Architectural Visualisation


About the project

VR is a great way to explore and inspect an environment - the viewer has a sense of space and control over their viewpoint that a screen + keyboard/mouse just does not provide. Spatial relationships and the scale of objects are understood intuitively when using a VR display.

Architectural Visualisation with gaze-based navigation and Virtual assistants

We have been working on a system for creating and distributing multi-room virtual environments, which include video-based 'virtual assistants'.

Uses of this platform might include property developers, architects, homebuilders and industrial or commercial users looking to provide easy visualisation of plans and/or proposals to their customers.

This product is aimed at the largest installed base of VR-capable devices - anyone with a modern phone and a cheap 'Cardboard' viewing device can enjoy the content, and the approach is trivially scalable to higher-end VR platforms.

Using a custom Unity3D-based stereo panorama viewer which offers high performance for a wide range of devices, viewers can use simple, gaze-based navigation markers to move from room to room, and to activate virtual assistant functions. Content creation is based around freely available 3D modelling tools, allowing the operator to set up viewpoints, and the possible paths a user can take through the building - this information is saved to a scene description file and packaged along with the stereoscopic renderings of each viewpoint for distribution.

Let us know if you are interested in trying out the tech demo - and feel free to talk to us about how our technology can help your business to leverage VR and immersive 3D environments.

Stereoscopic Panoramic Renderings

Most "360 degree" still and video imagery is monoscopic - it contains no stereo cues that are critically important for the perception of depth, so while it gives a good view of a space it does not give the viewer the feeling they are 'immersed' in the same way as a stereo view does.

It is quite difficult to create a stereoscopic panorama, since it is not possible to simply combine images from a camera rotated around a single point to produce an image with extremely wide field of view - The relationship between the two 'eye' viewpoints changes significantly as the camera pair rotates - and the relative position of left and right eyes are completely reversed when the camera pair has rotated 180 degrees from its starting point.

In general, it is very hard to create stereo panoramic imagery using physical cameras, though it is possible to synthesise arbitrary views of a scene given a large number of viewpoints. For our work, we use computer-rendered geometry, which makes it relatively easy to create an image that integrates the changing relationship between the eyes depending on the angle the of the viewer.

Some examples of stereoscopic panoramas



Current status

2016-07-13 Scene database and navigation-markers are now working nicely for multi-room navigation in viewer app.

2016-06-01 Happy that the basic concept seems to work, currently working on viewer app and pipeline for easy creation of multi-room environments.