JPH0935085A - Method and device for real-time video generation - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain real-time videos of high precision without using a Z value for a shadow decision. SOLUTION: A polygon number is given to each of all polygons present in the entire area of video display. A light-source side screen is supposed and the polygon numbers of polygons which are seen from the light source side screen are stored in a polygon memory corresponding to respective pixels. Pixel positions (p<e> , q<e> ) on a viewpoint side screen are converted into pixel positions (p<1> , q<1> ) on the light-source side screen. The pixel positions on the light-source side screen which are obtained by the conversion are used as addresses to read polygon numbers Id<1> out of the polygon memory. The polygon number Id<1> that a polygon displayed at a pixel position on the viewpoint side screen has is compared with the polygon number Id<1> on the light source side to decide that the pixel part viewed from the viewpoint is not a shadow when they are equal or a shadow when equal, thereby generating shadow videos. An independent circuit for shading is not required and common use with a texture mapping circuit is enabled.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to real-time generation of a shadow image close to a real scene in real-time image generation (or visual system or real-time computer graphics).

[0002]

2. Description of the Related Art FIG. 3 shows a hardware structure of conventional real-time computer graphics with high reality. The configuration is divided into two parts, a geometric calculation unit (Geometric Processor (GP)) 31 for performing geometric calculation, and a rendering processor (Rendering) for performing color calculation of pixels.
Processor (RP) 32. The geometric calculation unit 31 performs calculation for each frame, and the rendering processor 32 performs calculation for each pixel.

The geometric calculation unit 31 converts a three-dimensional model into two-dimensional polygon information and renders the rendering processor 32.
Send to A rasterizer (Rasterlizer) 321 of the rendering processor 32 decomposes the polygon into segments in the scan line direction. Next, the segment is decomposed into a span (for example, a plurality of pixels such as 4 pixels) in a span calculation unit (Span Calc.) 322. After that, the plurality of pixels are subjected to parallel processing, and the color of each pixel is simultaneously calculated in the texture generation circuit (Texture Calc.) 323. Furthermore, in order to make the edges look smooth, anti-aliasing (Anti-aliasing) processing is performed by the pixel calculation unit (Pixel Calc.) 324, and the pixel data is written in the frame memory. Finally, the brightness modulator (Video
The contents of the frame memory 3251 incorporated in the Output) 325 are DA converted and displayed on the CRT 33 as an image. As shown in FIG. 3, the highly realistic image generation apparatus has the texture generation circuit 323 as a standard.

The principle of texture mapping calculation is shown in FIG. Texture mapping is used to determine the color of each pixel on the screen.

(1) The address of the texture pattern is calculated by the texture mapping calculation shown in FIG.

[Outside 1] Decide. That is,

[Outside 2] Is extended to obtain an intersection point O with the polygon, and a perpendicular line is drawn from O to the plane Q to obtain an intersection point M.

(2) The contents of a texture pattern (a pattern made by digitizing a photograph or the like) stored in advance in a texture memory (not shown) in the texture generation circuit 323 is subtracted using this address.

(3) This color is used as the color of the screen pixel.

A large part of the calculation is the texture mapping calculation. Memory address by this mapping

[Outside 3] Is calculated by the following equation (1).

[0009]

[Equation 1]

However,

T: A matrix of 2 rows and 3 columns showing the posture of attaching the texture to the polygon.

[0012]

[Outside 4] : A vector that indicates the translation from the origin when the texture is pasted on the polygon.

[0013]

[Outside 5] : A position vector of the target pixel (p, q, 1) of texture mapping.

[0014]

[Outside 6] :vector

[Outside 7] Z value of the point that intersects the polygon when is extended.

Although the conventional real-time image generator is constructed as described above, the generation of a shadow image has not been put to practical use.
Some of them, which are performed in an experimental system, cannot be applied to applications such as flight simulators in which the shadow image generation accuracy is poor and the gaming area is wide and image accuracy is required.

The main causes of these are

(1) The accuracy of Z used for shadow determination differs for each polygon.

(2) The scale of hardware for shadow calculation becomes large, and it is difficult to put it into practical use in consideration of economical efficiency.

