KR100372901B1 - Ray-tracing acceleration method using ZF-buffer - Google Patents

Abstract

The present invention relates to a method for improving the speed of ray tracing using ZF-buffers. An image generation technique using ray tracing in the field of computer graphics, comprising: extracting geometric information of an object to be rendered and surface attributes of an object; Pretreatment step for ZF-buffer; Ray tracing method main processing step; And a resultant image output step.

Therefore, by using ZF-buffer, the speed of ray tracing method is more than three times faster than vista-buffer.

Description

Ray-tracing acceleration using ZF-buffer

The present invention relates to a method for improving the speed of the ray tracing method. In particular, ZF improves the speed of the conventional ray tracing method which pursues the actuality of the generated image as an image generation technique by the ray tracing method in the field of computer graphics. The present invention relates to a method of improving the speed of ray tracing using a buffer.

In general, the conventional ray tracing method is a method of determining the color of the pixel while tracking the light path by generating light toward the object from the observer's point of view every pixel. The ray tracing method requires a lot of calculations to determine whether an object exists on the path of light, and it takes a lot of time to produce an image. To improve this, technologies such as the direction cube [1] and Vista-buffer [2] have been developed. This is basically used to reduce the range of the object to be cross-checked by drawing out objects that can touch in the direction of light exiting from the observer's point of view.

The direction cube is a method of covering the cube at the observer's point of view, subdividing each direction of the cube's surface, and recording auxiliary data listing objects that can cross the light from the observer's point of view.

The Vista-buffer is similar to the direction cube, but instead of covering the cube at the viewer's point of view, it is a way to search for an intersection candidate by indexing screen coordinates by projecting an object onto the surface by placing a face that is configured identically to the screen buffer.

In the previous example, the direction cube and Vista-buffer reduced the cross-discrimination target, but the improved computation amount was not satisfactory because it used the discrimination between light and bounding volume. Even if the bounding volume to cross the light is determined, it must be determined whether to cross the elements constituting the object itself, which also requires a lot of calculations. The elements constituting the object have a longer calculation time than the bounding volume. Not only does it take more than the number of bounding volumes.

Therefore, the speed improvement in ray tracing method is basically to reduce the time spent on cross- discrimination between the elements constituting the object and light. Bounding volume is another object that represents an object and takes less time to cross-discriminate. As the bounding volume, spheres or cubes are used to completely wrap objects, and hierarchies between bounding volumes are used to reduce computation time.

However, the conventional ray tracing method consumes most of the execution time for cross discrimination between an object and light, and thus, the improved technology does not have a significant improvement effect, and thus is not suitable for applications requiring user interaction. This is because the number of objects to perform cross discrimination with light can be reduced, but it does not solve the problem of cross discrimination with the components constituting the object. Z-buffer rendering techniques are commonly used in this field. Although it is impossible to produce realistic video, the process is simple, and hardware accelerator can be applied to guarantee real time processing.

However, the ray tracing method has a problem in that it is impossible to manufacture special purpose hardware because of its complicated process.

The present invention has been proposed to solve the above problems of the prior art, and an object of the present invention is to simplify a part of the processing of the ray tracing method so that the operation of the CPU has sufficient acceleration effect, but existing Z-buffer hardware In order to maximize the efficiency by using, we have improved the cross-discrimination process, which consumes the most time during the ray tracing process, and as a result, the ray tracing method using the ZF-buffer which can perform the ray tracing method in real time. It provides a way to improve the speed.

1 is a target image.

2 is a comparison between Z-buffer rendering and ZF-buffer crossover.

3 is a recursive iteration for the reflection effect of the ZF-buffer method.

4 shows the opposite position X ′ of X on the plane P;

5 is a Utah kettle image (polygon number: 9216).

6 is a Utah kettle image reflected in a mirror.

7A is a flow chart illustrating a ray tracing method for an opaque / antireflective object in accordance with one embodiment of the present invention.

