Science Visualization

Science Visualization Introduction

Outline • Definitions • History • Visualization Goals • Visualization Quality Criteria • Visualization Level • Visualization Process • Visualization Examples • Visualization Principles SciVis 2013 - page 2

Definitions “Visualization is a method of computing. It transforms the symbolic into the geometric, enabling researchers to observe their simulations and computations. Visualization offers a method for seeing the unseen. It enriches the process of scientific discovery and fosters profound and unexpected insights. In many fields it is already revolutionizing the way scientists do science.” McCormick, B.H., T.A. DeFanti, M.D. Brown SciVis 2013 - page 3

Definitions “The standard argument to promote scientific visualization is that today's researchers must consume ever higher volumes of numbers that gush, as if from a fire hose, out of supercomputer simulations or high-powered scientific instruments. If researchers try to read the data, usually presented as vast numeric matrices, they will take in the information at snail's pace. If the information is rendered graphically, however, they can assimilate it at a much faster rate.” R.M. Friedhoff and T. Kiely SciVis 2013 - page 4

Definitions “The use of computer imaging technology as a tool for comprehending data obtained by simulation or physical measurement by integration of older technologies, including computer graphics, image processing, computer vision, computer-aided design, geometric modeling, approximation theory, perceptual psychology, and user interface studies.” R.B. Haber and D. A. McNabb SciVis 2013 - page 5

Definitions “Scientific data visualization supports scientists and relations, to prove or disprove hypotheses, and discover new phenomena using graphical techniques. The primary objective in data visualization is to gain insight into an information space by mapping data onto graphical primitives.” H. Senay and E. Ignatius SciVis 2013 - page 6

History • China, 1137 • Leonardo da Vinci (1452-1519) SciVis 2013 - page 7

History • Cartography Iso-lines: Compass deviations Wind flow SciVis 2013 - page 8

Visualization Goals • Identification of Goals • Determination of the Data Class • Typical User Tasks • Typical User Strategies • High‐Level Goals SciVis 2013 - page 9

Identification of Goals • Why are data observed/measured/simulated and analyzed? • Is the goal to • Classify the data? • Compare datasets with each other? • Which features are looked for in the data? • What should be conveyed by means of a visualization? • Which portions of the data should be in focus? • Which portions should be displayed to provide context for the interpretation? SciVis 2013 - page 10

Determination of the Data Class • Quantitative or scalar • measured or simulated values • represent (real valued) numbers • example: families without TVs in different regions • Ordinal • There is an order between values, but values are not quantitative. • Example: days of the week, month • Nominal • no order, data represent categories • Example: car manufacturers and countries SciVis 2013 - page 11

Typical User Tasks • Keller and Keller (1993) established a list of general tasks: • Identify: establish characteristics y by which an object is recognizable • Locate: convey the (absolute or relative) position • Distinguish: recognize as distinct or different • Categorize: place into divisions • Cluster: group similar objects • Rank: assign an order relative to other objects • Compare: notice similarities and differences • Associate: link or join in a relationship • Correlate: establish a direct connection SciVis 2013 - page 12

Typical User Strategies • Shneiderman’s Visualization Mantra: • Overview: Gain an overview of the entire data, e.g. by browsing • Zoom: Zoom in on interesting items for a more detailed understanding • Filter: Select relevant items, e.g., by applying a dynamic query • Details‐on‐demand: Select an item or group and get details when needed, e.g., in a pop‐up window with more information SciVis 2013 - page 13

Typical User Strategies • Additional strategies: • Relate: View relationships among items, e.g. show connected data elements • History: Keep a history to use undo, redo, … • Extract: Extract relevant items in a format usable for later use SciVis 2013 - page 14

High‐Level Goals • Creation of a visualization: • Identify visualization goals • Map (abstract) data to graphics primitives which are rendered and displayed (make the invisible visible) • Exploit human capabilities to perceive colors, contrast and shape • Consider typical user tasks and strategies in interpreting a visualization SciVis 2013 - page 15

High‐Level Goals • Schumann and Müller (2000) came up with p the following typical strategies: • Explorative analysis • Confirmative analysis • Presentation of results • Presentation for decision support SciVis 2013 - page 16