According to

[Conventional Shadow Generation Algorithm Using Z Value]

FIG. 5 is a diagram showing the relationship between the light source and the viewpoint. The same (a) shows the z value stored in the z buffer (z-buffer) when viewed from the light source, that is, when the light source is the viewpoint, and the same (b) shows the z value stored when the normal viewpoint is used.
Indicates a value. At this time, the pixel position (p,
q) and the memory address of the z buffer have a one-to-one correspondence. FIG. 5C is a general view of the whole, and shows the relationship between the viewpoint assuming the simulated field of view object and the screen coordinate system viewed from the light source.

In the image generation using the z-buffer (or depth-buffer) method for the invisible processing for expressing a solid, the light source is regarded as a virtual viewpoint, and FIG.
When the point O shown in is viewed, if it is not hidden, the depth information up to that point (normally the z value because the z axis is in the depth direction) is stored in the z buffer.

Therefore, the z value z _lightC after the conversion of the coordinate value of the point O in the viewpoint side screen coordinate system to the light source side screen coordinate system and the z value already stored in the z buffer when the corresponding point is seen from the light source. The shadow can be determined by comparing the magnitude with the z value z _lightM . z _lightM
When <z light _C , it means that there is an object between the light source and the point O, and the point O becomes a shadow.

The above calculation can be described by a practical equation using the inverse z ⁻¹ of z and the error ε. The reason for using the reciprocal of z will be described later.

[0025]

[Equation 2]

However,

[0027] ^{shodow (p e, q e)} :( p e, the logical variable indicating that shadow occurs in q ^e).

R: A domain of pixels in the screen coordinate system on the viewpoint side.

P ^e , q ^e : pixel positions (p ^e , q ^e ) in the screen coordinate system on the viewpoint side.

[0030]

[Outside 8]

Ε: comparison error

Z _light : A memory for storing z values (inverse numbers) of all pixels in the screen coordinate system on the light source side, assuming that the _light source is the viewpoint.

F ¹ (p ^e , q ^e ): a mapping function from the screen coordinate system of the viewpoint to the screen coordinate system of the light source, which corresponds to the p ^l direction.

F ² (p ^e , q ^e ): A mapping function from the screen coordinate system of the viewpoint to the screen coordinate system of the light source, which corresponds to the q ^l direction.

F ³ (p ^e , q ^e ): A mapping function from the screen coordinate system of the viewpoint to the screen coordinate system of the light source, which corresponds to the z ^l direction.

(P ^l , q ^l ): The pixel position in the light source side screen coordinate system corresponding to the point (p ^e , q ^e ) in the viewpoint side screen coordinate system.

[Defective shadow image using conventional z-buffer]

There is a problem due to sampling. Discrete pixel positions in perspective side screen coordinate system (p ^e, q ^e) the corresponding pixel position of the light source side screen coordinate system (p ^l, q ^l) when converted into, the pixel as shown in FIG. 6 The centers do not match, that is, the converted pixel position (p ^l , q ^l ) is 4 pixels in the light source coordinate system.

(Equation 3) Located in the area surrounded by.

As it is,

(Equation 4) Since the z value stored in the memory address indicated by is used, an error Err _z −1 for this is generated. Therefore,
It is necessary to read the z value of 4 pixels and correct the z value by linear interpolation or the like.

However, in many cases, the four points are not z on the same polygon and this linear interpolation cannot be performed, and it is difficult to estimate the true z.

The evaluation of the error Err _z -1 will be described.

When generating a real-time image, a model using polygons is used to represent the shape in order to speed up the calculation and reduce the amount of data.

Therefore, each pixel (p,
The z value z ^pq of the point on the polygon corresponding to q) is obtained by linear interpolation based on the inclination of the polygon.

Concretely, the interpolation calculation of the following equation (3) is carried out using z ^{r -1} , ∂z ^-1 / ∂p, ∂z ^-1 / ∂q, which are information peculiar to the polygon. . FIG. 7 shows the relationship between the parameters used for the interpolation calculation and the screen.

[0045]

(Equation 5)

[0046] In the interpolation calculation, the reason for using the inverse z ^-1 and z is the inverse z ^-1 not z is because having linearity with respect to the screen coordinates p, q.

Here,

(1) (p, q) is a pixel position on the screen, q is a scan line position, and p is a pixel position within the scan line.

(2) z _pq ^-1 is the reciprocal of the z value of the point on the polygon displayed at the pixel position (p, q).

