Screen Shots

	Tripod-based LiDAR scan of the UC Davis water tower and the Mondavi Center for the Performing Arts.
	Airborne LiDAR scan of a part of the northern California coast.
	LiDAR data set rendered using real-time point-based illumination. Local plane equations for every data point are calculated in a pre-processing step, and the resulting normal vectors are used to illuminate each point using the standard OpenGL lighting equation.
	Point-based illumination where point colors are modulated based on LiDAR return intensities.
	Point-based illumination from a different lighting angle.
	Airborne LiDAR scan with draped aerial photography of Google corporate headquarters, rendered using the 3.0-preview splat renderer.
	Another view of same.
	From the same LiDAR scan, Shoreline Amphitheatre.
	Airborne LiDAR scan of the Raplee Ridge anticline in Utah, with draped 1m resolution aerial photography.
	LiDAR scan of Mondavi Center visualized at a quality level of 1.5.
	LiDAR scan of Mondavi Center visualized at a quality level of 0.5.
	LiDAR scan of Mondavi Center visualized at a quality level of -0.5.
	LiDAR scan of Mondavi Center visualized at a quality level of -1.5.
	Using the 3D paint brush to select a subset of points for feature extraction. The brush is a sphere attached to a (tracked) input device, and selects all points inside the sphere whenever the associated input device button is pressed. Selected points are drawn in green.
	Here the user selected a set of points defining the front wall of the UC Davis Mondavi Center.
	Once the desired points are selected, the program is told to calculate the equation of the best-fit plane to the selected set. The plane equations can later be used to calculate other desired quantities.
	Extracting multiple planes allows a user to calculate the equation of intersection lines or the positions of intersection points at very high accuracies. For example, if each point in a LiDAR scan has an error interval of +-2mm, then the plane equation best fitting 10,000 points expected to lie in a plane can have an error interval as low as 20 micrometer (assuming that the error is normally distributed).
	Extracting a cylinder fitting a scanned air conditioning pipe. Although only the front half of the pipe was scanned, the fitting algorithm accurately determines the height and radius of the full cylinder, and completes the partial scan by filling in the previously invisible back half of the pipe.
	Illuminated rendering of a terrestrial LiDAR scan of a house overhanging a landslide scarp. After selecting point sets representing the house's walls and extracting planes and intersecting them, a very accurate base architectural model of the house has been reconstructed.
	A very large airborne LiDAR scan of the Cosumnes River south of Sacramento, California. The scan contains a total of 310 million points at about 1m horizontal resolution, and covers an area about 13km west to east and about 12km south to north. The multiresolution LiDAR viewer can show arbitrary views of this data set at interactive frame rates. The data set was kindly provided by Joshua Viers and Jeff Mount from the UC Davis Watershed Research Center.
	A perspective view of the Cosumnes River LiDAR scan, looking eastwards from Interstate 5 towards Highway 99.
	A user fixing a ground plane while analyzing the Cosumnes River LiDAR scan. This photograph was taken in the KeckCAVES immersive visualization environment.