JP2006195939A - Shadow silhouette anti-alias circuit - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce an alias which is generated on the silhouette line of a shadow concerning technique for smoothly drawing the silhouette line of the shadow in drawing the shadow with the use of shadow depth buffers in computer graphics. <P>SOLUTION: When a shadow polygon is drawn in a light source coordinate system in the first pass of a shadow silhouette anti-alias circuit, a pixel is defined as a rectangular grid area in addition to the x, y, z coordinate values of an outline in an outline interpolating circuit, so as to obtain information on the inclination, decimal point part, major/minor axis, right/left line, and start/end point, etc., of the polygonal outline. Edge information and the depth value of the polygon corresponding to the information are stored in a plurality of shadow depth buffers. In a second pass, a polygonal interpolation point to be drawn in a view point coordinate system is converted into the light source coordinate system. Then an area is determined through the use of the silhouette line buffer 22, and a means for determining whether the interpolation point belongs to the outside or inside of the shadow polygon using the shadow depth buffers. Comparison is performed with the depth value of the shadow depth buffer held in the area. The juggy of the silhouette line is reduced by a means which defines the interpolation point as the shadow when existing at a position apart from the light source and as a radiation point when existing at a position close to the light source. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to shadow drawing using a shadow depth buffer in computer graphics, and to a technique for smoothly drawing shadow silhouette lines.

  Concerning the method of drawing shadows by computer graphics, the object is drawn in the light source coordinate system as the first pass, the distance from the light source is stored in the shadow depth buffer, and then the object is drawn again in the viewpoint coordinate system as the second pass. In addition, there is a two-pass depth buffer method in which each drawing point is converted into a coordinate value of the light source system and it is determined whether or not it is a shadow from a comparison of the drawing point depth values. In the above-described method for determining the presence / absence of a shadow using a shadow depth buffer, the shadow polygon (shadow object stored in the shadow buffer) stored in the shadow depth buffer is enlarged and projected because of insufficient buffer resolution. In this case, jaggies (stair-like appearance) are generated on the silhouette lines (outer lines) of the shadows, which significantly reduce the image quality. When the light source is close to a point light source, the shadow drop position increases as the distance from the light-shielding object increases. Therefore, it is considered that this deterioration in image quality due to jaggies is unacceptable in practice. As a method for dealing with this problem, it is conceivable to implement a shadow depth buffer having a size (resolution) several times as large as the supersampling method. However, since the shadow depth buffer has a two-dimensional array structure, when the shadow is projected twice as large as the size 1 of the shadow polygon, it is usually 4 times more memory to draw with uniform smoothness. Capacity is required. That is, a capacity of nxn times the expansion ratio n times is required. This approach is not suitable for commercialization in terms of hardware costs. On the other hand, although there is a method called a shadow mapping method, real-time drawing becomes difficult when an object is complicated and complicated because a shadow map must be created in advance. On the other hand, there is also a method of reducing the jaggy effect by adding a linear filter or the like that can be seen in texture mapping to blur the shadow. The blurring of the shadow changes continuously from a position close to the light-shielding object to a position far from it, and filtering is effective for local (high-frequency component) blurring, but it is effective for large jaggy patterns. Absent. The present invention does not increase the resolution of the shadow buffer corresponding to the image buffer as found in the supersampling method, but relates to means for smoothly displaying silhouette lines of shadows using silhouette line information.

  For the smoothing of silhouette lines, a supersampling method for increasing the resolution at the interpolation point of the shadow polygon is known.

It has the problems pointed out in the section. An object of the present invention is to provide a means for drawing a silhouette line of a smooth shadow in real time at a small hardware cost compared to supersampling. The means for obtaining the silhouette line anti-aliasing of the present invention will be described below.

