Transcript SIGGRAPH Paper Reading 2011

SIGGRAPH Paper Reading 2011
Huang Haibin
2011.7.4
Paper list
• Procedural & Interactive Modeling






Interactive Furniture Layout Using Interior Design Guidelines
Converting 3D Furniture Models to Fabricatable Parts and Connectors
Make it Home: Automatic Optimization of Furniture Arrangement
Computer-Generated Residential Building Layouts
Interactive Architectural Modeling with Procedural Extrusions
Metropolis Procedural Modeling
• Image Processing
 Domain Transform for Edge-Aware Image and Video Processing
 Non-Rigid Dense Correspondence with Applications for Image
Enhancement
Layout Generation
• Interactive Furniture Layout Using Interior
Design Guidelines
• Computer-Generated Residential Building
Layouts
Interactive Furniture Layout Using Interior Design
Guidelines (SIGGRAPH 2011)
Eric Schkufza
Paul Merrell
Zeyang Li
Stanford University
Maneesh Agrawala
Vladlen Koltun
University of California, Berkeley
Main idea
Contributions
• 1. Identify and operationalize a set of design
guidelines for furniture layout
• 2. Develop an interactive system for creating
furniture arrangements based on these
guidelines
Furniture Layout Guidelines
• 1. Functional Criteria
• 2.Visual Criteria
• 3. Authoring
Functional Criteria
• 1. Clearance
• 2. Circulation
• 3. Pairwise relationships
• 4.Conversation
Functional Criteria
• 1. Balance
• 2. Alignment
• 3. Emphasis
Generation Suggestions
Monte Carlo Sampler
• Density Function and Sampling
Results
Computer-Generated Residential
Building Layouts(SIGRAPH Asia 2010)
Paul Merrell
Eric Schkufza
Stanford University
Zeyang Li
Main idea
A list of high-level requirements
Computer-Generated Residential Building Layouts
Contributions
• Data-driven generation of architectural
programs from high-level requirements.
• Fully automated generation of detailed multistory floor plans from architectural programs.
• An end-to-end approach to automated
generation of building layouts from high-level
requirements.
Building Layout Design
• 1. Architectural Programming
• 2.Floor Plan Optimization
• 3. Generating 3D models
Data- driven Architectural Programming
• 1. Bayesian Networks
• Structure Learning
Floor Plan Optimization
• Proposal Moves
1.Notation
2.Sliding a wall
3.Swapping rooms
Cost Function
• Accessibility
• Dimensions
• Floors
• Shapes
Generating 3D models
• Passageways
• Windows
• Staircases
• Roofs
Results
Procedural & Interactive Modeling
• Interactive Architectural Modeling with
Procedural Extrusions
• Metropolis Procedural Modeling
• 1. Grammar-based: L-system…
• 2. Other: 3DMax…
Metropolis Procedural Modeling
Main Idea
Contribution
• An algorithm for controlling grammar-based
procedural models
Solutions
Problems
• 1. Generate the space of productions from the
grammar
• 2.Define an objective function that quantifies
the similarity between a given production and
the specification
• 3. Optimization
PROBABILISTIC INFERENCE FOR
GRAMMARS
LIKELIHOOD FORMULATIONS
• Image- and volume-based modeling
• Mondrian modeling
Optimization
• MCMC jump MCMC
1. Reversibility
2.Dimension matching
3Acceptance probability
MCMC FOR GRAMMARS
• Diffusion moves
• Jump moves
Results
Interactive Architectural Modeling
with Procedural Extrusions
Main idea
• Model complex architectural features,
including overhanging roofs, dormer windows,
interior dormer windows, roof constructions
with vertical walls, buttresses, chimneys, bay
windows, columns, pilasters, and alcoves.
USER INTERFACE DESCRIPTION
•
•
•
•
•
•
Modeling With Profiles
Plans and Profiles
Overhangs
Anchors
Plan Edits
Positioning Procedural Details
COMPUTING PROCEDURAL
EXTRUSIONS
•
•
•
•
•
Generalized Intersection Event
Edge Direction Events
Profile Offset Events
Insertions into the Polygon
Ambiguities in Procedural Extrusions
Image Processing
• Domain Transform for Edge-Aware Image and
Video Processing
• Non-Rigid Dense Correspondence with
Applications for Image Enhancement
Domain Transform for Edge-Aware
Image and Video Processing
Authors
Main idea
Contributions
Transform for Edge-Preserving Filtering
1D
5D
2D
Domain Transform
Application to Edge-Preserving
Filtering
Filtering 2D Signals
Results
•
•
•
•
•
Detail Manipulation
Tone Mapping
Stylization
Joint Filtering
Colorization
Non-Rigid Dense Correspondence with
Applications for Image Enhancement
Authors
Yoav HaCohen
Eli Shechtman
Adobe Systems
The Hebrew University of Jerusalem
Dan Goldman
Adobe Systems
Dani Lischinski
The Hebrew University of Jerusalem
Main idea
• The images are close to each other in time
and in viewpoint, and a dense correspondence
field may be established using optical flow or
stereo reconstruction techniques.
• The difference in viewpoint may be large, but
the scene consists of mostly rigid objects
• The input images share some common
content, but may differ significantly due to a
variety of factors, such as non-rigid changes in
the scene, changes in lighting and tone
mapping, and different cameras and lenses.
Overview
Nearest-neighbor search
• Based on Generalized PatchMatch
algorithm(SIGGRAPH 2010)
Aggregating consistent regions
• Global color mapping
• Search constraints
Evaluation
Applications
• Local color transfer
• Deblurring
• Mask transfer