Visible Scalable Terrain (ViSTa) is a software format for production, interchange, and display of three-dimensional (3D) terrain data acquired by stereoscopic cameras of robotic vision systems. ViSTa is designed to support scalability of data, accuracy of displayed terrain images, and optimal utilization of computational resources. In a ViSTa file, an area of terrain is represented, at one or more levels of detail, by coordinates of isolated points and/or vertices of triangles derived from a texture map that, in turn, is derived from original terrain images. Unlike prior terrain-image software formats, ViSTa includes provisions to ensure accuracy of texture coordinates.
Whereas many such formats are based on 2.5-dimensional terrain models and impose additional regularity constraints on data, ViSTa is based on a 3D model without regularity constraints. Whereas many prior formats require external data for specifying image-data coordinate systems, ViSTa provides for the inclusion of coordinate-system data within data files. ViSTa admits high speed loading and display within a Java program. ViSTa is designed to minimize file sizes and maximize compressibility and to support straightforward reduction of resolution to reduce file size for Internet-based distribution.
This program was written by Paul Backes, Mark Powell, Marsette Vona, Jeffrey Norris, and Jack Morrison of Caltech for NASA's Jet Propulsion Laboratory. For further information, access the Technical Support Package (TSP) free on-line at www.techbriefs.com/tsp under the Software category. This software is available for commercial licensing. Please contact Don Hart of the California Institute of Technology at (818) 393- 3425. Refer to NPO-30600.
This Brief includes a Technical Support Package (TSP).
Format for Interchange and Display of 3D Terrain Data
(reference NPO-30600) is currently available for download from the TSP library.
Don't have an account?
Overview
The document presents the Visible Scalable Terrain (ViSTa) format, developed by Marsette Vona and Mark Powell, as a method for the production, interchange, and display of 3D terrain data, particularly for stereo vision-based robotic applications. Released in March 2002, ViSTa is designed to ensure data scalability, visualization accuracy, performance, and optimal resource usage.
ViSTa terrain data is composed of a set of triangles and isolated points, with textures derived from images captured by stereo vision hardware. Each point and triangle is associated with a specific texture map, ensuring that only visible elements are included in the display. The format supports multiple Levels of Detail (LOD), allowing for efficient visualization and resource management. A single ViSTa file can range from minimal data from a single stereo pair to extensive panoramas, but it cannot exceed 2 GB in size.
The document outlines the design goals of ViSTa, which include data accuracy, file size minimization, platform portability, and run-time efficiency. It aims to support both high-resolution and lower-resolution applications, making it versatile for various uses, including public outreach. ViSTa is also designed to be easily generated from existing stereo vision data and to integrate seamlessly with systems that handle terrain data, such as SAP, SUMMITT, and RSVP.
Key differences between ViSTa and the Adaptive Surface Data (ASD) format are highlighted, including the omission of certain features in ViSTa to reduce storage and processing requirements. For instance, ViSTa does not store Level of Detail (LOD) morphing information, surface normals, or vertex colors, and it requires explicit specification of texture map coordinates for all vertices.
The document emphasizes that while ViSTa is tailored for specific applications, it allows for implementation-dependent details, enabling flexibility in how ViSTa data is produced and displayed. Each implementation is identified by a unique 4-byte identifier, ensuring clarity in the use of different ViSTa formats.
Overall, the ViSTa format represents a significant advancement in the field of 3D terrain data management, providing a robust framework for applications in robotics and beyond, while maintaining a focus on efficiency and accuracy.
