Bundler software from the University of Washington

Photogrammetry deals with extraction of building and landscape geometry from photos.

OpenStreetMap project can directly benefit from photogrammetry software provided that it is easy to use and give reliable results
Here all related issues to the topic Photogrammetry & OpenStreetMap can be discussed: ideas, development, own experience of using photogrammety software.

We’ve allready done some tests with the Bundler (http://phototour.cs.washington.edu/bundler/) which is the original basis for Microsoft’s Photosynth.
Among other other things Bundler provides a functinonality to calculate sparse scene geometry and relative camera locations and orientations from matched points between a set of photos.

In order to simplify installation and further development we have rewritten Bundler’s script in Python. Details about their installation and running as well as best practices to take photos for Bundler input can be found at the OSM-wiki:
http://wiki.openstreetmap.org/wiki/Photogrammetry

Testing Bundler we’ve found that in many cases Bundler fails to match many photos automatically. This leads to unsatisfactory scene reconstruction (e.g. only one or two reconstructed building sides out of four).

At the moment we investigate possible ways to improve matching of photos.

Hi,

I’m a mapper, principally in my town Duitama:

http://www.openstreetmap.org/?lat=5.8067&lon=-73.0247&zoom=12&layers=M

For a few weeks, I’m looking for a free versions of photosynth, I took aerial pictures of some places with a canon camera, and CHDK software in a canon camera, and a KITE, then I processed this pictures with photosynth, and found that I can know the topography of this areas, but photosynth can’t download or work with the points cloud.

My work you can work here (spanish):

http://hobbiesleo.blogspot.com/2010_03_01_archive.html

Thanks, and I hope to contribute for our own development.

At the moment I am not sure which approach to take.

Here is an alternative to Photosynth: dense 3D points recontruction: http://grail.cs.washington.edu/software/cmvs/

A commercial implementation of a dense 3D points recontruction from a Swedish company: http://webtest.madlix.com/sthlm/

if you have Photosynth ready, try this: http://synthexport.codeplex.com/

It will export PLY file from photosynth url, ready for working in Meshlab for example…

Hi, guys,
I like the idea a lot and want to try this, but:

  1. Can I do photogrammetry in Ubuntu?
  2. Is Hugin and its components suitable for this?
  3. Can I restore the coordinates from reference points instead of having a GPS? (I have a smartphone that outputs way too incorrect coords)

Bundler on OS X?

I’ve installed PIL under OS X 10.6 but the RunBundler.py script does not work. Have you any experience getting it to work on the Mac like the blog says it should?

Sorbus