Dynamic WorldS in Miniature

Goals

The World in Miniature (WIM) metaphor allows users to interact and travel efficiently in virtual environments. In addition to the first-person perspective offered by typical VR applications, the WIM offers a second dynamic viewpoint through a hand-held miniature copy of the virtual environment. In the original WIM paper the miniature was a scaled down replica of the whole virtual environment, thus limiting the technique to simple models being manipulated at a single level of scale. Several WIM extensions have been proposed where the replica shows only a part of the virtual environment. This project aims at providing an improved visualization of the WIM that supports arbitrarily-complex, densely-occluded scenes. In particular, we will develop algorithms for selecting the region of the virtual environment which will be covered by the miniature copy and efficient algorithms for handling 3D occlusion from an exocentric viewpoint.

Movies

These movies (XviD codec required) show different visualization modes for the WIM.

The movies were recorded with stereo projection disabled to avoid double images. The WIM was shifted with respect to the user's hand position to avoid occlusion.

	WIM manually delimited with adjustable cube + transparency Rendering the WIM with alpha blending does not provide a clear view of the objects. Size: 2152448 bytes (2,05 MiB), duration: 00:00:12, avg.bitrate: 1435 kb/s Video: mpeg4, yuv420p, 880x576, 25,00 fps(r)
	WIM manually delimited with adjustable cube + opaque rendering Users are required to adjust the cube to avoid occlusion. Size: 2555904 bytes (2,44 MiB), duration: 00:00:12, avg.bitrate: 1704 kb/s Video: mpeg4, yuv420p, 880x576, 25,00 fps(r)
	WIM delimited automatically + transparency The WIM is adapted to the current room, but alpha blending still does not provide a clear view of the objects. Size: 1953792 bytes (1,86 MiB), duration: 00:00:12, avg.bitrate: 1303 kb/s Video: mpeg4, yuv420p, 880x576, 25,00 fps(r)
	WIM delimited automatically + zoffset (our approach) The WIM is adapted to the current room and revealed with z-offset. Size: 2897920 bytes (2,76 MiB), duration: 00:00:12, avg.bitrate: 1932 kb/s Video: mpeg4, yuv420p, 880x576, 25,00 fps(r)
	WIM delimited automatically + zoffset (our approach) Same as above but with another room. Size: 1437696 bytes (1,37 MiB), duration: 00:00:12, avg.bitrate: 958 kb/s Video: mpeg4, yuv420p, 720x576, 25,00 fps(r)
	Comparison of WIM rendering strategies Comparison of the four WIM rendering strategies we have implemented. Size: 1378304 bytes (1,31 MiB), duration: 00:00:12, avg.bitrate: 919 kb/s Video: mpeg4, yuv420p, 720x576, 25,00 fps(r)
	Example of selection and manipulation using DWIM DWIM provides a clear view of the region of interest, enabling accurate selection and manipulation of 3D objects through their proxies. Size: 1263616 bytes (1,21 MiB), duration: 00:00:12, avg.bitrate: 842 kb/s Video: mpeg4, yuv420p, 720x576, 25,00 fps(r)
	Example of selection and manipulation using DWIM Just another example. Size: 2068480 bytes (1,97 MiB), duration: 00:00:19, avg.bitrate: 871 kb/s Video: mpeg4, yuv420p, 720x576, 25,00 fps(r)
	DWIM automatic update during navigation The miniature replica is updated automatically when the user moves to another room. Size: 1466368 bytes (1,40 MiB), duration: 00:00:11, avg.bitrate: 1066 kb/s Video: mpeg4, yuv420p, 720x576, 25,00 fps(r)

Source code

Source code of a prototype implementation of the technique over VR Juggler will be available here soon.

Ramon Trueba, Carlos Andujar, Ferran Argelaguet 2009.