7B is a flowchart illustrating a recursive ray tracing method for the reflection / refraction effect according to an embodiment of the present invention.

In order to achieve the above object, the present invention provides a technique for generating an image by ray tracing in the field of computer graphics: geometric information such as the shape of an object to be rendered using a polygon mesh method, and the color and pattern of the object. Extracting surface properties such as; A preprocessing step for the ZF-buffer which records, in the F-buffer, a pointer of the face data which is the master of the depth value of the object determined in the Z-buffer after the step is performed; A ray tracing main processing step of checking whether the F-buffer value is null for all pixel positions (x, y) on the screen after the step is performed and determining the color of the current pixel according to the result; And outputting a resultant image after the above step is performed, and a method of improving the speed of the ray tracing method using the ZF-buffer.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

(1) ZF-buffer algorithm for the primary ray

ZF-buffer refers to two buffers: Z-buffer and Face buffer (hereinafter referred to as F-buffer). Z-buffer, F-buffer, and screen are all the same size and two-dimensional data structure. All coordinates have one-to-one correspondence. Z-buffer is a buffer used in conventional masking. The ZF-buffer method, applied in the preprocessing step of ray tracing, records the identifier of the face data that is the master of the depth value ('z' value) of the object determined by the Z-buffer in the F-buffer. .

As shown in FIG. 1, the conventional Z-buffer method for making a triangular figure calculates a distance to an object from a camera as shown in FIG. 2 (a), and then applies the pixel to which visibility is determined as shown in FIG. 2 (b). The shading model is applied to determine the pixel value of the screen to be displayed.

However, as shown in FIG. 2C, the process of determining a close pixel in the ZF-buffer method is the same as the Z-buffer method, but instead of determining the pixel value, a pointer of a surface including the pixel is written to the F-buffer. do . The algorithm for generating the F-buffer during the preprocessing is shown in Table 1 below.

Ray Tracing Pretreatment for ZF-Buffers for (all pixels (x, y)) Initialize Z-buffer [x] [y] to a very large value (infinity). Initialize F-buffer [x] [y] to NULL. // NULL means no one currently owns the z value} // transform the coordinate system for (vertex) {converts the vertex to screen coordinates using visual transformation and projection // the camera in the world coordinate system Convert Vertices to Coordinate System} // Determine the contents of the ZF-buffer for (All Faces) {Define all Faces (Polygons) for (All Pixel Positions on the Face (x, y)) {if (z of the current pixel Value <z value of the current pixel position recorded in Z-buffer [x, y]) {put z value of the current pixel in Z-buffer [x] [y] current in F-buffer [x] [y] Insert a memory pointer to a face, ie the polygon's geometry.}}}

After completing the process of Table 1, the Z-buffer has distance information from the camera for each pixel to be displayed on the screen, and the F-buffer has an identifier of a surface that is the owner of each pixel to be displayed on the screen. Therefore, the information on the plane of the ray tracing that meets the primary ray can be easily determined by using the identifier of the data contained in the F-buffer. The Z-buffer is used to determine the F-buffer and is no longer used during the rendering process.

Ray tracing method In this process, we apply shading techniques such as texture mapping using F-buffer contents. This process is the same as the existing algorithm applied to the nearest object after cross discrimination is performed in the conventional ray tracing method, and the algorithms are shown in Table II. ZF-buffers are basically similar to vista buffers, but they do not record the bounding volume, but rather the identifier of the side that will actually appear on the screen in the F-buffer.

Therefore, it is possible to reduce the additional cross discrimination time which is a problem when using the bounding volume, and to determine the pixel value of the surface visible in the primary ray by using only the information of the surface recorded in the F-buffer. Have

Ray tracing method of ZF-buffer method for (any pixel position on the screen (x, y)) {if (if value of F-buffer [x] [y] is not NULL) {Generates light exiting from the camera in the pixel direction. If the current plane is a reflection / refraction {if a ZF-buffer is not created on the current plane {set the new camera; set the screen corresponding to the new camera; the F corresponding to the new camera -Set the buffer. Recurse this function. Specify the resulting image as the texture of the current face.} Determine the color after reflection using the texture map and lighting information.}}

7A is a flowchart illustrating a ray tracing method for an opaque / antireflective object according to an embodiment of the present invention.