（１）式において前記傾きは（ｄｘ／ｄｙ）、メジャーマイナーはｄｘ−ｄｙの符号が示し、ライト／レフト・ラインは、本発明においてはアウトラインをＹ軸最小点から左右に分岐して補間するため、描画時の外郭線がその左右いずれかを示すフラグで示すことができる。また小数点以下の値とは傾きを除算した小数点以下の値（以下フラクションと言う）となる。例えばＸ軸メジャーとはｄｘ−ｄｙが正となり、外郭線の傾きが４５°以下の場合を言う。前記はＸ軸メジャー（ｄｘ≧ｄｙ）の場合であるが、ｄｙ＞ｄｘの場合では（１）式においてＸとＹが入れ替わる。（１）式でフラクションは補間点毎に変化する。１つの補間点は１つの画素に対応するが、１つの画素を１辺の大きさを１とする４角形空間（グリッド）と見なせば、傾きとフラクションはそのグリッドを交差する直線の傾きとグリッドの１辺との交差点を示す。これにメジャー・マイナーフラグとライト／レフト・ラインを考慮すれば、グリッドに交差する直線の内側と外側の面積が計算できる。グリッドに始点終点が含まれる場合は、シルエット・ラインバッファには２組のエッジ情報が記憶される。これらからグリッド内の外郭線の交点を求め、内挿補間点の領域の判定を行う。本発明では前記（１）〜（６）のそれぞれを、シルエット・ラインバッファを設けてこれに記憶する。一方、画素点には１つの直線だけでなく複数の直線あるいは接続点が含まれることがある。本発明では少なくとも２つの直線の前記（１）〜（６）を記憶する。フラクションおよび傾きの精度はグリッドの分割度を意味するが、それぞれ３ビット（８×８分割）程度で十分な効果を得る。この結果、シルエット・ラインバッファの容量はスーパーサンプリング法と異なり大きくならない。
シャドウポリゴンをシャドウ・デプスバッファに描画する際、光源にもっとも近いポリゴンのデプス値を記憶する。この際、前記アウトライン補間で得られるシルエットライン情報を含むポリゴンのデプス値が、後書きの他のポリゴンのデプス値と比較して光源よりも離れている場合、シルエットライン情報もデプス値同様に書き換えを行う。この結果、第１パスの終了時点ではシルエット・ラインバッファには光源に最も近いシャドウポリゴンの前記（１）〜（６）のライン情報のみが記憶される。
一方、オブジェクトのポリゴンはシルエットラインだけでなくポリゴン内部で隣接する接続面（ポリゴン境界）をもつ。このポリゴン境界はシルエットエッジではないため、シルエット・ラインバッファにはエッジフラグは立てない。ポリゴン境界であるためエッジフラグをリセットするには、前記（１）〜（６）の情報をシルエット・ラインバッファに書き込む際、すでに記憶されたエッジ情報と比較し、ライト／レフト・ラインが異なり、その他の情報が一致することで可能となる。When rendering a polygon (polygon) that forms an object plane in the first pass drawing in the light source system, the present invention first linearly interpolates vertex information along an outline (outline) connecting the polygon vertices. Thereafter, in the span processor, linear interpolation is performed between the left and right end points of the outline line in the horizontal direction to draw the entire polygon. When the polygon outline is interpolated, the following information (1) to (6) is generated or can be generated by the straight line generator (interpolation circuit) during the interpolation process. (1) silhouette line (edge) flag (2) slope, (3) major minor axis, (4) right / left line, (5) value after decimal point, and (6) start point or end point. The outline is drawn between two points connecting the vertices. Here, when each of the two points is (x ₀ , y ₀ , z ₀ ) and (x ₁ , y ₁ , z ₁ ) (1 ).

In the equation (1), the slope is (dx / dy), the major minor is indicated by the sign dx-dy, and the right / left line is interpolated by branching the outline left and right from the Y-axis minimum point in the present invention. Therefore, the outline at the time of drawing can be indicated by a flag indicating either the left or right. The value after the decimal point is the value after the decimal point (hereinafter referred to as a fraction) obtained by dividing the slope. For example, the X-axis measure refers to a case where dx-dy is positive and the outline slope is 45 ° or less. The above is the case of X-axis major (dx ≧ dy), but in the case of dy> dx, X and Y are interchanged in equation (1). In equation (1), the fraction changes for each interpolation point. One interpolation point corresponds to one pixel, but if one pixel is regarded as a quadrangular space (grid) with the size of one side being 1, the slope and the fraction are the slope of a straight line intersecting the grid. Indicates an intersection with one side of the grid. If the major / minor flag and right / left line are taken into consideration, the area inside and outside the straight line intersecting the grid can be calculated. When the start point and the end point are included in the grid, two sets of edge information are stored in the silhouette line buffer. From these, the intersection of the contour lines in the grid is obtained, and the area of the interpolation point is determined. In the present invention, each of the above (1) to (6) is stored in a silhouette line buffer. On the other hand, a pixel point may include not only one straight line but also a plurality of straight lines or connection points. In the present invention, at least two straight lines (1) to (6) are stored. The accuracy of the fraction and the inclination means the degree of division of the grid, but a sufficient effect is obtained with about 3 bits (8 × 8 divisions). As a result, the capacity of the silhouette line buffer does not increase unlike the supersampling method.
When drawing a shadow polygon in the shadow depth buffer, the depth value of the polygon closest to the light source is stored. At this time, if the depth value of the polygon including the silhouette line information obtained by the outline interpolation is farther from the light source compared to the depth value of other polygons in the afterword, the silhouette line information is also rewritten in the same way as the depth value. Do. As a result, at the end of the first pass, only the line information (1) to (6) of the shadow polygon closest to the light source is stored in the silhouette line buffer.
On the other hand, a polygon of an object has not only a silhouette line but also an adjacent connection surface (polygon boundary) inside the polygon. Since this polygon boundary is not a silhouette edge, no edge flag is set in the silhouette line buffer. In order to reset the edge flag because it is a polygon boundary, when writing the information (1) to (6) to the silhouette line buffer, the right / left line is different from the already stored edge information, This is possible because other information matches.

