Upload
leque
View
223
Download
2
Embed Size (px)
Citation preview
Ph.D Candidate Zhang Yuzhe
Supervisor Assoc Prof. Jianmin Zheng
Co-supervisor Prof. Nadia Magnenat Thalmann
2013-11
Customize garment pattern for made-to-measure
1
Outline
• Introduction
• Prior work
• Garment customization method
• Experiments
• Summary
2
Introduction
My PHD Project
Preposition
Adaptation
2D garment
Body model
Simulation
3D garment
Flatten
Made-to-Measure
For Virtual Try on For Manufacture
Flatten
Made-to-Measure
3
Problem statement
• Input:
– 2D garment composed of 2D patterns
– 3D body model
• Output:
– 2D patterns suitable for input body
• Goal
– According to measures of body
– Less force and pressure of garment on body
4
Fit on customer
Made-to-measure
Original 2D garment Customized garment
Motivation
• Fit customer made-to-measure
• Facilitate design procedure
• Manufacturing
5
2D Garment
Pattern
Body 1
Body i
Patterns set 1
Patterns set i
Body 2 Patterns set 2
…
…
Prior work
• Developable surface based approach [Huang12] [Meng12]
– Input: body embed with features
– Process: sketches of garment
– Output: 3D garment surface + flattened 2D patterns
– Drawbacks
• Have to redesign for each body model
• Difficult to locate darts and seams
[Huang12] H.Q Huang et al. “Block pattern generation: From parameterizing human
bodies to fit feature-aligned and flattenable 3D garments” Computers in Industry, 2012
[Meng12] Yuwei Meng et al. “Computer aided clothing pattern design with 3D editing and
pattern alteration”, CAD, 2012 6
Prior work
• Transfer based approach [Meng&Wang12][Brouet12]
– Input: standard garment on reference body + target body
– Output: transfer onto target body while optimizing shape features
– Advantage:
• Formulate 3D grading criteria
– Disadvantage:
• Not consider feature of 2D pattern
[Meng&Wang12] Yuwei Meng and Charlie C.L. Wang “Flexible shape control for automatic
resizing of apparel products”, CAD, 2012
[Brouet12] Remi Brouet et al. “Design preserving garment transfer”, SIGGRAPH, 2012 7
Challenges
• For developable surface modeling
– Curvature = 0 Non-linear
– How to linearize developability optimization
• For manufacturing
– Consider feature of 2D patterns
– How to formulate 2D grading criteria
8
Proposed workflow
Pre-fitting &
Simulation
Output
2D Pattern
Flattening
3D Garment
customization
Input
Sketch features
9
Pre-fitting & Simulation
• Pre-fitting
– Extract measurements of body
– Position patterns on body
– Fitness evaluation
– Roughly scale 2D patterns
• Follow original design
• Ready to suit for body
• Simulation an initial 3D garment surface
10
Fitness evaluation
Roughly scale Simulation
3D customization
• Sketch some features
– Tight part
– Loose part
– Feature curve
• Find a new 3D garment that
– Interpolate sketched features
– Preserve feature of 3D garment and 2D patterns
11
3D optimization for customization
• Optimize: 𝑎𝑟𝑔𝑚𝑖𝑛 (𝑤𝑠𝐸𝑠 + 𝑤𝐷𝐸𝐷 + 𝑤𝐵𝐸𝐵 + 𝑤𝐼𝐸𝐼)
12
Shape
preservation
Laplacian
Developability
Linearize using the
property that normals
on Gaussian sphere
form a continuous
curve
Boundary
Boundary curves own
similar 2nd difference
Fidelity
Interpolate features
2D Pattern flattening
• Apply ABF++[Sheffer05] to obtain an initial guess
• Perturb initial 2D patches to achieve
– Boundary smoothness
– Lengths of seam pairs are equal
– Adapt darts and seams if necessary
– Consider material parameter
[Sheffer05] Alla Sheffer et al. “ABF++: fast and robust angle based flattening”, TOG, 2005 13
Not finished yet
Experiments
• Evaluate
– Force: warp and weft (F/m)
– Pressure of garment on body (F/m2)
• Four cases:
– Skirt
– Pant
– T-shirt
– Dress
14
Experiments (Case1: Skirt)
15
Customization
Original garment Customized garment
Less force &
Less pressure
Force Pressure Force Pressure
Experiments (Case2: Pant)
16
Original garment Customized garment
Customization
Experiments (Case3: T-shirt)
17
Customization
Original garment Customized garment
Experiments (Case4: Dress)
18
Customization
Original garment Customized garment
Summary
• A new formulation of garment pattern customization
– 3D optimization model
– New linearization strategy using the normal criterion of developable
surfaces
– More features considered
• On progress
– 2D pattern flattening
– Comparison with prior art
– Comprehensive evaluation
19
Thank you !
20