CADIP Information Visualization Work: SFA

1
CADIP Information Visualization Work SFA

• David S. Ebert
• Computer Science Electrical Engineering
  Department
• University of Maryland Baltimore County
• ebert_at_.umbc.edu

Christopher D. Shaw Computer Science
Department University of Regina cdshaw_at_acm.org
2
Talk Outline

• Background
• Visualization Goals and Challenges
• SFA Overview
• Procedural Shape Visualization
• Results
• Recent Work with SFA
• New Features
• Interactive Lens Visualization
• Future Directions

3
Background Introduction

• Visualization
• Transforms the abstract and symbolic into the
  geometric
• Harnesses the human perception system (visual?)
• Glyph-based Volume Rendering
• Advantages of volume rendering
• Encodes multidimensional and multivariate
  information

4
Visualization Goal Strategy

• Goal
• Effectively convey information to the user
• Increase the quantity and clarity of the
  information
• Display only as much information as is
  perceptually understandable
• Strategy
• Use perceptual cues to aid understanding of
  multidimensional and multivariate data

5
Interface Goal Strategy

• Want 3D User Interfaces that are as easy to use
  as the WIMP style
• Use 3D Input devices
• Exploit 3D perception with animation
  interaction
• Enable fine manipulation
• Avoid user pain and fatigue

6
Challenge

• Rate of Information Increase Greater than Screen
  Resolution Increase
• Rapid increase in number of information sources
• Bandwidth of sources increasing
• Dimensionality increasing

7
Solutions

• More Effective Visualization Techniques
• Effective Use of Human Perception
• Utilize visual perception characteristics
• Add shading cues stereopsis to increase
  pre-attentive 3D perception
• Utilize proprioception - bodys innate sense of
  its position in space

8
Perceptive Senses Available

• Utilized in Near Term
• Visual
• 3D Spatialization
• Volume / Size
• Color
• Shape (curvature)
• Opacity
• Texture
• Haptic

• Experimental
• Proprioception
• Auditory
• Olfactory
• Ergonomics Issues
• Pain Fatigue

9
Our Glyph-based Visualization System SFA

• Minimally-Immersive VR Interface
• Multidimensional Multivariate Data
  Visualization
• Utilizes Many Perceptual Cues

10
SFA System Features

• Glyph Rendering
• Data dimensions mapped to glyph attributes
• Two-handed Interaction
• Stereo Viewing
• Multivariate, Multidimensional Time-varying Data
• Regular and Irregular Grids

11
Rendering Within SFA

• Visualizable Parameters
• Location (3)
• Color (1-3?)
• Transparency (1)
• Size (1-3)
• Shape (1-14)
• Surface Detail (1)

12
Perceptual Cues Used

• Shape / Texture
• Spatialization
• Color
• Volume / Size
• Proprioception
• Stereopsis

13
Minimally-ImmersiveUser Interface - Fishtank VR

• Access to environment
• Collaboration possible
• Low cost (lt 10K)
• Stereo viewing
• Two-handed interaction
• 3-space trackers with buttons
• Each hand has a distinct role
• Left hand sets up context
• Right hand performs fine manipulation

14
Two-Handed Minimal Immersion

• Left Hand
• Position and orientation of volume
• Selection of drawing context from 3D hierarchical
  menu
• Right Hand
• Volume subsetting
• Probe data volume
• Interactive lens visualization

• Takes Advantage of Proprioceptive Sense
• Hold Volume in 1 Hand, Operate on it with Other
  
• Hands Form a Kinematic Chain (Guiard)
• Left hand is base link, right hand operates
  relative to it Left is low frequency, Right is
  high frequency

15
Scientific Visualization Results

• Application
• Solar Magnetohydrodynamics Simulations
• Performance
• 2000 glyphs at interactive rates on an Indigo2
  High Impact

16
Information Visualization

• Visualize High Dimensional Abstract Data Spaces
• Examples
• Document similarities (50,000 dimensions)
• Database query routing, retrieval of meta-data
• Financial data
• Document Corpus Management
• Information analysis, not just retrieval
• Goals identify trends find anomalies, themes