  When the first pass rendered in the light source coordinate system ends, the second pass starts in the viewpoint coordinate system. In the second pass, the vertex coordinate value defined for the polygon vertex is interpolated to obtain the coordinate value inside the polygon. The interpolated coordinate value is converted into the light source coordinate system of the first pass for each pixel. In the present invention, assuming that the coordinate values converted from the viewpoint coordinate system to the light source coordinate system are xl and yl, the depth value zd stored in the shadow polygon and the silhouette line information are read using these coordinate values. The depth value zd is compared with the zl value of the interpolation point converted into the light source coordinate system, and when the zl value is far from the light source, it is a shadow, and when it is close, it is an irradiation point. At this time, if zd and zl are compared as they are, it is not possible to judge the depth exceeding the resolution of the shadow depth buffer. Here, in the present invention, when the interpolation point is converted into the light source coordinate system, not only the integer part of the xl and yl coordinate values but also the value after the decimal point (fraction) is output. The fraction indicates the interpolation point inside the pixel grid. As a result, the interpolation point including the fraction belongs to the grid area, and whether the shadow polygon outline belongs to the inside or outside of the area created by the intersection with the grid edge. Can be determined using the silhouette line buffer information of (1) to (6). If it is internal, the depth value of the pixel is compared with the external (peripheral pixel) depth value if it is external, and the presence or absence of a shadow is determined. When two lines intersect in the grid, such as a polygon connection point, a determination is made by calculating an area surrounded by the grid edges and two straight lines.

  The integer part of the xl and yl coordinate values indicates the pixel address, and the fraction part indicates the position in the pixel. When the pixel is a polygon edge, the shadow buffer depth value zd and the interpolation point zl are compared. did. When the interpolation point is located outside the silhouette line that intersects with the pixel grid, it cannot be immediately cut off from the irradiation part. An area outside the silhouette line may also be a shadow of another shading object. For this reason, a plurality of depth values are stored in the edge pixel and further comparison with them is required. Two or more sets of depth buffers are prepared as this means, and shadow polygon values closest to the light source are sequentially stored from the shadow polygon closest to the light source, and when the interpolation point is located outside the silhouette line, Compare with the depth value of the corresponding depth buffer. In each of the depth buffers, the depth value closer to the light source is an absolute value, and the other is capable of reducing the memory capacity by storing a difference value. Assuming that the second and subsequent buffers are stored as differential values of depth values, a plurality of depth buffers can be implemented in a system that allows memory cost. The practical minimum number of depth buffers is three sets.

  When polygons with different objects intersect in the same grid, it must be possible to determine which polygon region the plurality of depth buffers belong to. In the present invention, for this purpose, an identifier is added to the edge information, and the depth buffer for the edge information (inside) is made to correspond. This is achieved by adding 1 bit to the edge information. If there are two or more depth buffer systems, a few bits are used as an identifier.

