PhotoshopNews.com
Jul 21, 2008

Sweeping Panoramas, Courtesy of a Robot

Source: The New York Times

Written by Ann Eisenberg

ROBOTS already cut the grass and vacuum rugs. Now they are helping with a more artistic job: creating vast photographic panoramas with ordinary cameras.

A new, inexpensive robotic device from researchers at Carnegie Mellon University attaches snugly to almost any standard digital camera, tilting and panning it to fashion highly detailed panoramic vistas — whether of the Grand Canyon, a rain forest or a backyard Easter egg hunt. The robot is called GigaPan, named “giga” for the billion or more pixels it can marshal for a typical panorama. It creates the huge, high-resolution vista by extending its robotic finger and repeatedly clicking the camera shutter, taking tens, hundreds or even thousands of overlapping images, each at a slightly different angle, that are then stitched together by software to create one gigapixel shot.

Viewers can explore a panorama in detail when it is displayed on a computer screen, clicking on any part of the image and then zooming in for crisp close-ups. You can move from an overall shot of the forest, for instance, to an image of one small moth resting on the side of a single tree trunk.

The roboticized camera mount and related software were devised by a team led by Randy Sargent, a senior systems scientist at Carnegie Mellon West and the NASA Ames Research Center in Moffett Field, Calif., and Illah Nourbakhsh, an associate professor of robotics at Carnegie Mellon in Pittsburgh. The work was part of a project to introduce people to different countries and cultures through images.

The GigaPan provides a low-cost alternative to sophisticated motorized camera mounts on the market used to take panoramic photos, said Greg Downing, co-founder of the xRez Studio in Santa Monica, Calif., which specializes in gigapixel photography. The motorized mounts can cost thousands of dollars, he said, and typically require a high-end camera.

Read entire story

Comments are closed.