Geotags, aerial synths, street synths… capturing events as they happen. If you can mark it on a map, then this is the place to talk about it.
Hi Photosynth crew,
I do a lot of things off the grid and am always hunting around for good trail maps on the web. (Usually I have good luck with Google Image search.) One thing I'd love to see in Photosynth (given its potential to integrate with Bing Maps) is a view that would auto-stitch public domain maps from an image search.
As an example, maps of the whole length of the Bay Area Ridge Trail or the California Coastal Trail or Pacific Crest Trail would be an awesome showcase project.
But really, what I'm asking about is an auto-search feature that would let you zoom in on an area and find multiple maps that are nicely synthed for that locale.
This is a really cool idea.
Unfortunately, I believe that since most maps use completely different art styles, the way that Photosynth goes about matching parts of the images that are given to it won't really work for matching different maps together.
In other words, if you had two different maps of the same area and gave them both to Photosynth, Photosynth would probably identify all the corners in the roads as parts of the images to pay attention to, but when it examined a small square of imagery around each corner or blob, the area surrounding that feature wouldn't match to the texture from the different map. I'd be delighted to be proved wrong about that, though.
If you happen to know of existing maps, you could always use MapCruncher ( http://www.microsoft.com/maps/product/mapcruncher.aspx ) to add them to Bing Maps on your own, to then live in the same world as photosynths and panoramas as Read/Write World ( http://bit.ly/readwriteworld ) gears up.
Thanks for the boost, Nate!
Actually, it's MUCH more straightforward than you'd expect. Check out this search engine, then you'll see where this is going:
Every physical landmark in the USA has an assigned USGS LATLONG grid point, and it's public access. All the Photosynth dev team needs to do is write the code that helps Photosynth optically recognize a few place names on map image files, then assign those pixel positions on the image to that LATLONG reference grid, and you have virtually "told" Photosynth how to orient and re-scale the adjacent map images. Then add Photosynth's super-seamless polish, et voila!
I'd just like to add one further way that Photosynth could leap ahead of Google. Google has not yet indexed PDFs of maps into their image search. Yet most open space authorities' websites supply links to "printable" maps in PDF format. Do that, and you have a killer map engine.