表１は（２）式によって得た複数の内外判定結果とデプス値との関係を示す。表１は２つのエッジ情報との（２）式テスト結果Ｅｄｇｅ１とＥｄｇｅ２を、またＥｄｇｅ１に対するデプス値Ｄｅｐｔｈ１、Ｅｄｇｅ２のデプス価Ｄｅｐｔｈ２および背景のデプス値Ｄｅｐｔｈ３を示す。

表１で内挿補間点がいずれのシルエットの対しても外部となる場合は背景となるシャドウポリゴンのデプス値が比較判定結果の情報となり、この値と前記光源座標系の視点座標からの変換デプス値ｚｌとを比較する。
本発明ではｎ組のエッジ情報をシルエット・ラインバッファに記憶する場合、ｎ＋１個のデプスバッファを実装する。実用的には表１の例に示す２つのエッジ情報と３つのデプスバッファとなる。傾きとフラクションをそれぞれ３ビットとすると画素グリッドは６４に分割されたことを意味し、スーパーサンプリング法に比較してメモリ容量を１／３２にすることが可能となる。In the second pass, when the coordinate value converted into the light source coordinate system is determined to be a silhouette edge, a plurality of pieces of edge information stored in the first pass are read from the silhouette / line buffer. The inside / outside determination of the edge of the interpolation point is performed from the respective edge information using equation (2). Equation (2) shows the case where the x-axis is the long axis, (a) is the right edge (the inside is the upper edge line), and (b) is the left (the inside is the lower edge of the edge slice), and is read from the silhouette line buffer. The interpolation point (x ₀ , y ₀ ) is tested with respect to the slope ΔS _{i of} the i-th edge information and the fraction Y _fi . y _i is a linear expression into which the sampling x coordinate value x ₀ is substituted, and the obtained y _i performs a difference comparison with respect to y ₀ . When the y axis becomes the long axis, the x and y coordinates are switched.

Table 1 shows the relationship between a plurality of inside / outside determination results obtained by the expression (2) and the depth value. Table 1 shows the test results Edge1 and Edge2 of the expression (2) with the two edge information, the depth value Depth1 for Edge1, the depth value Depth2 of Edge2, and the depth value Depth3 of the background.

In Table 1, when the interpolation point is outside of any silhouette, the depth value of the shadow polygon as the background becomes the information of the comparison determination result, and this value and the conversion depth from the viewpoint coordinates of the light source coordinate system. Compare the value zl.
In the present invention, when n sets of edge information are stored in the silhouette line buffer, n + 1 depth buffers are mounted. Practically, the two edge information and three depth buffers shown in the example of Table 1 are used. If the slope and the fraction are each 3 bits, it means that the pixel grid is divided into 64, and the memory capacity can be reduced to 1/32 compared to the supersampling method.

  When a plurality of depth buffers are provided, the depth value usually gives an accuracy equivalent to 32 bits. If a large number of depth buffers are prepared with this bit precision, the hardware cost advantage over the supersampling method is lost. In order to avoid this, there is a method in which the second and subsequent depth buffers store a relative value (difference value) as a distance from the first depth value. If this difference value exceeds the prescribed buffer size, an overflow flag is set using one bit of the buffer exceeding the specified buffer size, and two depth buffer ranges are occupied. In this case, one depth buffer corresponding to the pixel is lost, but practically, in this case, the shadow behind the pixel is considerably far from the viewpoint, and the rendering farther away than the far is performed. The shadow of an object is rarely a problem as a reality. Therefore, there is no problem in expressing a complexly intersecting shadow adjacent to the shadow of the first object.

