work! How much years of development did it take?
Cool! This answered questions I never even knew I had.
I can't wait to see this implemented in video games come 2035.
We've had all this shit with mind-blowing demos with Relief Mapping, Digital Molecular Matter, Euphoria, Wiimote tracking, Geometrical Shaders, etc. It didn't get anywhere and was not used in any game (except like 5 ten year old games) and this one will die the same way.
Or can we even use this and for what price?
This looks freakin' awesome! What are your general use cases for this software?
When will this be available in Blender?
4:43 Why does the cloth move before contact?
Because it's soft and needs time to relax under gravity.
@@fyun0acm OK, thanks. With the focus on the liquid I assumed the cloth would be stretched tight at rest and didn't consider that it would need to settle as well.
it looks amazing and seems accurate, but for what you gonna use it? for wet t-shirt contest simulations?, exept that you can not do that because you never showed how two fabrics in contact transfer the liquid to each other (you know, the t-shirt and the bra)
aside from that, for commercials?
@Artur Terho yeah no, this kind of simulation is so heavy that any game dev would dare to use it
For CG character coming out of the water. They'll wear clothes. Have you watched Alita: Battle Angel and notice there's a second when Alita walked out of water?
@@fyun0acm oh i forgot that CGI was a thing, i completely retract from what i said
You should sell this technology to Sidefx Houdini
this looks amazing would you be willing to publish a hip file?
Yeah this is cool but i have one problem an that is that cloth when wet will stay on top of the water not go through is, that is an improovement yu can add
when will this be available for public use...…...damn
Can you make liquid water next
Would anyone be able to explain the process of how simulations like these are made? Is there certain software being used to create and render the animations? Are they coded from scratch, including the models themselves? Are the models created in 3D modeling software then imported? If so, how are they imported? Are the physics engines and rendering engines coded from scratch?
I apologize if I’m asking bad questions. I’ve always wanted to get into this kind of stuff but have never figured out where to even start.
Yes in the video info, you can find the technical details of our simulation, with source code accompanied on GitHub.
Oh okay, thank you!
Imagine what we could do with this if it were realtime?
Draw a dick? Somehow any amazing hi-tech makes people want to only draw dicks.
This is incredible!
how to import this itno blender?
Can i has this effects in Cyberpunk plz? K thx bye
can we start doing simulations with a real world physics as control? Setup a real world test that simulation is trying to mimic and show us how closely the simulation is matching with what actually happens in real world
Highest recorded GPU temp while doing this?
Could this also be used to develop better pads?
Nobody cares enough so shut the fuck up
This is awesome!!!
VR will be crazy immersive 15 years from now.
This is amazing what are the sim times like? and on what hardware?
You may find that info in our paper: www.cs.columbia.edu/cg/wetcloth/main.pdf
Generally the sim will take several seconds for a time step, on a 32-core workstation.
This seems like such an internal and proprietary software, with the channel likely existing for sales and demonstration purposes... which begs the question, why the hell are comments turned on? This has to be counter intuitive to whatever you hope to achieve here. I'd love to know though, what validates any of this beyond "you watched a video where it was similar in real life"... and how industry a term exactly is "fuzzy towel"?
Some smart dude: We have technology
Realistic Games/Animations: *Wet Blanket*
Nice project. What’s the music?
The music is done by Snow Music Studio. You may find their works on audiojungle.net/user/snowmusicstudio/portfolio
and one scene takes 5 days to fully render.....NICE!
What does it cost to produce (develop and render; time and hardware) one sim?
It depends on the size and length of the sim. For each example shown in the video it tooks in average 2~3 days for simulation on a 32-core workstation, and 1 day for the rendering on a small renderfarm with ~280 cores.
Raymond Yun Fei Thank you for the information. All this means that this clip is a result of hard work. Meanwhile it must bring a lot of plessure.
will this library be implemented in DirectX 13?
I want to see this in the next gen games. : D
4:54 it’s that me or the yarns cloth moved before contacting
Yes the clothes was relaxed at first, and then contacting. The initial state was unrelaxed so it will certainly moves during relaxing.
But if it's a simulation, how does the cloth relax if the vertices don't move, which they clearly don't in the video? It appears that the cloth is at rest and equilibrium under a constant gravitational force. Then before the water hits, without any external stimuli or the vertices moving, it dips before contact with the body of liquid.
Yes it's relaxed due to gravitational force. There's also a small gap between the liquid and cloth due to the finite resolution of the Eulerian grid.
towel simulator, the next big game after gta 5
Are we in a simulation of our own creation?
That's cool and all, but can you make the towel shrink and wrinkle when it dries? Yes. This is cool stuff
Nextgen!!!
no one asked for this but damn i want it in games
GTA VI will be only runnable on NASA computers
Awesome stuff
Can't wait for VR ShamWow ads.
i think these videos will be the dreams of advanced robots with artificial intelligents.
amazing
Impressive!!
2:48 got me feeling in some way
We are closer and closer to a simulation likely being our reality everyday...
Thanks for ruining my final "am I in the matrix" test
Somehow it looks like some type of liquidish gel in a high gravity environment, instead of watter...
v_p_n killed many, project page can't open, maybe zhi_hu?
You may access directly the following address: www.cs.columbia.edu/cg/wetcloth/
哈哈
Nice! How do you know it's right btw? Did you use any real life reference?
Yes, use some real life observation.
idk wtf you saying but, Here take my money, I want it!
im no expert but that liquid is lacking surface tension... thats why when you rolled the towel, less water came out than when it was held flat.
This is a very cool job but I'm pretty sure working on it can be frustrating.
I think there should have been validation of the simulation data with real experiments for each condition you guys tested for
You're right. The validation is left for future works.
The water at the end seems like slime :I whut !
4:50 is a perfect example of what is wrong with the simulation. The fluid is weird. No liquid when dropped in the shape with that volume would ever react that way unless it was some super viscous liquid. But then when it travels through the yarn you can tell it not super viscous skin the answer has to be. Your fluids dont behave enough like fluids. If you want more realistic fluid cloth interactions you need to start with realistic fluid
大神是用unity做模拟器吗？
没有，模拟器是我们自己开发的，在Git上有代码。
Defs needs some tweaking, but cool.
wow, the last one is kinda impressive. still needs work though...
ive remember days when i first saw havok in 3dsmax r3... maybe it was in later....
i honestly thought it was real for a second
cant wair for my Playstation 9 Pro Silver Edition
That's dope
oh look....more stuff we'll never see in games.
это аналог PhysX ?
Нет. Мы имитируем наш собственный код (см. Ссылку выше). В будущем физический движок в режиме реального времени, такой как PhysX, может его поддерживать.
Raymond Yun Fei круто!
well, so we live in a simulation...
Yeah probably
we can't prove we're not
what about this made you think that