SciVL: A Descriptive Language for 2D Multivariate Scientific Visualization Synthesis presented by Jason Sobel advisor: Prof. David Laidlaw
Feb 01, 2016
SciVL: A Descriptive Language for 2D Multivariate Scientific Visualization Synthesispresented by Jason Sobel
advisor: Prof. David Laidlaw
Road Map
Motivation and Introduction
Implementation
Language Specification
Conclusions and Future Work
Motivations
Good visualizations take time
1. Decide on “visual elements”
2. Code and debug
3. Evaluate and iterate
Motivations (cont.)
“Optimize” visualizations Find best combinations of visual properties
Our Question
Can we provide a fast and easy way to prototype visualizations that also allows optimization?
Proposed Solution
Define a language that can be used to represent a visualization
Create an instance in a text file
Apply an instance to a dataset to generate an image
Goals
The language should be:1. Simple
2. Expressive
3. Flexible
4. Hierarchical
5. Easily broken in to “genes”
Contributions
Understanding of “key” visual properties
Rapid prototyping system
Foundation for future work
Road Map
Motivation and Introduction
Implementation
Language Specification
Conclusions and Future Work
Layer System
Three types of layers: IconColorplaneStreamline
Each layer defines some number of visual elements
Rendering
A SciVL file specifies an arbitrary number of layers
They are combined to produce the final image
Values: Specifying “Numbers”
Visual properties are not given number values in the SciVL file
They are given abstract Values, one of:ConstantRandomData-driven
Realization
When rendering a layer, we realize a Value to get a numberUse location to map to data
Values Example
Icon Layer
Let’s look at all the properties of an icon layer
The following images were made using a gradient dataset0 on the left to 1 on the right
All Forms
Circle Form
Rectangle Form
Triangle Form
Multi-Offset Forms
Compound Forms
Color
Color (Partial Range)
Alpha
Borders
Border Color
Border Alpha & Width
Spacing
Orientation
Texture
Failures
Jitter
Example Icons
Colorplane Layer
Used for “regions” or “washes” of color
Colorplanes
Colorplanes in Use
Streamline Layer
Useful for visualizing vector data like velocity or vorticity
Streamlines Color & Alpha
Streamlines Width & Texture
Streamline Density
Road Map
Motivation and Introduction
Implementation
Language Specification
Conclusions and Future Work
Layer System
The language specifies visual elements layer by layer
The syntax is a simple interface to all the properties described above
Allows specifying a Value for each one
VisEl LayerBEGIN_LAYER VISELNVISELS 1BEGIN_VISELPOISSON POINT Constant .5 Constant .5 Constant 0NFAILS 0NFORMS 1BEGIN_FORMSTAGESHAPE Constant squareNOFFSETS 2 OFFSET POINT Constant 0 Constant 0 Constant 0 OFFSET POINT Constant 5 Constant 0 Constant 0BEGIN_STYLENCOLORS 1 POINT Variable gradient_x .4 .6 Constant .8 Constant .8NALPHAS 1 Constant .8NTEXTURES 0NORIENTATIONS 1 Random 0 .1 NBORDERS 1 COLOR POINT Variable gradient_y 0 .3 Constant .7 Constant .8 ALPHA Random .8 1 WIDTH Constant 2NSCALES 0NDIMENSIONS 1 POINT Variable gradient_y 3 6 Constant 0 Constant 0END_STYLEEND_FORMSTAGEEND_VISELEND_LAYER
Demo
Colorplane Layers
Similar syntax Can control, per vertex:
FailuresColorAlpha
Streamline Layers
Similar syntax Can control:
Failures Vector to follow Survival Density Color/Transparency Size Texture
Road Map
Motivation and Introduction
Implementation
Language Specification
Conclusions and Future Work
More Pictures
Success?
Goals were:1. Simple
2. Expressive
3. Flexible
4. Hierarchical
5. Easily broken in to “genes”
Did we accomplish these goals?
Anecdotal Feedback
A “design-expert” professor from RISD
A scientist with radar polarimetry data
Challenges
Allowing every possible combination
Interfacing with any kind of data
Finding “correct” visual elements & properties
Future Work
Genetic AlgorithmsCan we create the perfect visualization?Was man meant to play God?
Visualization “Rules”Can we find “The Do’s and Don’ts” of
Scientific Visualization?
Thanks
Prof. David Laidlaw
Daniel Acevedo
Cullen Jackson
Eileen Vote
David Karelitz
Daniel Keefe
Prof. Fritz Drury
Dean Turner
Prof. Andy van Dam
Morriah Horani
Sci Vis
Family and Friends