Gallery of Reconstructed Models

Reconstructions of the Arc de Trioumphe and the Taj Mahal (from a single photograph) by George Borshukov

(a,b) (c,d)

1. A photograph of the Campanile, with marked features shown in green.
   
2. A basic model of the Campanile recovered from the single photograph in (a).
   
3. The edges of the basic model projected into the original image through the recovered camera position.
   
4. A synthetic image of the Campanile generated using three photographs and view-dependent texture-mapping.
   

(a) (b) (c)

1. A photograph of the SFMOMA, with marked features shown in green.
   
2. A second marked photograph.
   
3. The model recovered from the lines in the two photographs.
   

1. Three of twelve photographs used to reconstruct Uni High, with marked features shown in green.
   
2. Recovered model edges reprojected through recovered camera positions into the three original images.
   
3. Three views of the recovered high school model and camera positions.
   
4. The process of projecting images onto the recovered model with view-dependent texture-mapping.
   
5. Computer-rendered fly-around of the high school model. (6MB QuickTime Movie)
   

(a) (b) (c)
(d) (e) (f)

1. One of four original photographs of the entrance.
   
2. A depth map for the entrance keyed to the photograph in (a), computed using a second photograph of the entrance and model-based stereo.
   
3. A computer-rendered view of the entrance using the image-based technique of forward pixel-splatting.
   
4. A computer-rendered fly-around movie of the entrance, using a single image projected onto flat geometry. (9MB QuickTime Movie)
   
5. A computer-rendered fly-around movie of the entrance, using view-dependent texture mapping of four images onto flat geometry. (9MB QuickTime Movie)
   
6. A computer-rendered fly-around movie of the entrance, using view-dependent texture mapping of four images onto complex geometry recovered using model-based stereo. (9MB QuickTime Movie)
   

(a) (b) (c)

1. A photograph of the cathedral, with the recovered model edges superimposed in blue.
   
2. A basic model of the cathedral's Façade recovered from features marked in three images.
   
3. A computer-rendered view of the cathedral using view-dependent texture mapping.
   

(a) (b) (c) (d)

1. A photograph of Hoover Tower.
   
2. A model of Hoover Tower recovered from the single photograph in (a).
   
3. Another view of the model, showing a user-placed camera that will be used to generate a novel view of the tower.
   
4. A synthetic view of the tower, obtained by texture-mapping the photograph in (a) onto the model in (b) from the novel vantage point shown in (c).