(3) ∂z ^-1 / ∂p and ∂z ^-1 / ∂q
Indicates the rate of change of z ⁻¹ when one pixel changes in the horizontal direction and one scan line changes in the vertical direction.

(4) z _r ^-1 is the reciprocal of the z value of the point on the corresponding polygon displayed at the point (p _r , q _r ) on the screen which is the basis of the interpolation calculation.

Therefore, the maximum value of the error Err _z −1 | Err
_max | is given by the following expression (4).

[0053]

(Equation 6)

When the angle between the projection direction of the light beam from the light source and the target surface is small (close to parallel) | ∂z ^-1 /
∂p | or | ∂z ^-1 / ∂q | becomes large, and the error | E
rr _max | cannot be ignored. This is the judgment of the shadow

(Equation 7) It means that the error ε of is different for each polygon.

For this reason, the method of generating a shadow using the z value is difficult to apply to a flight simulator or the like that needs to display the horizon.

FIG. 8 illustrates a defect example due to the error Err _max in the image generation using the z buffer. As shown in the figure, the original shadow 8 of the simulated building 8a on the simulated surface
In addition to b, a false shadow image 8c is displayed. Fake shadow 8c
It can be seen that is not local and covers a wide range.

[0057]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned inconvenience, and an object of the present invention is to determine a shadow without using the z value and obtain a shadow image with high accuracy. .

It is another object of the present invention to provide a method and apparatus capable of generating a real-time shadow image with a small hardware configuration.

Furthermore, a convenient method for generating an infrared image,
It is intended to provide a device.

[0060]

In order to solve the above problems, the present invention is configured as follows.

The real-time image generating method according to the present invention is for displaying an image of a simulated field of view object from a hypothesized viewpoint on the viewpoint side screen, and for each polygon existing in all areas of the image display. In advance, a polygon number is given to the light source side, and the light source side screen that virtually displays the field of view from the light source is virtualized, and the polygon number of the polygon forming the object virtually displayed on the light source side screen is It is stored in the polygon memory corresponding to each pixel on the light source side screen, and the pixel position (p
^e , q ^e ) is the pixel position (p ^l ,
converted to q ^l), the pixel position (p ^l of the light source side screen obtained by the conversion (coordinate), a polygon number is stored q ^l) in the polygon memory as an address Id ^l
Is read out, and the pixel position (p
Polygon number Id of the polygon displayed by ^e , q ^e )
and ^e, comparing the polygon number Id ^l, pixel portions having a polygon number Id ^e when they are equal to the non-shadow, a pixel portion having a polygon number Id ^e when they are not equal is shadow It produces a shadow image.

The real-time image generation apparatus according to the present invention converts the three-dimensional information of the simulated field of view object into the two-dimensional polygon information and performs the geometric calculation, and each of the polygons from the geometric calculation section. A rendering processor for calculating the color of pixels is provided, and in the image display of a simulated field of view object viewed from a hypothesized viewpoint on the screen on the viewpoint side, the geometric calculation unit assigns a polygon number to each polygon in advance, and The rendering processor is a texture generation circuit that associates a point on the viewpoint-side screen that displays an image of the field of view viewed from the viewpoint with a screen position on a virtual light-source-side screen that displays an image of the field of view that is virtually viewed from the light source; polygon number of polygons constituting the virtually objects displaying images on the light source side screen Id ^l And stored corresponding to each pixel of the light source side screen, and reading the polygon memory polygon number Id ^l screen position on the light source side screen that associates an address in the texture generating circuit, read from the polygon memory source Side polygon number Id ^l
If, compared with the polygon number Id ^e perspective side at a point on the viewpoint side screen corresponding to the address of the read out polygon memory, the polygon numbers when they are equal I
The pixel portion having d ^e is not a shadow, and when they are not equal, the pixel portion having a polygon number Id ^e is determined to be a shadow, and the polygon number determination unit determines that it is a shadow. Polygon number I on the viewpoint side
and a luminance modulation unit that casts a shadow on a pixel portion having d ^e .

Further, the texture generation circuit makes a texture mapping calculation for displaying an image of the field of view seen from the viewpoint and a point on the viewpoint side screen for displaying the image of the field of view seen from the viewpoint virtually seen from the light source. The mapping calculation corresponding to the screen position on the virtual light source side screen that displays the image of the field of view is configured to be performed, and the texture mapping calculation and the mapping calculation for the shadow calculation can be shared.

