JPH08263691A - Bump map circuit - Google Patents

Abstract

PURPOSE: To display a real-time virtual-reality video by making the inclination in a horizontal and a vertical direction of a new surface after mapping the input variable of a shading circuit. CONSTITUTION: A principal normal Nv and input β from a bump map table are summed by an adder 7a, and a sum is added to a ROM 8 in which a value sin(Nv+m) is registered. On the other hand, after the β component of a bump normal passes through the ROM 9 in which the functions of sinβ and cos' are stored, sinβ generates |t| by a multiplier 13a, and 1/Root(1-t<2> ) is outputted by a ROM 10 as making this value the input variable. This data is summed with the normal Nv by an arcsin function ROM 12 and the adder 7b after passing through the multiplier 13b together with cosβ, and Nh is generated. On the other hand, |t|generates the normal Nv' after passing through an arcsin ROM 11. As the result, the normals Nh', Nv' of the new surface after the mapping are obtained. Then, these two components are given to the shading circuit, and reflected light can be obtained.

Description

Detailed Description of the Invention

The present invention relates to hardware implementation of bump map processing for expressing the unevenness feeling of an object surface by computer graphics. The unevenness information is defined by a normal vector independently of the surface attribute of the object, and this is defined as a two-dimensional array. Create as a pattern, and use this pattern in RAM (Random A
access memory), and after that, the surface roughness of the object is expressed by mapping to a surface having an arbitrary inclination. The hardware of the bump map shown in the present invention enables real-time virtual reality images to be displayed. This technology is applied to computer graphics devices (amusements, simulators, etc.).

[Prior Art] The bump map effect in computer graphics defines a bump (protrusion) state by a normal vector, maps it on the surface of an object, and then compares it with light from a light source incident on these surfaces. , Can be obtained by calculating the diffused or specularly reflected light from the reflection angle. The mapping of the color information of the texture pattern has already been implemented in general hardware.However, a concrete method that targets the discontinuous inclination of a minute surface such as a bump map and the reflection calculation of each point after setting the light source is used. There is no hardware currently available. In other words, the rendering process that takes into account reflection calculations such as phon shading and bump mapping is currently dependent on software techniques.
It is difficult to obtain a depiction of hundreds of thousands of polygons per second. The present invention has been made in view of the above circumstances, and is characterized in that bump mapping is implemented as hardware and bump map calculation for each pixel on a curved surface is executed within one operation clock. The purpose is to obtain a high-speed image description corresponding to virtual reality by combining them. Where 1
"Within a clock" means that the time required for processing from input to output is only the switching delay of the circuit element and does not use repetitive calculation with a clock.

