Interaktiv 3D datorgrafik - mer än bara spel Stefan Seipel University of Gävle Stefan Seipel 2008-05-07
GraphiX @ HiG Teaching: Creative Computer Graphics Computer Science - Visiomatics Research: Computer Graphics Visualization Image Processing Decision support Stefan Seipel 2008-05-07
Graphics & Visualization Research Objectives: * Provide supportive tools for decision makers * Support group-work situations * Provide intuitive interaction * Gain insight by interaction (visual analytics) Stefan Seipel 2008-05-07
S upporting time-critical decision making in process industry Stefan Seipel 2008-05-07
Time varying scalar field data in pulp process industry The Lime Kiln – Part of the Chemical Process Process m onitoring � Efficiency � Energy savings � Safety Monitored variable � Temperature Stefan Seipel 2008-05-07
Monitoring of Temperatures - Termographic Imaging • Infrared line scanner • Kiln is revolving at 1/40 s -1 • Full surface scan every 40 sec. • Data matrix 123 x 512 scalar values – Value range: 50°C – 450°C – Instantaneous temperature Stefan Seipel 2008-05-07
Visual Monitoring of 2D Scalar Field Data Part of the operator’s environment The conventional visualization Stefan Seipel 2008-05-07
Critical situations to identify • Bonded deposits � slow dynamic process ( days-weeks ) � low variability of temperature • Variations of fuel and chemical composition � relative slow process ( minutes to hours ) � relative low variability of temperature • Localized damages of insulation � rapid dynamic process ( seconds to minutes ) � sudden increase of temperature Stefan Seipel 2008-05-07
Current form of visualization Problems perceived by process operators: • Instantaneous temperature • Assessment of deviations by comparison (page flipping) • Poor variation of color in region of interest Normal temperature range Stefan Seipel 2008-05-07
Current form of visualization Identified problems according to theory: • Existing palette is a fixed palette • Current color palette is quantized (approx.40 steps) • Δθ is not immediately visualized • Δθ and θ difficult to assess simultaneously Range of Interest 50 o 450 o Normal Temperature Range Stefan Seipel 2008-05-07
Color map improvements Gray scale Heated Iron -> Absolute Dual Ended Existing -> Deviation Min-Mean-Max Stefan Seipel 2008-05-07
General mapping plausible combinations V isualization Color Shape Same as used today, however with Heated Iron 2D map θ - scale 2D map Δθ - 3D field θ θ 3D field θ Δθ New 3M scale for Δθ 3D field Δθ θ 3D field Δθ Δθ 3D cylindrical map θ θ 3D cylindrical map θ Δθ 3D cylindrical map Δθ θ 3D cylindrical map Δθ Δθ Natural mapping to real situation Stefan Seipel 2008-05-07
Visualization Results θ -> Color Δ θ -> Color θ -> 3D Shape θ -> Color Δ θ -> 3D Shape Stefan Seipel 2008-05-07
Visualization Results Stefan Seipel 2008-05-07
What are the effects in real use? Field study in real working environment: • Real working teams in sharp environment • Additional virtual lime kiln • Simulated local insulation defects Temperature increase Temperature increase modulates 40s intensity of defect pattern 400 350 300 Degrees C 250 200 150 100 50 0 0 1 2 3 4 5 6 7 8 9 Interval Stefan Seipel 2008-05-07
Simulator study Experimental design: • 2 Teams (B and F) • Each team tested all 3 conditions (3 afternoon shifts) • 6 simulated incidents during 4 hours shift • Depending variable: Detection time • Main factor: Visualization • 2 x 6 Observations per condition Stefan Seipel 2008-05-07
Results of the Simulator Study Pooled Detection Times (Both Teams) 250 225 200 Detection Time [s] 175 150 Median 125 Mean 100 75 50 25 0 2D Map 3D Field 3D Tube Stefan Seipel 2008-05-07
Results of the Simulator Study Detection Times (Median) Between Teams 250 225 200 Detection Time [s] 175 150 B-Team 125 F-Team 100 75 50 25 0 2D Map 3D Field 3D Tube Stefan Seipel 2008-05-07
Results of the Simulator Study Detection Intervals 6 4 B-Team F-Team 2 0 2D Map 3D Field 3D Tube Stefan Seipel 2008-05-07
Conclusion • Operators appreciate the visualization (in daily use since January 2006) • New color scale for visualization of deviations • 3D gave better performance better than 2D • 3D cylinder intuitive and natural mapping • 3D map was better than 3D cylinder • No factorial analysis of shape and color Stefan Seipel 2008-05-07
