KR100283071B1 - Fast Texture Mapping Method - Google Patents

Abstract

1. TECHNICAL FIELD OF THE INVENTION

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to video generation by computer graphics, and more particularly to a texture mapping method.

2. Technical problem to be solved by the invention

The present invention is to provide a high-speed texture mapping method that performs a texture mapping at a high speed without using a dedicated board by significantly reducing the amount of computation.

3. Summary of the Solution of the Invention

According to the present invention, in the fast texture mapping method in a video generation system, when coordinates of a circular triangle and a modified triangle vertex are input, the modified triangle is classified, a vertex data structure of the modified triangle is set, and the data structure is determined. A first step of obtaining a start point and an end point of the x-axis scan line of the triangular side while scanning along the y-axis; obtaining a Z value by addition for each pixel along the x-axis scan line; A third step of adding the corresponding pixels in the triangle by matrix operation; And a fourth step of interpolating the luminance values using the table and repeating until all the pixels in the triangle are scanned.

4. Important uses of the invention

The present invention is used for fast texture mapping.

Description

Fast Texture Mapping Method

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the generation of moving images by computer graphics, and more particularly to a computer readable recording medium recording a texture mapping method and a program for realizing the same.

In general, in the field of computer graphics, video animation is generated using texture mapping using the hardware configuration as shown in FIG. 1.

1 is a configuration diagram of a video generation system using a conventional texture mapping board.

That is, various animations are produced by texture mapping using a picture input to the camera 101 or the scanner 102 or an image generated in the computer 103.

Fast texture mapping is required for real-time animation. Since high speed texture mapping currently requires only a large amount of computation with only the central processing unit (CPU) 104, high speed texture mapping is difficult. Therefore, high speed texture mapping is realized using a special dedicated board 105.

2 is a flowchart illustrating an embodiment of generating a video animation using a conventional texture mapping.

After deforming the 3D shape model 202 (203), the texture of the original image 201 is mapped (204) to generate a video animation (205). That is, by mapping the texture of the original image to the deformed shape model, the deformed image may be synthesized.

As the motion or viewpoint of the 3D image model changes, the 3D model may be gradually transformed and texture mapped to obtain a video animation.

The present invention has developed a method of texture mapping at a high speed without using a dedicated board 105 by greatly reducing the amount of computation in the conventional texture mapping method. 3 is a detailed flowchart of general texture mapping.

3 is a detailed flowchart of an embodiment of a conventional texture mapping.

It is determined whether the direction of the triangle that is the texture mapping object is facing the front or the rear (301). That is, it is determined whether the Z value of the triangle normal vector is positive or negative.

Positive values are toward the front and negative sides toward the back. If facing toward the front side, the front and rear relationship of the three-dimensional image is determined using the Z buffer for the pixel in the polygon (303). After calculating the correspondence point between pixels with respect to the pixel which is the frontmost among the pixels drawn so far, the luminance value is interpolated.

In the present invention, a high-speed algorithm is proposed for the calculation of the correspondence point between the pixels 302, the calculation of the Z-buffer and Z value 303, and the interpolation 304 of the luminance value.

Figures 4a and 4b is a conceptual diagram of an embodiment of the texture mapping of a conventional triangle, Figure 4a is for the triangle of the original shape model, Figure 4b is for the triangle of the modified shape model.

That is, FIG. 4A maps a pixel belonging to the triangle A B C and FIG. 4B maps a pixel belonging to the triangle A 'B' C '.

Triangle A B C is part of the circular model, and triangle A 'B' C 'corresponds to triangle A B C in the modified shape model.

Points A, B, and C are the vertices of triangle A B C, and points A ', B', and C 'are the vertices of triangle A'B'C'.

First, the corresponding position in triangle A B C with respect to the pixel in triangle A'B'C 'is obtained. Equations 1 and 2 are used to calculate the corresponding positions.

p = s u + t

p ′ = s u ′ + t v ′

U, v, p in (Equation 1) and u ', v', p 'in (Equation 2) is the same as the following (Equation 3) and (Equation 4).

u = A-C, v = B-C, p = P-C

u ′ = A′-C ′, v ′ = B′-C ′, p ′ = P′-C ′

In addition, u, v in Equation 3 and u ', v' in Equation 4 are as follows.

In other words, u = (v _x , u _y ) ^T , v = (v _x , v _y ) ^T, u ′ = (u ′ _x , u ′ _y ) ^T , v ′ = (v ′ _x , v _y ) ^T to be.

In order to seamlessly texture-map the pixels in triangle A 'B' C ', we use inverse mapping by scanning along the y-axis of the modified triangle A' B 'C.

That is, s and t are calculated from (Equation 2), and then the corresponding point P of P 'is calculated using (Equation 1). In Equation 2, s and t are obtained as in Equation 5 below.