The geometry of the object to be rendered and the surface properties of the object use polygon meshes as one of the methods in computer graphics. A polygon is a plane divided by straight lines connecting n vertices. A polygon mesh is a polygon mesh that is composed of several polygons. In addition to the shape of the object, another element representing the object is surface property information of the object. This includes colors and patterns.

Set the Z-buffer [x, y] to infinity for all pixel positions (x, y) and set the F-buffer [x, y] to set the geometry of the object to be rendered and the surface properties of the object to visualize the object. Is set to null to initialize the ZF-buffer (step S1).

Z-buffers and F-buffers are two-dimensional arrays with the same size as the screen. The Z-buffer is an array with depth information of the object closest to the camera, and puts infinity (or a large value close to infinity) as an initial value. The F-buffer is an array with the identifier of the object closest to the camera, with a NULL value indicating nothing at all.

Convert the coordinate system that transforms vertices from the world coordinate system that is the absolute coordinate where the object can be placed for all vertices to the camera coordinate system (visual coordinate system) viewed from the observer's point of view (step S2) and set the boundary of all polygons. (Step S3).

For all pixel positions (x, y) belonging to the polygon, check if the value of the Z-buffer [x, y] is greater than the depth value of the current pixel (step S4), and if it is large, the Z-buffer [x, y] The depth value of the current pixel is assigned to the value, and the memory pointer of the current polygon data structure is assigned to the F-buffer [x, y] value (step S5).

The content determination of ZF-buffer is a method of determining the object to intersect the light by Z-buffer method. The Z-buffer method is a method of rendering an object. The Z-buffer method compensates for the loss of dimensional information generated when a 3D object is displayed in a 2D screen by using an auxiliary buffer called a Z-buffer. The Z-buffer has a one-to-one correspondence with each pixel on the screen and has depth information of the pixels represented on the screen.

When using ray tracing for opaque / anti-reflective objects Ray tracing This processing step checks whether the F-buffer [x, y] value = NULL for every pixel position (x, y) on the screen (step S6). ) If not null (if there is a face identifier at the current position), it generates light exiting from the camera in the pixel direction, calculates the position where this light meets the object, and calculates the position of the current pixel from the surface properties of the object obtained by referring to the position value and the F-buffer. The color is determined (step S7), and if the F-buffer [x, y] value = NULL, the color of the current pixel position is determined as the background color (step S8).

After determining the color of the current pixel, the resultant image is output (step S9).

○ Main processing of ray tracing method

The ray tracing method is a method of determining the color of a pixel based on the surface information of an object that meets the light generated from the camera toward all the pixels on the screen. At this time, a lot of execution time is required in the process of judging the object that meets the light, and the speed is improved by referring to the contents of the F-buffer made during the preprocessing. If a non-NULL value is recorded in the F-buffer [x, y], the ray meets an object and the value recorded in the F-buffer is the object's identifier. If the value of the F-buffer [x, y] is NULL, it means that no object meets the rays.

○ Recursive ray tracing

The recursive ray tracing method is a method in which a ray incident on an object having a reflective / refractive property is reflected / refractored in order to express a reflective / refractive object and moves in a new direction to search for another object. Recursive ray tracing using the ZF-buffer method improves the existing method of individually navigating all rays reflected / refracted by a single polygon, placing a new camera on the opposite side of the polygon and rendering the resulting image We implemented a recursive ray tracing method by mapping to.

