JP2001101445A - Soft shadow casting circuit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate a soft shadow effect for blurring a shadow and the periphery of the edge of an irradiated part corresponding to a natural phenomenon in real time. SOLUTION: On a first path, a coordinate value is interpolated by a circuit 2c and after a distance from a light source and a shadow area flag are stored in a buffer 10, the area flag is filtered while using a spline filter 14. On a second path, a distance provided by simultaneously interpolating a viewpoint and the coordinate value of a light source system is compared by a shadow polygon cache 11, distances with a light shielding object, a display object and the light source and a coefficient for determining the area of Penumbra and Umbra parts of the shadow and the luminance in a luminance determining circuit 12 are found. The light reflection luminance of the object itself, after the luminance is found by a multiplier 8 and a coefficient is found by the shadow luminance determining circuit 12, are multiplied by a multiplier 13 in an inverse mapping converting circuit 5 for the interpolation coordinate value, bump normal memory 4, texture pattern memory 6, shading circuit 3, texture color and shading luminance, and the result is stored through a pixel cache 9 into the buffer.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001] The present invention relates to a display circuit for rendering a shadow projected by a three-dimensional object by a light source in computer graphics. The shadow casting of the present invention is indispensable for expressing reality in a virtual reality system, especially in a walk-through system, by realizing a real-time image having a smoothing effect that eliminates hard edges generated in shadow portions and illuminated portions. Technology.

2. Description of the Related Art Scanline, shadow clipping, shadow volume, and the like have been proposed as shadow casting algorithms in computer graphics technology. Has been put to practical use. At the time of shadow processing, coordinate values and various attributes are defined at the vertices of a polygon constituting an object. The renderer first linearly interpolates the outline of the polygon using the vertex coordinate values of the polygon as interpolation end points, and then interpolates between the two left and right points where the interpolated polygon outline line intersects the horizontal axis. In the first process (pass), the vertex coordinate values are defined in a coordinate system with the light source as the projection center. The resulting interpolation point (L ₁
For (x, L ₁ y, L ₁ z), the distance L ₁ z from the light source is stored in the shadow polygon buffer as (L ₁ x, L ₁ y) as the writing position. For writing in this buffer, a hidden surface elimination method in which the distance closest to the light source is stored is used. When all the objects have been rendered in the light source coordinate system and stored in the shadow polygon buffer, the interpolation process is started by defining the polygon vertex coordinate values in the viewpoint coordinate system as a second process (pass). In the rendering of this second pass, texture mapping and shading processing are usually performed simultaneously and in parallel. At the time of this polygon interpolation, the distance from the light source stored in the shadow polygon buffer is compared with the distance in the light source coordinate system corresponding to the point currently being interpolated for each interpolation point. For this purpose, coordinate values (L ₂ x, L ₂ y, L ₂ z) of the light source coordinate system with respect to one interpolation point (Ex, Ey, Ez) of the viewpoint coordinate system are obtained, and these (L ₂ x, L ₂ y) are obtained. ) Is given as the address of the shadow polygon buffer obtained in the first pass, and L ₁ z is read. If the L ₁ z value of the first pass is closer to the light source than the transformed L ₂ z value of the interpolation point, it is determined to be a shadow. If the result of the comparison is a shadow, the brightness of the display pixel at the interpolation point is reduced. The following are questions concerning the hardware implementation of this conventional method. (1) In the second pass, conversion into light source system coordinate values must be obtained for each viewpoint system coordinate value to be interpolated. (2) In the method using the shadow polygon buffer, the resolution of the image memory causes a geometrical error in the determination of the shadow. (3) It is difficult to obtain smoothing and soft shadow effects at shadow edges. (1)
The problem with the term is that a 3x3 or 4x4 rotation matrix is generally required to convert from the viewpoint coordinate system to the light source coordinate system in the second pass, and it is not possible to process this in 10 ns for each interpolation point in order to display this in real time. It is difficult without sacrificing significant hardware costs. The problem of the item (2) is that when the polygonal surface is inclined substantially parallel to the z-axis in the first pass, the data stored in the shadow buffer is reduced to the distance closest to the light source by performing hidden surface elimination, that is, Only polygon edges are stored. As a result, when the same polygon surface is orthogonal to the viewpoint axis in the viewpoint coordinate system, most of the light source system coordinate values obtained in the second pass are behind the data in the shadow buffer, and these points are Considered a shadow. This is an erroneous determination. The problem of item (3) is that sampling is performed by comparing the shadow and the irradiation by each pixel, so that the brightness at the boundary between the shadow and the irradiation changes suddenly, resulting in jaggy and an unnatural image. . This can generally be addressed by filtering, but real-time processing requires an efficient and reproducible filter circuit. In the real-time polygon rendering method based on pipeline processing, pixels are generated by referring only to given data, so that it is difficult to perform soft shadows (environmental conditions) that are affected by surrounding objects (environmental conditions). Conventionally, there is no published hardware technology for solving these problems, and a memory called a stencil buffer or accumulation buffer is used as a shadow buffer for shadow projection. Is actually dependent on software, especially (3)
Smooth soft shadows can only be achieved by software.