  As described above, in the present invention, the shadow depth buffer for sequentially storing the depth values of the shadow polygon in order from the light source and the silhouette line buffer for storing the edge information on the outline (silhouette line) of the shadow polygon are provided. In the first pass, means for storing at least three depth values of shadow polygons by hidden surface removal processing, and storing two or more contour line edge information in each of the shadow depth buffer and the silhouette line buffer, and the second pass In this example, the pixel unit of the shadow depth buffer is a rectangular grid space, and the contour line is assumed to intersect this space, and the inside and outside of the shadow polygon in the interpolation interpolation point pixel are determined from the edge information of the contour line. And each depth stored in the shadow depth buffer. Means for determining whether the interpolation interpolation point of the visible polygon is a shadow or an irradiation point using the value, and silhouette line information (1) presence flag (2) of the silhouette line for each pixel in the silhouette line buffer Means for storing each of (1) inclination, (3) long / short axis flag, (4) right / left edge flag, (5) cross coordinate decimal point value, and (6) connection point flag, and the depth value of the shadow depth buffer. The silhouette line information of the silhouette line buffer is a means that makes it possible to determine the belonging by an identifier, and when the silhouette line information is stored in the silhouette line buffer in the first pass, of the silhouette line information, the right / left All the information except for the edge flag matches the information already stored in the silhouette line buffer. In this case, the information to be newly stored resets the presence / absence flag already stored in the silhouette line buffer as the outline of the adjacent polygon, and does not store the new silhouette line information in the silhouette line buffer. Means and three shadow depth buffers for storing from the depth value closest to the viewpoint to at least the depth value closest to the viewpoint in the first pass, and for the pixels on the silhouette line, the depth value of the silhouette line closer to the viewpoint When there is a new intersection, the depth value already stored in the shadow depth buffer is stored in sequence, and if there is a silhouette line that is erased from the shadow depth buffer by the reduction, the silhouette line buffer To the silhouette line The means for erasing the corresponding information, the means for storing new silhouette line information in the silhouette line buffer, and the depth values stored in the second and subsequent buffers of the shadow depth buffer are the values of the previous numbered buffer. As a relative value from the depth value, each means for reducing the capacity of the second and subsequent shadow depth buffers is provided.

  According to the present invention, in computer graphics, a shadow, which is the most important function for recognizing a geometric relationship between objects, can be drawn with a smooth silhouette line, and a high-quality image can be drawn.

  The circuit of the present invention is mounted on a graphics LSI or implemented as an IP (Intelligent Property).

  Examples of the present invention will be described below. FIG. 1 shows one pixel grid of the present invention and a silhouette line intersecting the grid. 1 in FIG. 1 shows a case where one silhouette line 15 intersects, and has an inclination ΔS and an intersection fy between the y-axis. If the grid size is 1, when there is an interpolation point in the grid, the silhouette line has information about which is left or right, and the interpolation point is inside the silhouette line and Whether it is outside or not can be determined by the equation (2). Each of the grids 11 to 13 is a case where two silhouette lines (inclinations ΔS0 and ΔS1) intersect, and fx and fy indicate intersections with the x and y axes, respectively (in the grid 13, the intersections fy0 and fy1 are 2). Crosses the y axis). In any of these states, the region to which the interpolation point belongs can be determined by performing the inside / outside determination on each line. On the other hand, the grid 14 shows a case where a plurality of silhouette lines intersect. In such a case, the interpolation interpolation point area determination is not performed. A multi-line flag (two or more silhouette lines intersect) is set on the target pixel grid. In this case, the average value of the luminance values of the surrounding pixels is set as the luminance value of the interpolation point.

  FIG. 2 shows an overall view of the shadow silhouette anti-aliasing circuit of the present invention. The outline interpolation circuit 20 linearly interpolates the information defined in the polygon vertices along the outline, and information (x, y, z coordinate values, texture mapping coordinate values, surface normals) of each pixel on the outline. Etc.). Further, the span interpolation circuit 21 obtains all vertex definition information inside the polygon by interpolating the end point information of the outline line where the outline line and the horizontal axis intersect. In the outline interpolation circuit 20, in addition to the vertex definition information on the contour line,

