JP2008152743A - Circuit for calculating texture pattern size - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To allow accurate texture mapping even on a polygon outline. <P>SOLUTION: This circuit interpolates linearly a geometric coordinate value defined on a polygon apex, a texture mapping address and the like. The circuit outputs crossing information of the outline and a pixel to an FIFO memory. The circuit reads out outline information, when receiving the outline information to be compared with an x-axis coordinate value and when a point under interpolation is one on the outline, to find an x-axial coordinate value or a y-axial coordinate value, a texture mapping coordinate value and a perspective transformation relation value, in a point where the outline is crossed with a grid. Distances between the x- and y-coordinate values and between the texture coordinate values are calculated using not grid crossing information but adjacent pixel information, even on the outline, and further a polygon magnification and a texture ratio are found by dividing those distances, respectively. A texture RAM address generator receives calculation results therein to select the optimum pattern. Brightness in an interpolation point is determined based on the texture information and the like. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a logic circuit relating to a means for selecting an optimum texture pattern size for a polygon size when mapping a texture pattern onto a polygon surface in computer graphics drawing technology, and belongs to the field of devices equipped with the circuit.

  In computer graphics technology, texture mapping is one of the most important drawing techniques. In texture mapping, a drawing object is configured as a set of polygonal surfaces (hereinafter referred to as polygons), and three-dimensional coordinate values and texture pattern (hereinafter referred to as pattern) coordinate values are defined at the vertices of each polygon. Using each coordinate value obtained by interpolating on the polygon surface, color information or the like previously stored in the memory is read, and this is mapped to the interpolation point to display the texture. At this time, the geometrical size of the polygon in the drawing space changes due to field conversion or the like, while the pattern coordinate value is not affected. As a result, the polygon to pattern size ratio dynamically changes with the movement of the viewpoint, for example. This change also occurs within a single polygon region under perspective projection. Therefore, the condition for obtaining high image quality is to calculate the ratio for each interpolation point (pixel) of the polygon and select a pattern having an optimum size with respect to the geometric size of the polygon. In texture mapping technology, when the optimal ratio is calculated and the pattern size is determined, two patterns that are closest to the size are selected, the plurality of pattern information is referred to, and further, the filtered information is divided. A trilinear texture mapping technique that obtains a pattern value of an interpolation point by weighting with the ratio is widely implemented in commercial machines. Under perspective projection, not only the polygon-to-pattern size changes, but also the ratio of pattern coordinate axes (hereinafter referred to as u and v coordinates) may change. To cope with this, the RIP map method that selects the most suitable pattern for mapping by calculating the difference in the ratio between polygon and pattern and pattern coordinate axis components in advance by preparing patterns with different uv ratios is also put into practical use. Has been. On the other hand, a large calculation cost is required to obtain the change rate (magnification) of the u and v axes for each pixel with respect to the change rate of the x and y axes inside the polygon. For example, in span (horizontal) line interpolation processing in polygon interpolation, the ratio for each pixel cannot be calculated accurately. This is because span interpolation makes the y-coordinate value constant. Therefore, obtaining a uv ratio using only the change of the x-axis for the change of the polygon shape cannot obtain a highly accurate magnification. In order to avoid this, two horizontal lines are interpolated at the same time, the distances u and v are obtained with respect to the inter-pixel distances on the x and y axes, and the divided values of these distances (hereinafter referred to as polygon magnification), u and There is a means for calculating a division value (hereinafter referred to as uv ratio) of the shift amount of v. However, on the outline of the polygon, there is always an inclined shape in which the two horizontal axes cannot be interpolated simultaneously in parallel, and an accurate ratio cannot be obtained. Thus, depending on the shape, a pattern mapping alias occurs at the polygon boundary (on the outline) where the polygon side contacts.

  It is an object of the present invention to provide a means for obtaining u and v-axis change amounts relative to x- and y-axis shift amounts on the polygon outline, that is, a ratio of the polygon and uv, and circuit structure means for mounting the means. .

