Photo Uncrop

Qi Shan, Brian Curless, Yasutaka Furukawa, Carlos Hernandez, and Steven M. Seitz

Paper [pdf 17M]
Supplementary video [YouTube][720p 66M]
ECCV 14 Poster [pdf 15M]
Result Gallery [Link]
A Seattle Times article

We address the problem of extending the field of view of a photo--an operation we call uncrop. Given a reference photograph to be uncropped, our approach selects, reprojects, and composites a subset of Internet imagery into a larger image around the reference using the underlying scene geometry. The proposed Markov Random Field based approach is capable of handling large Internet photo collections with arbitrary viewpoints, dramatic appearance variation, and complicated scene layout. We show results that are visually compelling on a wide range of real world landmarks.
1) Faithful reconstruction of the actual scene
2) Dramatic range expansion
A typical scenario:
Capturing family photos with the desired background in the image frame can be tricky. Landmark: Stravinsky Fountain in Paris.
Our approach expands the FOV of a user photo thus enables better spatial context.
Image source: Flickr
User input A set of Internet photos from the same landmark Uncrop result
A few results:
Image source: Flickr
User input
Uncrop result. Landmark: Place de la Concorde in Paris. Image source: Flickr
Landmark: Institut de France. Image source: Flickr
Nomura, Y., Zhang, L., Nayar, S.: Scene collages and flexible camera arrays. In: Eurographics Symposium on Rendering. (2007)

*** End ***