Dynamic Data Sculpture
Dynamic Data Sculpture
Physical embodiment of information in a tangible presence
John Capogna, IBM Research Intern
Mauro Martino, Intern Mentor
A display that changes and deforms its shape dynamically according to information being shown on it
Volume and depth are considered, must be able to move around it and view different angles
Little to no direct physical manipulation by the user; movement is automated
How can we combine these ideas to make something useful to people who work with information?
We conceptualized a new kind of display which can change dynamically as new data is being shown on it.
For this, it’s important that volume and depth are preserved, so that people can walk around it and view it from many angles.
it’s also important for the interaction to be seamless. a person shouldn’t have to operate any complex mechanisms to see their results.
Let’s look at some of the problems we can identify with our current models.
Let’s also see what works already with existing models.
Screen vs Installation
Two ways to show the number of people on earth
Of All the People
Source: Analyzing the Design Approaches of Physical Data Sculptures, University of Sydney, Andrew Vande Moore and Stephanie Patel, University of Sydney, Australia
Here is a screen-based versus physical data representation of an identical data set.
on the left is a pixel-based representation of the number of people on earth. you are the tiny pixel at the top left, and you can scroll to the right for a very long time to reach the last pixel at the bottom right.
on the right is a large warehouse filled with grains of rice, arranged into piles according to category, “number of astronauts, number of mothers, etc”
The findings showed that aside from the emotional context the physical embodiment provides, its main success is that it communicates insight. the physical installation’s design rationale is motivated by exploiting physical qualities to capture the attention of lay people to data-related issues that would otherwise be disregarded if shown factual descriptions, journalistic articles, infographic illustrations or data graphics.
What about Spatial Visualizations?
Point-based displays and information landscape
Source: Spatialization Design:Comparing Points and Landscapes, Melanie Tory, David W. SPrague, Fuqu Wu, Wing Yan So, Tamara Munzer, University of Victoria, University of British Columbia
This research, conducted at the university of victoria and the university of british columbia presented an experiment in which participants were given different visual representations of spatialized data (representations of non-spatial data in a spatial layout similar to a map). The participants were asked to identify groups of points, and based on how quickly they were able to do so and the speed at which they were able to do so, it was determined which spatializations were most effective. The findings were that point based specializations were far superior to landscapes, and that 2D was superior to 3D.
This paper didn’t resonate with me so well. First of all, the huge oversight here is that these 3D landscapes were presented on 2D screens. It isn’t surprising that participants found the landscapes confusing or distracting because there are natural occlusions that occur when trying to view information hidden behind things. However, the researchers didn’t have any other type of display to present to people, so they didn’t have many options. Likewise, they weren’t able to see this bias because all they know are the screens to which they are accustomed.
My argument is that if we were to ask these same participants to observe the 2D point cloud on a 2D surface and the 3D landscape on a 3D surface the results would have been very different.
We are going to try and create a new type of display
We wanted to try and solve this problem, and to create a display that would show this same type of volumetric data on a volumetric surface.
we would then conduct tests similar to those done at the university of victoria and the university of british columbia to see if I could prove somehting different.
the goal is to demonstrate that everyday people can recognize information patterns quickly and easily when allowed to use the spatial parts of their brains, which we do unconsciously.
So how are we going to make this display?
We are going to use the spandex membrane I showed you earlier because of it’s flexible properties.
We are going to use motors so we can create a platform for automatic and dynamic movement with little human interference.
We are going to figure out how to attach the motors to the membrane so that they hold a precise numerical position that can later be correlated to numbers from a dataset.
So let’s get started.
12 pages from slides with building process
Short term goals for improvement
Increase modularity with removable headers
Motors are noisy! Will try microstepping (also improves holding torque)
Write more robust software to prevent blocking
Create more elaborate visualizations using real data sets
Sound to reinforce visual feedback(?)
We have been working on this display for the last ten weeks, and we have roughly another four weeks to make progress.
Some short term goals I would like to implement are:
Increase modularity with removable headers. This way, if a driver fails we can easily replace it.
The motors are still noisy, though some immediate measures can be taken to reduce this. Microstepping is one of them, which has the added benefit of improving the motors’ holding torque.
In my current understanding of the motor software, the functions are blocking. This means that a motor doesn’t move to its new position until the previous motor has finished. There are ways around this, but we are still gaining familiarity with the software phase.
The very next step is to get communication working so that visuals can be projected in real time.
And if there’s time, maybe we’ll include some sound feedback as well.
(cont.) Steps for the next few weeks
With a working prototype, collaborate with other members of IBM Research to discover the kind of content that’s best suited for this type of display
User Testing, compare results with comparison studies
Properly Document (Video, Web)
Publish Paper with findings
I’d like to take the coming weeks to work with members of the IBM Research team on this floor to discover the kind of content that’s most ideally suited for this type of display.
With user testing, we can also compare our findings with those of the earlier research I mentioned involving spatialization studies to see if we’ve made a difference in the way people can perceive visual information.
It’ll be important to document this project properly, with a good vide, and publish it to the web.
And lastly we’ll publish our findings in a paper.
Continue to make spatial work and think volumetrically
In the long-term, I would like to make this bigger. It would be interesting to see this on an architectural scale.
We can create better resolution by increasing the number of nodes.
As a personal long-term goal, I hope to create more spatial work and continue to think volumetrically.
This is something I consider to be a lifelong journey.
Questions That Will Be Answered
How can we physically embody data in a tangible presence?
How can we draw connections between tangible features and the information they represent? How do shape, movement and space inform these connections?
How can we engage and teach audiences through interaction?
How can we explore information as a medium for communication beyond pixel-based screens?