17
Results

• Visualization of Wall Street Journal Corpus SFA
  Results IVEE Results

18
Shape Visualization

• Utilizing Pre-attentive Ability to Understand
  Shape (Parker et al.)
• Shapes Shouldnt Detract from Spatialization
• Intuitive Shape Mappings (curvature)

19
Parameterized Procedural Shape Visualization

• Automatic Generation of Shapes for Data
  Visualization
• Map Data Range to the Parameter Range
• Easy
• Supports continuous data ranges
• Parameterization generates a visual order
• May be very slow to generate shapes

20
Parameterized Procedural Shape Visualization

• Exploring Three Approaches
• Fractal Detail
• Vary surface roughness from smooth to mountainous
• Data value determines amount of fractal
  perturbation to the surface
• Superellipsoids
• Implicit Surfaces

21
Superquadrics - Superellipsoids

• Introduced to Graphics by Barr in 1981
• Extends Quadric Surfaces
• With two exponents that control the overall
  roundness or pointiness of the shape.

22
Superellipsoids

23
Perception of Superellipsoid Shapes

• Just Noticeable Difference Experiment Completed
  (Chris Shaw)
• Both Exponents Changed Together
• Result 21 Superellipsoid Shapes
  Perceivable

24
Scientific Visualization -Solar Wind MHD
Simulation

• Opacity j vorticity
• Shape j velocity -1
• convex positive velocity
• diamonds zero j velocity
• concave negative velocity

25
Scientific Visualization II

• Shape, color, opacity mapped to j vorticity

Shape, color, opacity mapped to j vorticity
26
Information Visualization

• X Gold
• Y US
• Z Federal Reserve
• Shape Stock Prices
• Color Noriega

27
Implicit Shape Visualization (Rohrer, Ebert,
Sibert)

• Implicit Surface Modeling (ISM)
• Blobby models, metaballs, soft objects
• Arbitrarily shaped functions
• Automatic blending, CSG
• Procedurally-generated
• Global Density Field (F)
• Produced from sum of potential field source
  functions

F(p) ? fi dist(p)
28
Implicit Shape Visualization Techniques

• 1. Document Content
• Shape (blobby text)
• 2. Corpus Relationships
• Proximity clustering
• 3. Combine
• Content (shape)
• Relationships (cluster shells)
• Focuscontext

29
Document Visualization Blobby Text

• Map n Dimensions to Equally-spaced Spherical
  Directions in 3-space
• n directions emanating from origin of sphere
• Magnitude of directional vectors proportional to
  corresponding data value
• Point source field function at end of vectors

30
(No Transcript)
31
(No Transcript)
32
New Examples
33
Transparency Textured Implicits
34
Corpus Visualization

• Compute Document Similarities
• Relationship graph
• 3D Proximity Clustering
• Mass-spring simulation
• Visualization
• Fit implicit surface model to cluster space

35
Combined Visualization

• 3D Proximity Clustering
• Semi-transparent cluster shells
• Individual Documents
• Content-based shape
• FocusContext
• Examples
  

InfoViz and Shakespeares Richard II and Richard
III
36
(No Transcript)
37
Recent Work with SFA

• New Features
• Multi-vector / glyph display added (3 data
  dimensions mapped to each vector direction)
• Multiple dataset display added
• Data filtering for each data dimension
• Communication with Fanatic Telltale
• Communication with Jackal agent completed

• Standard 2HI techniques now overloaded on mouse
  interface (subsetting, scale, rotation, zoom)
• Only 2HI feature not available in mouse
  interface is the Slice View
• Substantial documentation
• Interactive lens visualization
• Multi-threaded interaction

38
Interactive Lens Visualization Techniques

• Three Types
• Contour plate
• Texture-based volume rendering
• X-ray projection (thin slab)
• Allows Display of More Data Variables /
  Dimensions
• To Appear at IEEE Visualization 99

39
Contour Plate of Wall Street Journal
40
X-Ray Plate of WSJ Data
41
SFA Initial Perceptual Studies

• Perceptual Study Experiment Completed this Summer
• Initial Results
• 10 steps in transparency
• 21 in superquadric shape
• Preattentive test -- Wheres Waldo?
• Experiment Completed, analysis under way