Explorative analysis • Starting Point: data and no specific hypothesis • Process: undirected search • Goals: • Gain insight into the structure of the data: cluster analysis • Quality control • Gain insight into relations and trends • Identification, differentiation and classification of the data • (e.g. possible diagnosis or differential diagnosis) • Applications: • Exploration of large amounts of data: DB queries, simulations SciVis 2013 - page 17

Confirmative analysis • Starting Point: hypothesis concerning the data • Process: goal‐directed verification of hypothesis • Goals: • Comparison of values • Rejection or confirmation of hypothesis • Refinement of an initial diagnosis • Applications: • Diagnostic support : medical, debugging tools, … SciVis 2013 - page 18

Presentation of results • Starting Point: data has been carefully analyzed • Process: selection of visualization parameters which achieve the visualization goals effectively • Goals: • Emphasis of certain structures, • Comparison of structures, • Clarification of trends • Applications: • Generation of visualizations for education and publication SciVis 2013 - page 19

Visualization Quality Criteria • Questions: • Is the visualization method appropriate to achieve a certain visualization goal? • Are important structures and relations perceivable and recognizable? • Is it likely that important information is recognized efficiently (by preattentive vision)? • Is it likely that viewers draw wrong conclusions? SciVis 2013 - page 20

Visualization Quality Criteria • Influencing Factors: • Structure of the data (2D or 3D, spatial resolution, …) • Previous knowledge of users • Visual capabilities and preferences of users • Output medium (video, screen, paper, …) SciVis 2013 - page 21

Visualization Quality Criteria • Expressiveness: • Information is presented faithfully (unaltered) • Effectiveness: • An expressive visualization is effective if it employs the properties of the output medium effectively and if it considers the visual capabilities of the users. • Criteria: cognitive effort, interaction effort • Appropriateness: • The visualization technique is appropriate with respect to the computational effort. • Criteria: memory consumption and computational effort SciVis 2013 - page 22

Expressiveness • Example: car manufacturers and countries Source: Schumann and Müller (2000) SciVis 2013 - page 23

Effectiveness • Example: land prices in different regions Source: Schumann and Müller (2000) SciVis 2013 - page 24

Visualization Level • Elementary level • Each ( measured/simulated) value is / ) mapped to visual attributes (e.g. several diagram displays) • Abstraction of elementary data • Data are filtered or aggregated (e.g. histograms, isolines) • Visualization of high‐level results • Visualization of trends, emphasis of critical data (e.g. weather forecast) SciVis 2013 - page 25

Visualization Process Data Image Filtering Rendering Mapping Filtered Data Graphic Primitives SciVis 2013 - page 26

Filtering • Care must be taken in order to avoid misleading the viewer. • Types: • Signal processing: noise reduction, smoothing & sharping, histogram equalization, … • Resampling, e.g., on a different-resolution grid • Data interpolation, e.g., linear, cubic, … • Data derivation, e.g., gradients, curvature • Error correction • Reduction of data SciVis 2013 - page 27

Mapping • Make data “renderable” • Choice, placement, scale and parameterization of graphic primitives and visualization techniques • Definition of geometry, color, and material properties. • Visual Attributes: • Grey values/Intensities, Colors, Transparency • Textures • Lines (with color, width, pattern, e.g. Isolines) • Arrows (e.g., to visualize vector information) • Diagrams • Glyphs (iconic representations) SciVis 2013 - page 28

Rendering • Rendering = image generation with computer graphics • Visibility calculation • Adjustment of camera parameters • view direction, camera position, focal length, zoom, … • Adjustment of illumination parameters • number, type, position, and color of light sources • Rendering styles • Shading techniques (Gouraud or Phong shading) • Photo‐realistic or Illustrate visualization • Compositing (combine transparent objects,…) • Animation SciVis 2013 - page 29

Author vs. Viewer: Scenarios Author Viewer Looks at (the) final image(s). Modifies camera, clipping planes and illumination to explore the geometric model. Modifies filtering, mapping, and rendering parameters. Completes the whole visualization pipeline. • Completes all stages of the visualization pipeline. • Completes filtering and mapping (→ geometry model). • Completes filtering and provides a variety of mapping techniques. • Delivers data only. SciVis 2013 - page 30