[Means for Solving the Problem] That of the previous section
The present invention solves each problem by the following method.
You. In the present invention, the problem of the item (1) is in the second pass.
Not only the coordinates of the viewpoint system but also
Define the light source system coordinate values and perform interpolation with the viewpoint system coordinate values.
Sometimes light source system coordinate values are also interpolated. But the coordinate system is different
The interpolation from the main axis of each coordinate system (interpolation distance is the most
Long coordinate axes) independently or of all defined values
Light source system corresponding to the viewpoint coordinate system interpolation point when performed on the longest axis of
The benchmark cannot be obtained synchronously. Therefore, in the present invention
Is the coordinate value of the viewpoint coordinate system (A
X- or y-axis, horizontal-axis parallel interpolation
Is the main axis of the interpolation circuit. As a result,
Light source system coordinate values corresponding to the interpolated viewpoint coordinate system interpolation points of
Can be output in accordance with the speed of the interpolation circuit.
For this purpose, three sets of light source system coordinate value interpolation circuits (each axis) are required.
It becomes important. Whether the interpolated point is a shadow or an irradiation point
The interpolated point L₂read from z and shadow buffer
L issued₁The shadow is determined by comparing z. In this method
The problem of item (2) in the preceding section [0002] occurs. In the present invention
Solves this problem by providing an identifier for each polygon.
You. This identifier is uniquely assigned to each object.
Therefore, the polygons that make up the same object have the same
You will have a separate child. Put this identifier ID in the first pass
And distance L₁Recorded in the shadow polygon buffer with z
Remember In the second pass, we interpolate the polygon
Distance L₂z and stored in the shadow polygon buffer
Distance L₁z, and if it is considered a shadow,
Compare the offspring. Cast shadow if identifiers match
The object and the object currently being rendered are the same,
The interpolation point is not considered as a shadow. Therefore, in the present invention, interpolation
If the point is determined to be a backward distance and the identifier is different
First, the interpolation point is regarded as a shadow point. This method
The problem is that the back side of the object viewed from the direction of the light source is regarded as a shadow.
There is no problem, but this is due to the shading circuit (for example,
Phon shading) can be used together with the
It can be obtained by the luminance expression calculated at the angle between the line and the light source.
Can be. In addition, the shadow
Current percentages for the smoothing method
-The closed filter method is known. In the present invention,
Shadows on the first pass to get a smoothing effect
When storing a polygon, (L₁x, L₁y) position
A flag having a predetermined value is simultaneously stored. This flag
A 1-bit shadow flag and an n-bit area flag
Consist of lugs. In other words, the position where the shadow is projected
Are initialized for each pixel. Shadow polygo
The distance from the light source of the buffer and the area flag are all
After the body polygon is interpolated, the shadow polygon buffer
The entire area is scanned for filtering. As an example
Distance is the distance with the identifier contained in the sampling area
Injuries are used and averaged by the number of
Lag performs a three-dimensional spatial filter. By this method first
By directly filtering the shadow polygon after passing
Distance error can be minimized. Filtering is a sump
The distance included in the area of 3x3 pixels or more near the ring point
Use the area flag. This method is similar to reading video signals.
In the direction of polygon interpolation and shadow polygons.
Does not depend on the amount of memory Also multiple filtering
Can be realized by repeating the same method. Filter
The flag value before shadowing is the shadow polygon area (shadow initial
This flag value is stored in the
Filtering around shadow edges
It has an intermediate distribution value that changes from the initial value to the zero value.
You. In the present invention, 7F (hexadecimal) is used as an example of this initial value.
Set as The filters are, for example, each in a 4x4 area
Is In this case, the intermediate value is approximately
Take 16 levels. Medium depending on how many times the filter is repeated
The inter-value area diffuses and the quantization level further increases. The present invention
Now we can get the effect of shadow Umbra and Penumbra
To generate two values. One is that the shadow projection plane is a shading object
The further away, the darker (due to ambient light)
To determine the decay rate of the shadow in the second pass
Compare with the distance stored in the bufferMultiply the luminance of the display pixel. Where a0 and a1 are light
The scattering coefficient, which is a constant that depends on the environment. Decrease darkness
The decay rate is given by (1-D).The other one is a Penumbra value that expresses the blur around the shadow
It is. A local brightness graph around the boundary between the shadow and the illuminated area.
