JP4440632B2 - Translucent object display circuit - Google Patents

Description

  The present invention relates to a rendering technique for expressing the thickness effect of a translucent object by computer graphics, and is applied to an LSI as a real-time visualization circuit of a virtual reality system.

For computer graphics drawing of a semi-transparent object, the transmittance of the object is defined together with the texture pattern, and this is mapped to the inside of the polygon. There are interpolating methods and the like, and the color of the polygon and the color of the object as the background are synthesized using the transmittance for each pixel obtained by these methods. This composition makes it possible to change the transparency for each pixel. However, in the case of an object such as a cloth, the surface normal of the cloth viewed from the viewpoint direction changes from a parallel state to an orthogonal state due to movement fluctuation. When the viewpoint direction and the surface normal are orthogonal to each other, the cloth has a state of overlapping and thickness, and the transparency is reduced. On the other hand, in the case of a lace curtain, the transparent space (area) between the meshes is the maximum when the surface normals are parallel when viewed from the viewpoint direction, and has the highest transmittance. When orthogonal, the transparent space becomes zero and becomes opaque. For such an object, the transmittance is proportional to the size of the mesh space viewed from the viewpoint. The conventional transmittance definition method described above cannot express a change in transparency related to the viewpoint and the direction of the surface. As a result, the video is different from reality.
On the other hand, a light reflection reasoning method such as a phone model that is often used cannot be applied to a translucent object or a lace cloth. This is because the angle formed by the light source incident angle and the surface normal is 9
This is because the diffusion component and the specular reflection component are both zero on the surface exceeding 0 degree. Luminance bias by ambient light is possible, but when this model method is applied to a translucent object, there is a sense of discomfort in the display effect that is not natural. For translucent objects, it is necessary to consider the irradiation of the opposite surface by the passing light.

JP 2001-056868 A JP 2001-229403 A JP-A-11-203486

  When the translucent object is in a state orthogonal to the viewpoint direction due to fluctuations or the like, the transmittance decreases. In order to obtain this effect as an image, it is necessary to dynamically change the transmittance according to the angle formed by the viewpoint direction and the surface normal. The present invention is a technique for obtaining an opacity effect by generating a scale value of transmittance using a positional relationship between a translucent object surface and a viewpoint. On the other hand, the present invention also solves the generation of luminance by irradiating transmitted light on the back surface with respect to the light source incident surface by defining a model with an expanded light reflection range.

In the present invention, a transmittance, a surface normal, and a line-of-sight vector are defined at each vertex of the polygon. The normal and line-of-sight vector inside the polygon are obtained by interpolation of the polygon, and the angle formed by these two vectors is obtained. The direction cosine cos θ formed by the surface normal and the viewpoint vector is obtained from the equation (1). In equation (1), (Nx, Ny, Nz) and (Ex, Ey, Ez) are unit vectors in the normal and viewpoint directions, respectively. When the transmittance defined at the vertex is α, the transmittance t in consideration of the angle formed by the viewpoint and the surface normal is given by equation (2). Here, k and n are a coefficient and an index value related to the transmittance of the object and the mesh free area. The t function of the equation (2) is only one variable of cos θ, and each equation can be directly constituted by a multiplier and an adder / subtracter. Alternatively, a RAM table having cos θ as an address can be used. The synthesis of the color P of the polygon itself and the background color B is obtained by equation (3). By making each of the equations (1) to (3) into a circuit, a change in transmittance depending on the relationship between the object plane and the viewpoint direction can be expressed in real time. cos θ = Nx · Ex + Ny · Ey + Nz · Ez (1) t = α × (1−k · sin ²ⁿ θ) (2) Ip = (1−t) · P + t · B (3)

  In computer graphics, the phone inference model defined by equation (4) is often applied to light reflection expression. I = Kd · (N · L) + Ks · {2N (N · L) −L} V (4) In Equation (4), N, L, and V are a surface normal, a light source incident vector, and a line-of-sight vector, respectively. Kd and Ks are diffuse reflection and specular reflection coefficients. Here, the range of the inner product cos θ of N and L is given by 0 ≦ θ ≦ 90 °, and when the angle formed by N and L exceeds 90 °, both diffuse reflection and specular reflection components become zero. This setting is effective for opaque objects, but for semi-transparent objects, a reflection model different from that for opaque objects is created on the surface opposite to the direction of the light source because the light passing through the surface on the light source side illuminates the back surface. Necessary. In the present invention, equation (5) is set as means for obtaining the effect of illuminating the back surface.

ここでＫｄ１≦Ｋｄ０およびＫｓ１≦Ｋｓ０とする。拡散Ｋｄ１および鏡面Ｋｓ１反射係数は透明度に依存する。以上から本発明では物体の幾何学的な厚みを計算しなくとも、面の法線と視点角を用いて半透明物体の厚み効果を出すことができる。

Here, Kd1 ≦ Kd0 and Ks1 ≦ Ks0. The diffusion Kd1 and the specular Ks1 reflection coefficient depend on the transparency. As described above, in the present invention, the thickness effect of the translucent object can be obtained by using the surface normal and the viewpoint angle without calculating the geometric thickness of the object.

  According to the present invention, a realistic image can be drawn in real time when the transparency of a portion where an object such as a lace or a semi-transparent cloth shakes and leans is changed.

  The circuit of the present invention is implemented in an LSI as a computer graphics renderer.

