JPH0575150B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an object image synthesis device that synthesizes images for displaying objects.

[Conventional technology]

A method called the ray search method is used to synthesize digital images for displaying objects. In this method, the path of the light ray from the light source to the viewpoint is traced in the opposite direction, a process is performed to determine the intersection between the light ray and the object, and the brightness of each pixel making up the image is calculated based on the result. An example of this ray search method can be found in the Information Processing Society of Japan, Volume 25, No. 6, Hiroshi Deguchi, Hitoshi Nishimura, Hiroshi Yoshimura, Toru Kawata, Isao Shirakawa, and Koichi Omura, in the paper "Computer Graphics System
``How to speed up image generation in LINKS-1''
It is described in.

In order to generate images quickly using such a ray search method, the space in which the object is defined is divided into multiple regions, each region is assigned to one computer, and each computer is Methods have been proposed for processing light rays passing through a region. For more information on this method, see Computer Graphics,
Volume 18, Issue 3, Mark Deitzpe
Dippe), John Swensen
Swensen), in his paper “An Adaptive Subdivision Algorithm and Parallel Architecture for Realistic Image Synthesis”.
Subdivision Algorithm and Parallel
Architecture for Realistic Image Synthesis)
It is described in.

In the above method, the space is first divided into a plurality of rectangular parallelepiped regions, and each region has one
assigned to one computer. Next, each computer stores data on objects included in the assigned rectangular parallelepiped area. Then, each computer performs a process of determining the intersection between the light beam passing through the assigned area and the object. Through this intersection determination processing, when the object and the light ray intersect, reflection and transmission processing on the object surface is performed. Information about rays that do not intersect objects within that region is transferred to computers assigned to adjacent regions.

When performing such processing, there is a possibility that the processing amount of each computer will vary considerably. In other words, a computer assigned to an area containing many objects has a very large amount of processing, while a computer assigned to an area containing no objects does almost no processing at all.

In order to reduce such variations in load between computers, a method has been proposed in which the vertices of a region divided into rectangular parallelepipeds are arbitrarily moved to change the volume and shape of the region. That is, by moving the vertices, the amount of computer processing allocated to the area whose volume has decreased is reduced. Then, the computer assigned to the area whose volume has increased due to the movement of the vertex is assigned the corresponding processing load. In this way, the load is redistributed between the computers.

[Problem that the invention seeks to solve]

In such conventional load redistribution methods, in order to redistribute the load by moving the vertices of the region,
The shape of the area changes in various ways. That is,
Since the deformation is due to the movement of vertices, the shape of the region is kept as a hexahedron, but the orientation and shape of each boundary surface of the region can be arbitrary. This causes the following problems.

First, when transmitting light ray information from a divided region to an adjacent region, the region to be transferred is determined by determining the intersection between the boundary surface of the region and the light ray.
However, since the orientation and shape of each boundary surface of the region are arbitrary, the processing amount of this intersection determination process becomes extremely large. Therefore, there is a problem in that the overall processing time is also delayed.

Second, when the shape of a region is changed, information about objects included in the region must also be changed. That is, in the case of a region whose volume has decreased, there is a possibility that some objects that were included in the region before the change are no longer included in the region after the change. Furthermore, in the case of a region whose volume has increased, there is a possibility that a new object will be included in that region. When changing object information like this,
Since the shape of the region varies, the amount of processing and processing time required to determine the objects included in the region will be
The problem is that there are too many.

As described above, the conventional method of redistributing the load by moving the vertices of the area has the problem that the redistribution requires a very large amount of processing, and the effect of the redistribution cannot be sufficiently obtained.

The present invention simplifies the process of changing the area shape for load redistribution, thereby reducing the processing involved in changing the area shape and sufficiently increasing the effect of redistribution. The purpose of the present invention is to provide an image synthesis device.