[Means for Solving the Problems] This invention is directed to a normal vector of each point in a polygonal plane (hereinafter referred to as a main normal line).
On the other hand, it is a circuit for mapping a discontinuous normal vector (hereinafter referred to as a bump normal) that defines a bump surface that is two-dimensionally arranged independently of the inclination of the surface to the rendering surface. The main normal line at each point in the polygonal plane is the same as the coordinate value. When the polygon is interpolated (filled), nonlinear or linear DDA (Digit
al Different Analyzer)
You can ask by In this method, a normal vector is given as an attribute to a polygonal vertex or inside a vertex, and this is first interpolated into an outline and then inside. On the other hand, the bump normal is defined as the inclination of each point in a planar array in the bump coordinate form as in the pattern definition for texture mapping, and this is defined as a RAM (Random Acces).
s Memory). The difference between the texture pattern and the bump normal is that the former is color information,
The latter is to set a predetermined value based on the inclination information of the horizontal and vertical components with respect to the Z axis when the plane is the XY space, for example. The contents of the RAM storing this bump normal component are displayed in the screen (screen) in order to correspond to the mapping, because the mapped figure is subjected to coordinate conversion by scaling or perspective conversion, similar to the texture pattern. Inverse mapping transformation from the coordinate system to the bump coordinate system is required, and the read address obtained as a result causes the D
The registration data is read from the RAM in synchronization with DA.
The bump normal is the horizontal direction Bh with respect to the viewpoint coordinate axis (Z axis).
(XZ plane) and the vertical direction Bv (XY plane). On the other hand, the main normal is also defined by the horizontal Nh and vertical Nv direction components.
Japanese Patent Application No. 4-25531 for phone shading
As shown in FIG. 3, for the horizontal and vertical components, components of diffused and specularly reflected light were calculated using independent function memory elements, and then these were combined to calculate the reflected light at an arbitrary point. All of these conditions require that the pixel shading generation rate be executed within one clock. The bump map obtains horizontal and vertical inclinations Nh 'and Nv' of the new surface after the mapping based on the respective two sets of inclination components Bh and Bv and the main normals Nh and Nv, and calculates these. It is to be an input variable of the shading circuit. These Nh 'and Nv' are 3 by mapping the bump normal on the polygonal surface.
It is equivalent to being rotated by the amount of inclination of the main normal in the dimensional space. As a rotation calculation method at this time, the present invention is characterized in that each function term included in the rotation coordinate conversion formula is expressed by only one variable among the input variables. The bump normal is defined by two inclination components Bh and Bv with respect to the viewpoint coordinate axis (Z axis). This normal is mapped by being subjected to the inclination of the surface to be mapped, that is, the rotation of the main normals Nh and Nv. If the main normals after this new mapping are Nh 'and Nv', this relationship is expressed by equation (1). Nv '= arcsin (cosNvsinBv + sinNvcosBhcos Bv) Hh' = arcsin {sinBhcosBv / Root (1-sin 2 Nv ')} + Nh (1) where β = arccos (sin│Bh│cosBv) sig ( Bh) m = arcsin (sinBv / Sin│β│) (2) sig (Bh): 1 Bh> 0 0 Bh = 0 1 Bh <0 Nh ′ = arcsin {cos | β│ / Root (1-sin ² Nv ′)} ・ S (-1) ^u + (-1) ^s uπ + Nhu u: 0 | Nv + m│ <= π / 2 1 | Nv + m│> π s: 0 −β <= 0 1 −β> 0 where t = sin (Nv + m) When sin│β│ Nh '= Nh + arcsin { cos│β│ / Root (1-t 2)} · s (-1) u + (- 1) uπ Nv '= arcsin {sin ( Nv + m) sin│β│} become (3). Expression (2) is a function defined by normals of horizontal and vertical angles, and this value is stored in the bump map table. In equation (3), the function parameters of each term included in Nh ′ and Nv ′ are Nh, cos β, 1 / Root (1-t ² ), and sin | β | It is given as one or a sum of variables such as sin (Nv + m). This is a triangular function or 1 /
The Root (1-t ² ) value is stored in RAM or ROM (Re
It means that it is easy to make an ad only memory). As a result, this function calculation can be performed using a plurality of small capacity ROMs. A multiplier and an adder are used for the multiplication and addition of each term. At this time, the bump normal component is tabulated in advance, and it is preferable to use an expression that can be directly substituted into each term included in the rotation calculation formula as much as possible at the time of registering the pattern. Therefore, the two functions of interest (defined as β and m) having the bump normals Bh and Bv as variables are given as input variables. The circuit obtains new inclinations in the horizontal and vertical directions after the bump map by obtaining each term in the conversion formula mainly by the function table of the ROM and multiplying and adding these terms. This two component Nh '
And Nv 'are input variables of the shading circuit shown in Japanese Patent Application No. 4-255313.

[Embodiment] FIG. 1 shows a system of a bump map shading circuit according to the present invention. In FIG. 1, the interpolation processor DDA1 outputs the main normals Nh and Nv together with the coordinate values of each point inside the polygon to the bump map circuit 2 according to the present invention. In synchronization with this DDA1, the texture mapping circuit 3 generates an address of the memory in which the texture pattern is registered by the inverse mapping conversion from the coordinate value from the DDA and a predetermined attribute, and the address 4 is also applied to the bump map table 5 at the same time. Added. This is a process necessary for mapping from the bump normal definition coordinate system to the screen (display) coordinate system whose shape has been changed by coordinate conversion. The bump map table 5 stores two inclination components m and β composed of the bump normals shown in the equation (2). These two values are sequentially output to the circuit 2 by the address 4 in synchronization with the DDA operation. The bump map table is usually composed of RAM. In the bump map circuit, the normals Nh 'and N on the new plane of the bump normals that have been rotated by the main normals Nv and Nh are subjected to the same process as the calculation process shown in the equation (3).
get v '. This value is applied to the shading circuit 6. The shading circuit 6 calculates diffused and specularly reflected light on the basis of a light source incident angle set in advance using Nh 'and Nv' as input variables to obtain a reflected light Ip. FIG. 2 shows a concrete implementation circuit inside the bump map circuit 2 shown in FIG. 1 of the present invention. This circuit constitutes a bump map circuit derived from the equation (3). In FIG. 2, 8 to 12 are all composed of storage elements. A RAM can be used, but a ROM is usually used. The main normal line Nv and the input β from the bump map table are added by the adder 7a and added to the ROM 8 in which the sin (Nv + m) value is registered. On the other hand, the β component of the bump normal passes through the ROM 9 in which sin β and cos β functions are stored, and then sin β
Generates │t│ in multiplier 13a, and outputs the input variable and to ROM10 at ^{1 / Root (1-t 2} ) this. This data passes through the cos β and the multiplier 13b, and is then added to the main normal line Nh by the arcsin function ROM 12 and the adder 7b to generate Nh '. On the other hand, | t | generates the normal line Nv ′ via the arcsin ROM 11.
In FIG. 2, the broken line from the adder 7a to the ROMs 11 and 12 is s, the code u is (π / 2) bit, and the broken line from β is the code β. As a result, new normals Nh 'and Nv' of the surface after mapping are obtained. These two components are given to the shading circuit 6 shown in FIG. 1 and the reflected light Ip can be obtained. As is clear from the above specific examples, according to the present invention, the main normal of the polygon and the bump normal defined independently are defined by the horizontal and vertical angle components from the viewpoint axis, respectively, and from these, the function table and the multiplication are calculated. And the adder are used to determine the normal of the surface after mapping.
A feature of the present invention is that bump map processing is performed for each pixel 1
It is to execute within the clock. In the present invention, only the delay of the mapping circuit element itself from the bump normal line to the main normal line is required, and repeated calculation is not required. In FIG. 2, the route of maximum delay is the respective circuit 9, 13a, 1
This is the case where 0, 13b, 12, and 7b are processed in this order.
Since these circuits do not have feedback, the delay is the sum of the circuit delays in the previous route. There is a case in which a pipeline is provided via a register before and after each storage element in order to speed up the circuit, but the clock applied to the register is not a clock for arithmetic operation. As a result, a bump map circuit of about 10 ns / pixel becomes possible.

[Effect] By implementing the hardware of the bump map processing that expresses the smoothness and roughness of the surface of the object, more realistic computer graphics images can be realized several thousand times faster than the software. As apparent from the embodiments, this circuit is small in size in consideration of ASIC compatibility, and can be used as an arithmetic circuit of an inexpensive and high-performance graphic processor from amusement to simulation.

[Brief description of the figure]

FIG. 1 is a bump map circuit of the present invention.

[Explanation of symbols]

 1 DDA circuit 2 Bump map circuit 3 Texture mapping circuit 4 Bump map table address 5 Bump map table 6 Shading circuit

FIG. 2 is a bump map circuit of the present invention.

[Explanation of symbols]

7a, 7b adder 8 sin function ROM 9 sin / cos function ^{ROM 10 1 / Root (1-} t 2) 11 arcsin function ROM 12 arcsin function ROM 13a, 13b multipliers

Claims (1)

[Claims] The normal vector of each point on the curved surface (hereinafter referred to as the main normal) and the light incident angle from the light source to each point are defined by the horizontal and vertical angles with respect to the viewpoint coordinate axis, and these two tilt components are input variables. In the shading method for obtaining diffused light and specular reflected light on a surface, a bump surface normal vector (hereinafter referred to as a bump normal) is used as the main normal as bump map processing for giving unevenness expression on a curved surface. Similarly, these two tilt variables, which are defined by horizontal and vertical angles and patterned in a two-dimensional space, are registered in a storage element, and in this storage element, the inverse of the bump normal definition coordinate system is mapped to the display coordinate system. The first means for reading the registered value by giving the mapping conversion address and the means for rotating the bump normal read from the storage element at the inclination of the main normal during mapping are And flop normal component, storage elements of an input variable principal normal and respectively, provided with a computation circuit consisting of multipliers and adders, from the input variables in the calculation circuit,
A second means for obtaining two new horizontal and vertical angles after mapping on the curved surface for each point on the curved surface in synchronism with the in-plane interpolation calculator is combined, and the new horizontal and vertical angles are calculated. Bump map circuit characterized by being used as an input variable of a shading circuit