Continual surface-based multi-projector blending for moving objects

Continual surface-based multi-projector blending for moving objects,10.1109/VR.2011.5759447,Peter Lincoln,Greg Welch,Henry Fuchs

Continual surface-based multi-projector blending for moving objects  
BibTex | RIS | RefWorks Download
We introduce a general technique for blending imagery from multiple projectors on a tracked, moving, non-planar object. Our technique continuously computes visibility of pixels over the surfaces of the object and dynamically computes the per-pixel weights for each projector. This approach supports smooth transitions between areas of the object illuminated by different number of projectors, down to the illumination contribution of individual pixels within each polygon. To achieve real-time performance, we take advantage of graphics hardware, implementing much of the technique with a custom dynamic blending shader program within the GPU associated with each projector. We demonstrate the technique with some tracked objects being illuminated by three projectors. Index Terms: H.5.1 [Multimedia Information Systems]: Animations—Artificial, augmented, and virtual realities I.3.7 [Computer Graphics]: Three Dimensional Graphics and Realism— Virtual Reality; I.3.8 [Computer Graphics]: Applications; 1I NTRODUCTION One long-term dream of VR has been to integrate synthetic objects with a user’s immediate surroundings and to enable the user to examine, manipulate, and change aspects of those synthetic objects. One part of this dream that is close to being realized is the control of the appearance of a physical object by projection techniques. To achieve this effect, a projector need only render a virtual model of the object from the point of view of the projector. In order to allow a user to view the physical object from a wide range of viewing positions, multiple projectors need to be used, each projecting from sufficiently different angles to illuminate all the visible parts of the object. To achieve a consistent appearance, the intensity from any projector needs to be attenuated in surface regions illuminated by multiple projectors. This blending of multiple projected images has to be performed precisely, and the shape and pose of the object needs to be accurately known, because even slight errors in illumination patterns can result in obvious and disturbing artifacts. If the user also wants to move and manipulate the object, the required real-time computation increases significantly as the attenuation mask of every projector could change for every pixel. However, we believe that the resulting natural object-manipulation capability will be widely useful if it can be achieved without creating disturbing visual artifacts on the surface of the object. It is toward this end that the efforts of this paper are aimed. This paper presents preliminary work. Future efforts in geometric and latency analyses should lead to further reductions in the remaining artifacts that appear during object motion.
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.