Andrew Larcombe <[EMAIL PROTECTED]> wrote:
> Sounds like Microsoft Photosynth: > http://www.youtube.com/watch?v=p16frKJLVi0
I hadn't seen that before - very cool. It will be interesting to see how 'automated' it can be. It's nice to see a use for all the Gigabytes of data that is now produced on a daily basis at many sites. We're digitally archiving the world, now make it virtual tourable. :)
Certainly, Photosynth could be an output for all of the photos identified as a cluster, but there is still the fundamental question of how the clustering gets done. Do you cluster the locations of the photographer, or of the thing being photographed? The former is of some interest I guess, but the latter is potentially far more useful. And sadly far more difficult to do.
With a large enough set of data, clustering and averaging would figure out what people were taking the photo *of*. Similar to the photos of the mountain from various countries example. However, problems arise when the mountain may only be visible from a single side, or small arc of the total circular ring. It would be nice if cameras recorded estimated focus distance to EXIF. Then this could be applied to help determine the photo's subject was nearby the photographers location, or very far away. Andrew (T) -- Andrew Turner [EMAIL PROTECTED] 42.4266N x 83.4931W http://highearthorbit.com Northville, Michigan, USA _______________________________________________ Geowanking mailing list [email protected] http://lists.burri.to/mailman/listinfo/geowanking
