Wadi Hafir Survey

Gigapixel Imaging


About Gigapixel images


     A gigapixel image is, by definition, an image of 1 billion pixels or greater, that is to say 1000 megapixels or greater. The robotic Gigapan head is a commercial product that was developed by a partnership between NASA Ames Research Center and Carnegie Mellon University that stemmed from a project to develop panoramic imaging to be deployed on the Mars Rovers. Later developed as a commercial project, Gigapan images can now be uploaded to their website and shared on the web. Viewers can also create their own snapshots of details found within the images. While the Gigapan head is easy to deploy, manual capture of gigapixel images is possible with a pano-gimbal head. The only requirement is that the camera be pivoted around its no-parallax point, which is the point around which the camera can spin without parallax error. This aids in stitching the constituent images. While this is the ideal situation, we have had success creating these images with lenses as long as 400mm. When using such a long lens (a 400mm f2.8 lens weighs more than 5kg) it is impossible to rotate it around the no-parallax point, which is located behind the sensor plane on most lenses longer than 200mm. A gigapixel image of a boulder face may also be stitched as a photo mosaic by using a macro lens and overlapping images (the very same that are used to create a photogrammetric strip project). These types of images can be used in three ways in an archeological context like that of the Wadi Hafir:




     A gigapixel image can capture a landscape in up to 360 degrees. While one could achieve a 360 degree image with a fisheye and 3 or 4 photographs, capturing hundreds of images with a telephoto lens and merging them into one allows a viewer to zoom in to detail not available in an individual image. In addition, using the snapshot feature of the gigapan viewer, one can point to a myriad of details within an indicidual images. In the case of the Wadi Hafir, a snapshot can bring a viewer's attention to specific carvings, boulders, or areas of interest within the context of the overall landscape.




     While an overall landscape panorama is useful, another way in which this technique can be applied is to focus on an individual carved boulder face. By capturing dozens of images of a single face and merging them, one can zoom in on a macro level to individual characters or see the bigger picture. While capturing these mosaics with the Gigapan head is ideal, it has been mentioned that manual capture is possible by two methods. The first is the same concept, but with a dual axis pano-gimal head. The second is simply a by-product of strip photogrammetry. By capturing overlapping strips of images, a photo mosaic can be created in Microsoft Ice or Adobe Photoshop very easily.




     When a gigapixel image is captured of a specific boulder face, it is the same as capturing a camera station for use in a photogrammetric fan project. By capturing two or more of these camera stations in a convergent geometry, the constituent images can be brought into ADAMtech to create a 3D model. This is very useful if a boulder is too large to be captured by 2 or 3 individual photographs at a suitable resolution, or if it is too tall or inaccesable for strip photogrammetry. (see: Photogrammetry)

Copyright 2014