Reviews of 3-D Smooth Zooming
- Project by Ninad Jog & Bruce Lin
- Reviewed by Claudio Esperanca and Greg Baratoff
Claudio Esperanca's Review
The problem tackled by the authors is apparently trivial, given that it has been around for so much
time and so many techniques have already been devised to solve it. The authors however were astute
enough to understand that, in practice, a simple approach yields frequently a better result than a
more complicated (albeit more "technical") one. The original implementation of FilmFinder used
arrays for storing the database and the authors chose not to modify this structure - which was
probably a wise decision. Trying improvements in increasing order of complexity
also typifies this "simplicity oriented" approach.
It must be said however that authors missed some interesting (and possibly more promising)
alternatives. Instead of scheme #4, which consists in using a simple regular grid data structure,
the authors could have tried, for instance, a hierarchical data structure such as a range tree or
a priority-search tree, which are spatial data structures usually prescribed for this kind of
query (See section 2.5 of "The Design and Analysis of Spatial Data Structures", by Hanan Samet,
Addison-Wesley, 1990). Even if not actually implemented, a more detailed survey of applicable
techniques should have been presented in the report.
Greg Baratoff's Review
The goal of this project is clearly stated, and the challenges in
achieving the goal are well specified. The existing FilmFinder application
is described in a concise manner in terms of its functionality, motivation,
and the underlying algorithms used. A description of its limitations leads
to the identification of collection of possible improvements, of which a
few effective ones are explicitly formulated as reasonable, and thus,
achievable intermediate goals.
The article rightly states that the exploration of large information
spaces is greatly facilitated by these 'star-displays', but does not
discuss the extension of the technique to other applications. Also,
maybe a more rigorous argument about the complexity of the algorithm
could be made. A starting point for this could be the analysis,
suggested in the conclusion of the article, of the relative costs of data
access technique and rendering strategy as a function of the number of
items in the database.