The edge information of the terms (1) to (6) is also output. The edge information is stored in the silhouette line buffer 22. One pixel grid can store up to two sets of edge information. The depth value (z coordinate value) on the contour line is compared with the depth value of the shadow polygon stored in each of the three shadow depth buffers 25, 26 and 27 by the comparison circuit 23. The comparison circuit 23 performs the following determination processes (A) to (F).
(A) If the depth value of the contour line is closest to the light source (is closer to the light source than all the depth values stored in the shadow depth buffer), the edge information is stored in the silhouette line buffer. . Further, the depth value is stored in the shadow depth buffer 1 (25), the depth value of the buffer 1 (25) is moved to the buffer 2 (26), and the depth value of the buffer 2 (26) is moved to the buffer 3 (27). To do. At this time, if the edge information is already stored in the silhouette line buffer 22, the first edge information is updated, and the stored first edge information is moved as the second edge information. .
(B) When the depth value of the contour line is farther than the depth value of the shadow depth buffer 1 (25) and is closer to the light source than the depth buffer 2 (26) storing the depth value closest to the light source. Updates the second edge information, updates the depth value of the shadow depth buffer 2 (26), and moves to the buffer 3 (27).
(C) When the depth value of the contour line is farther than the depth value of the shadow depth buffer 2 (26), and closer to the light source than the depth buffer 3 (27) storing the depth value closest to the light source. Assumes that three edges intersect one pixel grid and sets a multiline bit in the silhouette line buffer. The depth value of the shadow depth buffer 3 (27) is updated.
(D) If the edge information is not stored in the silhouette line buffer (the flag is not set) and it is located farther than all or any of the depth values of the shadow depth buffer, The edge information is not stored in the silhouette line buffer, assuming that the outline is located behind.
(E) The span interpolation circuit 21 obtains vertex definition information inside the polygon. The depth value of the polygon is also passed through the depth values of the shadow depth buffers 25, 26 and 27 and the multiplexer 24 in the same manner as the contour line depth value. The depth comparison circuit 23 compares the magnitude relation, and the depth value from the span interpolation circuit 21 is located closer to the light source than the depth value of each shadow depth buffer, and is not a value on the edge. In this case, it is assumed that the edge stored in the silhouette line buffer is located behind the pixel being interpolated, and all the edge flags already stored are reset.
(F) In (E), the depth value from the span interpolation circuit 21 is farther from the depth value of the shadow depth buffer 1 (25), closer to the light source than the depth buffer 2 (26), and on the edge. If it is not a value, the first edge information stored in the silhouette line buffer is left as it is, and the second edge flag (if stored) is reset.
On the other hand, the pixels inside the polygon are output from the span interpolation circuit 21. When the depth value is closest to the light source, the pixel is replaced with the depth value of the shadow depth buffer 1 (25). Further, the depth value of the buffer 1 (25) is moved to the buffer 2 (26), and the depth value of the buffer 2 (26) is moved to the buffer 3 (27). As a result, the depth value previously stored in the buffer 3 (27) is deleted. When the depth value from the span interpolation circuit is the second closest to the light source, the buffer 25 is left as it is, the new depth value is replaced in the buffer 26, and the depth value in the buffer 26 is moved to the buffer 27. If it is close to the third, only buffer 3 is replaced.
As a result of the above processing (A) to (F), when an edge flag is set in the silhouette line buffer, this edge information is the outline of the shadow polygon closest to the light source, or two sets of edge information. When the edge information is stored (determined by checking the edge flag), the depth values stored in the shadow depth buffers 25 and 26 and the first and second values stored in the silhouette line buffer are stored. The edge information is on the outline of the polygon closest to the light source and the polygon closest to the second.
On the other hand, the pixels from the span interpolation circuit 21 also include edge pixels. In this case, the new depth value from the span interpolation circuit coincides with the depth value of the shadow depth buffer. In this case, it is necessary to determine whether the edges of the same polygon currently being interpolated are present. This can be determined by storing the identifier added to the polygon or object in units of pixels, and comparing the new pixel with the already stored pixel.

FIG. 3 shows the shadow silhouette anti-aliasing circuit of the present invention for the second pass. In the second pass, polygon interpolation is performed in the viewpoint coordinate system, and the x, y, and z coordinate values from the span interpolation circuit 30 are converted into light source coordinate values by the light source coordinate conversion circuit 31. The light source system coordinate value becomes the address of the silhouette line buffer 33 and the shadow depth buffer 34, and the edge information and depth value obtained in the first pass are read out. The buffers 33 and 34 correspond to the silhouette line buffer 22 and the depth buffers 1 to 3 (25 to 27) in FIG. 2, respectively. When it is determined from the edge flag of the read silhouette / line buffer that the pixel point is an edge, the area calculation and luminance determination circuit 32 reads the silhouette / line buffer from the silhouette / line buffer.

Using (2) to (6) described in the section, the area determination between the interpolation point and the edge and the depth value of the area are compared. It was determined that the interpolation point of the polygon in the viewpoint coordinate system may be a shadow (the depth value from the span interpolation circuit is farther from the light source than one of the depth values in the shadow / depth buffer) If it is determined that the interpolated interpolation point belongs to the inside or outside of the first and second contour lines, and the inside interpolation point is determined to be inside, the shadow depth that is inside the interpolation point is determined. Compared to the depth value of the buffer, if it is far from the depth value, it is considered as a shadow. On the other hand, when it is near the light source, it is set as an irradiation point. If both are determined to be external in the inside / outside determination of the contour line based on the two edge information, the shadow point is far from the light source in the comparison with the depth value of the depth buffer 3 (27), and the irradiation point is near the light source. And On the other hand, when the interpolation point is located closer to the light source than the depth value of any shadow depth buffer, the interpolation point is regarded as an irradiation point. The region calculation and luminance determination circuit 32 generates a shadow by outputting a luminance scale value to a shader that determines the luminance of the interpolation point for the presence or absence of the shadow obtained by the determination.

