Summaries for week 7
A Practical Multi-viewer Tabletop Autostereoscopic Display
Autostereoscopic display allows stereo perception without HMD or glasses, and allows people or objects to interact as if they were in the same place. However, it did not meet success for the while because of its price. The technique used here is the Random Hole Display (RHD) concept, but several issues exist: the image quality is not optimal, viewpoints are not mobile and there is interference because of inter-reflection between barrier and display. The work described in this paper solves several problems: the image quality is optimal, viewpoints are mobile; and produced evaluation and analysis of the system for multi-user situations.
The configuration used is an LCD display panel, a random-hole mask printed on polyester film and a polycarbonate sheet as the separating layer, eliminating internal reflection. The algorithm used is the Poisson disk sampling algorithm. Concerning rendering, only a part of each pixel can be seen through the mask. Therefore, some pixels are seen from several viewpoints and are conflicting, this is why their color is blended and this approximation is spread to their neighbors. The hardware-accelerated algorithm is made of 4 passes: the scene is rendered for each viewpoint, then a visibility mask is applied to the image for each viewpoint, all images are mixed and the color for the conflict pixels is computed, this error is applied to their neighbors. Finally, a priority could be set, for active users for instance or for a surgeon, so that the image quality is better. For mobile viewpoints, the user has to be tracked, thus the system has to be calibrated, as well as the orientation and position of the barrier, and the refractive index and dimensions of the polycarbonate sheet.
An evaluation has been led for two points: the enhancement of the image quality with error diffusion and the degradation o the image quality with additional viewpoints. The first feature has been evaluated with a camera to compare the brightnesses of three situations: a single viewpoint, two viewpoints without error diffusion and two viewpoints with error diffusion. It appears as the method is not efficient if too many pixels are in conflict. In order to prevent it, the hole density has to be inversely proportional to the number of viewpoints. The same system has been used for the second feature but with situations going from one to six viewpoints and with four different models: cityscape, room, head and cubes. Naturally, the mean absolute error never decreases as the number of viewpoints increase but it can be the same, heavily depending on the model
Thus, the work achieved is very positive, and its applications are navigation or objects manipulation (real or virtual). Future work will focus on using brighter panels or several projectors, as well as making the barrier less noticeable. But the final aim might be to get rid of the barrier and the layer, to allow a better immersion.
Scape: Supporting Stereoscopic Collaboration in Augmented and Projective Environments
This paper presents an efficient system for remote collaboration, in order to be able to work and communicate with people as if they were in the the same place. Similar systems already exist, using for instance LCD-shutter glasses, but they only track one user. Some solutions have been brought, but they came with drawbacks, such as a maximum number of users or limited movements.
The display system used here is a HPMD (head-mounted projective display), which has projective lenses and a retro-reflective screen. This technology allow as many as viewpoints as wanted and is a natural way of communicating face-to-face. Thus, the first feature allows several viewpoints for a single user who will be able to fully observe the environment, by switching between miniature view and immersion. The workbench is a table with a retro-reflective film on it, inside an immersive cage. For tracking, magnetic and vision-based trackers are used.
However, some issues exist in Scape. First, depending on the angle, the image quality can be lowered because of a bad reflection. Also, some tradeoffs have to be made concerning the screen shape (concave or convex), the screen position according to the size of the room, the number of applications available, the field of view and ambient brightness. Finally, the HPMD prototype is quite heavy (compared to glasses), which may reduce the mobility of the users.
For what concerns collaboration, a tool has been developed to help users building their own environments. Three kinds of collaboration exist: interactive local collaboration where users have to be in the same place to work on an environments, passive remote collaboration where remote users can see the environment, and interactive remote collaboration (which will be developed later) where all users can work on the environment.
The first work to be done in the future is the miniaturization of the HPMD and the development of the interactive remote collaboration and a better design, as well as building easier tools to build environments.
Comments are closed.