一方、外郭線の傾きによっては、２ライン分のスパン並行補間が不可能な領域があり、描画点を基点とする隣接する対角線上の画素点が同時処理として得られない。この場合、１スパンライン分のｘ軸の変移量だけでΔｕとΔｖを求めなければならない。よってポリゴンの投影形状あるいは画素位置によっては、ポリゴン倍率やｕｖ比に大きな誤差を生じる。これに対して３角形の３頂点の幾何学情報からポリゴン全体のスケール比を求める３点法や、スパン補間段階でスパン両端点でのｕおよびｖの差分値を用いる手段も考えられるが、スパン区間でｕおよびｖの変化率は透視変換補正のため画素毎に透視変換関係値で除算されているため線形でない。In the complete interior of the polygon (the region where the drawing point is surrounded by adjacent horizontal and vertical pixels), x and y on the diagonal line adjacent to the drawing point (x, y) are obtained by simultaneous interpolation of two lines in the span direction. The rate of change of u and v with respect to the y-coordinate value (x + 1, y + 1) can be obtained by equation (1). Here, the inter-pixel distance is 1, the polygon magnification is P, and the uv ratio is T. Δu and Δv are obtained by dividing the u and v coordinate values of two diagonal points (x, y) (x + 1, y + 1) by the perspective transformation relation value (homogeneous coordinate system) w at each point, It is the value.

On the other hand, depending on the inclination of the outline, there is an area in which span parallel interpolation for two lines is impossible, and pixel points on adjacent diagonal lines with the drawing point as a base point cannot be obtained as simultaneous processing. In this case, Δu and Δv must be obtained only by the amount of x-axis shift for one span line. Therefore, depending on the projected shape of the polygon or the pixel position, a large error occurs in the polygon magnification and uv ratio. On the other hand, a three-point method for obtaining the scale ratio of the entire polygon from the geometric information of the three vertices of the triangle and a means of using the difference values of u and v at both ends of the span in the span interpolation stage can be considered. The rate of change of u and v in the interval is not linear because it is divided by the perspective transformation relation value for each pixel for perspective transformation correction.

  In the present invention, the change rate of the x and y coordinates, the u and v coordinate values on the outline that causes the problem of insufficient information to be referred to, and the change rate inside the polygon are obtained by individual means. Japanese Patent Application No. 2005-359568 shows a means for storing outline line information in a buffer for anti-aliasing during outline interpolation. In this patent application, the outline line information includes a plane buffer consisting of two bits of a two-dimensional array, and a plane. The outline of the outline, the cross axis, the long / short axis, the x-axis direction, the z value, the color information, and the like are stored in the outline information buffer indicated by the indirect address obtained from the address having the outline line flag. In the patent application, the information stored in the buffer is deleted when the polygon side is shared by subsequent drawing, and the hidden surface is deleted at the span interpolation stage. As a result, only the silhouette line is stored in the buffer after drawing. It was remembered.

In the present invention, similarly to the above-mentioned patent application, interpolation of polygons proceeds by pipeline processing of an outline interpolation circuit and a span interpolation circuit, but differs from the means of the above-mentioned patent application in the following points.
(1) The outline information is stored only until the texture mapping address is determined from the outline interpolation to the span interpolation. Therefore, a buffer for storing outline line information for all outline lines is not required.
(2) No buffer processing due to polygon shared edge deletion or hidden surface deletion is performed, and outline information is processed in a pipeline register or FIFO (First IN First Out) from the outline interpolation circuit to the texture address generation circuit in synchronization with each process. ).
(3) As contour line information, the inclination ΔSxy with respect to the xy coordinate axis, x, y grid contour line intersection coordinate value, xy long / short axis flag, x axis direction, u and v axis grid intersection coordinate values (u0, v0), perspective The grid intersection value w0, x or y of the transformation relation value includes the slopes of u and v axes (ΔSxu, ΔSxv) with respect to the major axis, and the slope ΔSxw of the perspective transformation relation value w with respect to the major axis.