(2) ZF-buffer application for secondary rays

The ZF-buffer method is basically applicable only to the primary ray because it can be used to predict the direction of the ray in advance and to easily calculate the direction of the ray. However, the ZF-buffer can be applied to the light beam after the secondary light for reflection / refraction, etc., if the same conditions as the primary light beam are made. In the conventional ray tracing method, when the primary ray touches the reflector, the reflected ray is generated according to the correlation between the reflector and the ray, and the reflection process is realized by repeating the same process from the starting point of the reflected ray. In contrast, the recursive ray tracing method using the ZF-buffer method does not use a single ray of reflection, but sets the reflective surface to a new screen and sets a new camera behind the reflective surface to apply the same process as the primary ray.

3 illustrates a reflection process that occurs when one side f of the reflector is viewed from the position of the camera C ₀ . When the primary ray generated by the first camera C ₀ that generates the primary ray meets the reflection plane f , it generates a secondary ray r . The new camera C ₁ created on the opposite side of the reflection plane f with r is located opposite the same distance as the original camera C ₀ . The ray generated by C ₁ becomes the secondary ray reflected by plane f for the first camera C ₀ , but is the primary ray for camera C ₁ . To find C ₁ , we need to find the opposite position of the given point with respect to the arbitrary plane. The process is as follows.

Referring to Figure 4, t is to say the distance between the plane near jeomyigo, X and t on the P in X α, have a relation of the following Equation (1).

Since t is a point on the plane P , P (t) = 0 and t = x−α * normal (P), and α may be calculated by the following equation (2).

Also, And X ' is the opposite direction of P normal with respect to t, and the point X' in the opposite direction is determined by Equation (3).

At this time, the equation of the plane P is called ax + by + cz + d. Using equations (2) and (3), the following equation (4) is obtained.

However, when calculating the U, V, and N direction vectors representing the axes of the camera coordinate system, d is not meaningful in the plane equations of P. Therefore, the position component X ' and the direction component X'' are calculated by the following equation (5). .

In the above equation, if the equation of (4) is f (x) and (5) is g (x), the camera matrix of C ₁ is determined as follows using Equation (6).

By applying the ZF-buffer method from the camera position C ₁ determined as described above, the reflection effect from C ₀ can be realized. Table 3 shows pseudo codes for the recursive ray tracing method of the ZF-buffer method.

7B is a flowchart illustrating a recursive ray tracing method for the reflection / refraction effect according to an embodiment of the present invention.

In case of using the recursive ray tracing method for the reflection / refraction effect, the ray tracing method processing step checks whether the F-buffer [x, y] value = NULL for every pixel position (x, y) on the screen. (Step S16) If it is not null, it generates light exiting from the camera in the pixel direction, calculates the position where the light meets the object (step S17), and checks whether the current plane is a reflection / refractive object (step S18).

Setting the color of the current pixel position to a background color if the F-buffer [x, y] value = NULL (S22) and outputting the resultant image (S23); And

If the current surface is a reflection / refraction, it is checked whether a ZF-buffer is created on the current surface (step S19). If the ZF-buffer is not created, a new camera is set, a screen setting corresponding to the new camera, and a new camera Set the corresponding F-buffer, recursively execute this function, specify the resulting image as the texture map of the current surface (step S20), and if the ZF-buffer is generated, determine the color after reflection using the texture map and lighting information. (Step S21) Next, the resultant image is output (step S23).

ZF-Buffer Recursive Ray Tracing for (any pixel position on the screen (x, y)) {if (if the value of the F-buffer [x, y] is not NULL) {Creates light exiting from the camera in the pixel direction. If (if the current plane is a reflection / refraction) {if (if a ZF-buffer is not created on the current plane) { Set up a new camera. Set the screen corresponding to the new camera. Set the F-buffer corresponding to the new camera. Recurse this function. Set the resulting image to the texture of the current face. } Use the texture beam and lighting information to determine the color after reflection.

FIG. 5 shows a Utah kettle image (polygon number 9216), and FIG. 6 shows a Utah kettle image reflected in a mirror.

The ray tracing method using the ZF-buffer according to the present invention has the result of speed improvement compared to the existing method such as vista-buffer as shown in Tables 4 and 5. This is the result of testing without hardware acceleration.