  Examples of the present invention will be described below. FIG. 1 shows a translucent object display circuit of the present invention. In FIG. 1, the normal, line-of-sight vector, and light source incident vector of the surface of the semitransparent object are N, V, and L, respectively. These vectors are defined for each vertex of a polygon (usually a triangle), and values inside the polygon are obtained by linear interpolation using these vertex data. In obtaining the luminance of an arbitrary point in the polygon thus obtained, the surface normal and the line-of-sight vector are obtained in the inner product circuit 1 by the direction cosine (N · V) formed by each. The direction cosine cos θ is zero when the viewpoint and the direction of the surface are orthogonal, and is parallel and has a maximum value of 1.0. The cos θ is given to the RAM table 2, and the output is obtained by scaling the transmittance α of the material itself defined as the translucent object by the multiplier 5a. That is, the output of the multiplier 5a becomes a transmittance influenced by the relationship between the viewpoint and the inclination of the object plane. On the other hand, the surface normal N and the light source incident angle vector L are obtained by calculating the direction cosine (N · L) by the vector inner product circuit 3 and then by the specular reflection angle circuit 4 by R = 2N (N · L). The reflection vector R is obtained from the relational expression of −L, and this is taken as an inner product with the line-of-sight vector V. What is supplied from the inner product circuit 3 and the specular reflection circuit 4 to the multipliers 5b and 5c becomes a diffuse reflection component and a specular reflection component on the translucent object, respectively. The inner product circuit 3 that performs the inner product of the surface normal and the light source incident vector also outputs a code when the angle exceeds 90 °. For example, 0 is set when the angle formed with the light source is within ± 90 °, and 1 is set when the angle is not less than ± 90 °. This signal is supplied to the multiplexers 6a and 6b, and the diffuse reflection coefficients Kd0 and Kd1 and the specular reflection coefficients Ks0 and Ks1 are selected. Kd0 and Ks0 are within ± 90 °, and Kd1 and Ks1 are as described above. Each reflection component is multiplied by a coefficient by multipliers 5b and 5c, and then added by adder 7a. In an actual circuit, the diffuse reflection component is further multiplied by the light source color, texture color, and diffuse reflection coefficient, and the specular reflection component is multiplied by the light source color, specular reflection color, and specular reflection coefficient. Since it is different from the new subject of the invention, it is omitted.

  In the present invention, the combined value Cs of the diffuse and specular reflection components is further linearly combined with the background color Cb using the transmittance obtained in 5a. This combination is given by I = Cs × αs + Cb × (1−αs) where αs is the transmittance and I is the combined transmittance value. Multiplication of Cs and αs is performed by a multiplier 5d, subtraction of (1-αs) is performed by a subtractor 8, and multiplication of Cb × (1-αs) is performed by a multiplier, and these are added by an adder 7b. A composite value is obtained in consideration of the transmittance with the background color.

  FIG. 2 shows the transmittance stored in the RAM table of the present invention. The horizontal axis represents the angle formed by N and V, and the vertical axis represents the transmittance scale value I. If the direction cosine cos θ obtained by the inner product circuit 1 in FIG. 1 exceeds 90 °, it is assumed that the signed value is 1. When the value is 90 ° or more in FIG. It will be. When the transmissivity of the translucent object is determined in a planar gap space such as a lace cloth, for example, a value obtained by the above equation (2) is calculated in advance and stored. . Since the mathematical model of the transmittance varies depending on the object, for example, curves a and b, the table is RAM, and the table contents are changed every time the object is different. In FIG. 2, the curve differs from 90 ° as a boundary. When the angle is 90 ° or more, the attenuated luminance of the surface opposite to the direction of the light source illuminated by the luminance of the transmitted light is attenuated. Is not symmetric. In FIG. 1, the input coefficients Kd0, Kd1, Ks0, and Ks1 of the multiplexers 6a and 6b are respectively switched between the surface in the light source direction and the opposite surface. Is set to a low value.

  The circuit of the present invention is mounted on a computer graphics LSI in order to obtain real time and realism at the same time.

1 shows a translucent object display circuit of the present invention. The characteristic of the transmittance | permeability value memorize | stored in the RAM table of this invention is shown.

        1 surface normal / line-of-sight vector inner product circuit 2 RAM table 3 surface normal / light source incident vector inner product circuit 4 specular reflection circuit 5a, 5b, 5c multiplier 6a, 6b multiplexer 7a, 7b adder 8 subtractor a, b transmittance curve

Claims (3)

An interpolation means for defining a transmittance, a line-of-sight vector, a surface normal, and a light source incident vector at each vertex of the polygon, and linearly interpolating the line-of-sight vector, the surface normal, and the light source incident vector in a polygonal plane,
From the interpolated value, an angle formed by the line-of-sight vector and the surface normal at an arbitrary point in the polygonal surface is obtained, and a transmittance calculating means for calculating the transmittance by changing the transmittance with a function using this angle as a parameter When,
An angle formed by the surface normal and the light source incident vector is obtained, and a reflection component is calculated using different reflection coefficients when the angle is not less than 0 degrees and not more than 90 degrees and when the angle is more than 90 degrees and not more than 180 degrees. Reflection component calculation means;
A translucent object display circuit that takes into account that the light passing through the light source side surface illuminates the back surface. A memory having a table for storing a value for scaling a transmittance defined in advance as a transmittance of a polygon surface, with an inner product value formed by the line-of-sight vector and the surface normal as an address,
The transmittance value stored in the memory table is asymmetric about 90 °,
The translucent object display circuit according to claim 1, wherein the transmittance changing unit reads the value from the memory and changes the transmittance.   The translucent object display circuit according to claim 1, wherein a reflection coefficient of greater than 90 degrees and less than or equal to 180 degrees is set to be smaller than a reflection coefficient of not less than 0 degrees and not more than 90 degrees.

Images