  The circuit of the present invention is provided as an IP (Intelligent property) or mounted on a graphics processor LSI so that it can be used for CG video production and an amusement system.

FIG. 1 “Shows a cross-relationship diagram between a pixel grid and a polygon outline relating to the present invention.”
FIG. 2 “Shows a shadow silhouette anti-alias circuit in the first pass of the present invention.”
FIG. 3 “Shows a shadow silhouette anti-alias circuit in the second pass of the present invention.”

Explanation of symbols

FIG.
10-14 pixel grid 15 polygon outline diagram 2
20 Outline Interpolation Circuit 21 Span Interpolation Circuit 22 Silhouette Line Buffer 23 Depth Comparison Circuit 24 Multiplexer 25 Shadow Depth Buffer 1
26 Shadow Depth Buffer 2
27 Shadow Depth Buffer 3
FIG.
30 Span interpolation circuit 31 Light source system coordinate conversion circuit 32 Interpolation point area calculation and brightness determination circuit 33 Silhouette line buffer 34 Shadow depth buffer

Claims (4)

Regarding the drawing of shadows in computer graphics, in the first pass, the shadow polygon is drawn in the light source coordinate system, and in the second pass, the visible polygon is drawn in the viewpoint coordinate system. A shadow depth buffer for sequentially storing values and a silhouette line buffer for storing edge information on contour lines (silhouette lines) of shadow polygons are provided. In the first pass, the shadow polygon depth value is obtained by hidden surface removal processing. Means to store at least three or more contour line edge information in the shadow depth buffer and silhouette line buffer respectively, and in the second pass, draws a visible polygon in the viewpoint coordinate system and makes it visible The interpolation interpolation coordinate value of the polygon As a means for reading each information of the shadow depth buffer and the silhouette line buffer stored in the first pass by converting to a value, and determining whether the interpolation point belongs inside or outside the shadow polygon, Assuming that the pixel unit of the shadow depth buffer is a rectangular grid space, the outline is considered to intersect this space, and the inside / outside of the shadow polygon in the interpolation interpolation point pixel is determined from the edge information of the outline. And a shadow silhouette anti-alias circuit for determining whether the interpolation interpolation point of the visible polygon is a shadow or an irradiation point using each depth value stored in the shadow depth buffer.   In the circuit of claim 1, as contour line edge (hereinafter referred to as silhouette line) information, the silhouette line buffer stores (1) presence flag (2) slope, (3) long and short axis flag for each pixel, 4) Means for storing right / left edge flag, (5) decimal point value of intersection coordinates, and (6) connection point flag, and at least two sets of the silhouette line information in the silhouette / line buffer. A shadow silhouette anti-alias circuit having means for storing, a depth value of a shadow depth buffer, and silhouette line information of a silhouette line buffer, each of which can be attributed by an identifier.   3. When the silhouette line information is stored in the silhouette line buffer in the first pass in the circuits of claim 1 and 2, all of the information excluding the right / left edge flag in the silhouette line information has already been If it matches the information stored in the line buffer, the information to be newly stored resets the presence / absence flag already stored in the silhouette line buffer as the outline of the adjacent polygon, Shadow silhouette anti-alias circuit with a means of not storing silhouette line information in the silhouette line buffer.   3. The circuit according to claim 1, wherein three shadow depth buffers for storing from the depth value closest to the viewpoint to at least the depth value closest to the viewpoint in the first pass are provided, and the silhouette closer to the viewpoint in the pixels on the silhouette line. If there is a new crossing of the depth value of the line, if there is a means to sequentially store the depth values already stored in the shadow depth buffer, and there is a silhouette line that is erased from the shadow depth buffer by the reduction Means for erasing information corresponding to the silhouette line from the silhouette line buffer; means for storing new silhouette line information in the silhouette line buffer; and second and subsequent buffers of the shadow depth buffer. The depth values stored in the Shadow Silhouette anti-aliasing circuit having means for reducing the capacity of the second and subsequent shadow depth buffer by defining as a relative value from the depth value of the buffer.

Images