The gradation is the filtered file obtained above.
Use the lag value. However, Area Sun does the whole buffer
In the pulling, changes around the boundary area are compared with the flag value 7F.
There is a uniform width change between the 0 values at the projected part. Therefore
Direct use of this flag value for inter-point brightness will result in light blocking
Penumbra width near the object and width at a far place
Cannot produce different effects. Light shield
Distance between body and shadow projection pointConsidering the method of rounding the flag intermediate value using the distance to the body)
available. In the present invention, And (1-D) obtained from the above equation (1),
This is a coefficient of the luminance of the display pixel. Pen by the above method
The umbra and Umbra effects are the shadow cast table
This is true if there is one shading object before the indicating object
However, when multiple shadows are projected,
Is calculated from the object closest to the light source within a certain range.
The 0p value is reflected and becomes an erroneous value. Limited buff
Two or more shadows can be used to increase
Prepare a polygon buffer and use the light in the first pass
The distance between the object near the source and the next object near the light source
Memorize each separately and compare them with the distance of the second pass
And the interpolation point of the display object is between these buffer distances
In some cases, the process is located behind these two objects.
The Penumbra function is ignored and the darkness decay
Only the rate (1-D) is used as the coefficient of the luminance of the display object.
Therefore, the error of 0p can be eliminated. The present invention
From two or more shadow polygon buffers
prepare. 0p is generated for each shadow polygon
If so, the distinction is not required. On the other hand,
Ming's circuit also has the ability to create shadows from translucent objects.
One. In the shadowing of translucent objects, patterns (textures)
Pattern) is used to determine the pattern using light transmittance.
Memorize the shadow reflecting the texture to project the image into the shadow
A new buffer is needed. This is called a shadow in the present invention.
C Image buffer. In the first pass, light transmission
The light attenuation factor ε, which determines the excess, is defined for each polygon vertex, and
Is interpolated inside the polygon, and the pixels of the shadow image are
Store in units. As a result, in the first pass, the polygon
Of the light source system (L₁x, L₁y, L
₁z, L₁w) plus texture mapping coordinates
(U, v) and the decay rate ε
Therefore, the object has a shading phenomenon.
As in the second pass, each vertex of the polygon has a normal slope,
It is also necessary to set the angle with the light source. Text in the first pass
Char mapping is inverse mapping transformation u / L₁w, v / L₁w
Using the texture address obtained by
Read the char pattern and map it to a shadow polygon.
Ping. In the second pass, the viewpoint coordinates and L₁w is Ew
And the damping rate ε is the blending parameter
The processing of the second pass is performed. Shad on the first pass
C) Brightness of the shadow image stored in the image buffer
Degree S, decreaseAt the interpolation point of the polygon pixel in the
Brightness I shade and its shadow
Obtained luminance I shadow and the shadow is projected
If the polygon to be translucent has a translucent
The parameter is α and the frame bar overlaps with
The brightness of the pixels in the buffer image to dest
And the finally obtained luminance I The image is expressed by equation (2)
And (3).The problem with shadowing translucent objects is that they are translucent and opaque.
Depending on the relative positions of the body and the display object being interpolated.
In other words, correct processing is not performed. On the first pass
According to the hidden surface elimination method in
Translucent and opaque (light-blocking) objects
Distance L from the light source closest to the light source₁z
Must be remembered. Even with this method,
Relationship where multiple bright objects overlap and display objects are sandwiched between them
, The two buffers may not be able to handle
is there. In other words, the number of buffers, the display
A certain number of buffers are required to accurately represent the positional relationship.
It becomes important. The distance from the shading object to the shadow projection plane is large
Once againThe distance of the light-blocking object closest to the light source to the normal hidden surface
Using the past method, the first passDistance. The light blocking object has a long width on the z-axis
If it is a body, the projected shadow is located immediately after the object.
However, it is regarded as a considerable distance. As a result, equation (1)
The shadow attenuation coefficient (1-D) becomes an incorrect value. this
To avoid this, the object closest to the light source should be the light blocking object
On the other hand, farthest from the light source in the same object
Must be stored in the shadow buffer. This
To handle this, the identifier given for each object must be
Used, usually between objects (if the identifier is different)
Between the polygons of the same object
The front is to be erased. From the above, according to the present invention,
Ucasting processing is hardware-based, real-time
Enables time shadow casting.