[0064]

The operation of the real-time image generation method is as follows. A polygon number is given in advance to every polygon existing in all areas of the image display. Virtualize the screen on the light source side that displays the image of the field of vision virtually seen from the light source,
A polygon number of a polygon forming an object virtually displayed on the light source side screen is stored in a polygon memory corresponding to each pixel of the light source side screen. Pixel position in perspective side screen (p ^e, q ^e) pixel positions of the light source side screen (p ^l, q ^l) into a. The polygon number Id ^l stored in the polygon memory is read using the pixel position (p ^l , q ^l ) of the light source side screen (coordinates) obtained by the conversion as an address. The pixel position in the view point side screen (p ^e, q ^e) pixel having a polygon number Id ^e with the polygons are displayed, the comparison between the read polygon number Id ^l, polygon numbers Id ^e when they are equal moiety is not a shadow, a pixel portion having a polygon number Id ^e when they are not equal is shadow. A shadow image is generated based on this determination.

The real-time image generator according to the present invention operates as follows. The geometric calculation unit converts the three-dimensional information of the simulated field of view object into two-dimensional polygon information and performs geometric calculation. At this time, each polygon is given a polygon number in advance. The rendering processor calculates the color of each pixel of the polygon from the geometric calculation unit.
Further, the texture generation circuit of the rendering processor performs a texture mapping calculation for displaying an image of a field of view virtually viewed from the light source. At that time, the polygon number,
Write to polygon memory which is a part of texture pattern memory. Then, texture mapping calculation is performed for displaying the visual field of view from the viewpoint. At this time, mapping calculation is performed so that a point on the viewpoint side screen that displays the image of the field of view viewed from the viewpoint corresponds to a screen position on the virtual light source side screen that displays the image of the field of view virtually viewed from the light source. In the polygon memory, the position on the light source side screen obtained by this mapping calculation is used as an address, and the polygon number Id ¹ corresponding to the position is read. Polygon number determination unit compares the polygon number Id ^l of the read light source side from the polygon memory and a polygon number Id ^e perspective side at a point on the viewpoint side screen corresponding to the address of the read out polygon memory. If they are not equal, the pixel portion having a polygon number Id ^e viewpoint side, because different from that illuminated the polygon i.e. light visible from the light source side, and a shadow. When they are equal, the pixel portion having a polygon number Id ^e viewpoint side are the same as the those illuminated polygon i.e. light visible from the light source is determined not to be shadow. The brightness modulation unit shadows the viewpoint side image obtained by the texture generation circuit on the pixel portion having the viewpoint side polygon number determined to be a shadow by the polygon number determination unit based on the determination by the polygon number determination unit. Attach.

[0066]

EXAMPLE The present invention uses a polygon number (a number uniquely assigned to all polygons forming a gaming area), which is a unique identifier for the gaming area, as a model instead of using z ^-1 in the past. Grant,
By using this number as a comparison target, the problem due to accuracy can be solved. The modified shadow calculation algorithm is shown below.

[0067]

(Equation 8)

However,

Id ^l : The polygon number for each pixel stored in the memory (polygon memory PolygonM provided for storing polygons) corresponding to the screen when an image is displayed with the light source as the viewpoint.

[0070] Id ^e: subject to pixel position pixel positions with shadow perspective side (p ^e, q ^e) polygon numbers with the.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a flowchart showing the flow of processing for explaining the method of generating a shadow image. The calculation of the shadow image is roughly divided into (a) generating a polygon number from the viewpoint of the light source and (b) calculating the image using the normal viewpoint. Further, (b) is subdivided into the following calculations.

(1) Mapping for mapping the point O ^e (p ^e _o , q ^e _o , 1) on the screen coordinate system at the viewpoint to the screen position O ^l (p ^l _o , q ^l _o , 1) of the corresponding light source. Calculation, (corresponding to (5-1) in FIG. 1 and (5-1) to (5-3) in the formula described in the above algorithm)

And (2) its screen position (p ^l _o ,
_{^{q l o, 1) p l}} o, Find polygon numbers Id ^l a q ^l _o as the memory address of the polygon number storage, ((2 in Figure 1) and, in the formula described in the above-described algorithm (5- 4))

(3) The polygon numbers Id ^l and O
determination of the shadow by comparison with the polygon number Id ^e in ^e, (the (3) and FIG. 1, in the formula described in the above-described algorithm (5-5), corresponding to (5-6))

Finally, (4) the pixel (p
^e , q ^e ) can be divided into luminance modulation of pixel values (corresponding to (4) in FIG. 1).