ZF-buffer vs. vista-buffer execution time comparison Polygons ZF-buffer (sec) vista-buffer (sec) Speed improvement (times) 1248230440969216 0.2250.2300.2310.240 0.6950.7200.7540.820 3.083.133.263.41

ZF-buffer vs. vista-buffer execution time comparison Reflective surface size world coordinate system ＼ polygon number 200 x 200 400 x 400 1248230440969216 0.741 sec 0.745 sec 0.750 sec 0.760 sec 2.010 sec 2.011 sec 2.015 sec 2.021 sec (Figure 6)

Therefore, by using ZF-buffer to improve the speed of ray tracing method more than 3 times compared to vista-buffer, it is possible to shorten the time to produce the image when applied to the image production tool.

As described above, the method of improving the speed of the ray tracing method using the ZF-buffer according to the present invention has improved more than three times in comparison with the vista-buffer in the processing time, and as the number of objects to be processed increases, the speed improving effect is achieved. There is.

In addition, if the speed improvement technique of the ray tracing method is applied to the image production tool using the ZF-buffer, it is expected that the production time of the image can be shortened and the real-time processing can be performed by using the hardware accelerator.

Although the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art will be able to variously modify and change the present invention without departing from the spirit and scope of the invention as set forth in the claims below. It will be appreciated.

Claims (4)

In ray tracing method of computer graphics, Extracting geometric information such as a shape of an object to be rendered and surface properties such as a color and a pattern of the object by using a polygon mesh method; A preprocessing step for the ZF-buffer which records, in the F-buffer, a pointer of the face data which is the master of the depth value of the object determined in the Z-buffer after the step is performed; A ray tracing main processing step of checking whether the F-buffer value is null for all pixel positions (x, y) on the screen after the step is performed and determining the color of the current pixel according to the result; And And outputting the resultant image after the step is performed. The method of claim 1, The pretreatment step for the ZF-buffer is To set the geometry of the object to be rendered and the surface properties of the object, set Z-buffer [x, y] to infinity for all pixel positions (x, y) and set F-buffer [x, y] to null. Initializing the ZF-buffer by setting (S1); Transforming a coordinate system for vertex transformation from a world coordinate system to a camera coordinate system for all vertices (S2) and setting a boundary of all polygons (S3); And For all pixel positions (x, y) belonging to the polygon, check if the value of the Z-buffer [x, y] is greater than the depth value of the current pixel (S4). Allocating a depth value of the current pixel to and assigning a memory pointer of the current polygon data structure to an F-buffer [x, y] value (S5). How to improve speed. The method of claim 1, When using ray tracing for opaque / reflective objects, the ray tracing main processing step For every pixel position (x, y) on the screen, we check if the F-buffer [x, y] value is NULL (S6) and if it is not null (if there is a face identifier at the current position), the light exiting from the camera in the pixel direction is emitted. Create, calculate the location where this light meets the object, determine the color of the current pixel from the surface properties of the object obtained by referring to the position value and the F-buffer (S7), and if the F-buffer [x, y] value = NULL, And (S8) determining the color of the pixel position as the background color. The method of claim 1, When using the recursive ray tracing method for the reflection / refraction effect, the ray tracing main processing step is Check if the F-buffer [x, y] value = NULL for every pixel position (x, y) on the screen (S16), if it is not null, generate the light exiting from the camera in the pixel direction and where the object meets this light. Calculating (S17) and checking whether the current surface is a reflection / refractive object (S18); Setting the color of the current pixel position to a background color if the F-buffer [x, y] value = NULL (S22) and outputting the resultant image (S23); And If it is a reflection / refraction of the current surface, it is checked whether a ZF-buffer is created on the current surface (S19). Setting the F-buffer, recursively executing this function, and assigning the resultant image as the texture map of the current plane (S20). And then outputting the resultant image (S23), wherein the speed of the ray tracing method is improved by using the ZF-buffer.