"We propose to systematically address the question of using cinematic effects in visualizing data about land-use change in large regions as simulated in LEAM. We will document those cinematic effects that relate to space and time in the landscape. We will also document the kinds of inferences we must support through visualizing LEAM output in virtual worlds. We will then map between these two sets to identify what cinematic effects support what kinds of inferences, as well as conditions in the underlying data that warrant triggering a particular cinematic effect. Finally, we will implement these effects and data triggers, and then assess the extent to which they are successful in enhancing visualization in virtual worlds."
While we have increasingly sophisticated ways of modeling how complex systems operate over space and time, we do not as yet have visualization tools that support the different kinds of inferences that can be made using these models.
For
instance, the Land-use Evolution and impact Assessment Model
(LEAM; http://www.leam.uiuc.edu/)
simulates the different complex ways in which urban development and change can
transform a region over time. Data produced in a LEAM simulation can be visualized
using maps, which allows us to make some kinds of assessments:
Maps are much less effective in answering a different class of questions:
Virtual worlds are perhaps the best way of answering this class of questions, those that an individual can only consider by navigating through space and in time. The 'egocentric' views that are afforded by virtual worlds allow many assessments and inferences that are not possible through maps, which are 'exocentric' views.
A virtual world provides an 'egocentric' view
Our recent work has addressed some technical and artistic issues in visualizing in CUBE-based virtual worlds the dynamic spatial data produced by LEAM simulations. For instance, how do we deal with the large quantity of data associated with large regions? Our approach involves reading in only the portion of the data needed to construct the world immediately surrounding the viewer, the amount of data being determined by the capacity of the system to handle rendering and display. An artistic question we have addressed: How do we translate the numeric model output into structures and landscapes that make up a meaningful world? We have considered different realistic and iconic representations, and are finding that perhaps a combination of iconic and realistic representations are most effective.
non-realistic 'iconic' view with realistic trees
Our work has not, we believe, fully exploited the artistic potential of the medium. In particular, we have not tapped into the wealth of experience and knowledge, about controlling sequences of scenes to evoke certain responses in viewers, that has been assembled in over a century of cinema. This not only consists of standard effects (mise-en-scene, transitions, cutting, camera movements such as panning, travelling, etc) but also instances where these standards have been violated to great effect. Cinematic effects have been used in virtual worlds, but in their role as a narrative medium [1][2][3](more to come...). That is, these effects are used to enhance the evocative quality of one or more narrative threads. To our knowledge these effects have not been considered as a way of enhancing virtual worlds as a visualization medium. The effects have not been used to enhance the ability of people to discover answers to questions, or even raise new questions, as they encounter the data.
For example, to make judgments involving the passage of time, an individual has to be made aware of the passage of time in the scene they are watching. How is this done in cinema? How can we use that knowledge to convey the passage of time to an individual in the virtual world? How can we trigger this effect automatically based on changes in the underlying data or relationships among components?
We propose to systematically address the question of using cinematic effects in visualizing data about land-use change in large regions as simulated in LEAM. We will document those cinematic effects that relate to space and time in the landscape. We will also document the kinds of inferences we must support through visualizing LEAM output in virtual worlds. We will then map between these two sets to identify what cinematic effects support what kinds of inferences, as well as conditions in the underlying data that warrant triggering a particular cinematic effect. Finally, we will implement these effects and data triggers, and then assess the extent to which they are successful in enhancing visualization in virtual worlds.
Cinematic Techniques, non-interactive, already scripted
[1] Mitologies: Medieval Labyrinth Narratives in Virtual Reality, Maria Roussos and Hisham Bizri, UIC (http://www.evl.uic.edu/mariar/DOCS/VW98/vw98.html)
Cinematic Techniques, camera-controller interfaces
[2] CINEMA: A System for Procedural Camera Movements (1992) Steven M. Drucker, Tinsley A. Galyean, David ZeltzerProceedings of the 1992 Symposium on Interactive 3D Graphics, Special Issue of Computer Graphics, Vol. 26.
[3] MIKE: a Multimodal Cinematographic Editor for Virtual Worlds http://www.inesc-id.pt/pt/indicadores/Ficheiros/971.pdf
[4] Field Cinematography Techniques for Virtual Reality Applications, Michael Naimark VSMM98 Conference Proceedings (http://www.naimark.net/writing/gifu.html)
[5]...
[6] Lev Manovich, The Language of New Media
Great Links: