Terrain rendering in a three-dimensional environment

Details Terrain rendering in a three-dimensional environment Terrain rendering in a three-dimensional environment

2002-03-19

2004-06-08

Jankus, Almis R. (Department: 2671)

Computer graphics processing and selective visual display system

Computer graphics processing

Adjusting level of detail

C345S587000

Reexamination Certificate

active

06747649

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to improved terrain mapping in interactive 3D computer graphics applications.
2. Description of the Related Art
Flight simulation is the quintessential example of interactive 3D computer graphics applications. The purpose of any flight simulator Image Generator (IG) is to provide one or more pilots with a virtual representation of the real world (especially outside the cockpit) to the maximum degree of fidelity possible with available technology. The more realistic the rendering is, the more effective the training. Since the earliest days of 3D flight simulators, there has been some representation of the terrain (i.e. the ground, lakes, rivers, ocean, etc.). At first, this was just a simple horizon dividing a blue sky from green earth, providing not even the illusion of traveling over the earth. Later came systems using “geotypical” terrain, in which swatches of real-looking terrain were stitched together to provide the illusion of flying over an actual place (e.g. landscapes similar to what one might see in the vicinity of Darmstadt, Germany). These systems generally use actual elevation data from the region being simulated, but generic texture maps. For example, the locations of farmland, desert, and cities might be consistent with reality, while the images used to represent these types of terrain do not correspond to the actual farms and cities over which flight is being simulated.
The current state of the art is known as geospecific terrain, in which both the elevation data and the texture maps that are combined to form the terrain are derived from the real world. The elevation data is typically obtained using radar from satellites (e.g., DTED-1, DTED-2), or other conventional imaging systems. The texture data is also typically obtained from satellite imagery, although aerial photography is commonly used as well. The quality of both elevation and texture data is measured in terms of its resolution. For elevation, this means the distance between “postings,” meaning the locations where the height of the earth is sampled. For example, DTED-1 elevation data provides elevation at roughly 90-meter postings. Imagery resolution is measured similarly. For example, in 16-meter imagery, a single texel (texture element) corresponds to a 16 m×16 m square of the ground. Imagery is available in much higher resolutions, such as 20 cm and beyond.
Additionally, flight simulation is no longer limited to the visual spectrum. For military flight simulators, it is important to also simulate sensors, such as FLIR (Forward Looking Infra Red), NVG (Night Vision Goggles), low light level television, and Radar. In a geospecific simulator, multispectral or hyperspectral (i.e. some or many wavelengths) imagery may be used in conjunction with an accurate model of the sensor's spectral response characteristics to create a physics-based simulation of how a particular sensor will behave during a simulated mission.
For any flight simulator, whether geospecific or geotypical, the distance to the horizon is an important factor in realism. Some simulators have very short horizons, e.g., 5 nautical miles (nmi), which causes the experience to be like constantly flying through dense fog, regardless of the simulated weather conditions. For jet simulators, a horizon of 50 nmi is generally acceptable, and for helicopters, 30 nmi is generally acceptable for effective training using commercially available flight simulators consistent with the current state of the art. In the real world, however, horizons can be much farther, reaching as far as 100 nmi for turbine aircraft. Keeping a large horizon range while showing realistic close range detail at real time interactive update rates is one significant challenge faced by designers of terrain rendering algorithms.
Transforming raw source data into a visual simulation that a user can experience typically involves two main steps: Database Creation and Interactive Rendering. The scene database is a collection of all the data fed into the IG to produce a realistic simulation of a particular place, and particular tactical data (such as specific enemy planes, tanks, etc.), in a format that the image generator can understand and efficiently render. While there are several common, or “open” database formats, it is often necessary to translate these databases into proprietary formats that are optimized for the technical characteristics of a specific IG. Terrain is the largest element of a large geospecific database. The texture and elevation data can range in size from a few gigabytes to over a terabyte for a single scene database. Since the early days of image generation, the position of the pilot's viewpoint (also known as the “ownship”) has been used to control reading, or “paging” of data from an input/output device such as a disk storage system or network. This is especially important for large scene databases because it is impractical or impossible to store the entire scene database in high-speed memory such as RAM.
For geospecific terrains, the Database Creation process includes orthorectification of elevation and imagery, creating 3D geometry from elevation data, assembling individual images into larger texture maps, and converting these texture maps into a format usable by the IG. Generally, the amount of data decreases during this process, as elevation is subsampled to match the geometry performance of the IG and the imagery's precision (the number of bits used to represent each texel) is reduced to match bandwidth constraints of the IG. Multiple Levels of Detail (LOD) of the elevation geometry are generally created, and texture MIP levels are also created.
The Interactive Rendering process is where the IG must render the “area of regard” of the pilot (i.e. the pilot's current surrounding environs) at a rate equal to the display's refresh rate, typically 60 Hz. It is important that the IG constantly meet this refresh rate target, or the scene will appear to stutter or jitter, which is unacceptable for effectively training pilots. The 3D scene can comprise many elements, including sky, sun, clouds, terrain (including ocean), air targets (i.e., other aircraft), ground targets (e.g., tanks, trucks), special effects (e.g., explosions, smoke columns), and cultural features (e.g., buildings, trees). Some of these elements (targets, terrain, cultural features) are contained in the scene database. Other elements (such as special effects, sea) are generated procedurally within the Image Generator.
The method used for rendering the terrain depends on how the data is organized in the scene database. Most geospecific flight simulators organize the terrain geometry into “cells.” In Aechelon Technology's C-Nova Image Generator, for example, which is a runtime engine for flight training applications that transforms commercially available platforms into high-resolution, multi-spectral, multi-channel, geo-specific image generators, the terrain is subdivided into 5 km square cells. For a 1000 km×1000 km database, there are 40,000 cells. The structure is similar to a quilt, where each cell is a “square.” The pilot flies over the quilt. Generally, each cell can be rendered at multiple LODs, with the finest levels of detail drawn nearest the viewer (“ownship”) and the coarsest LODs drawn at the horizon. One purpose for using a LOD mechanism is to maximize the finite geometry capability of the graphics subsystem. It is not necessary to render fine LODs far from the viewer because the fine and coarse LODs are essentially indistinguishable at that range: Beyond a certain range, the cells are not drawn at all. Modern IGs perform “view frustum culling,” in which objects (including terrain cells) that fall entirely outside the viewport (display) are excluded from rendering at the per-object or greater level. In some systems, terrain cells are merged into larger cells in the coarser LODs. This technique can be desirable because it allows a greater variance in th

Affiliated with

Also associated with

Rating

Terrain rendering in a three-dimensional environment does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Terrain rendering in a three-dimensional environment, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Terrain rendering in a three-dimensional environment will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3297535