With the advent of digital cameras, taking many pictures of a scene
from different viewpoints has turned into a common practice. The
abundance of such image sets has motivated researchers to develop
algorithms for generating photomosaics with wide fields of view. In
this project, we developed an approach to generating a Hockney-style
collage from a set of input images, which we call a scene collage. In
a scene collage, images of a scene are laid out using similarity
transforms computed based on feature matching; the layer ordering of
the images is determined automatically via an optimization procedure
to make the collage appear least fragmented. The final collage can be
displayed with opaque layers, transparent layers, or blended
For many scenes, scene collage provides a richer experience than what
a seamless mosaic does.
When looking at such a collage, a person can comfortably perceive the
overall structure of the scene as well as imagine the camera's motion
during the capture of the images. The scene collage can also be
generated in a nested manner, enabling a user to browse photos at
various levels of details. Finally, scene collage can be generated
from images of a scene with strong parallax, making it potentially
have a wider operating range than photomosaics. An algorithm for generating
scene collages that is very similar to ours has been simultaneously
Lihi Zelnik-Manor and Pietro Perona at Caltech.
To create video collages of dynamic scenes, we present new designs of
camera arrays for capturing multiple videos of a scene from different
viewpoints. Our design consists of a plastic frame onto which a set of
cameras can be easily attached, very much like Lego(R) building
blocks. The spatial layout of the cameras can be reconfigured in a
matter of minutes to achieve a variety of configurations, such as "L"
and "T" shaped ones. The plastic frame can be physically flexed to
vary the shape of the array. This gives a photographer significant
creative control &mdash the composition of the scene can be smoothly varied
as the scene changes.
If you are having trouble viewing these videos in your browser, please save them to your computer first (by right-clicking and choosing "Save Target As..."), and then open them.
||EGSR 2007 Video:
This video is the supplemental video for our EGSR 2007
paper. It contains a brief summary of our approach, a comparison of
scene collage and photomosaic, flexible camera array designs, and some
examples of dynamic collages.
||Photo-browsing using scene collage:
The collage representation is suitable for browsing
collections of photos taken from the same scene. As the user rolls
over the collage, the component images are highlighted. The user can
choose to view the most desirable image.
||Photo-browing using nested collage:
The collage can be generated in a nested fashion to organize pictures taken with a wide range of resolutions.
||Flexible Camera Array:
This video demonstrates the construction of flexible camera array and
how they can be flexed to vary the composition of the scene during