Shane’s Week 7 Summaries

Multi-viewer Tabletop:

This paper is about the construction and inner workings of  a new type of Autostereoscopic table top display.  The basic goal is to devise a display that will render different views to a number of users all position at various positions around the display.  This problem has been attacked before using s few different approaches mostly focusing on using barriers to block transmission of light from all the pixels while only allowing a smaller number of pixels to be displayed to the users.

The implement approach is similar to those that have come before.  To build the display they choose a very high resolution and bright LCD display.  Then they took two sheets different films and stacked them on top of the LCD monitor.  One of the films had been perforated using a pseudo random algorithm.  The films will obscure most of the pixels from view allowing only certain pixels to be seen by a user.  The system then tracks(using markers) the location of each users eyes.  The system then uses that information in conjunction with the map of all the holes locations to compute which pixels are visible to each user.  The system then draws the display using  a four pass rendering algorithm.  Sampling the actual images where pixels are visible.  Problems occur when two users can see the same pixel in which case the system averages the colours that would be displayed.  As users are added the picture quality is reduced.

–I must say I felt like the glossed over possible applications of the display.

Scrape:

This paper is focused on creating different views for every user in a collaborative environment.  The main goals are to not block the users view of the real world allowing them to see the other participates, and to allow for an unspecified number of users to use the system at one.  To that end they built a new type of HMD called Head Mounted Projective Display(HMPD).  It works by actually using a small projector to project an image out where the user is looking and then bouncing that image off of a retro-reflector back the users eyes.  They track every user using the HMPD in the room using a high ball sensor.  The HMPD only work on surfaces that are retro reflectors which limits the mobility of the system.  The best case scenario is to have a mobile retro reflector sheet of workbench.  The system also tracks the users hand motion using s sensor glove.  They state in their paper that there is no cross talk between users but that there cross talk between the images meant for the left and right eyes.  Which only happen once the user is over 9 meters away from the retro reflector.  One of the problems they ran into was having to decrease the light in the test system because the beam splitter reduces the incoming light making harder to see the projected image in a bright environment.

–I have some doubts on the cross talk assumption.  They over ever built one device so while theoretically it should work without actually testing it… I have a feeling it will fail when users are close together, while retro reflectors do bounce light straight back at the source they do not so perfectly mean than some light will be visible to some one other than the user it’s meant for.  It wasn’t clear to me whether the lenses on the system were polarized to prevent that.

 

Comments are closed.