The major part of the processing is the point O ^e in the viewpoint.
^Is a mapping calculation that transforms to a point O ^l at the corresponding light source.

Mapping calculation (f
Equations (6) and (7) are obtained by obtaining equations suitable for hardware implementation of ( ¹ , f ² , f ³ ). Of equation (7)

[Outside 9] Component (p ^l , q ^l ) gives the memory address of the polygon memory corresponding to the pixel position (p ^e , q ^e ) on the viewpoint side screen.

[0078]

[Equation 9]

[0079]

(Equation 10)

However,

[0081]

[Outside 10]

T _eye : 3 of viewpoints in the world coordinate system
Posture matrix with 3 rows. The conversion direction is from the world coordinate system to the viewpoint coordinate system.

[0083]

[Outside 11]

T _light : 3 × 3 posture matrix of the point light source in the world coordinate system. The conversion direction is from the world coordinate system to the light source coordinate system.

[0085]

[Outside 12]

[0086]

[Outside 13]

[0087]

[Outside 14]

[0088]

[Outside 15]

[0089]

[Outside 16]

Here, the calculation suitable for hardware means that (1) the calculation amount is small, and (2) the pipeline is easy (that is, the calculation has regularity). There is. Also, in equations (6) and (7), per p
"ixel" indicates the calculation for each pixel, and the per frame
Indicates calculation for each frame.

The degree of calculation time per pixel for generating a shadow in real time is 1024 × 1024, and the overwrite ratio is 4 in a screen having the resolution.
1/30 × 1/4 × 1/102 when (translucent display is quadruple)
4 × 1/1024 ≈8 ns. This is a time shorter than the SRAM access time of 15 ns. Therefore, the processing must be parallelized or pipelined, and the circuit scale increases.

Here, by using the similarity between the shadow calculation and the texture mapping calculation, means for performing the shadow calculation without a large increase in hardware will be described.

First, the calculation amounts will be compared according to the circuit scale. A pipeline calculation circuit (formula (1)) for texture mapping is a scalar calculation if it is configured using a general-purpose LSI.

[0094]

[Equation 11] Equation (8) AX + BY + CZ + D

Substrate of VME12U size (approximately 40 cm x 30
cm) occupies about 0.5 sheets. The texture mapping calculation requires two scalar calculations, and in order to perform parallel calculation, the above-mentioned substrate size requires about one sheet, which is a calculation amount. Also, the shadow calculation requires three scalar calculations, and about 1.5 sheets are required under the same conditions. Independent calculation is not preferable because the circuit scale increases. That is why real-time generation of shadow images is not common.

If the following correspondence is taken, there is similarity between the calculation of texture mapping (formula (1)) and the calculation used in shadow calculation (formula (9)).

[0097]

(Equation 12)

[0098]

[Outside 17] When

(Equation 13) , T and

[Equation 14] Are regarded as two parameters in correspondence with each other. This parameter is calculated for each frame in the geometric calculation unit (GP),
It is given to the rendering processor (RP) as a parameter unique to the processing polygon. As a result, the texture generation circuit can be shared by the texture mapping calculation and the mapping for shadow calculation.

FIG. 2 is a functional block diagram of a texture generation circuit used in the present invention. This texture generation circuit is the texture generation circuit 32 shown in FIG.
3 and the circuit of the former stage and the latter stage connected thereto is the same as that of FIG. The texture generation circuit
It constitutes a part of the rendering processor 32 described in FIG. 3, and the rendering processor 32 is connected to the geometric calculation unit 32 and the CRT 33. In FIG. 2, 21
Is an input register, 22 is a register file, 23
Is an address calculator, 231 is an x component calculator, 232 is y
A component calculation unit, 233 is a z component calculation unit, 234 is an address output unit, 24 is a texture memory with a polygon memory,
Reference numeral 241 is a polygon memory, 25 is a comparator functioning as a polygon number determination unit, and 26 is a delay device.

