Adjust, Just Adjust Eduard Gröller Institute of Visual Computing & Human ‐ Centered Technology TU Wien, Austria
Dagstuhl 2007 Eduard Gröller 2
Action without Interaction Autonomous Vehicles [1] [2] Artificial Intelligence (machine learning, deep learning, CNNs, …) [3] Eduard Gröller 3
Chameleon Dynamic Color Mapping for Multi ‐ Scale Structural Biology Models N. Waldin, M. Le Muzic, M. Waldner, E. Gröller, D. Goodsell, A. Ludovic, I. Viola
Multi ‐ Scale Structural Biology Models Nicholas Waldin
Multi ‐ Scale Structural Biology Models Nicholas Waldin
Chameleon: Overview Initial color assignment zoom View ‐ dependent color adjustments Hierarchical color Nicholas Waldin
Chameleon Nicholas Waldin
Output ‐ Sensitive Interaction Peter Mindek, Gabriel Mistelbauer, Eduard Gröller, Stefan Bruckner TU Wien, Austria University of Magdeburg, Germany VRVis Research Center, Austria University of Bergen, Norway
Output ‐ Sensitive Interaction Input (I), Interaction ( Δ I), Paramters (P), Output (O), Transformation (T) I P O Δ I Δ P Δ O Δ I Δ O / / P = T(I) Δ I Δ O Peter Mindek, Eduard Gröller 10
Mapping of the Slider to the Image Stack Positon Eduard Gröller, Peter Mindek 11
ArcBall: Data ‐ Sensitive Guidance (3D Rotation) Peter Mindek, Eduard Gröller 14
Output ‐ sensitive Interaction: Summary Δ I Δ O Output ‐ sensitive Interaction Data ‐ sensitive navigation Manipulation Guidance Peter Mindek, Eduard Gröller 16
Visualization of 4D Ultrasound Data [Stefan Bruckner] HD live – together with Kretztechnik (GE) [Varchola et al. 2012] [Karimov et al. 2016] Live Fetascopic Rendering [6] Eduard Gröller 17
Eduard Gröller 18
Visualizations for Broad Audiences ‐ Lessons Learned Interaction constrained by Hardware/software Hardware design with a few available knobs Minimize interaction to pre ‐ sets Few view positions Few lighting stages Heterogeneous audience Doctors do diagnosis Grandparents enjoy images [6] [7] Aesthetics important, more than realism Eduard Gröller 19
Visualizations for Broad Audiences ‐ Lessons Learned System must be Interactive Robust (graceful degradation) No [costly] pre ‐ processing possible Engineering effort Just good enough solution [6] [7] Not 100% realistic, not 100% correct Glanceable visualization, graspable interaction, small learning effort Adhere to established preconceptions Users got accustomed to a fast, heuristic model Acceptance of more elaborate model difficult Eduard Gröller 20
Eduard Gröller 21
Developing Visualizations for Broad Audiences Automatic, context ‐ aware adaptation of visual/interaction channels Simple, intuitive output ‐ sensitive interaction Δ I Δ O Glanceable visualization, graspable interaction Eduard Gröller 22
Δ I Δ O Adjust, Just Adjust Acknowledgements Ludovic Autin Mathieu Le Muzic Andrej Varchola Stefan Bruckner Peter Mindek Ivan Viola David Goodsell Gabriel Mistelbauer Nicholas Waldin Alexey Karimov Gerald Schröcker Manuela Waldner
References [1] https://tctechcrunch2011.files.wordpress.com/2017/10/efp_170925_toyota ‐ autonomous ‐ car_0085 ‐ edit ‐ 2.jpg?w=738 [2] http://vehiclepassion.com/wp ‐ content/uploads/2017/02/ehang ‐ 184 ‐ aav ‐ flying ‐ taxi ‐ in ‐ UAE.jpg [3] https://medium.com/@xenonstack/log ‐ analytics ‐ with ‐ deep ‐ learning ‐ and ‐ machine ‐ learning ‐ 20a1891ff70e [4] Waldin, N., Le Muzic, M., Waldner, M., Gröller, E., Goodsell, D., Ludovic, A., Viola, I.: Chameleon ‐ Dynamic Color Mapping for Multi ‐ Scale Structural Biology Models. Proceedings of Eurographics Workshop on Visual Computing for Biology and Medicine (EG VCBM), Sep 7 ‐ 9, 2016, pp 11–20. (Honorable Mention Award) [5] Mindek, P., Mistelbauer, G., Gröller, E., Bruckner, S.: Data ‐ Sensitive Visual Navigation. Computers & Graphics 67: 77 ‐ 85 (2017) doi: 10.1016/j.cag.2017.05.012. (Special Section on SCCG 2017, Best Paper Award) [6] Varchola, A.: Live Fetoscopic Visualization of 4D Ultrasound Data. PhD Thesis, TU Wien (2012). [7] GE Healthcare. Voluson E8 Expert, available from http://www3.gehealthcare.com/en/Products/Categories/Ultrasound/Voluson/. Accessed: 2018 ‐ 01 ‐ 15. Eduard Gröller 25
Abstract Adjust, Just Adjust Developing visualizations for broad audiences requires glanceable graphics and graspable interactions. This talk will concentrate on interaction facilitation through automatic adjustments. The first example illustrates an automatic color scale adjustment in a bio ‐ molecular setting to accommodate contradicting and overlapping color schemes across scales. The second example discusses output ‐ sensitive interaction to make changes in the input proportional to changes in the output, or to visually indicate the sensitivity of input changes with respect to output changes. The third example deals with visualization of 4D ultrasound data, which is targeted to a broad audience in prenatal imaging and diagnosis. Lessons learned during this project are presented. The talk makes a case for automatically reducing interaction complexity in visualizations for broad audiences. Eduard Gröller 26
Recommend
More recommend