[Embodiments] Specific embodiments will be described below.
I will tell. FIG. 1 shows a shadow casting circuit according to the present invention.
The upper and lower rows show the first and second paths, respectively.
You. Variables shown in Table 1 are defined at the vertices of the polygon.In FIG. 1, the interpolation circuit 1 for interpolating a polygon
The values shown in Table 1 except for the
After interpolating linearly between the vertices using the interpolation circuit 1,
Interpolation between two points where the polygon outline intersects the horizontal axis
Interpolate further by roads 2a-2c and values inside all polygons
Ask for. In the interpolation circuit 2c, the light source system coordinate values Lx, Ly,
Each of Lz and Lw is interpolated in the first pass, and then inverse mapping
In the conversion circuit 7, L₁x = Lx / Lw, L₁y = Ly
/ Lw and the distance L from the light source₁z is the pixel cap
Address (L₁x, L₁y)
Write in Fa10. Write the written data to the shadow port.
It is called Rigon. Lw is a scalar containing the Lz value during projection conversion
-Value. The buffer 10 has a shadow poly in the area.
Gon, display image and various attributes
It is a memory to remember. Therefore, in the present invention, the information stored
Distinguishes the name of the buffer by the area name. Shad
U polygon value L₁When z is stored in the buffer 10,
For each interpolation, compare the value already stored in the buffer with the value of the interpolation point.
Comparison, the one closer to the light source is selected and finally the closest to the light source
Store the value. In the present invention, the value closest to the light source is stored.
Shadow polygon and the shadow with the second closest value
Two shadow buffers to store each polygon
prepare. Therefore, the distance between the interpolation points is
When it is determined that the light source is closer to the light source than the
Move the value stored in the second buffer to the second buffer.
You. If the polygon being interpolated is a translucent object,
Define a value for each vertex. Normal (Nh, Nv),
Light source angles (Lh, Lv), texture coordinate values (u, v) and
And the attenuation factor ε are interpolated along the polygon outline by the interpolation circuit 1.
After that, the interpolation circuit 2a and the interpolation circuit 2a interpolate between two points intersecting the horizontal axis.
And 2b are used to determine all interpolated values within the polygon. these
Is given to the shading processor 3 and the polygon surface
Used for light reflection calculations above. Texture coordinate value u '= u
/ L₁w and v'v / L₁w is the bump normal
Memory 4 and the memory where the texture pattern is stored
Give to 6. Bump normal is shading processor 3
And the surface normal (Nh, Nv) and the light source angle (Lh, L
v) and from the bump-up vector
The degree Ip is determined. This luminance Ip is defined as a texture pattern.
The data is merged by the multiplier 8 and passed through the pixel cache 9.
Write to the shadow image buffer 10. Opaque objects
In the interpolation in the first pass, 2a, 2b, 3, 4, 5 and
6 and 6 are not used. Second pass smell
And the light source system coordinate value L of the interpolation circuit 2c.₂x, L₂y and
L₂z is given to the shadow polygon cache 11. Sha
The dough polygon cache 11 has a distance L obtained in the first pass.
₁z and other relevant data every time a cache miss occurs
The data is transferred from the buffer 10 and stored. Cash 11 is L
₂x, L₂read out, shadowed, addressed by y
L in the luminance determination circuit 12₂Compare with z. Shadow brightness
When the display pixel is shadowed, the output value of the
It becomes a value that attenuates the degree and is given to the multiplier 13. This result
The result is stored in the buffer 10 through the pixel cache 9 (table).
Shown) It is written in the image buffer area. With this invention
Is the distance L from the light source in the first pass₁z, flag and
And the identifier are stored in the buffer 10. These identifiers and
And flag are the same address where the shadow polygon is stored
Written in The flag indicates the shadow area found in the first pass
Consists of a 1-bit flag and an area flag consisting of n bits
I do. The initial value of the area flag is 7F (hexadecimal) as an example.
To be stored. Interpolate all polygons in the first pass
After that, the area flag and the distance L₁z in circuit 14
The buffer 10 is sequentially scanned and filtered. FIG. 2 shows the circuit 1
FIG. 4 is a detailed view of FIG. 4, showing a 4 × 4 correlation filter circuit and an average value
Circuit. The filter circuit is a spline circuit 16
(16a-16e), the input line buffer 15 and the output line
An in-buffer 17 and an averaging circuit 18
You. Read the area flag and the identifier from the buffer 10;
Load to input line buffer 15. In this invention example
Simultaneous sampling of 4x4 area (P00-P33) sample values
After loading the roads 16a-16d and correlating each column,
Correlate at 16e in the row direction. 4x4 from sampling point
Is the distance x to each sample point of
The basis function is stored in the memory 19, and the function value and the
Multiplier 20a-20 with pull data (P00-P33)
d, and then summed by adders 21a-21d.
Get the filter value. In this filter processing,
It has the following procedure. (1) Only one type in the sampling area
If there is an identifier of
Function filtering, while the distance
After summing the values, divide by the number of identifiers present (average
This is used as a filter value. (2) Sampling
If a region has multiple different identifiers,
Distance closest to the light source among the distances included in the ring area
Is detected and the identifier of the identifier with the same name as the identifier of the distance is detected.
Area flags, and all other identifier flags
