JP3214389B2 - Data converter - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an image information processing system, and more particularly, to a data conversion device for realizing texture mapping.

[0002]

2. Description of the Related Art Texture mapping is a typical technique for expressing the texture of three-dimensional computer graphics.
In this texture mapping technique, a texture pattern of a wood grain pattern or a marble pattern defined separately is pasted (mapping) on the surface of a three-dimensional shape so that the texture of an object is obtained. There are several methods for mapping a two-dimensional image to a three-dimensional shape. In the following, data conversion in the case of performing “UV mapping” will be described as a typical example. For example, reference is made to JP-A-5-324842.

[0003] First, a shape surface to be mapped is represented by a two-dimensional UV coordinate system. At this time, the texture pattern and the two-dimensional coordinate system of the display are set as an ST coordinate system and an XY coordinate system, respectively. Also, the ideal coordinates of the texture pattern are defined as an st coordinate system.

[0004] The color and luminance of the point (U, V) on the surface of the three-dimensional shape that can be seen at the point (X, Y) in the display coordinate system are the color and luminance of the corresponding coordinate system (S, T) of the texture pattern. Decide from. At the time of mapping, this (X, Y) →
The coordinate calculation of (U, V) → (s, t) may be performed.
However, at this time, if (s, t) of the calculation result does not match the texture pattern coordinate system (S, T) (for example, if s, t takes a decimal value), the color or luminance is calculated. Interpolation calculation is required.

Conventionally, bilinear interpolation has been required as a type of interpolation of data displayed on a display plane. In the bilinear interpolation, weighting is performed based on a distance from four points around a point to be displayed, and an average is obtained. For example, as shown in FIG.
p, q, (0 ≦ p ≦ 1, 0 ≦ q ≦ 1), and D ₀ (4
Assuming that 3), D ₁ (44), D ₂ (45), and D ₃ (46), weighting is performed by the following equation (1) to obtain D
The data of (47) is obtained. In FIG. 4, 4
Reference numerals 1 and 42 denote S coordinates and T coordinates in the texture pattern coordinate system.

D = pqD ₃ + p (1-q) D ₁ + (1-
p) qD ₂ + (1-p) (1-q) D ₀ (1)

If this is to be realized on a circuit, a multiplier is required, a long time is required for the calculation, and the number of accesses to the memory storing the texture pattern is also required four times (that is, D ₀ to D ₀ ). access to read the D _3).

For this reason, compared to the case of simple surface coating,
There was a problem that the processing was remarkably slow.

As a second prior art, as shown in FIG. 5, a point sampling in which data of a point D ₀ (53) closest to a point D (57) to be displayed is selected. Can also be taken. However, in the case of this point sampling, the value of (s, t) is only replaced by the nearest (S, T), ie, the decimal value of (s, t) is rounded up or down, It has not been possible to achieve higher quality texture mapping.

[0010]

As described above, in the conventional method, the point sampling (p
In the case of using point sampling, there is a problem that the quality of an image is deteriorated instead of solving the problem of the processing speed.

On the other hand, the above-mentioned bilinear
ar) In the case of using interpolation, a high-quality image can be obtained, but the processing speed is slow. In order to solve the problem of speed in bilinear interpolation, However, there is a problem that the chip area of the semiconductor chip on which the interpolation operation circuit is mounted becomes large.

Accordingly, the present invention has been made in view of the above problems, and an object of the present invention is to provide a texture data reading circuit used for texture mapping, which performs processing for color and luminance interpolation and memory access. It is an object of the present invention to provide a data conversion device which achieves high speed by using a simple interpolation and a special logic for an address generation unit, and also reduces the circuit scale.

[0013]

In order to achieve the above object, the present invention provides a texture data reading circuit used for texture mapping, wherein a texture to be read from ideal coordinates on a texture pattern of image data to be used for a display image is provided. first means for determining the actual coordinates of the pattern, and a second means for selecting a simple interpolation calculation from the ideal coordinates, the read color values (color, brightness) designated by the second means, adding And a third means for performing a simple interpolation operation comprising a bit shift operation .

[0014]

Embodiments of the present invention will be described below. In a preferred embodiment, referring to FIG. 1, the data conversion device of the present invention issues a read instruction to the texture pattern memory 12 and a calculation instruction when performing simple interpolation of the read data from the texture pattern memory 12. It comprises address generation and simple interpolation operation control means 11, texture pattern memory 12 for storing texel data, and simple interpolation operation means 13 for converting texels read from texture pattern memory 12 into display data. ing.

The present invention, in a preferred embodiment thereof, subdivides each coordinate area of texture mapping,
It is represented by the position coordinates of the subdivided area including the ideal coordinates, and therefore, can be expressed more accurately than the original coordinate area, and a multiplier for performing an operation for ensuring accuracy This also eliminates the need for this, thereby achieving high-speed arithmetic processing and a reduction in circuit scale.

[0016]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of the present invention;

FIG. 2 is a diagram showing the configuration of one embodiment of the present invention. Referring to FIG. 2, reference numeral 21 denotes a drawing control unit,
It includes an address generation and simple interpolation calculation control unit, and issues a read instruction to the texture pattern memory and a calculation instruction for simple interpolation of read data. Reference numeral 22 denotes a texture pattern memory, which is a memory unit for storing texel data. A data operation unit 23 performs a simple interpolation operation for converting a texel read from the texture pattern memory into data to be written to an FRB (frame buffer). Reference numeral 24 denotes a frame buffer unit for storing display data, and reference numeral 25 denotes a display.