ここで、長軸（上式ではｘ軸）に対するｕ、ｖおよびｗの傾きは、外郭線補間回路において補間計算の際に求めた値をそのまま、スパンおよびテキスチャーマッピング回路へ転送する。（２）式においてｙ軸が長軸となる場合は、ΔＳｘｙの分母分子が逆になり、またグリッド交差軸がｘ長軸においてｘ軸（例えばｘｃ点）となる場合は、それぞれの傾きは、前記それぞれの傾きに対して（１−ｘｃ）を乗算した値となる。グリッド内の交差軸間の距離ΔＬｘｙに対しても（１−ｘｃ）を乗算した値となる。ｙ軸が長軸で、ｙ軸に交差する場合も同様の考えで処理を行う。前記透視変換関係値ｗは、視点と投影面との距離および視点からレンダリング点までの距離の逆数１／ｚを含み、それぞれのポリゴン頂点のｕおよびｖ値はｗ値で除算された値として定義する。これは透視投影での消点に向かっての図形の非線形な変化を表現するための一般的に知られた処理である。In the present invention, the polygon magnification and the uv ratio are obtained from the outline information on the outline by Equation (2). In equation (2), the x axis is the long axis and the cross coordinate value is the y axis.

Here, the slopes of u, v, and w with respect to the long axis (the x axis in the above equation) are transferred to the span and texture mapping circuit as they are, with the values obtained during the interpolation calculation in the outline interpolation circuit. In equation (2), when the y-axis is the major axis, the denominator of ΔSxy is reversed, and when the grid crossing axis is the x-axis (for example, the xc point) in the x-major axis, the respective slopes are It becomes a value obtained by multiplying each inclination by (1−xc). The distance ΔLxy between the intersecting axes in the grid is also a value obtained by multiplying (1−xc). When the y-axis is the long axis and intersects the y-axis, the same process is performed. The perspective transformation relation value w includes the distance between the viewpoint and the projection plane and the reciprocal 1 / z of the distance from the viewpoint to the rendering point, and the u and v values of each polygon vertex are defined as values divided by the w value. To do. This is a generally known process for expressing a non-linear change of a figure toward a vanishing point in perspective projection.

（３）式は補間画素点が外郭線上であり、スパン補間において隣接するスパンラインが存在しない場合に適用し、それ以外では少なくとも２スパンラインの隣接する画素の最大距離（対角線上）とその距離間でのパターン座標値の差分値である（１）式を用いる。In the present invention, the polygon magnification P and the uv ratio T are obtained from the respective values obtained from the expression (2) according to the expression (3).

Equation (3) is applied when the interpolation pixel point is on the outline and there is no adjacent span line in span interpolation. Otherwise, the maximum distance (on the diagonal line) of adjacent pixels of at least two span lines and its distance Equation (1), which is the difference value of the pattern coordinate values between them, is used.

  The calculation of the ratio between the xy geometric coordinate value of the polygon and the texture pattern coordinate value directly defines the pattern coordinate value at the polygon vertex. On the other hand, as one of computer graphics drawing technologies, a peripheral image surrounding an object is created as a six-sided texture pattern, and the line-of-sight vector is reflected from the line-of-sight reflection vector reflected by a drawing point on the polygon. There is an environment mapping that calculates the plane intersecting the pattern coordinates and maps the intersection pattern information to the drawing point. In the environment mapping, the pattern coordinate values (u, v) are replaced with the line-of-sight reflection vector on the polygonal surface as compared with the texture mapping described above, and the corresponding surface and pattern coordinate values (u, v) are converted from the line-of-sight reflection vector. Except for the required points, the processing is the same as the texture mapping. That is, in contrast to the information in [006] (3), in environment mapping, as the outline information, the gradient ΔSxy, x, y grid intersection coordinate values, long and short axis flags, x-axis direction, line-of-sight reflection with respect to the x and y axes The grid intersection coordinate value R0 of the vector R, the grid intersection value w0 of the perspective transformation relation value, x, y, the slope ΔSxr of the line-of-sight reflection vector R with respect to the major axis, and the slope ΔSxw of the perspective transformation relation value w with respect to the major axis. Here, the line-of-sight reflection vector is shown in bold, including x, y, and z components, for example, R means R (Rx, Ry, Rz).