After applying the correlation filter,
Only the distances with the filtering identifier
Add and divide by the number of added identifiers. This distance
Averaging is performed by an adder, a memory table for division, and
And a multiplier that multiplies the output of the adder and the table value.
After the area flag filter, the flag value 0p is 0 ≦ 0p
≦ 7F. 0p passes through the output line buffer 17
The data is written into the buffer 10. On the other hand, a 1-bit
The shadow flag f is stored in the shadow polygon buffer.
Identification of the sampling point
Set to 1 if it is the same as the child, otherwise set to 0.
The shadow point closest to the light source at the end of the first pass
It is possible to save the shadow of the Ligon area and the boundary line of the irradiation part.
You. FIG. 3 shows a shadow polygon cache of the present invention.
In the second pass, the vertex coordinate values (Ex, E
y, Ez, Ew) and the vertex coordinate value (L₂x, L
₂y, L₂z) is simultaneously interpolated. As shown in FIG.
In the interpolation circuit 2c, the long axis of the coordinate value of the viewpoint coordinate system (in the figure,
Ex), the light source system coordinate values are interpolated. Interpolated coordinates
Value (L₂x, L₂y) is a shadow polygon cache 11
Give to. As shown in FIG.
A memory 22 and a cache controller 23
The distance from the light source of the shadow polygon to the memory 22
(L₁z), area flag 0p, polygon identifier ID,
Shadow image S and decay rate ε (for translucent objects)
Is controlled each time a cache miss occurs from the buffer 10.
By way 23, cache miss address peripheral data (for example,
(8 × 8 pixel area). Shad
U polygon cache 11 is buffer 10, shadow brightness
Distance and identifier comparator 24 included in decision circuit 12
And a shadow luminance determination consisting of a function memory,
Connection module 25. FIG. 4 shows the Pen of the present invention.
Show the umbra function. In the figure, horizontal axis is 0p, vertical
Axis outputs function valuesIf there is no change, the change between 0 and 1 is large.
The digital change shown inSelect a line. The filtered area flags are shaded in the figure on the right.
(Straight line). This curve/ L₁z) The function can be approximated by hypertangent.
Put this function in a memory tableCan be obtained. FIG. 5 shows the shadow luminance determination time of FIG.
2 shows a detailed circuit example of the road 12. Shadow brightness determination circuit
From number multipliers, adders, memories and multiplexers
Each of formulas (1) and (2) in [0003]
Do this. Distance of the interpolation point from the light source in the second pass
Release L₂z is the distance Lz of the opaque object from the light source opq
And the distance Lz of the translucent object from the light source transp-D is stored. Each table output value is raised to the power of 27
After the calculation, the luminance I from the shading processor 3 is calculated.
The merge is performed by the shade and the multiplier 30c. Translucent object
The shaded brightness I shade and semi-transparent
Use the attenuation rate ε for the texture image S of a bright objectWhen the comparator 24 determines that the interpolation point is a shadow, the sign is A
It is provided to ND circuit 28. If Cash 11 Polygo
Identifier IDd matches identifier IDs of polygon being interpolated
Otherwise, multiplexer 29 is raised to
The output of the calculator 30c is selected. Irradiation point if coincident
I Select "shade". Follow the above steps
I with a mood change shadow is obtained. Comparison
The detailed circuit of the device 24 is shown in the lower part. The figure shows two opaque objects
The distances Lz0 and Lz1 from the body light source (the distances closest to the light source)
From each buffer that stores the separation and the next closest distance
), And the distance Lz of the translucent object from the light source tra
nsp is given from the cache 11 and these distances are interpolated.
Point distance L₂z and comparators 24b1, 24b2 and 24
Compare with b3. The comparison result is given to the control circuit 24d.
In this case, a selection signal for the multiplexer 24c is used. 24c
Is either the closest or next closest light source
Select This distance L₁Of the z and the sign, the sign (shadow
In the case of possibility, 1) is given to the AND circuit 28 and the ratio with the identifier is given.
Comparison result (1 when the identifier is different) and AND
Determine whether to irradiate shadow or conditionTo the file 26a. FIG. 6 shows details of the buffer 10 of the present invention.
An example of the configuration is shown. Closest to the light source of the opaque object in the figure
The object's shadow polygon (distance) and the object next to the light source
2 sets of shadow polygons (F3, F4) and translucent
Shadow polygon (F5) of object, polygon identifier (F
6), area flag (F6), one-bit flag (F6),
The shadow image (F7) and the attenuation rate ε (F7) are used.
The buffer for the display object is the display image (F1), the table
The indicated image z value (F2) and the transparency (F1) are obtained. In the figure
Here, the word is, for example, 32 bits. These bits
Length and number of words are gray for display screen resolution and brightness.
Depends on In FIG. 1, a shading processor
The luminance Ip obtained in step 3 is the texture color and the multiplier.
8 and the output value of the multiplier 8 is
Sa3 I shade. I shadow is pi
Once buffered in xell cache 9, buffer 1
Written to 0. This I cash shadow 9
Is transferred to the buffer 10 from the
Read the data (F1, F2 brightness, z value, transmittance in Fig. 6)
Read (read modify write operation), read
(3) of the previous section [0003] using the obtained luminance and transmittance.
Perform an expression. Shadin through all the above processes
The interpolation of the polygon to which the shadow and the shadow are added is completed.

