Summary

  • CU-SeeMe VR placed live video on moving flat panels and adjusted voices to the listener’s virtual position.
  • Its viewing radius governed stream requests, while compatibility kept ordinary video callers in the scene.

A tedious video conversation acquired an intriguing escape route in 1996: walk away. There was one wonderfully human complication. The other person could follow.

Jefferson Han and Brian Smith described that possibility in CU-SeeMe VR, their Cornell prototype. The laboratory’s publication record places the work at ACM Multimedia ’96. It offered a desktop room to navigate with arrows or keys, populated by flat panels carrying participants’ live video. People slid around the space with their faces attached. Office conversation had acquired furniture that talked back.

The volume control was a place

Approach a participant and their voice grew louder. Move farther away and it weakened. Left and right channels reflected position, and the speaker’s facing direction also affected the sound. A person’s place in the virtual room helped select whose conversation occupied the listener’s attention.

This was an ambitious use for an ordinary navigation gesture. Moving towards somebody became an invitation to listen more closely. Moving away supplied distance. The resulting social choreography could be as familiar as the colleague who keeps talking while accompanying you down a corridor.

The prototype’s flat video surfaces were an economical visual device. A separate map helped users locate themselves, while navigation controls moved or turned their viewpoint. The room made the organisation of a conversation visible.

Turning your back kept the connection warm

Distance also reached into media delivery. Outside a specified viewing radius, the client removed its request for a participant’s audio and video streams from the reflector. That saved network traffic and decoding work.

Nearby people received different treatment. Their streams remained requested even when they were behind the viewer, outside the current field of view, or facing away. The reason was practical: a user could turn faster than the conference machinery could restore a stream. The software kept nearby company ready for a quick glance.

That is the revealing distinction in the paper. Turning and walking can feel like variations on leaving someone’s company. The receiver assigned them different costs and consequences.

A room with visiting callers

The experiment kept a route back to ordinary CU-SeeMe. Its reflector stayed unchanged, and the additional position data travelled in an auxiliary stream that regular clients ignored. Within the VR scene, regular callers appeared on a fixed, rotating, multifaceted cylinder. The new room could receive guests with older furniture.

The authors then put a pointed question to their own design. Seat four people in a virtual circle and each gets a good view across it, with awkward views of the neighbours. A screen could arrange the other three faces side by side. Which arrangement better serves the conversation?

Their user testing was still limited, and chance encounters were an aim for further investigation. What the prototype already made concrete was a design choice: a convincing room brings the inconveniences of rooms along with it. Even the person you were trying to drift away from.