0:11

So, the problem we want to address here,

Â is a design of a new novel original garment.

Â And again, this is very difficult for inexperienced people.

Â So, you have 2D cross pattern,

Â and then you have three dimensional garment on the body.

Â it's, what you want is three dimensional shape.

Â But what you have to control is two dimensional shape.

Â And it's very difficult to imagine what 2D pattern you have to

Â get to get the desired 3D shape.

Â So, that's a problem we what to address.

Â And our approach is the same as the previous one.

Â So we learn continuous concurrent simulation behind the scene.

Â So user continuously edits are saved 2D pattern with mass operation and

Â the system continuously applies physical draping across simulation.

Â So user can directly edit to decide while watching the clothing in

Â the 3D configuration.

Â Well, that's

Â the idea.

Â So the same as before, on the left side,

Â you see 2D clothes patterns the user is editing.

Â And on the ride si, on the right side you see clothing, putting onto the 3D body.

Â And here the simulation is much more accurate than the one I showed for

Â the stuffed animal design.

Â So you can actually predict details, wrinkles and so on.

Â So you can interactively edit the 2D pattern.

Â And the system interactively updates clothing draping simulation results.

Â So you can actually see some more wrinkles here.

Â And then you can edit the shape.

Â So as to reduce wrinkles if necessary.

Â 2:11

So traditionally, if you want to do this, you have to actually physically cut paper

Â or a cloth material, stitch together and put on the mannequin, physical mannequin.

Â And learn draping and see the result and then go back to the pattern and

Â cut it again.

Â It can be very, very long, long time, it's very slow.

Â And those are computer based simulations also available.

Â It's similar.

Â You have to move the simulation engine simulation and then go back to the design.

Â That's a problem.

Â And also, technically, it's very,

Â challenge to run physical simulations so rapidly.

Â Of course there are real time rapid cloth simulation existing already.

Â But they, these physical simulation is for animation.

Â So they assume fixed clothes pattern.

Â So clothes geometry doesn't change.

Â And the system learns lots of precomputation for

Â efficient computation real-time.

Â And then just draping learn animation.

Â To learn simulation to create animation.

Â However, here, the system, the user continuously change the rest of shape.

Â Which is, violates assumption of standard animation,

Â animation simulation [INAUDIBLE] so that's a challenge we have to address.

Â 3:55

So, yeah, so design of t-shirt or

Â garment for human character is not actually so difficult.

Â There are many standard ways to design a garment.

Â However, if you want to design a pajama for

Â armadillo like here, it's a different program.

Â No one knows how to design a character shirt.

Â And especially if you wanted to buy a physically bodied garment,

Â it's very difficult, and this kind of design tool can be very, very useful.

Â 4:24

So previously in computer graphics, garment of this character is

Â a very cross mesh, there's no physics, just a polynomial mesh.

Â But recently, game engines are very fast, more efficient.

Â So we use, people starting to use physical assimilation for these garments.

Â Now then, it's necessary to design physical bodied garments.

Â So that's these are why you need this kind of tools.

Â 4:55

So here is a comparison with a real one.

Â The left side is a computer graphics view in your simulation, and

Â the right hand side is the physical construction which we 3D printed.

Â This kind of character and also we physically stitched together this garment.

Â And as you see, of course, it is not 100 millimeter size exact match.

Â However, you see the general pattern like sleeve folds here and

Â then you see full sleeve folds here.

Â So I think this is enough for our initial design.

Â 6:25

Okay.

Â So let me briefly describe algorithm.

Â So, again, so oh, complete implementation is beyond the scope of this short video.

Â But I can briefly describe the basic idea behind it.

Â How to make it faster.

Â So this is a fundamental equation.

Â Mathematical representation of the problem we trying solve here.

Â So we have 2D input pattern.

Â This is the input from the user.

Â So user specify 2D clothes pattern.

Â And then what we, ha want to compute is 3D cloth shape.

Â So, yeah.

Â So input these XY coordinate positions or we input clothes, and

Â the output is XYZ position or or when you give your body shape.

Â So 3D cloth shape, this is output you want to get.

Â And then we cover function, that represents an additional set

Â of 2D cloth pattern input and 3D cloth shape output.

Â And this function returns 0,

Â if it is satisfy the physicality, body natural shape.

Â And this function is called the resi, residual.

Â So if the 3D cloth shape is not in the desired,

Â appropriate physically balanced shape.

Â This zero becomes bigger, larger.

Â And then we, we gradually, this residual are smaller.

Â They get more physically realistic results.

Â So that's the problem we try to solve.

Â And visually, the situation looks like this.

Â So you have a space of varied designs.

Â And on this view,

Â on this axis, you have many different two dimensional cloth patterns.

Â So use the space for that starts from this pattern and

Â then moves to another pattern, and use the edit.

Â 8:03

And for each shape, you have a single iso surface, or

Â space-specific region, where R becomes 0.

Â So here, it's represented as a single line, single cloth line.

Â So, for

Â the given 2D clothes pattern, we take a look at the region where the R equals 0.

Â And then look at the 3D shape, and then you will get the proper 3D output.

Â So this is a, very very simplified view of what's going on internally.

Â The program is at, notice here.

Â So R, the residual is very highly non-linear and very, very slow to compute.

Â So, yes, so R 0 is very curved and very complicated.

Â When you change the pattern, you have to recompute R, which takes too much time.

Â So we, what we use is a kind of a stanadard.

Â But we use a linear appro, approximation around the current state.

Â So visually, it looks like this.

Â 9:01

So, this a current design and then you compute the linear appro,

Â approximation of this residual function.

Â And this method called, called sensitivity analysis in structural analysis.

Â And then after computing linear approximation it's kind of easy,

Â to predict the, compute from 2D input to 3D shape.

Â Just by solving our linear system.

Â Which is faster.

Â Of course, this is an approximation near this design.

Â So, if you go farther away, then it's, it becomes too different.

Â So and your system.

Â So single linear appropriate is nothing now.

Â Users who're walking far away from the linear state.

Â It's a difference from the real physically realistic result gets bigger.

Â So in that case, we compute another approximation here right this way.

Â And then, so

Â we cache multiple linear approximations occasionally and then blend them.

Â So, in this case, if you have two examples and

Â if you blend these two, you get a very, very close approximation.

Â And this is still very fast to compute.

Â So that's what we do in internally.

Â So caching happens occasionally.

Â So, the user, the system continuously monitors.

Â Users the dragging operation.

Â And then when the users dragging getting far away, you create another cache.

Â Now you have two cache, and

Â you move far away from these two caches, the system will generate another.

Â So in this way, the system incrementally, the more and more caches.

Â So this is a summary.

Â So we presented garment design with concurrent physical simulation.

Â So we edit a 2D pattern.

Â And the system presents 3D draping results.

Â And then we have uprising sensite,

Â sensiivity analysis which is linear approximation.

Â And those are multiple caches before providing rapid

Â feedback without random simulation each time.

Â 10:59

So, to learn more the original paper was published on Sensitive Couture for

Â Interactive Garment Editing and Modeling.

Â And the garment design is also a hot, popular topic in graphics community.

Â And one example is Virtual Garments, A Fully Geometric Approach for

Â Clothing Design.

Â So, in this project, they did not use any physical simulation.

Â Purely geometry approach, but

Â the, but they can rapidly generate reasonable shapes this way.

Â