[Means for solving problems]

The object image synthesis device of the present invention includes: an initial ray generation section that generates information about a plurality of rays passing through each pixel from a viewpoint; an object information setting section that sets information about the object;
Each of them is in charge of one region out of a plurality of regions generated by dividing the space in which the object is defined, and performs intersection determination processing between a ray passing through this region and the object included in the assigned region. and a plurality of brightness calculation units that calculate the brightness of the pixels, and an image storage unit that stores the brightness calculated by the brightness calculation units as an image, and each of the brightness calculation units stores the object definition space. area information storage means for storing the range of an area in charge of one of a plurality of areas generated by dividing on a plane perpendicular to the coordinate axis; a load determining means for determining the load of the brightness calculation section; and the coordinate axis. A mutual communication means for mutually communicating with the adjacent brightness calculation unit in charge of the area in charge and an adjacent area in parallel directions, and the load of the adjacent brightness calculation unit obtained through the mutual communication means and the load of the adjacent brightness calculation unit itself. expansion determining means for determining whether or not to expand the respective areas in charge in the direction of the areas in charge of these adjacent brightness calculation units by comparing the loads; Area enlarging means for changing the range of the area in charge stored in the area information storage means is provided.

[Effect]

A method for redistributing the load of the luminance calculation unit in the present invention will be described. Each brightness calculation unit is assigned one area out of a plurality of areas generated by dividing the object definition space along a plane perpendicular to the coordinate axes as the area in charge of that brightness calculation unit. The range of this area in charge is stored in area information storage means provided in the brightness calculation section.

Here, a brightness calculation unit in charge of an area adjacent to the area in charge of this brightness calculation unit is called an adjacent brightness calculation unit.

The load of each brightness calculation section is determined by the load determining means and transferred to each adjacent brightness calculation section via the mutual communication means. The enlargement determining means provided in the brightness calculating section that receives these compares the load determined by the load determining means with the load of each adjacent brightness calculating section obtained via the mutual communication means. Based on these comparison results, the expansion determining means determines whether or not to expand the area of this brightness calculation unit in the direction of the area in charge of each adjacent brightness calculation unit so that the load on each brightness calculation unit is smoothed. Decide each.

Finally, the area enlarging means changes the range of the area in its duty stored in the area information storage means in accordance with the determination by the expansion determining means, and executes the expansion of the area in its duty.

In this way, by expanding the assigned area based on the load comparison results, the load is redistributed between the brightness calculation unit and its adjacent brightness calculation units.

〔Example〕

DESCRIPTION OF THE PREFERRED EMBODIMENTS An object image synthesis apparatus as an embodiment of the present invention will be described below with reference to the drawings. In order to simplify the explanation, the explanation will be limited to the process of redistributing the load between the brightness calculation units adjacent in the direction parallel to a specific coordinate axis, but this does not impair generality.

FIGS. 1a and 1b are block diagrams showing an object image synthesis device as an embodiment of the present invention, and FIG.
is an overall configuration diagram showing the entire object image synthesis device,
FIG. 1b is a block diagram showing the detailed structure of the brightness calculation section 3. As shown in FIG. 1a, an initial light ray generating section 1 is provided which generates information on a plurality of rays passing through each pixel of an image to be synthesized from a preset viewpoint. Further, an object information setting section 2 is provided for setting information about the object to be displayed.

Furthermore, taking charge of one region out of a plurality of regions generated by dividing the space in which the object is defined, and performing intersection determination processing between the light ray passing through this region and the object included in the assigned region. A plurality of brightness calculation units 3 calculate the brightness of each pixel by
is provided. An image storage section 4 is provided to store the luminance calculated by the luminance calculation section 3 as an image. The image storage unit 4 sets all luminances to 0 before combining images.

A connection line 5 is provided for transmitting information between the initial light generation section 1, the object information setting section 2, the plurality of brightness calculation sections 3, and the image storage section 4. An information input section 6 for inputting information from, for example, a keyboard is provided via this connection line 5.

