New technology from the University of Washington allows for millions of photos taken by different people over the duration of years and years to be sorted through, selected, and pieced together to create jaw-dropping time-lapse videos. Using photos from public databases, they were able to watch the progression and/or degradation of growing city skylines, receding glaciers, monuments, and more. Check out the video to see their amazing work!
For the colleagues at the University of Washington to accomplish this feat, they faced two major hurdles. The first was to cull through the literal millions of photos in Google's public photo database. They chose time-lapse locations where they expected to see a massive collection of photos spanning through the years as far back as the Internet made possible. Tourist attractions, of course, fit the bill perfectly. It was important to have a large sample to be able to account for all the possible alternate perspectives of the shooter in the public photos. Photos from inconsistent angles were immediately discarded.
Figure showing examples of discarded and accepted photos.
The second major hurdle was aligning the chosen photos. As you can see in the above figure, even the two selected photo's angles do not match perfectly. From this point, the process was similar to how we understand 3D imaging. First by aligning perspective, and second, by accounting for depth.
Dual processes taken to align photos properly.
For a seriously in-depth review of the steps taken to create these time-lapses, visit the University of Washington's published paper.
wow! this time lapse looks great!