We present a new GPU-based rendering method for ray casting of multiple geometric data. Our approach supports a polygonal data to calculate scattered light. Terrain is represented for the base of scalar perturbation functions. The geometric model is based on non-polygonal representation. We present a new representation scheme for freeform surfaces it is possible to combine basic surface and perturbation functions. To recognize the place it is often sufficient to fill areas with the generalized texture patterns, such as "themes". Themes designed beforehand and consumption of memory for storing them is not huge.
Using traditional polygonal representation for the example
complex surface give rise to a range of problems such as
visible surface determination, depth complexity handling,
controlling levels of details, clipping polygons by viewing
frustum, geometry transformations of large number of
polygons [
A method for constructing a triangle mesh whose vertices
coincide with the zero-valued isosurface is the Marching
Cubes algorithm [
Texturing and shading of voxel-based terrain is more difficult than it is for height-based terrain. In the cases that triangle meshes are generated for multiple resolutions, arises the cracking problem.
A method for patching cracks on the boundary plane
between cells triangulated at different voxel resolutions was
described in [
The proposed method includes the following main features: rendering second-order surfaces; rendering surfaces defined on a regular altitude grid (Shape texture); rendering freeform surfaces; scattered light visualization.
Along with possibility to rasterizing 2D space, the main
feature of the proposed method is rasterizing made by 3D
space quadtree subdivision of pyramids of different levels,
which constitute the whole pyramid of vision. Then a
pyramid of the lowest level is binary-tree subdivided into
voxels of the lowest level - Recursive Multilevel Ray Casting
(RMLRC). In the latter case, extent space regions (in depth)
can masked out. Depth subdivision of space performed on the
logarithmic basis. The technique of RMLRC allows
determining an intersection of a ray (pyramid) of any level
with a surface effectively. It is also suitable for fast culling of
a spatial region outside an object. The core of this approach is
effective search of volume elements (hereinafter voxels),
involved in current frame generation, fused with direct
projection [
Traditionally, the parametric form represents each patch in
a freeform surface as a mapping from 2D parameter space to
3D space. Although parametric patches are powerful for
constructing freeform surfaces, processing these patches
poses fundamental problems, only two - constructive solid
geometry (CSG) and boundary representation (Brep)
commonly represent solids exactly, that is, without
approximation. In particular, "separation" and other such
problems associated with curved halfspaces are hard to solve
in parametric patches. We present a new representation
scheme for freeform surfaces - it is possible to combine basic
surface F(x,y,z) and shape-driving function R(x,y,z) or
perturbation function, where shape-driving function
represents smooth deformation of basic surface [
Because of application of the 3D space, subdivision algorithm the possibility appears to render surfaces, which usually consist of huge amount of patches such as freeform surfaces.
Non-regular terrain algorithms are more complex but
have the potential to reduce the number of polygons that the
system must process [
Volume oriented rendering and uniformity of object
processing result in an efficient hidden surface removal and
detection of spatial collisions. Chosen representation of
terrain data is based on regular multi-level elevation map
complemented with levels of detail [
Systematically analyzing geographer’s work, Barr [
Simulation of big areas of terrain in the training system image generators requires the large capacity of memory for the photographic data storage. The usual approach to reduce this information is to compress the data with the help of the methods such as JPEG, etc. In the approach of the texture composition it supposed, that the observer (pilot) often is not interested in the exact, photographic information about the areas, covered by, for example, forest, water, mountains, etc. To recognize the place it is often sufficient to fill these areas with the generalized texture patterns (we will call them "themes" afterwards). Themes designed beforehand and consumption of memory for storing them is not huge. Area of the terrain photo-texture with the same theme approximated by the polygons, which are bounded by curves. Each of the polygons has the pointer to the texture theme and to the procedure of generating the boundaries between the areas with the different themes. The procedures of generating the boundaries do not reproduce true boundaries exactly, but have the goal to keep their features. There are two main problems in the technique of the texture composition. The first is the generation of the boundaries between the areas with the different themes. To resolve this problem, the algorithms, which allows generating the different kinds of boundaries, designed. Among the possible kinds of the boundaries the following can be selected: fractal boundaries, following natural boundaries in the maps of the texture themes (for example, following the streets in the texture map of the town), procedural blend zones (beaches, etc.). The second problem of the technique of the texture composition is the filling the areas by the homogeneous texture pattern. Filling the big area of the terrain by the small texture pattern leads to the undesirable periodic, unnatural picture. The introduction of the different kinds of non-regularity, which keeps the features of the texture pattern, allows obtaining more natural picture.
The first stage is the usual rendering of polygons, describing the texture, and filling the service buffer by the pointers to the texture themes and kinds of boundaries. The second stage is access to the buffer through the perturbation map, to obtain the color of each texel. Perturbation map calculated beforehand. Note that approach is adapted for the generation of the fractal boundaries, but it allows to use the procedural blend zones also, and to generate boundaries, which coincide with the natural boundaries in the maps. To generate fractal boundaries the following actions must be done (some are simplified): access to the perturbation map by address, defined by the U and V coordinates of the given texel, to obtain the additions to the values of U and V. Obtaining the new values of U and V by the addition with the values from the perturbation map. Access to the buffer of the source map by the address, defined by new values of U and V, to obtain the index of the texture theme; access to memory of the texture themes, to obtain the color of the texel. Boundaries between regions, obtained by the method described, looks fractal due to the fractal character of the perturbation map used. Turbulent noise used to fill this map. As it is well known, the higher spatial frequencies of the turbulent noise have the smaller amplitude (and on the contrary). All of the boundaries, obtained by this method, are similar, but different themes can have different boundaries, with different widths (amplitudes) and "crooknesses" (frequencies). Therefore, the sequence of actions, described above, must be little more complex: access to the source map to obtain the characteristics of boundary; scaling of U and V coordinates to obtain required frequency of the noise, then access to the noise map (perturbation map). Multiplication of the values, obtained from the noise map, by the constant value to provide required amplitude of the noise, then modification of U and V; access to the source map to obtain the index of the texture map; access to memory of the texture themes to obtain the color of the texel.