As shown in FIG. 1b, the brightness calculation unit 3 is provided with mutual communication means 31 for communicating with the brightness calculation units 3 in charge of areas adjacent in six directions. Further, a communication means 32 for communicating via the connection line 5 is provided. Furthermore, the brightness calculation section 3
A connection line 101 is provided for transmitting information within the network.

FIG. 2 is an explanatory diagram showing a method of calculating the brightness I of a pixel p(i,j), and a light source L, objects O, O', and a viewpoint E are arranged as shown in the figure. As shown in Figure 2, in the ray search method, the brightness of pixel p(i, j) is determined by tracing the path of the ray from light source L through pixel p(i, j) to viewpoint E in the opposite direction. Perform the calculation. Here, the brightness I of the pixel p(i,j) will be referred to as the brightness of the light ray R generated in the opposite direction from the viewpoint E through the pixel p(i,j). In Figure 2 and the following explanation, R, R', R _L , etc. indicate rays, R→, R→', R _L , etc. indicate directions, direction R indicates the direction of ray R, and so on. .
The brightness I of this light ray R is the intensity I of light that enters the viewpoint E through the pixel p(i,j). Ray R
The brightness I of is calculated using the following formula. (〓 indicates multiplication) I=ref〓I′+dif〓(N→・R _L )〓I _L ref: Reflection coefficient of object O dif: Diffusion coefficient of object O I′: Incident light from R→′ direction Intensity I _L : Incident light intensity from light source L N→: Unit normal vector of the surface of object O R→': Specular reflection direction vector in R→ direction R _L : Direction vector of light source L These two incident light intensities In order to obtain I' and I _L , two light rays R' and R L _{whose starting point is the intersection point CP between the light ray R and the object O and whose directions are R→' and R L are} generated. In addition, the coefficients of I' and I _L for determining the brightness I are
The attenuation rates G′ and G _L of the rays R′ and R _L are set. That is, the light ray R is attenuated by colliding with the object O, and the light rays R' and R _L are generated.
These attenuation rates G' and G _L are as follows: G'=G ref G _L = G dif (N→, R _L ) G: attenuation rate of light ray R (=1). By using these attenuation rates G' and G _L , the brightness I of the light ray R, that is, the pixel p (i, j)
The brightness I of is determined as follows.

I=G'〓I'+G _L 〓I _L In this way, when new rays R' and R _L are generated, the information on ray R is no longer necessary.

Furthermore, as shown in Figure 2, the ray R'
When intersecting O', rays R'' and R _L are generated in the same way.The attenuation rates G'' and G _L ' of these are similarly calculated using the following equations.

G″=G′〓ref′ G _L ′=G′〓dif′〓(N→′, R _L ′) ref′: Reflection coefficient of object O′ dif′: Diffusion coefficient of object O N→′: Object O Unit normal vector R _L ′ of the surface: directional vector of light source L In this way, the attenuation rate G′ is integrated into the attenuation factors G″ and G _L ′.

However, the processing of the light ray R _L is different from that of the light rays R and R'. If the ray R _L intersects the object, the intersection point
CP becomes a shadow of the object and cannot receive the illumination light from the light source L. Therefore, the brightness I _L of the light ray R _L is zero. If the ray _RL does not intersect any object, the brightness I _L of the ray RL will be _the brightness of the light source L. In this way, the light rays R and R' directed toward the object and the light ray R _L directed toward the light source L are handled differently, so it is necessary to distinguish between the types of light rays. Therefore, the ray R
Type C is set for distinguishing the types of light rays. This type C is given a value of 0 when the ray R is directed toward an object, and 1 when the ray R is directed toward a light source.

As shown in FIG. 2, a ray of light directed toward an object generates a new ray of light directed toward the object each time it collides with an object. Therefore, in order to calculate the brightness I of one pixel p, many light rays may have to be processed. However, each time the light ray collides with an object, it is attenuated, so if the number of collisions increases,
The influence of the light beam on the brightness I becomes almost negligible.