In order to obtain the above equation (5), 6 multiplications and 3 additions are required. In order to find P, four multiplications and two additions are required.

In other words, a total of 10 multiplications and 5 additions are required to find the corresponding points between the two triangles. An object of the present invention is to speed up texture mapping by developing an algorithm for finding correspondences while eliminating multiplication and reducing the number of additions.

For Z buffering, the Z value at P 'is obtained from Z' _A , Z ' _B , Z' _C as in the following (Equation 6). Z ' _A , Z' _B and Z ' _C are the Z coordinates of points A', B 'and C', respectively.

z ′ ₁ = Z ′ _A -Z ′ _C , z ′ ₂ = Z ′ _B -Z ′ _C

z ′ = sz ′ ₁ + tz ′ ₂

Z 'in Equation 7 is the Z value at P'. Two multiplications and two additions are needed to find the Z value.

In general, the position of point P is not an integer pixel position (see FIG. 4).

Where m and n are integers not exceeding x and y, respectively. The luminance value of the position of the point P from the pixel value of the peripheral constant position performs bilinear interpolation as shown in FIG.

5 is a conceptual diagram of one embodiment of conventional bilinear interpolation of luminance values.

In other words, I (m, n) and I (m + 1, n) are used to interpolate I (x, n) and I (m + 1, n) and I (m + 1, n + 1) After I (x, n + 1) is interpolated, I (x, n) is interpolated using I (x, n) and I (x, n + 1). This is expressed by the following equation (8).

In Equation (8) above, I (x, y) is a luminance value at P = (x, y) ^T point. The computation required is six multiplications and three additions. The total computations required for this texture mapping process are 18 multiplications and 10 additions.

In other words, as described above, high-speed texture mapping is required for real-time animation. However, in the conventional technology, high-speed texture mapping requires a large amount of computation using only a general CPU, and high-speed texture mapping is difficult. There was a problem that should be used.

SUMMARY OF THE INVENTION The present invention devised to solve the above-mentioned problems has a high-speed texture mapping method and a program for realizing the texture mapping at high speed without using a dedicated board by greatly reducing the amount of computation in the conventional texture mapping method. Its purpose is to provide a computer readable recording medium having recorded thereon.

1 is a configuration diagram of an embodiment of a video generating system using a conventional texture mapping board.

2 is a flowchart illustrating an embodiment of generating a video animation using a conventional texture mapping.

Figure 3 is a detailed flowchart of one embodiment of a conventional texture mapping.

4A and 4B are conceptual diagrams of one embodiment of a conventional triangular texture mapping.

5 is a conceptual diagram of one embodiment of conventional bilinear interpolation of luminance values.

6 is a flow diagram of one embodiment of a texture mapping in accordance with the present invention.

Figure 7 is a classification diagram of one embodiment of the triangle type to which the present invention is applied.

8 is an exemplary data structure diagram of a triangle to which the present invention is applied.

9 is a conceptual diagram of an embodiment of luminance value interpolation to which the present invention is applied.

10 is an embodiment table for luminance value interpolation to which the present invention is applied.

11 is an embodiment address table for retrieving interpolated luminance values to which the present invention is applied.

According to the present invention for achieving the above object, in the fast texture mapping method in the video generation system, when the coordinates of the original triangle and the modified triangle vertices are input, the modified triangle is classified, and the vertex data structure of the modified triangle is classified. A first step of obtaining a start point and an end point of an x-axis scan line of a triangular side while scanning along the y-axis using the data structure; obtaining a Z value by addition for each pixel along the x-axis scan line; A third step of adding the corresponding pixels in the triangle by matrix operation; And a fourth step of interpolating the luminance values using the table and repeating from the first step until all the pixels in the triangle are scanned.

In addition, the present invention, if the coordinates of the original triangle and the modified triangle vertices are input to the video generation system having a large capacity processor for fast texture mapping in the video generation system, classify the modified triangle, and A first function of setting a vertex data structure and finding a start point and an end point of an x-axis scan line of a triangle side while scanning along the y-axis using the data structure; a second function of obtaining a Z value by addition for each pixel along the x-axis scan line; A third function of obtaining corresponding pixels in the triangle by addition by matrix operation; And a computer-readable recording medium having recorded thereon a program for realizing a fourth function which interpolates the luminance values using a table and repeats the first function until all the pixels in the triangle are scanned.

Hereinafter, a preferred embodiment of the present invention will be described in detail with reference to FIGS. 6 to 11.

6 is a flowchart of an embodiment of a texture mapping according to the present invention.

As shown in FIG. 4, when coordinates of a circle triangle and a modified triangle vertex are input (601), the coordinates of the vertex of the triangle are integerized (602), and the modified triangle is classified into five types as shown in FIG. 7 ( 603).