Obviously, blending can become correct only if textural areas have coinciding borders. To satisfy a condition an additional database is introduced which describes faces of blending areas. These faces do not differ from others and their special role is to connect textural areas with different borders. For this purpose, such face owns border description of one area, and a texture of the other area. While the database created, these faces constructed by expanding polygons, i.e. by moving their edges. By the way, procedurally generated areas can obtained exactly the same way: if a certain polygon with any textural theme has a prescribed blended border, the procedure described above executed and a newly created polygon assigned descriptions of the area created.
The boundaries coinciding with natural borders within a textural picture are considered. Such borders can formed by processing the image from the original map buffer after all textural polygons have been rasterized. Beside difficulties with forming the borders between areas of different textures, there is a problem concerning inner filling of these areas by a homogeneous picture. Therefore, in the proposed approach the textural coordinates disturbed by a noise-map to provide singularities on the picture. It is possible to reuse noise values previously obtained while determining the textural theme index. Let us estimate memory and time consumption within the proposed approach. Note, that original map resolution can be less than final texture map resolution. Maximal spatial frequency of the final image determined by a spatial frequency of textural themes and a noise-map; therefore, they can be scaled, so that one texel of original map is transformed into an area filled by a certain picture with a fractal border. Thus, minimal size of original map determined by a required precision of border presentation. The image kept in an original map consists of contiguous areas each filled by a certain textural theme index, that is why these data can be safely compressed using, for instance, Color Cell Compression method (this will ensure up to six times compression). Besides, this task can accomplished during textural polygon rasterizing with no additional time consumption. As noise-map, contents do not explicitly emerge in the final image, periodically wrapped map can used for this purpose and unwanted regularity will not noticed. For instance, fractal border has a repeating structure only if it originally presented as a straight line parallel to one of textural coordinate axes. The task of rasterizing textured polygons and filling original map is not extremely difficult and can be handled by any sufficiently powerful universal processor, moreover the time is limited only by a needed uploading due to the camera’s motion. To determine the color of a texel several steps are completed: fetch operation from original map, from noise-map, then, modification of U and V coordinates, fetch operation from original map and from textural theme memory. As one may see, these operations are quite simple and, therefore, a conventional textural fetch operation can substituted. This implementation has several significant advantages, as follows: overall memory requirements are less then those for a global texture; texture animation features become available by simple means, as fractal border can made animated, if a shift is applied to a noise-map at each frame. In addition, noise magnitude can change resulting in an interesting effect. In analogous way, texture inside the distinct regions can animated. Besides, this approach would be naturally used also for all other feature textures (fetch operation from original texture memory should be excluded). All this could be useful while rendering, for instance, water surface, flames, snow, etc.
It should be also mentioned that implementation described above requires original map, noise-map and textural theme (as MIP map) storing, Generation of levels of detail for textural themes and noise-maps is of no difficulty, however levels of detail for original map is not clear, because these data are indices, which cannot be blended like colors. A straightforward solution of this problem is to eliminate details that are smaller than corresponding texel size, i.e. only wholly covered texels handled during rasterizing original textural polygons.
The vertex shaders compute the light reaching the eye from a source or a reflective object and fog component. The inputs to the vertex shader are vertex position, transformation matrices, sunlight intensity, the sun direction, the various extinction and scattering coefficients. In this implementation using calculate them per-vertex in a vertex shader. If the Sun is not too low in the sky, this approximation gives good results at a low cost. For these assumptions, the illumination of the ground plane and light reaching the eye from it may easily derived (Fig. 3). To find the color perceived along a line of sight, we must consider both the light reflected by objects along the line of sight, attenuated by the inverting fog, and the light scattered towards the eye by fog along the line of sight.
The visualization time is reduced by using the computational resources of a graphics processing unit with compute unified device architecture (CUDA). The result of running the programs on different processing units is the same even if they may have a different number of streaming multiprocessors. A large portion of the cube will be computed in parallel. Among the functions of the graphics processing unit was to calculate the coordinates of points of the surfaces, normals, and illumination. Geometric transformations were performed by the central processing unit (CPU). Rendering results using CPU (i7-2700K) and GPU (GTX 550Ti, GTX 750
Resolution i7-2700K GTX 550TiGTX 750 Ti GTX 950
The surface obtained will be smooth (Fig. 1 and Fig. 3), and a small number of perturbation functions (16) will be necessary to create complex surface forms. The figure shows a result of modeling a scene object by means of free forms, whose description required 2Kbyte information, which is 500 times less than the polygonal description that would take 1Mbyte information.
The main advantages include ease of calculation of points on the surface with quick search and rejection of the regions not occupied by the scene objects. A factor of 100 or more decrease in the number of surfaces for describing curved objects. Operations of the Geometry Processor (CPU) becomes significantly easier and data stream from the Geometry Processor (CPU) to the Renderer (GPU) reduced. Paralleling of the system becomes easier because the total system performance determined by the performance of the Renderer (GPU) which operations can be easily distributed between different screen regions. All problems connected to terrain generation solved. The opportunity to describe objects with grid values, i.e., with shape texture map; it becomes possible to morph objects; it is achieved by plain interpolation or animation of shape texture. It can used to render clouds, explosions, waves on water, etc.