At first I was like "eh it doesn't have the same visual look of having individual threads" but then I realised that it can do the same thing as physically modelling each thread... Without actually having to do so. Plus being able to accurately simulate higher stitch densities with a trivial amount of processing time when compared to the standard method? That's killer. Great work!
@@logitech4873 True, but displacement mapping can't simulate tensile properties like this method can. Different weave patterns can make cloth stretch in one way but not another, whereas with a displacement map it'd stretch uniformly and regardless of which way the thread patterns face (unless you make a quick hack where you limit your softbody from stretching along a specific axis). That's why this method is so magic.
@@MajatekYT I meant that you could apply displacement mapping onto the flat sheet simulated with this method. I.e. simulate with this method + render with displacement = visuals close to ground truth
while you do lose detail if its for something larger thread, its still an AMAZING time cost saver and you could likely map a 3d model onto it and with some tweaking it could work pretty well
First I thought it failed due to missing the 3d graphics of the yarn model but its clearly better for small threads where those details would be too small.
Would it be possible to map a model of the thread onto simulations made using this method, rather than just projecting an image of the thread onto a plane?
Yes, that should be possible. However, you won't get any yarn-scale physics details, like knit loops tightening if the pattern is stretched, since that is as expensive as doing a yarn-level simulation.
The method fits a material model for a 'normal' cloth sim, such that the simulated cloth behaves approximately like a full yarn simulation would. Simpler existing cloth materials would not allow for the same rich behavior (e.g., anisotropic stretching and bending resistance, as well as interactions between the two).
it allows simulation of pattern-dependant elasticity. the same thing could almost certainly be solved with similar results with anisotropic elasticity, but this does it in a more specific and physically accurate way.
@@ldskjfhslkjdhflkjdhf then please use English to create a virtual representation of a cloth, don’t forget to tell every single molecule exactly how it is supposed to react in relation to every other molecule and what position in space it is supposed to end up in when it’s finished falling because otherwise the simulation won’t be realistic. Tell it to “settle lightly over the table like fallen snow” and see where it gets you. Or you could just use math, which can tell you exactly what the cloth will do, exactly how it can be done, and more importantly be used to perfectly recreate the physics of how it works virtually. I truly cannot believe the astonishing ignorance of this answer. Were you not directed to this video by math?
@@dodiswatchbobobo Math isn't a universal language. In fact, almost all proofs are still written in English and all solvers are written in some programming language. Math isn't uniquely powerful. That's it, that's my entire point.
![Homogenized Yarn-Level Cloth [SIGGRAPH 2020 Presentation]](http://i.ytimg.com/vi/qrDczRcuB3Y/mqdefault.jpg)
![Homogenized Yarn-Level Cloth [SIGGRAPH 2020 Presentation]](/img/tr.png)
![Mechanics-Aware Deformation of Yarn Pattern Geometry [SIGGRAPH 2021]](http://i.ytimg.com/vi/rO10fjzOzWM/mqdefault.jpg)
![[SIGGRAPH 2020] AnisoMPM: Animating Anisotropic Damage Mechanics](http://i.ytimg.com/vi/eXJi7pkUZn0/mqdefault.jpg)
![[SIGGRAPH 2018] A Multi-Scale Model for Simulating Liquid-Fabric Interactions](/img/n.gif)
At first I was like "eh it doesn't have the same visual look of having individual threads" but then I realised that it can do the same thing as physically modelling each thread... Without actually having to do so. Plus being able to accurately simulate higher stitch densities with a trivial amount of processing time when compared to the standard method? That's killer. Great work!
You can use displacement mapping to make it visually appear very similar :)
@@logitech4873 True, but displacement mapping can't simulate tensile properties like this method can. Different weave patterns can make cloth stretch in one way but not another, whereas with a displacement map it'd stretch uniformly and regardless of which way the thread patterns face (unless you make a quick hack where you limit your softbody from stretching along a specific axis). That's why this method is so magic.
@@MajatekYT I meant that you could apply displacement mapping onto the flat sheet simulated with this method.
I.e. simulate with this method + render with displacement = visuals close to ground truth
@@logitech4873 Ah yeah. It might not get the same close-to-ground-truth thread pulling that this method tries to reproduce, though.
while you do lose detail if its for something larger thread, its still an AMAZING time cost saver
and you could likely map a 3d model onto it and with some tweaking it could work pretty well
yarnmadilo!
First I thought it failed due to missing the 3d graphics of the yarn model but its clearly better for small threads where those details would be too small.
They should simulate the fabric draping over two spheres.
Great work!
I hope to have these solvers in Sidefx Houdini...!
It may hold up for non-close ups, but for filmic quality the approximation looks live a uv projection.
This is just the simulation aspect though. I'm sure you could render complex yarn geometry still, but use this method for the simulation.
Thank
But can it render outfits from Crisis at 60fps?
Would it be possible to map a model of the thread onto simulations made using this method, rather than just projecting an image of the thread onto a plane?
Yes, that should be possible. However, you won't get any yarn-scale physics details, like knit loops tightening if the pattern is stretched, since that is as expensive as doing a yarn-level simulation.
I think my advisor reviewed this!
I guess they really love armadillos in this industry.
Whats the diferance between this ans a normal cloth sim ?
The method fits a material model for a 'normal' cloth sim, such that the simulated cloth behaves approximately like a full yarn simulation would. Simpler existing cloth materials would not allow for the same rich behavior (e.g., anisotropic stretching and bending resistance, as well as interactions between the two).
it allows simulation of pattern-dependant elasticity. the same thing could almost certainly be solved with similar results with anisotropic elasticity, but this does it in a more specific and physically accurate way.
yarnmadillo :)
Math is the language we use to understand the universe.
Lots of languages are used to understand the universe, and spoken languages are by far the common sort of language used for that purpose.
@@ldskjfhslkjdhflkjdhf then please use English to create a virtual representation of a cloth, don’t forget to tell every single molecule exactly how it is supposed to react in relation to every other molecule and what position in space it is supposed to end up in when it’s finished falling because otherwise the simulation won’t be realistic. Tell it to “settle lightly over the table like fallen snow” and see where it gets you.
Or you could just use math, which can tell you exactly what the cloth will do, exactly how it can be done, and more importantly be used to perfectly recreate the physics of how it works virtually.
I truly cannot believe the astonishing ignorance of this answer. Were you not directed to this video by math?
@@dodiswatchbobobo Pro tip: simulations are written in programming languages, not math.
@@ldskjfhslkjdhflkjdhf the programming language defines mathematical parameters for the simulation to follow.
@@dodiswatchbobobo Math isn't a universal language. In fact, almost all proofs are still written in English and all solvers are written in some programming language. Math isn't uniquely powerful. That's it, that's my entire point.