We introduce a new and versatile web-based system which allows users to dynamically generate surfaces of three-dimensional (3D) scenes from stereo image pairs taken with monocular and stereo cameras or web-cams.
The system uses the Internet to communicates with potential user around the world and accepts data from different imaging sources via direct or indirect uploads of static individual images or live video sequences.
If necessary the uploaded images are automatically co-aligned into stereo pairs using calibrated or uncalibrated rectification techniques depending on the input source.
A rectified pair is processed (remotely or locally) by selecting from a number of different stereo matching algorithms in order to reconstruct the scene and return it to the user as a disparity map, a virtual Java3D/webGL scene, a 3D .OBJ file, or a live depth video.
The system is portable, simple to set up and operate, and is currently available online at https://www.ivs.auckland.ac.nz/quick_stereo.
Potential applications of the proposed system, e.g. remote camera control, on-line calibration and rectification, simple object and 3D avatar reconstruction, biometrics, collaborative on-line research meta-tools are discussed.

All the above 3D scenes are recontructed by Stereo Vision techniques, scene is changed every 10-20s, or press F5 to see a new sample set.
Click here to view our 3D photo gallery

If the above canvas does not show 3D properly,
please refer to the below section on how to enable WebGL on your browsers.