The texture generation circuit performs the texture mapping calculation in the same manner as the conventional texture mapping calculation in order to obtain the image with the light source as the viewpoint. That is, from a geometric calculation unit (not shown) (the same as the geometric calculation unit 31 of FIG. 3 described as a conventional technique is used) as parameters t ₁₁ , t ₁₂ , ... T ₃₂ , t ₃₃ , t _x , t. _y , t
_z , pixel information X, Y, Z ⁻¹ are stored in the input register 21 and the register file 22. Tag information (Tag)
Is the texture mapping calculation, this parameter is used for the texture calculation as the posture and the pixel information X, Y, Z ^{−1 in} which the three-dimensional information of the simulated visual field object is converted into the two-dimensional polygon information and the texture is attached to the polygon. The position vector of the target pixel (p ¹ , q ^l , 1) is shown. The address calculator 23 calculates the address of the texture memory 24 using these parameters. This address calculation is performed by applying the above equation (1) from the viewpoint of the light source. At this time, the parameters t ₁₁ , t ₁₂ , ... T ₃₂ , t ₃₃ correspond to T in the equation (1), and the parameters t _x , t _y , t _z are

[Outside 18] Corresponding to. Using these parameters, the x component calculation unit 233

[0101]

(Equation 15) The x component of the texture memory 24 is calculated by The y component calculation unit 234

[0102]

(Equation 16) The y component is calculated by Also, the z component calculation unit 235
Is

[Equation 17]

Calculate The address output unit 234 uses the calculation result of the previous stage as

(Equation 18)

By substituting into m as the x and y components of the addresses of the texture memory 24 and the polygon memory 241.
_{Get x} and m _y . For the address of this polygon memory 241, the target pixel (p ^l ,
Write the polygon number Id ^l for q ^l , 1).

Next, the texture generation circuit performs the texture mapping calculation in the same manner as the conventional texture mapping calculation in order to obtain the image using the regular viewpoint. That is, a geometric calculation unit (not shown) sets t ₁₁ ,
t ₁₂ , ... t ₃₂ , t ₃₃ , t _x , t _y , t _z , pixel information X,
The Y and Z ⁻¹ are stored in the input register 21 and the register file 22. Since the tag information (Tag) is the texture mapping calculation, this parameter is the posture and the pixel information X where the three-dimensional information of the simulated field of view object is converted into the two-dimensional polygon information and the texture is attached to the polygon.
Target pixels (p ^e , q ^e ,) of texture calculation as Y and Z ⁻¹
The position vector of 1) is shown. Furthermore, the position of the target pixel (p ^e, q ^e) the polygon number Id identifies the polygon
^e is input from a geometric calculation unit (not shown) and stored in the register file 22. The address calculator 23 calculates the address of the texture memory 24 using these parameters. This address calculation is performed by applying the above-mentioned equation (1) with the viewpoint set to the normal viewpoint position. At this time,
The parameters t ₁₁ , t ₁₂ , ... t ₃₂ , t ₃₃ correspond to T in the equation (1), and the parameters t _x , t _y , t _z are

[Outside 19] Corresponding to. Using these parameters, the x-component calculator 233 calculates the x-component of the texture memory 24 by the equation (10). The y component calculation unit 234 uses the equation (11).
The y component is calculated by Also, the z component calculation unit 235
Calculates the z component according to equation (12). The address output unit 234 substitutes the calculation result of the previous stage into Expression (13) to obtain m _x and m _y as the x and y components of the address of the texture memory 24. This address reads the contents of the texture pattern, the target pixel this color (p ^e, q ^e)
The pixel position of the color (p ^e, q ^e) 3 for explaining the luminance modulation portion (not shown) as an address (as the prior art
The same one as the luminance modulator 325 of is used. ) Write to the frame memory.