[Effects of the Invention] The present invention enables shadow casting with a soft shadow effect, which has been difficult with hardware, and realizes real-time video in a polygon rendering method by applying it to a virtual reality system. Can express.

[Brief description of the drawings]

FIG. 1 shows a shadow casting circuit according to the present invention.

[Explanation of symbols]

1: outline interpolation circuit 2a-2c: horizontal interpolation circuit 3: shading processor 4: bump normal memory 5: inverse mapping conversion circuit 6: texture pattern memory 7: inverse mapping conversion circuit 8: multiplier 9: pixel cache 10: buffer 11: Shadow polygon buffer 12: Shadow brightness determination circuit 13: Multiplier 14: Filter circuit

FIG. 2 shows a filter circuit according to the present invention.

[Explanation of symbols]

10: Buffer 15: Input line buffer 16a-16e: Spline correlation circuit 17: Output line buffer 18: Distance averaging circuit 19: Basis function table 20a-20d: Multiplier 21a-21
d: adder

FIG. 3 is a shadow polygon cache of the present invention.

[Explanation of symbols]

10: Buffer 22: Cache memory 23: Cache control circuit 24: Comparator 25: Shadow luminance determination module

FIG. 4 is a Penumbra function of the present invention.

[Explanation of symbols]

ΔLz: distance between display object and light-shielding object L1: distance between light-shielding object and light source

FIG. 5 is a circuit for determining a shadow luminance according to the present invention;

[Explanation of symbols]

3: Shading processor 11: Shadow polygon cache 24: Shadow brightness determination circuit 24a: Opaque object shadow polygon buffer 24b1 to 24b3: Subtractor 24c: Multiplexer 24d: Control circuit 24e: Subtractor 24f: Memory table 24g: Multiplier

FIG. 6 shows a buffer according to the invention.

[Explanation of symbols]

F1: Display image and transparency buffer F2:
Display image z value F3, F4: opaque object shadow polygon buffer F5: translucent object shadow polygon buffer F6: polygon identifier, area flag, one-bit flag buffer F7: shadow image, attenuation factor ε buffer

Claims (9)

[Claims] 1. A polygon defined by a three-dimensional light source coordinate system having a light source as a projection center is interpolated and the distance between each interpolation point and the light source is stored in a shadow polygon buffer using hidden surface elimination. The first processing is to interpolate a polygon defined in a three-dimensional viewpoint coordinate system with the viewpoint being the projection center, and at the same time, coordinate values in the light source coordinate system corresponding to the interpolation points of the polygon in the viewpoint coordinate system. A second process for determining whether the interpolation point is a shadow or an irradiation point by comparing the distance from the light source of the light source coordinate system coordinate value with the distance stored in the shadow polygon buffer in the first process. In the two-pass z-buffer shadow casting method, an identifier is assigned to each polygon in the first processing, and the distance from the light source, the identifier, and the flag are stored in the shadow polygon buffer for each interpolation point. In the first means and the second processing, an identifier is provided for each polygon, and not only the vertex coordinate value of the viewpoint coordinate system but also the vertex coordinate value of the light source coordinate system is defined for each vertex, and these coordinate values are interpolated. A second means for simultaneously interpolating by a circuit, and interpolating a distance in a light source coordinate system corresponding to each viewpoint coordinate system interpolation point obtained in the second processing and a distance stored in a shadow polygon buffer. When the distance from the light source stored in the shadow polygon buffer is closer to the light source, the identifier stored in the shadow polygon buffer in the first processing and the polygon in the second processing are further compared. A shadow circuit having third means for setting the interpolation point as a shadow when the identifiers do not match, and as an irradiation point when the identifiers match, by a comparison circuit for comparing the identifiers Computing circuit. 2. The shadow casting circuit according to claim 1, wherein the coordinate values of the polygon vertices of the light source coordinate system are calculated in a homogeneous coordinate system (Lx, Ly, Lz, Lw) in the first processing and in the second processing. Is defined in a normal coordinate system (Lx, Ly, Lz), and in the first processing, each coordinate value of Lx and Ly is converted into a scalar value by an inverse mapping conversion circuit for each polygon interpolation. The value divided by Lw is used as the address of the shadow polygon buffer as the distance Lz from the light source.
In addition to storing the value, in the second processing, the interpolated Lx
Means for reading out the distance stored in the shadow polygon buffer using the same as Ly and comparing the distance with the distance from the light source in the second processing to determine the presence or absence of a shadow by the point light source; A shadow casting circuit having an interpolation circuit for linearly interpolating each vertex coordinate value of a light source coordinate system with respect to a principal axis using a viewpoint coordinate value with respect to interpolation of a polygon. 3. The shadow casting circuit according to claim 1, further comprising: a filter circuit for filtering a distance from said light source and a flag stored in a shadow polygon buffer, respectively, in at least a 3 × 3 pixel area near a filter sampling point. The distance closest to the light source among the distances stored in this area and the identifier added to the distance are obtained, and the distances having the same identifier as the identifier are averaged, while the flag is composed of 1 bit. Means for performing filtering after setting a shadow flag and an area flag consisting of at least two bits or more, leaving an area flag value having the same identifier as the identifier as it is, and setting area flag values of other identifiers to a zero value,
When the identifier of the sampling point is different from the identifier including the closest distance from the light source with respect to the shadow flag, resetting is performed, and when the identifier is the same, the respective means for setting are used to determine the distance of the sampling point, the area flag, and the identifier. Shadow casting circuit. 4. The shadow casting circuit according to claim 1, wherein in the second processing, a distance from a light source corresponding to an interpolation point of each viewpoint coordinate system obtained in the second processing, and a shadow in the first processing. A difference value from the distance stored in the polygon buffer is obtained for each interpolation, and a constant value depending on ambient light is added as a coefficient to this difference value, and a shadow attenuation value for attenuating a shadow as the difference value increases becomes larger. A shadow casting circuit having a circuit to be sought. 