Role of the Viewer SciVis 2013 - page 31

Data, Visualization, Interaction • Coupling varies considerably: • Data generation (data acquisition): • Measuring, simulation, modeling • Can take very long (measuring, simulation) • Can be very costly (simulation, modeling) • Visualization (rest of visualization pipeline): • Data enhancement, visualization mapping, rendering • Depending on computer, implementation: fast or slow • Interaction (user feedback): • How can the user intervene, vary parameters SciVis 2013 - page 32

Passive Visualization • All three steps separated: • Off-line data generation • Off-line Visualization • Previously generated data are visualized • Result: video or images/animation • Passive Visualization • Viewing of the visualization results SciVis 2013 - page 33

Interactive Visualization • Only data generation is separated: • Off-line data generation • Measurements, Simulation, Modeling • Interactive visualization • Previously generated data are available • Visualization program allows interactive visualization of the data • Possibilities: choice, variation, parameterization of the visualization technique • Nowadays widespread SciVis 2013 - page 34

Interactive Steering • All three steps coupled: • Interactive steering • Simulation and/or modelling (measuring) generate data “on the fly” • Interactive visualization allows “real-time” insight into the data • Extended possibilities: user can interfere with the simulation and/or the modeling, change the design, ... • Often requires lots of efforts, very costly SciVis 2013 - page 35

Data: General Information • Focus of visualization, everything is centered around the data. • Driving factor (besides user) in choice and attribution of the visualization technique • Important questions: • Where do the data “live” (data space) • Type of the data • Which representation makes sense (secondary aspect) SciVis 2013 - page 36

Data Space • Where do the data “live”? • inherent spatial domain (Science Visualization): • 2D/3D data space given • examples: medical data, flow simulation data, GIS data, etc. • no inherent spatial reference (Information Visualization): • abstract data, spatial embedding through visualization • example: data bases • aspects: dimensionality (data space), coordinates, region of influence (local, global), domain SciVis 2013 - page 37

Data Characteristics • What type of data? • Data types: • Scalar = numerical value (natural, rational, real, complex numbers) • Non-numerical values • Multi-dimensional values (N-dimentional vectors, N×N-dimentional tensors of data from same type) • Multi-modal values (vectors of data with varying type [e.g., row in a table]) • Aspects: dimensionality, co-domain (range) SciVis 2013 - page 38

Examples • Scalar field visualization • Iso-surface rendering • Volume rendering • Transfer functions • Volume lighting • Unstructured grid visualization SciVis 2013 - page 39

Examples • Vector field and flow visualization • Direct vs. indirect techniques • Particle tracing • Integral curves and surfaces • Dense flow visualization techniques SciVis 2013 - page 40

Examples • Tensor visualization • Tensors are geometric objects that describe linear relations between vectors, scalars, and other tensors. • Elementary examples of such relations include the dot product, the cross product, and linear maps. • Vectors and scalars themselves are also tensors. • A tensor can be represented as a multi-dimensional array of numerical values. SciVis 2013 - page 41

Examples SciVis 2013 - page 42

Examples • N1→R1 SciVis 2013 - page 43

Examples • R1→R1 SciVis 2013 - page 44

Visualization Principles • Suitability of a visualization technique depends on: • the data (dimension, range, scalar or vector data), • the primary visualization goal, • and (less important) the application area. • Visualization techniques which are wide‐spread in one application area may also be useful in others. • Often, data have to be converted. • This is a frequent source of inaccuracy! • The results must be carefully checked. SciVis 2013 - page 49

Visualization Principles • It is important to consider data AND its context • Data visualization: convert a given portion of the data to an image by means of some visualization technique. • Context visualization: essential to enhance the interpretation of the entire visualization • Figure captions (what is displayed?), • Coordinate axes (indicate perspective), • Scales (legends), • Labels, • Contours, • Landmarks (anatomic, geographic). SciVis 2013 - page 50

Science Visualization