Further, subsequently, the tag information (Tag) becomes shadow calculation, and in response to this, the texture generating circuit receives the pixel position (p ^e , q ^e ) viewed from the normal viewpoint as the pixel position (p ^e viewed from the light source). ^l , q ^l ). The algorithm for this is given by the above-mentioned equations (5-1), (5-2), (5-3). Then, as described above, the texture generation circuit can be shared by the texture mapping calculation and the shadow calculation. That is, a geometric calculation unit (not shown) uses parameters t ₁₁ , t ₁₂ , ... T ₃₂ , t ₃₃ , t as parameters.
_{The x} , t _y , t _z and pixel information X, Y, Z ⁻¹ are stored in the input register 21 and the register file 22. Since the tag information (Tag) is a shadow calculation, this parameter is for mapping from the screen coordinates of the viewpoint to the screen coordinates of the light source, and the pixel information X, Y, Z.
Target pixels as ^{-1 (p e, q e,} 1) indicating the position vector. Further, input from the position of the target pixel (p ^e, q ^e) the geometric calculation section (not shown) polygon number Id ^e identifying the polygons, are stored in the register file 22. The address calculator 23 calculates the position (p ^l , q ^l ) of the screen coordinate of the corresponding light source using these parameters. This position (p ^l , q ^l ) becomes the address of the polygon memory 241. This address calculation is performed by applying the above-mentioned equation (1) with the viewpoint set to the normal viewpoint position. At this time, the parameters t ₁₁ , t ₁₂ , ... T ₃₂ , t ₃₃ are expressed by the formula (1).
Corresponding to T, and the parameters t _x , t _y , and t _z are

[Outside 20] Corresponding to. Using these parameters, the x component calculation unit 233 calculates the x component according to the equation (10). The y-component calculator 234 calculates the y-component according to equation (11). In addition, the z component calculation unit 235 calculates z by the equation (12).
Calculate the ingredients. The address output unit 234 substitutes the calculation result of the previous stage into the equation (13) to generate the polygon memory 241.
M _x and m _y are obtained as the x and y components of the address. It reads the polygon number Id ^l of screen coordinates of light sources corresponding to the screen coordinates of the viewpoint from this address.

The comparator 25 outputs the read polygon number Id.
and ^l, compared with the polygon number Id ^e of the pixel position obtained from the register file 22 (p, q). As a result of the comparison, when both are not equal, the pixel portion having a polygon number Id ^e viewpoint side, because different from that illuminated the polygon i.e. light visible from the light source side, and a shadow. When they are equal, the pixel portion having a polygon number Id ^e viewpoint side are the same as the those illuminated polygon i.e. light visible from the light source is determined not to be shadow.

The brightness modulation section (not shown) reads the viewpoint side image obtained and written by the texture generation circuit, and shadows the polygon number judging section based on the judgment of the comparator 25 as the polygon number judging section. Signal processing is performed so that a pixel portion having the polygon number on the viewpoint side determined to be shaded is shaded, and further DA converted and output. A CRT (not shown) displays this signal as an image. FIG. 9 shows an example of an image displayed by implementing the present invention.

In the above description, it was premised that a shade of visible light was added to add a gray shade to the shade. However, since the shade can be accurately simulated as described above, it is not limited to visible rays. For example, an infrared image can be processed, and an infrared image can be generated.

[0110]

As described above, according to the present invention, since the shadow is determined without using the Z value, there is no problem that the shadow is caused by an error due to the Z value. Since it is determined whether or not it is a shadow only by comparing polygon numbers, it is possible to obtain a highly accurate shadow image.

Since the real-time shadow calculation can be shared with the texture mapping calculation, the real-time shadow image can be generated with a small hardware configuration.

Furthermore, since the shadow can be correctly simulated,
Infrared images can be generated.

[Brief description of drawings]

FIG. 1 is a flow chart showing a flow of a method according to an embodiment of the present invention.

FIG. 2 is an example of a functional block diagram of a texture generation circuit used in the present invention.

FIG. 3 shows a hardware configuration of conventional real-time computer graphics.

FIG. 4 is a diagram showing the principle of texture mapping calculation.

FIG. 5 is a diagram showing a relationship between a light source and a viewpoint.

FIG. 6 is a diagram for explaining a deviation of a light source side pixel center corresponding to a viewpoint side pixel center.

FIG. 7 is a diagram showing a relationship between parameters used for interpolation calculation and a screen.

FIG. 8 is a diagram showing an example of a defect of an image by a conventional method.

FIG. 9 is a diagram showing an example of an image displayed by implementing the present invention.