The read instruction and the operation instruction are shown in Table 1 below.
Is performed according to the logic shown in FIG.

In Table 1, s and t (s and t are real numbers)
Are ideal coordinates for reading a texture pattern. S ₋₁ , s ₋₂ , t ₋₁ , t ₋₂ are s,
The first and second decimal bits of t are the first and second bits. S, T (S, T
Is an integer) is the actual coordinates for reading the texture pattern. A in Table _{1 0 ~A 8, D 0 ~D} 3 is, A ₀ in FIG. 3
ＡA ₈ , D ₀ 0D ₃ . Further, n and m (n and m are integers) are integer parts of s and t.

More specifically, in Table 1, s ₋₁ , s ₋₂
Represents the value of the X coordinate in FIG. When the coordinate area is subdivided into three parts, for example, as shown in FIG. 3 as viewed from the X coordinate, these two bits are required to represent each X coordinate.

For s _-1 and s _-2 , points A0 and A
3, A6 are all 0, 0, and represent the left end of the coordinate area.

Further, for s _-1 and s _-2 , point A
2, A5 and A8 are 1, 1 and represent the right end.

Since the points A1, A4, and A7 are in the middle of these, if expressed by 2 bits, either s _-1 or s _-2 is 1, and in Table 1, this is represented by s _-1 xor s
_-2 = 1 (xor is exclusive OR operation).

Similarly, t _-1 and t _-2 represent the values of the Y coordinate.

N and m represent the position of the coordinate area with respect to the entire coordinates of the texture mapping. Color value D
Can be calculated on the texture mapping by the calculation formula shown in the column of “calculation” in Table 1. That is, as shown in the "operation" column of Table 1, the operation includes addition operation and division by 2,4.

That is, a simple interpolation operation (“operation” in Table 1)
) Can be realized only by the functions of the addition operation and the bit shift (that is, 1/2 and 1 / of Table 1).
4 are realized by right 1-bit and right 2-bit shift operations, respectively), enabling high-speed operation processing and reducing the circuit scale.

As can be seen from Table 1, the probability of the number of accesses to the texture pattern memory is reduced.

More specifically, each of A0, A2, A6, and A8 is one memory access (that is, one read of D0, D1, D2, and D3, respectively).

Each of A1, A3, A5, and A7 requires two memory accesses.

A4 is four times of memory access.

From the above, the bit rate of memory access is:
The average number of memory accesses per pixel (pixel) is 2.25 times / pixel, which reduces the memory access rate.

[0032]

[Table 1]

[0033]

As described above, according to the present invention,
By using the first and second decimal bits of (s, t) obtained as a result of the address operation as a criterion, a simple interpolation operation of color and luminance data is simplified, and a high-speed operation is achieved and a circuit scale is reduced. This has the effect of realizing the conversion. Further, according to the present invention, there is an advantage that high speed can be expected by reducing the number of times of accessing the texture pattern memory.

[Brief description of the drawings]

FIG. 1 is a block diagram for explaining an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of an embodiment of the present invention.

FIG. 3 is a diagram illustrating a coordinate system according to an embodiment of the present invention.

FIG. 4 is a diagram for explaining conventional bilinear interpolation.

FIG. 5 shows conventional point sampling (point s);
FIG. 3 is a diagram for explaining (ampling).

[Explanation of symbols]

11, 22 Address generation and calculation control unit 12, 24 Texture pattern memory 13, 23 Data calculation unit 21 Drawing control unit 26 Display 31, 41, 51 S coordinate 32, 42, 52 T coordinate 33, 43, 53 D ₀ (texel 34, 44, 54 D ₁ (texel) 35, 45, 55 D ₂ (texel) 36, 46, 56 D ₃ (texel) 47, 57 Ideal data

Claims (3)

(57) [Claims] (1) Each coordinate area of texture mapping is represented by N
Texture data subdivided by the actual coordinates (ST coordinates) of a bit-length point is stored in a texture pattern memory, and data of a point of N + 2 bits or more on a texture pattern of image data to be used for a display image is stored.
The ideal coordinates (st coordinates) of the
Approximated by approximating data points, read one or more data from the texture pattern memory to determine the actual coordinates for reading the texture pattern, before
When the approximate data is not at the actual coordinates, an interpolation operation for the read one or more data is determined, and the determined addition and bit shift operation are performed on the read plurality of data. , Approximation by interpolation
A data conversion device for obtaining data and outputting the approximate data as display data. 2. A texture data reading circuit used for texture mapping, wherein N + 2 bits on a texture pattern of image data to be used for a display image
The ideal coordinate of the data of the above points is
First means for approximating the actual coordinates of an N-bit long point of the texture pattern to be read out by approximating the data with the approximate data of the bit length ,
If not, a second means for selecting a simple interpolation operation based on the ideal coordinates, and a simple operation comprising adding and bit-shifting the read color value, that is, color and luminance, designated by the second means. Third means for performing an interpolation operation;
A data conversion device comprising: 3. A texture data read circuit used for texture mapping stores conversion values for N-bit point data in a table,
The data at the point of N + 2 bits or more is converted to the nearest N + 1
A data conversion device, wherein data is approximated by bit-length point data and output at points not in the table is obtained by an interpolation operation consisting of addition and bit shift operation.