views:

136

answers:

4

I've read about systems which use the Flickr database of photos to fill in gaps in photos (http://blogs.zdnet.com/emergingtech/?p=629).

How feasible is a system like this? I was toying with the idea (not just a way of killing time but as a good addition to something I am coding) of using Flickr to get photos of a certain entity (in this case, race tracks) and reconstruct a model. My biggest concern is that there aren't enough photos of a particular track and even then, it would be difficult to tell if two photos are of the same part of the racetrack, in which case one of them may be irrelevant.

How feasible is something like this? Is it worth attempting by a sole developer?

A: 

LINQ to Flickr on codeplex has a great API and would be helpful for your task.

Chris Ballance
+1  A: 

Sounds like you're wanting to build a Photosynth style system - check out Blaise Aguera y Arcas' demo at TED back in 2007. There's a section about 4 minutes in where he builds a model of the Sagrada Família from photographs.

ninesided
A: 

I say +1 for photosynth answer, its a great tool. Not sure how well you could incorporate it into your own app though.

Its definately feasable. Anything is possible. And yes, doable for a single developer, just depends how much free time you have. It would be great to see something like this integrated into Virtual Earth or Google Maps Street View. Someone who could nail some software like this could help 3D model the entire world based purely on photographs. That would be a great product and make any single developer rich and famous.

So get coding. :)

DarkwingDuck
A: 

I have plenty of free time, as I am in between jobs.

One way to do it is to get an overhead view of the track layout, make a blueprint based on this model, and then get one photo of the track and mimic the track's road colour. That would be a start.

dotnetdev