[Explanation of Codes] 21 ... Input Register, 22 ... Register File,
23 ... Address calculator, 231, ... x component calculator, 232
... y component calculation unit, 233 ... z component calculation unit, 234 ... address output unit, 24 ... texture memory with polygon memory, 241 ... polygon memory, 25 ... comparator (polygon number determination unit), 26 ... delay device, 31 ... Geometric calculator, 32
... Rendering processor, 321 ... Rasterizer, 3
22 ... Span calculator, 323 ... Texture generation circuit, 3
24 ... Pixel calculator, 325 ... Luminance modulator, 3251
… Frame memory, 33… CRT.

Claims (3)

[Claims] 1. A real-time image generation method for displaying an image of a simulated field of view from a hypothesized viewpoint on a screen on the viewpoint side, in which polygon numbers are assigned in advance to all polygons existing in all areas of the image display. The light source side screen that virtually displays the visual field from the light source is virtualized, and the polygon number of the polygon that constitutes the object virtually displayed on the light source side screen is assigned to each pixel of the light source side screen. stored in the polygon memory corresponding pixel position in the view point side screen (p ^e, q ^e) is converted the pixel position of the light source side screen (p ^l, q ^l), the light source side screen obtained by the conversion The polygon number Id ^l stored in the polygon memory is read by using the pixel position (p ^l , q ^l ) of (coordinates) as an address, and the polygon number Id ^l is read on the viewpoint side screen. A polygon number Id ^e to the pixel position (p ^e, q ^e) has a polygon to be displayed that, comparing the polygon number Id ^l, pixel portions having a polygon number Id ^e when they are equal to the non-shadow, the method of real-time image generation pixel portion having a polygon number Id ^e when they are not equal, for generating a shadow image as a shadow. 2. A geometric calculation section for converting three-dimensional information of a simulated field of view object into two-dimensional polygon information and geometric calculation, and a rendering processor for performing color calculation of each pixel of the polygon from the geometric calculation section. In the real-time image generation device for displaying a simulated field of view object viewed from a hypothesized viewpoint on the viewpoint side screen, the geometric calculation unit assigns a polygon number to each polygon in advance, and the rendering processor includes: A texture generation circuit that associates a point on the viewpoint-side screen that displays the image of the field of view seen from the viewpoint with a screen position on the virtual light-source-side screen that displays the image of the field of view that is virtually viewed from the light source; The polygon number Id ^l of the polygon forming the object virtually displayed on the screen is displayed on the light source side screen. A polygon memory for storing the polygon number Id ^l corresponding to each pixel of the screen and reading the polygon number Id ¹ with the screen position on the light source side screen corresponding to the texture generation circuit as an address; pixel having a polygon number Id ^l, wherein comparing the polygon number Id ^e perspective side at a point on the viewpoint side screen corresponding to the read out polygon memory addresses, polygon numbers Id ^e when they are equal the perspective side The part is not a shadow,
And determining the polygon number checker and pixel portion is a shadow having a polygon number Id ^e perspective side when they are not equal, the polygon number of the polygon number checker is determined to be shadow viewpoint side Id ^e A real-time image generation device, comprising: 3. A geometric calculation unit for converting three-dimensional information of a simulated field of view object into two-dimensional polygon information and performing geometric calculation, and a rendering processor for performing color calculation of each pixel of the polygon from the geometric calculation unit. In the real-time image generator that displays the simulated field of view object from the hypothesized viewpoint on the viewpoint side screen, the geometric calculation unit assigns a polygon number to each polygon in advance, and the rendering processor A texture mapping calculation for displaying the image of the field of view viewed from the point, and a point on the viewpoint side screen that displays the image of the field of view viewed from the viewpoint, on the virtual light source side screen that displays the image of the field of view virtually viewed from the light source. Texture generation circuit for performing mapping calculation corresponding to the screen position of the Polygon number Id ^l of polygons constituting the objects are virtually displayed picture is stored corresponding to each pixel of the light source side screen, the screen position on the light source side screen to correspond with the texture generating circuit and reading polygon memory polygon number Id ^l as an address, a polygon number Id ^l of the read light source side from the polygon memory, a polygon number perspective side at a point on the viewpoint side screen corresponding to the address of the read-out polygon memory Id ^e is compared, and when they are equal to each other, the pixel portion having the polygon number Id ^e on the viewpoint side is not a shadow,
And determining the polygon number checker and pixel portion is a shadow having a polygon number Id ^e perspective side when they are not equal, the polygon number of the polygon number checker is determined to be shadow viewpoint side Id ^e A real-time image generation device, comprising: