Abstract: In an example embodiment, a first set of estimated locations for calibration points in a roof is identified in a first satellite image, the first satellite image having a first zoom level and a first angle relative to ground. Then a second set of estimate locations for the calibration points is identified in a second satellite image, the second satellite image having a second zoom level and a second angle relative to ground. Actual locations for the calibration points are derived using the first and second sets of estimated locations, the first and second zoom levels, and the first and second angles. Then dimension information is calculated for the roof based on the derived actual locations for the calibration points.

Abstract: In an example embodiment, a first set of estimated locations for calibration points in a roof is identified in a first satellite image, the first satellite image having a first zoom level and a first angle relative to ground. Then a second set of estimate locations for the calibration points is identified in a second satellite image, the second satellite image having a second zoom level and a second angle relative to ground. Actual locations for the calibration points are derived using the first and second sets of estimated locations, the first and second zoom levels, and the first and second angles. Then dimension information is calculated for the roof based on the derived actual locations for the calibration points.

Abstract: In an example embodiment, a first view of a real-world object is displayed from a first angle. Then a two dimensional shape is overlaid over the first view. Then user interface manipulations of the two dimensional shape are received from a user, defining the boundaries of a first side of the real-world object in the first view. Then a second view of the real-world object is displayed from a second angle. Following that, the manipulated two dimensional shape may be overlaid over the second view. Additional user interface manipulations of the two dimensional shape may be received from the user, defining the boundaries of the first side of the real-world object in the second view. Then dimensions of the first side of the real-world object are derived from the manipulated two-dimensional object from the second view.