A data structure of the vertices of the classified triangle is set as shown in FIG. 8 (604). The conversion matrix between the original triangle and the modified triangle is calculated (605).

The x coordinate of the start point and the end point is calculated while descending along the y axis with respect to the triangle (606). For example, in the triangle A'B'C 'of FIG. 4, along the y-axis, the x coordinates of the start point and the end point are obtained at the sides A'B' and the sides A'C ', respectively.

Scanning between the start and end points along the x axis to the right (607), the existing Z value of the pixel of interest is obtained by adding one time to the Z value of the pixel of interest (608), and the new Z value is compared by comparing the new Z value. If it is, the next pixel is scanned and the corresponding position of the circular triangle is calculated (610). In the corresponding position calculation, only two additions are used.

Since the corresponding positions of the circle triangle are generally not integer pixels, pair primary interpolation is performed using the luminance values table (611). This is to reduce the multiplication of (8).

After the scan in the x direction is finished, the next line in the y axis direction is scanned (612). When all the pixels in the triangle are scanned (613), the processing proceeds to the next triangle (614).

The fast texture mapping method of the present invention as described above will be described in more detail for each section with reference to FIGS. 7 to 11.

7 is a diagram illustrating an embodiment of triangular types to which the present invention is applied, and FIG. 8 is a diagram showing an embodiment data structure of a triangle to which the present invention is applied. 9 is a conceptual diagram of luminance interpolation to which the present invention is applied, FIG. 10 is an exemplary table for luminance value interpolation to which the present invention is applied, and FIG. 11 is an interpolated luminance to which the present invention is applied. Example address table for retrieval of values.

First, the positions of points A ', B', and C 'are integerized (602). Types of triangles are classified for scan conversion of triangles A ', B', and C '(603).

Triangles are classified into five shapes as shown in FIG. 7.

In other words, Type I is when the lower side is parallel to the horizontal line and the remaining vertices are on the upper side, and Type II is when the upper side is parallel to the horizontal line and the remaining vertices are on the lower side of the base. By value, when the second vertex is listed in size order, it is on the right side of the straight line connecting the remaining vertices. In addition, type IV is similar to type III, but the second vertex is on the left side, and type V is when three points are in a straight line.

When triangles are classified according to types, they have a data structure as shown in FIG. 8 for scan conversion. According to the data structure of FIG. 8, scan conversion is performed as follows (606).

Type I scans along y axis from y _a to y _b Find the x-coordinate of the viewpoint and the midpoint of the image respectively.

Type II is straight Find the x-coordinate of the top.

Type III is _a straight line from y _a to y _b Find the x-coordinate of the phase and then the straight line from y _b +1 to y _c Find the x-coordinate of the top.

Type IV is _a straight line from y _a to y _b similarly to type III. Scan along the line, and from y _b +1 to y _c Find the x-coordinate of the top.

Type V does not scan convert.

For example, straight line of type I The x-coordinates of the start point and the end point of the scan line in are obtained using the Digital Differential Algorithm (DDA) as shown in Equations 9 and 10 below (606). That is, straight When the point of the k-th scanning lines X _k d

Y _k = Y _k-1 +1

X ₀ = X _a , Y ₀ = Y _a and m _ab is a straight line Is the slope of. When the coordinates of the starting point and the ending point on the y axis are obtained, the Z value is obtained by the following equation (11) in each pixel. In other words, if you modify (7) and substitute (5)

z ′ = a ₁ x ′ + a ₂ y ′

Where a1 and a2 are as shown in Equation 12 below.

z ′ _i = a ₁ x ′ _i + a ₂ y ′ _k

z ′ _{i + 1} = a ₁ x ′ _{i + 1} + a ₂ y ′ _k = z ′ _i + a ₁

z ′ _o = a ₁ x ′ _o + a ₂ y ′ _k

In the above formula, the scan along the x-axis is x _{i + 1} = x _i +1, so only one addition is necessary to calculate z ' _i as shown in Equation (14). Z ' ₀ in (15) is the Z value of the first pixel of the k-th line.

For the Figure 4 P '= (x', y ') corresponding to the location of the ^T P = (x, y) to obtain a ^T (Equation 1) to clean up to, substituted for (Equation 5) instead of s, t If the following equations (16) to (19).

B ₁₁ , b ₁₂ , b ₂₁ , and b ₂₂ are represented by Equation 20 below.

Since the scan along the x-axis is x _{k + 1} = x _k +1, the position of the corresponding point in the circle triangle is calculated as shown in Equation (18). That is, the corresponding position is calculated only by two additions (610).

