[-empyre-] Augmented reality as public art, mobile location based monuments and virtual memorials

[Julian Oliver]

..on Sat, Apr 16, 2011 at 09:25:46AM -0400, Alan Sondheim wrote:
>
> Hi - I'm not sure how to reply to this; I've been thinking about it. One  
> thing about locative art is its oddly inert quality - it's _there_ and  
> remains there, is fixed there. It's _there_ in the sense of geographic  
> location, and _there_ in the sense of specific technology needed to 
> reveal it, almost as if it's embedded in the technology, welded to it. 
> The ephemerality lies in the fact that it takes a specific, soon-to-be-  
> outdated technology to run, as well as energy; unlike a physical public  
> monument, the energy is meted out within a specific regime of capital and 
> control. So the 'We' in electracy you talk about is inextricably mixed  
> with capital, with enclaving, and with the specifics of location; only 
> the last is accessible to everyone. In this sense, what you call 'this 
> virtual public sphere' is a 'real private sphere' whose manifestation or 
> represen- tation is is virtual.

Indeed. I've mentioned this problem in a couple of talks, albeit to Dutchies who
are of course very proud of the company at large (sorry Mediamatic, V2_)!

We're seeing a real platform dependency set in the artistic AR scene, akin to
all those countless artists depending on Macromedia Director, back in the day of
CDROM art, much of which is completely unplayable now on modern systems. 

Forward compatibility will a nightmare as will archiving any of this stuff -
destined to become space-junk unless the LayAR team actively take an interest in
the problem. Naturally this is only an ethical issue if artwork deployed using
LayAR is publically funded.

In any case, AR on handhelds is still in a fairly impoverished, unconvincing
place: LayARs UX is not so good really, at least for the time being. This is one
of the issues of trying to do pose estimation using a digital compass rather
than true NFT style tracking which is concerned with things like ground plane
detection, even occlusion. These features are the difference between AR as 3D
art wobbling around on a plane of video and augmentation that truly transforms a
sense of place, of being somewhere. 

In any case, we're still not there yet on handhelds. Georg (Oxford) got pretty
close however with PTAM before working for Microsoft:

	http://www.youtube.com/watch?v=pBI5HwitBX4

From:

	http://www.youtube.com/watch?v=Y9HMn6bd-v8

Sadly while the source code is open (the right to read it) it's entirely
proprietary (restricted use rights, no modification).

It would certainly be good to gather together and dedicate some time writing a
free and open source alternative to LayAR but with working ground plane
detection and a better renderer (a la PTAM). Perhaps this is something that
could be kickstarted. By the end of the year I will be able to contribute quite
some code to such a project as a result of a collaboration with V2_ and
Lighthouse, Brighton. 

Regarding other proprietary AR suites, Damian Stewart and I have an alternative
to Junaio's proprietary image tracking solution. It is robust and (it seems)
more performant than their own. It allows augmenting any complex image with
images, video and 3D models:

	http://selectparks.net/~julian/theartvertiser/install.html

A version for Android is on it's way thanks to the help of colleague Arturo
Castro. It may in fact use a more robust algorithm, allowing detection of images
that are very difficult to track due to having fewer feature points, like logos.

Cheers,

-- 
Julian Oliver
home: New Zealand
based: Berlin, Germany 
currently: Berlin, Germany 
about: http://julianoliver.com
follow: http://twitter.com/julian0liver