42
Multi-Threaded Interaction Software
Architectures

• The usual drawing process is
• Draw on BackRGB while FrontRGB refreshes CRT
• Clear BackRGB frame buffer and Z-buffer
• Draw all objects in BackRGBZ
• Swap Back and Front RGB buffers
• Front buffer is used to refresh CRT
• All graphical objects updated simultaneously
• Many polygons to draw implies a slow update rate

43
Drawing with 3D Trackers

• Usual Process
• Get 3D Tracker position orientation
• Process button hits and interaction with 3D scene
• Draw 3D scene
• Draw 3D cursors
• Swap Buffers
• MR Toolkit
• Written by Shaw Green at U of Alberta
• Licensed by 600 academic research sites
  worldwide

44
3D Interaction

• For True Interaction, Need
• gt10 updates/second
• Lags lt 100ms
• Lag -- Still the Most Significant Problem
• Lags gt 50ms are noticeable
• Lags gt100ms start to affect performance
• Lags gt500ms destroy interaction
• Users adopt move-and-wait strategy

45
Fast Update

• How to Get Update Rates gt 10 Updates/sec?
• Buy a fast graphics computer
• Separate application from graphics
• SGI Performer uses a 3-stage pipeline
• MR Toolkit Computation process
• Decoupled simulation
• Draw less stuff
• Decimate polygonal model
• Draw cheap approximations
• Textures approximate object, perhaps

46
Lag

• Lag Is the Time From User Input to Graphics
  Response
• Sum of Lags in System
• Tracker lag
• Data transmission lag
• Drawing lag
• How to Get Low Lags?
• Predictive filters
• Tuning tracker data timing
• Fast update rate

47
Multi-Threaded Drawing

• Another Solution Class
• Draw in multiple frame buffer segments
  simultaneously
• Draw 3D Cursors on Overlay Planes
• Perform 3D interaction in real time
• Draw Full 3D Scene in RGB Z Planes
• Allow different update rates on the screen
• Same advantage as mouse cursor menus on a 2D
  frame buffer

48
Video

• Video Shows
• SFA with multi-threaded 3D trackers

49
Multi-Thread Advantages

• Allows Real-time Update of 3D Tracker Cursors
• Overlay update can do menu interaction
• Interaction with syntactic elements
• RGB update can take as long as it likes
• Limited semantic update is possible
• picking on the front-buffer data set
• Both RGB and Overlay threads are separate from
  any Computation process or thread

50
Plans

• Create More Java-based Tools (e.g., Alpha,
  Colormap Editors)
• Continue to Develop Version That Communicates
  With Current Agency Tools
• Finish Time-sequence Support
• Experiment with Performance of Java-based Glyph
  Visualization

• Add Isosurface Rendering for Cluster and Metadata
  Display
• Explore Metadata Visualization Techniques and
  Visualization of LSI Results
• Extend Perceptual Studies to User Studies of SFA

51
Conclusion

• Glyph Rendering Allows
• Comprehensible display of multiple variables
• Spatialization of complex relationships
• Minimally-immersive Visualization Aids
• 3D perception and feature detection
• Procedural Shape Generation
• Useful for encoding multiple data dimensions
  (1-14)
• Allows visual exploration of information spaces
• More Effective Use of the Human Perceptual System
  is Essential for Visualization

52
Acknowledgments - Collaborators

• UMBC / Baltimore / D.C.
• Randy Rohrer (DoD)
• Aaron Roberts (NASA)
• Jim Kukla, Pradyut Panda, Ted Bedwell, Amen Zwa
• Chris Morris, Alex Eller, Joe Romano, Ian
  Soboroff
• Funding DoD, NASA, NSF, NSERC

• U. Regina / U. Washington
• James Hall, Dee Jay Randall, Brook Bakay,
  Christine Blahut, Ben Korvemaker, Aaron Jones

53
MR Toolkit Structure

• Distributed Processes

Tracker Servers
Computation
3D Model
Inter-Process Communication
Shared memory
Master Process
Display
54
Many Update Loops

• The overlay scheme introduces 2 loops

Tracker Servers
Computation
3D Model
Tracker Thread
Shared memory
Shared memory
OverlayThread
RGB Thread
Display
55
Many Update Loops

• Overlays may access 3D Model

Tracker Servers
Computation
3D Model
Tracker Thread
Shared memory
Shared memory
OverlayThread
RGB Thread
Display