When the corresponding position (X _i , Y _i ) is calculated, since (X _i , Y _i ) is not an integer position, the luminance value is interpolated using a table from surrounding integer pixels (see Fig. 5).

9 is a conceptual diagram for luminance value interpolation. In calculating the luminance value of I (x, y), multiplication is reduced and interpolated using a table (611).

That is, the luminance value I (x, n) is the luminance value I (m, n) at the peripheral integer pixel position (m, n) and the luminance value I (m + 1, n) at the pixel position (m + 1, n). Interpolated from. Where m is an integer that does not exceed real x. First, x _f multiplies the value of the decimal point of x by 100 and then takes only an integer value as shown in Equation 23.

For this interpolation, a table IT for luminance value interpolation as shown in FIG. 10 and a table AT for address search of the table of interpolated luminance values as shown in FIG. 11 are prepared.

A table (IT) for interpolation of luminance values is a table for storing a value equal to 100 from 1 to 255 in each line.

The table AT for retrieving an address of a table of interpolated luminance values is a table for finding an address of a table IT including interpolated luminance values, and contains 0 to 254 multiplied by 100. The interpolation method of the luminance values is shown in Equations 23 and 24 below.

x _f = [(xm) * 100]

I _d = I (m + 1, n) -I (m, n)

Addr is called | I _d | th value of the table AT of FIG. If Value is the (Addr + x _f ) th value of the table IT, I (x, n) is I (m, n) + Value if I _d1 > 0, and I (m if I _d1 <0. , n)-Value, and if I _d1 = 0, then I (m, n). I (x, n + 1) can be found in the same way. Furthermore, if I (x, n + 1) and I (x, n) are interpolated in the same way, I (x, y) can be obtained.

By calculating in this way, the luminance value can be interpolated twice by two multiplications and seven additions. For more detailed interpolation, the luminance value interpolation table (IT) can be further refined.

Using this method, texture mapping for one pixel is possible with two multiplications and ten additions.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit of the present invention for those skilled in the art to which the present invention pertains, and the above-described embodiments and accompanying It is not limited to the drawing.

As described above, since the high speed texture mapping is possible using only a central processing unit (CPU) built into the computer, a special dedicated board may not be used in real-time video animation, and it may be used for high speed image synthesis. It has an excellent effect.

Claims (6)

In the fast texture mapping method in the video generation system, When the coordinates of the original triangle and the modified triangle vertices are input, the modified triangles are classified, the vertex data structures of the modified triangles are set, and the data triangles are scanned along the y axis, and the x-axis scan lines of the triangle sides are scanned. A first step of finding start and end points; obtaining a Z value by addition for each pixel along the x-axis scan line; A third step of adding the corresponding pixels in the triangle by matrix operation; And Fourth step of interpolating luminance values using a table and repeating from the first step until all pixels in the triangle are scanned Fast texture mapping method comprising a. The method of claim 1, The first step is, A fifth step of integerizing coordinates of the vertex of the modified triangle when the circle triangle and the modified triangle vertex are inputted; A sixth step of classifying the modified triangle into each type; A seventh step of setting a data structure of vertices of the sorted triangles; An eighth step of calculating a transformation matrix between the circular triangle and the modified triangle; And The ninth step of calculating the x-coordinates of the start and end points while descending along the y axis with respect to the triangle Fast texture mapping method comprising a. The method of claim 2, The sixth step of classification, Type 1 means that the lower edge is parallel to the horizontal line and the remaining vertices are on top of the base, while type 2 means that the upper edge is parallel to the horizon and the remaining vertices are on the bottom of the base, and type 3 represents the vertex of the triangle as the y value. According to size, if the second vertex is to the right of the straight line connecting the remaining vertices, type 4 is similar to type 3, but if the second vertex is to the left, type 5 Fast texture mapping method characterized in that is classified as the case where three points are in a straight line. The method of claim 1, 2. The method of claim 2, wherein the Z value is obtained by one addition for each pixel along the x-axis scan line. The method according to any one of claims 1 to 4, The third step is the fast texture mapping method, characterized in that to obtain the corresponding pixel in the triangle by two additions by matrix operation. For a high speed texture mapping in a video generation system, in a video generation system having a large processor, When the coordinates of the original triangle and the modified triangle vertices are input, the modified triangles are classified, the vertex data structures of the modified triangles are set, and the data triangles are scanned along the y axis, and the x-axis scan lines of the triangle sides are scanned. A first function of finding a start point and an end point; a second function of obtaining a Z value by addition for each pixel along the x-axis scan line; A third function of obtaining corresponding pixels in the triangle by addition by matrix operation; And Fourth function that interpolates the luminance values using a table and repeats from the first function until all pixels in the triangle are scanned A computer-readable recording medium having recorded thereon a program for realizing this.