Matthias Trapp, Tassilo Glander, Henrik Buchholz, Jrgen Dllner

1
3D Generalization Lensesfor Interactive Focus
Context Visualizationof Virtual City Models

• Matthias Trapp, Tassilo Glander, Henrik Buchholz,
  Jürgen Döllner
• Hasso-Plattner-Institut,
• University of Potsdam Germany
• 12th International ConferenceInformation
  Visualization9-11 July 2008

2
Motivation Virtual 3D City Models

• Properties
• Large number of objects
• High degree of visual detail
• Tool for communicate complex 3D geoinformation
• Numerous applications
• ? Can lead to perceptional and cognitive overload

3
Motivation Generalized Virtual 3D City Models
Glander, ACMGIS 2007, ICA WS 2008
4
Motivation Combined Generalization Levels

• Simple case
• Single arbitrarily shaped, non-convex volume
  (lens shape)
• Two Levels of Abstraction (LOA)
• Model Complexity lt 500 MB (Geometry Textures)

Context (LOA1)
3D Lens Shape
5
Motivation Combined Generalization Levels

• Complex case
• 3 intersecting lens shapes (multiple foci)
• Three Levels of Abstraction
• Model Complexity gt 2GB

• Challenges
• Interactive manipulation of lens shapes
• Combined usage (intersection, nesting) of
  multiple lenses
• Enabling arbitrary lens shapes
• Handling spatial data complexity

6
Outline

• Related Work
• Conceptual Overview
• Preprocessing Focus Context Mappings
• Real-Time Rendering
• Application Examples
• Future Work Conclusions

7
Related Work

• Interactive 3D Focus Context Visualization with
  Lenses
• 3D Magic LensesViega et. al, UIST 1996
• A Solution for the Focus and Context Problem in
  Geo-Virtual EnvironmentsRopinski et. al, DMGIS
  2005
• Real-Time Volumetric Tests Using Layered Depth
  ImagesTrapp and Döllner, Eurographics 2008
• Generalization
• 2D Building Simplification Aggregatione.g.,
  Mayer, ISPRS 1998
• Single 3D Building Simplificatione.g., Kada,
  ISPRS WS 2006
• 3D Cell-Based City Model GeneralizationGlander
  and Döllner, ACMGIS 2007

8
Our Approach Conceptual Overview

• Preprocessing Phase
• Create levels of abstraction (LOA) and
  volumetric depth sprites (VDS)
• Rendering Phase
• Create / modify focus context mapping FNC
  map(VDS, LOA)
• Real-time image synthesis render(FNC)

9
Automatic Generalization of Virtual 3D City Models

• Input City Model CM
• Output Levels of Abstraction LOA
• Process

10
Preprocessing of Lens Volumes

• Input Derived or modeled solid polygonal shapes
  S
• Output Volumetric Depth Sprites VDS
• Processing

11
Mapping Generalization Levels to Lens Volumes

• Mapping for n lenses and generalization levels
• Mapping Properties
• Prioritized, hierarchical one-to-one mapping (i
  priority)
• Additional attributes (colors, lens positioning
  scaling,)
• Mapping can be changed at run-time

12
Real-time Rendering of Focus Context Mapping

• Multi-pass rendering clipping against VDS
• First pass render context
• Successively one pass per LOA
• Start with lowest priority i n

13
Applications Examples Usage Scenarios

• Scene Lens with intersecting foci of multiple
  non-convex volumes

14
Applications Examples Usage Scenarios

• Camera Lens with nested foci of the same convex
  volume

15
Limitations Future Work

• Current conceptual limitations
• Only one-to-one mapping possible
• Model complexity requires out-of-core rendering
• Clipping limitations
• Under-sampling / aliasing artifacts
• No capping of clipped areas
• Future Work
• Extend mapping mechanism one-to-many
• Compensate sampling artifacts
• Enable capping of clipped areas
• Use cases that exploit technical potential

16
Conclusions

• Wrap-up
• Concept technique for combing different levels
  of generalization
• Two phase process preprocessing rendering
• Hierarchical focus content mapping
• Interactive multi-pass rendering
• Potential for future work

17
Thank You

• Contact
• Matthias Trapp matthias.trapp_at_hpi.uni-potsdam.de
  
• Tassilo Glander
• tassilo.glander_at_hpi.uni-potsdam.de
• Henrik Buchholz
• henrik.buchholz_at_hpi.uni-potsdam.de
• Computer Graphics Systems Group Prof. Dr.
  Jürgen Döllner www.hpi.uni-potsdam.de/3d
• Researchgroup 3D-Geoinformation www.3dgi.de

18
Main References

• Glander, Döllner, Cell-Based Generalization of 3D
  Building Groups with Outlier Management, ACMGIS,
  2007
• Kada 3D Building Generalization Based on
  Half-Space Modeling, Proceedings of the ISPRS
  Workshop on Multiple Representation and
  Interoperability of Spatial Data, 2006
• Mayer Model-Generalization of Building Outlines
  on Scale-Spaces and Scale-Space Events,
  International Archives of Photogrammetry and
  Remote Sensing, Vol. 33, 1998
• Trapp, Döllner, Real-Time Volumetric Tests Using
  Layered Depth Images, Eurographics 2008