以上の、本発明のそれぞれの手段によって、ポリゴン倍率とｕｖ比を、外郭線上においても外郭線１画素から得られることと、ポリゴン共有辺でのテキスチャーパターンの選択サイズの不一致を避けることにより、高品質の画像を得ることができる。In the outline information, the line-of-sight reflection vector R is usually obtained from the surface normal N of the polygon at the interpolation point and the line-of-sight vector V using the equation (4). When the line and line-of-sight vector are to be interpolated, and the line-of-sight reflection vector R is calculated after span interpolation, R0 as contour line intersection information replaces the grid normal values and line-of-sight vector grid intersection values N0 and V0. In this case, the surface normal and the line-of-sight vector are usually divided by the perspective transformation relation value w, and then the line-of-sight reflection vector is obtained. Therefore, the division of R by w after the line-of-sight reflection vector calculation is unnecessary. In the environment mapping, from the line-of-sight reflection vector R in the equation (4), a plane f where the vectors intersect and (u, v) in the plane are obtained. In the present invention, on the contour line having only one grid reference point, the polygon magnification and the uv ratio are obtained by taking the difference of f, u, v between two intersections intersecting in the grid.

By each means of the present invention, the polygon magnification and the uv ratio can be obtained from one pixel of the contour line even on the contour line, and by avoiding the mismatch in the selected size of the texture pattern on the polygon shared side, A quality image can be obtained.

  According to the present invention, it is possible to draw an alias-free video in real time.

  The technique of the present invention is implemented in an LSI, a programmable logic circuit or the like or in the form of IP (Intelligent Property), and is applied to a computer graphics processor.

  Examples relating to the present invention will be described below. FIG. 1 shows a texture mapping magnification calculation circuit according to the present invention. In FIG. 1, three-dimensional geometric coordinate values (x, y, z), texture mapping addresses (u, v), normals, light source incident vectors, etc. are defined for the polygon vertices. First, along the outline of the polygon, The vertex definition information is linearly interpolated by the outline interpolation circuit 10. This interpolation proceeds two contour lines branching left and right from the vertex (the bottom of the polygon) that is the minimum value of the y axis, and simultaneously proceeds upward in the y axis. When the left and right interpolation points on the same horizontal axis are aligned, the span is reached. The vertex definition information of both ends of the contour line is output to the circuit 11. By executing span interpolation sequentially from the bottom to the top of the polygon, all interpolation information in the polygon is obtained. Of the interpolation information, a polygon coordinate value, a surface normal for determining the luminance inside the polygon, a light source incident vector, and the like are sent to the shader 15 via the span interpolation circuit 11. On the other hand, the outline interpolation circuit 10 outputs information on the intersection of the outline and the pixel (the pixel is regarded as a square area called a grid and the outline intersects this area) to the FIFO memory 12. 2a in FIG. 2 shows an example of a data format stored in the FIFO. In FIG. 2a, it consists of a valid flag F bit 20 indicating an outline, a bit field of an outline x-axis coordinate value 21, and outline information 22. The contour line information includes the slope ΔSxy, x, y grid crossing coordinate values, xy long and short axis flags, x-axis direction, u- and v-axis grid crossing coordinate values (u0, v0), and perspective transformation-related grid crossings. The values w0, x or y include the inclinations of the u and v axes (ΔSxu, ΔSxv) with respect to the long axis, and the inclination ΔSxw of the perspective transformation relation value w with respect to the long axis.

In FIG. 1, the span circuit 11 further linearly interpolates the contour line interpolation information at the left and right end points from the left end toward the right end, and includes pixel points on the contour line. In the span interpolation process, it cannot be determined by the span circuit itself which pixel is on the contour line or the polygon complete internal point. This relationship is shown in 2b of FIG. In FIG. 2b, there are two outlines 23 and 24. In section A, the outlines 23 and 24 overlap. In section A, there is no y + 1 pixel (or line) adjacent on the outline, and there is no pixel point (x + 1, y + 1) to be referenced. On the other hand, in the pixel of the outline line 24 in the section B, adjacent pixels on the y + 1 line can be obtained by the span processing of two lines simultaneously. As described above, there are an arrangement in which adjacent pixel points on two diagonal lines or in the y-axis direction are obtained and an arrangement in which the adjacent pixel points cannot be obtained. In the section where adjacent pixels are obtained, interpolation points on the diagonal are obtained as shown in FIG. 2c, and interpolation information at points A and B can be obtained. That is, the respective distances x, y, and uv at the center point of the grid on the diagonal line are obtained, and the processing of the above equation (1) is performed. Since the grid size is 1, the distance between x and y is the square root of 2 on the diagonal line.
The span circuit 11 outputs the interpolated x-axis coordinate value to the contour line detection and ratio calculation circuit 13, and the circuit 13 receives the contour line information including the x-axis coordinate value from the FIFO 12, and receives these two circuits. To the x-axis coordinate value from. If the outline line valid flag is set and the x-axis coordinate values match, it can be determined that the span interpolation point being interpolated is a point on the outline line. If the circuit 13 knows that the point currently being interpolated is on the outline, the outline information (22 in FIG. 2a) from the FIFO 12 is read out from the FIFO, and the x at the point where the outline intersects the grid. Alternatively, a y-axis coordinate value, a texture mapping coordinate value, and a perspective transformation relation value are obtained. If the span interpolation circuit 11 can process two span lines simultaneously in parallel even on the outline line, the outline line detection and ratio calculation circuit 13 does not use grid intersection information but adjacent pixel information (FIG. 2c). ).

