Floating up to the Spatial Web

(Originally posted 20 April 2009).

Google Street View is an option in Google Earth that shows sequential 2D images down the world’s main streets. This tool provides for a simple 3D effect if you traverse up or down a particular street. Photosynth is MSFT Live Labs software that creates a 3D effect from multiple 2D images of the same scene taken from different angles. You can zoom in/move around the scene if there are enough 2D images of it available. It is more sophisticated than Street View since it is not tied to a particular vector i.e. a street and also because it performs image extrapolation to complete partial views.

Moving through alternate realities like this holds a natural attraction for people; training, POS, medical diagnosis, national security, gaming, film-making, virtual tourism and urban planning would all immediately benefit. Photographs of existing places and objects are increasing at a huge rate through (camera phone) image uploads to social networking sites and the thousands of commercial and Government camera installations throughout the world. There will surely eventually be blanket photograph coverage of pretty much everything everywhere (and once that is achieved - at every time). Geo-tagging will help connect these pictures together.

These technologies basically link photographs of existing scenes up. Your brain mainly adds missing spatial orientation. There is no 3D model of structures behind them unlike say as with Google Sketch-up, AutoCAD or the various First Person Shoot-em-up (FPS) games e.g. Call of Duty etc.

Does there need to be a 3D model behind them? You could ignore surfaces/obstacles, simply walking through them as a disembodied ghost. This might be acceptable for many applications but for gaming it will not. 3D models are time consuming to plot/maintain with changing actual landscapes. The reason why models are created now is primarily for edge detection e.g. you see a wall and because the 3D model has been defined at design-time to identify it as a wall, you are prevented from moving through it. It is much better to do edge detection at run-time using some algorithm to identify the wall as an edge (or barrier), preventing it from being moved through. Gaming will likely be the driver for this new standard. This will be accessible through some cloud-based service and other applications will simply adopt it.

Development of the Internet has shown us that competing infrastructure technologies will co-exist for a while but they will eventually “float-up” to the highest common denominator e.g. the most extensible, cost-effective, egalitarian and “fit-for-purpose” technology. Applications built on selected-out infrastructure will eventually switch to the new standard. It is easy to see a gaming driven cloud-based service, regularly updated with near-real time images as being adopted by all other applications requiring spatial understanding.

In this way, we will see a single alternate reality develop where synergies and opportunities are created by markets colliding (as with actual reality) and the “Spatial Web” described last year by MSFT’s Craig Mundie will become normal (if not “real”!).

UPDATE: Swiss Computer Scientists use Flickr to do this now http://www.openculture.com/2010/12/3d_rome_was_built_in_a_day.html.