S upporting decision makers in group-work situations Stefan Seipel 2008-05-07
Collaborative & interactive 3D graphics Traditional geographical maps are 2D Showing a 2D image on a 2D display is trivial No problems for existing SW to visualize in table-top displays Most geographical map viewers are 2D or at best 2.5D Early versions of TT displays in com m and and control Stefan Seipel 2008-05-07
What if we would show real 3D features ? …we would get a hole lot of trouble…technically speaking Eye’s positions needed (tracking) Multiple individualized views Dynamic rendering even for static content Stereo cues (optionally) Requirem ents to visualization of 3 D features Stefan Seipel 2008-05-07
Is it feasible ? Com puter Sw eden, 2 0 th of October 2 0 0 4 Stefan Seipel 2008-05-07
Is it feasible ? …there is a technical challenge… Network Network A schem atic of Mim er’s w ell Stefan Seipel 2008-05-07
This is what we can achieve… I ndividual content Dynam ic view ing conditions Pettersson, Spak and Seipel, 2005 : “Collaborative 3D Visualizations of Geo-Spatial Information for Command and Control”, proceedings of Sigrad 2004 Stefan Seipel 2008-05-07
GIS and 3D TT Displays GIS applications do largely not exploit the functionality of advanced (3D) table-top displays Dynamic and observer dependent projections still a mystery Potential benefits not yet clearly perceived by GIS community When is observer dependent (3D) content useful e.g. for GIS? Stefan Seipel 2008-05-07
a) Improved text and symbolic presentations Experiment: Comparing the efficiency in identifying symbols presented using three different forms of visualization. Viewing angle Viewing angle 1. 1. Display surface Display surface 3. 3. 2. 2. Visualization conditions as on screen 1. Frontoparallel 2. Flat 3. Reference condition Covers the same display area as frontoparallel Pettersson, L.W.; Lind, M.; Spak, U. & Seipel, S. (2005),Visualization of symbols in a horizontal multiple viewer 3D display environment, in 'Proceedings of IEEE Information Visualization 2005', pp. 357-362. Stefan Seipel 2008-05-07
a) Improved text and symbolic presentations 3400 3400 Test leader Test leader 3200 3200 Test subject Test subject 3000 3000 Polarizing Polarizing 2800 2800 Time in milliseconds Time in milliseconds eyewear eyewear 2600 2600 Cheek mount Cheek mount 2400 2400 2200 2200 Response gear Response gear 2000 2000 1800 1800 1600 1600 1400 1400 1200 1200 1000 1000 Frontoparallel Frontoparallel Flat Flat Ref. Condition Ref. Condition Experim ental procedure Results 12 subjects Frontoparallel versus 2D normal flat: (t= -2.31, N=11, p < 0.022) 50 trials in each of the three conditions Frontoparallel versus 2D control condition: Score: Average time of lat 10 trials per condition (t= -2.93, N=11, p < 0.007). Stefan Seipel 2008-05-07
b) Collaborating in complex tasks Experiment: Solving visually complex tasks by collaborating using partitioned subtasks. Partitioned Views Integral View Pettersson, L.W.; Kjellin, A.; Lind, M. & Seipel, S. (2006),'Effects of Layer Partitioning in Collaborative 3D Visualizations''Proceedings of the 2nd International Symposium on Visual Computing ISVC 2006', Lake Tahoe, Nevada. Stefan Seipel 2008-05-07
b) Collaborating in complex tasks Views of the actual experimental task – a 3D and logical/combinatorial task Integral view (first person view) Partitioned view for user A Experim ental procedure: 2 0 persons in 1 0 Dyads 5 trial stim uli + 2 x 1 5 stim uli Blocks of 1 5 stim uli is random ized Stefan Seipel 2008-05-07
b) Collaborating in complex tasks Results in a nutshell: ANOVA: main effect of condition (F(1,8)=14.30, p<0.0054) main effect of order (F(1,8)=10.54, p<0.012) For the details see: Pettersson, L.W.; Kjellin, A.; Lind, M. & Seipel, S. (2006),'Effects of Layer Partitioning in Collaborative 3D Visualizations''Proceedings of the 2nd International Symposium on Visual Computing ISVC 2006', Lake Tahoe, Nevada. Stefan Seipel 2008-05-07
c) Understanding time and geography Series of experiments investigating the space-time cube : The space-time cube uses elevation over ground to depict time 2D trajectory of moving targets 3D trajectory of moving targets W orks subm itted to I nform ation Visualization Journal and Journal of Applied Perception Stefan Seipel 2008-05-07
Recommend
More recommend