5. The shadow casting circuit according to claim 3, wherein the filtering of the area flag is performed in the first processing as means for obtaining a display in which the vicinity of a shadow close to a light-shielding object is sharp and the periphery of a distant shadow is blurred. Performed at least once after the interpolation process, and using a memory table that uses the filtered flag value and the difference value of the distance and the distance between the light-shielding object and the light source as an address variable, when the difference value is small. Stores a function that makes the rate of change between the flag initial setting value and the zero value steep, and that, when the difference value is large, makes the rate of change gradual. This function value and the shadow attenuation value obtained in claim 4 Is a scale value for the luminance of the display pixel, and determines the luminance around the shadow. 6. The shadow casting circuit according to claim 1, further comprising at least two or more shadow polygon buffers for sequentially storing a distance from an object closest to the light source to a next closest distance, and a light source for each of these shadow polygon buffers. Is compared with the distance of the interpolation point from the light source in the second processing, and if the interpolation point is located between the object closest to the light source and another light-shielding object, Means for multiplying the brightness of the display object by the shadow attenuation value and the function value of 5 and 5, and means for multiplying the brightness of the display object only by the shadow attenuation value when the interpolation point is located farther than two or more light-shielding objects. Shadow casting circuit 7. The shadow casting circuit according to claim 1, wherein when the polygon to be interpolated in the first processing is a translucent object, the polygon vertices include not only vertex coordinate values and identifiers in the light source system but also surface methods. An interpolation circuit that defines lines, light source angles, texture mapping coordinate values, and light attenuation factors, respectively, and interpolates them at the same time, and a texture mapping circuit that maps a texture pattern to an interpolated polygon. And a shadow image buffer for storing the obtained shadow texture pixel and the light attenuation factor. In the second process, when determining the brightness of the display pixel, the brightness of the display pixel shaded in the viewpoint coordinate system and the shadow image Means for synthesizing the luminance of the pixel stored in the buffer using the light attenuation rate. Shadow casting circuit 8. The shadow casting circuit according to claim 4, wherein the distance of the translucent object from the light source;
It has at least two or more shadow polygon buffers each having a different distance from the opaque object, and stores the interpolation values in the first processing, respectively, and the light source for the polygon interpolation interpolation point obtained in the second processing. Is compared with at least two or more distances stored in the shadow polygon buffer, and if the interpolation point is closer to the light source than to the opaque object and farther than to the translucent object, 7. The shadow attenuation buffer and the shaded display pixel are synthesized by the means of item 7 using the light attenuation factor, and the shadow attenuation value and the function obtained in claims 4 and 5 are obtained when the shadow image buffer is farther than the opaque object. Shadow casting circuit having means for determining luminance using respective values 9. The hidden surface elimination method according to claim 1, wherein the distance from the light source already stored in the shadow polygon buffer is compared with the distance of the interpolation point, and one of the distances is selectively stored. For the circuit, if the distance stored in the shadow polygon buffer is behind the distance of the interpolation point with respect to the light source and the identifier stored in the buffer and the identifier of the interpolation point are different, select the interpolation point distance. Means for selecting the distance stored in the shadow buffer when the identifiers are the same while writing the new data in the shadow polygon buffer, and the distance stored in the shadow buffer is closer to the light source than the distance of the interpolation point. If the identifiers are different, the distance stored in the shadow polygon buffer is selected. If the identifiers are the same, the distance of the interpolation point is selected. Shadow casting circuit having means to write to the shadow polygon buffer