  FIG. 3 shows an intersection of a grid and an outline line intersecting the grid. FIGS. 3a and 3b are examples in which the x-axis is the long axis, and the y-axis (FIG. 3a) and the x-axis (FIG. 3b) are the axes where the intruding contour lines intersect. After obtaining x, y, u, v, and w at grid intersections P0 and P1 in the figure, and then calculating u / w and v / w for each point, the respective difference values between the two points The (distance) is obtained, and the polygon magnification and the uv ratio are calculated by the equations (2) and (3). That is, the polygon magnification and the uv ratio are obtained by using the xy coordinate distance between P0 and P1 in the grid and the uv coordinate system distance, respectively. As a result, the polygon magnification and the uv ratio can be obtained with only information in one grid on the outline. On the other hand, if the u and v values of the point (x + 1, y + 1) or (x + 1, y-1) adjacent to the span interpolation point (x, y) are obtained for each interpolation, the calculation in one grid is performed. First, the polygon magnification and the uv ratio are obtained by the above equation (1). This is because, when the distance calculation point is between the grid center points on the diagonal line, the distance width increases as compared with the distance inside the grid, and the magnification and the ratio accuracy increase. Rather than calculating the distance between xy and uv between adjacent diagonal pixels as in equation (1) above, the grid side size is 1, so u and v between the four vertices of the grid are differentiated. Although there is means for directly defining Δu and Δv which are the maximum values among them as addresses of the RIP pattern, the accuracy becomes higher by using the uv value for a larger xy distance.

Returning to FIG. 1, in the circuit 13 of FIG. 1, the x coordinate value 21 from the FIFO 12 of FIG. 1 shown in FIG. 2 matches the x coordinate value from the span interpolation circuit 11 of FIG. If 20 is valid and it is determined that the span interpolation point is an outline grid, the in-grid intersection is obtained using the outline information 22 in FIG. 2a, and the polygon magnification and the uv ratio are calculated. The details are shown in FIG. In FIG. 4, the x coordinate value 21 shown in FIG. 2a from the FIFO 12 of FIG. 1 is compared with the x coordinate value from the span interpolation circuit by the coincidence detection circuit 41, and the coincidence information is the grid intersection and perspective correction division circuit 40. To give. In the circuit 40, the x and y coordinate values (x, y) and (x + 1, y + 1) for two span lines from the span interpolation circuit, the texture pattern addresses u and v at those points, and the perspective transformation relation value w are further obtained. The outline information from the FIFO is received for each step. That is, it is assumed that the span interpolation circuit 11 in FIG. 1 simultaneously interpolates two horizontal axes on the y and y + 1 axes. In the circuit 40, it is determined whether or not simultaneous interpolation of (x, y) and (x + 1, y + 1) is possible in each outline line. On the outline line where simultaneous interpolation is not possible, the outline of the grid is determined from the outline information. The points where the xy axis and the outline line intersect, the x and y coordinate values and u, v and w values at those points, and u and v at (x, y) and (x + 1, y + 1) inside the polygon And w are obtained, the u and v values are divided by the w value, and the difference between these values is output to the distance calculation circuit 42. The circuit 42 performs the calculation shown in the above equation (2), and further obtains the polygon magnification and the uv ratio by the division shown in the equation (3) by the ratio calculation circuit 43 including a divider. On the other hand, if the u, v, and w values on the two-span line diagonal are obtained even inside the polygon or on the outline, it is not necessary to calculate the intersection in the grid. Further, the distance of the xy coordinates is constant, and the u and v values are subtracted from each other after being divided by the w value. This is the relationship of the formula (1).
In FIG. 3, in the case of FIG. 3c in which the outline has the start point or the end point inside the grid, the longer distance between P0-P1 or P0-P2 is selected, or between P0-P1 or P1-P2 Choose the longer one. The point P1 in FIG. 3c can be obtained from the respective outline information as the intersection of the two outlines P0 and P2.

Returning to FIG. 1, the texture RAM address generator 14 receives the texture coordinate values (u, v), the uv ratio Δv / Δu, and the polygon magnification ΔLuv / ΔLxy as output information of the ratio calculation circuit 43 of FIG. A predetermined address of the RAM 16 in which a texture pattern having an optimum size and a given uv ratio is stored is generated. The texture information read from the RAM 16 is given to the shader 15 to determine the luminance of the interpolation point.
As is apparent from the above description, in the present invention, in the texture mapping to the polygon, as a means for selecting the optimum size of the texture pattern with respect to the polygon size, the outline line information in the pixel grid and the interval between a plurality of span lines are used. Two methods of distance are used.

  By mounting the circuit of the present invention as an IP (Intelligent Property) or a graphics processor LSI chip, real-time drawing can be performed in a system such as video production or CG game.

“Shows the texture mapping magnification calculation circuit of the present invention.” “Shows outline data, outline relation, and pixel spacing of the present invention.” “Shows the outline grid crossing aspect of the present invention.” “The outline detection and ratio calculation circuit of the present invention is shown.”

Explanation of symbols

FIG.
10 outline interpolation circuit 11 span interpolation circuit 12 FIFO memory 13 outline detection and ratio calculation circuit 14 texture RAM address generator 16 shader 16 texture RAM
FIG.
FIG.
20 outline line valid flag 21 x-axis coordinate value 22 outline line information Figure 2b
23 outline 24 outline 2 figure 2c
25 Diagonal pixel relationship diagram 3
3a y axis crossing outline 3b x axis crossing outline 3c grid end point outline 4
40 Grid intersection and perspective correction division circuit 41 Match detection circuit 42 Distance calculation circuit 43 Ratio calculation circuit

Claims (3)

  Regarding computer graphics / texture mapping technology, as means for determining the texture pattern size adapted to the polygon size for each pixel, in polygon interpolation, means for successively performing interpolation along the span line from polygon outline interpolation Each interpolation point is regarded as a square area (hereinafter referred to as a grid) for each pixel, and contour line information is stored for each interpolation as a means for obtaining a coordinate value at which a polygon outline intersects the grid side. Means for sending the contour line information to the circuit, and in the texture mapping circuit, the contour line information is read, and the distance between the intersections where the polygon contour line intersects the grid side and the texture coordinates between the intersections Means for determining each of the above and each of the above Divide the distance to determine the ratio of the texture pattern size to the polygon size in the grid, and further divide the length of each texture mapping coordinate axis component to obtain the aspect ratio of the texture pattern mapped under perspective transformation. In a polygon interpolating region capable of parallel stepping in the interpolation of two or more span lines, the distance between two pixels adjacent on the diagonal line between the two or more span lines, A texture pattern size calculation circuit having both means for obtaining the polygon size to texture pattern size ratio and the texture pattern aspect ratio using the texture coordinate distance between pixels.   2. The circuit according to claim 1, wherein the contour line information includes an inclination with respect to an xy coordinate axis, a grid contour line cross coordinate value, a long / short axis flag, an x axis direction, a grid cross coordinate value of a texture coordinate axis, a grid cross value of a perspective transformation relation value, a polygon A texture pattern comprising means for setting the inclination of the texture coordinate axis with respect to the coordinate long axis and the inclination of the perspective transformation relation value with respect to the long axis, and means for storing the respective outline information in a FIFO (First IN First Out) memory. Size calculation circuit.   A computer graphics image device using the texture pattern size calculation circuit according to claim 1.

Images