Therefore, in order to limit the number of collisions of the ray R,
A number of times T is set for the light ray R. This number T indicates the number of possible collisions of the ray R. A ray of light R with a number of times T collides with an object and heads toward the object.
If R′ is generated, the number T′ of the ray R′ is (T−
1). If a ray R whose number of times T is 0 collides with an object, the ray toward the object
R' is not generated, but only the light ray R _L directed toward the light source L is generated.

In addition, although there is only one light source L in Figure 2, if there are multiple light sources L _i (i = 1, 2, ...), it is necessary to generate light rays directed to all light sources L _i . Must be. By the way, in the case of the light ray R _L heading toward the light source, no new light ray is generated even if it intersects with an object, as described above, so the number of times T as information about the light source R _L has no meaning. Therefore, if there are multiple light sources L _i , the number of times _T
, the number i of the light source to which the ray R _L is directed is set (in terms of computer processing, the number i of the light source for the ray R _L is set to correspond to the setting of the number of times T for the rays R and R').

In this way, by setting the number i of the light source L _i instead of the number of times T for the ray R _L , even if there are multiple light sources L _i , the light source to which each ray R _L is directed
Since the number i of L _i is known, the process can be performed correctly.

Also, here, for simplicity, only reflection on the surface of the object was considered. If transmission through the object is also taken into consideration, it is sufficient to generate a ray in the transmission direction at the intersection of the object and the ray R. However, this type of ray is a ray that heads toward an object, and is similar to the rays mentioned above.
If you process it in the same way as R', you can accurately represent the transparency of the object.

FIG. 3 is an explanatory diagram showing information on the light ray R that should be set when the light ray R is generated. When the ray R is generated as shown in Fig. 3, the information about the ray R is the starting point position coordinates (S _x ,
S _y , S _z ) and direction (d _x , d _y , d _z ) are set.
Furthermore, the pixel p(i,
The position (i, j) of j) is also set as information on the ray R. Furthermore, the attenuation rate G of the light ray R, the number of times T of the light ray R, and the type C of the light ray R are also set as information about the light ray R.

FIG. 4 is an explanatory diagram for explaining the operation of the initial light beam generating section 1. As shown in FIG. 4, the initial light ray generating section 1 generates a ray R on an extension of a half straight line passing through each pixel p(i, j) constituting the image P, starting from the viewpoint E.

For this purpose, the position coordinates (E _x , E _y , E _z ) of the viewpoint E are inputted to the initial ray generating unit 1 in FIG. 1A by the information input unit 6. Further, as information defining the image P to be synthesized, parameters indicating the plane and range of the image P are input from the information input unit 6. The initial light beam generating section 1 generates information about the light beam R based on these parameters. An example of how to find a half-line indicating this ray R is given by T. Whitted, Communication of ACM, Volume 23, No. 6,
pp. 343-349, the paper “Unimproved
Illumination Model Huo-Dated Display (An Improved
Illumination Model for Shaded Display).

Next, among the intersections between this half-line and the object definition space, the position coordinates of the intersection closest to the viewpoint (S _x , S _y ,
S _z ). These position coordinates (S _x , S _y , S _z ) are
These are the starting point position coordinates (S _x , S _y , S _z ) of the ray R. That is, the position where the ray R first enters the object definition space becomes the starting point of the ray R. Also, the direction from the viewpoint E toward the pixel p is the direction of the ray R (d _x , d _y ,
d _z ). Further, the position (i, j) of the pixel p (i, j) is set as the pixel position (i, j) of the light ray R.

In the initial beam generating section 1, the attenuation rate G of the generated beam R is set to 1, that is, the beam R is not attenuated completely. Further, as the number of times T, a constant value set in advance is given to the initial light beam generation section 1 from the information input section 6. Further, as type C, a value of 0 is given, which indicates a ray of light directed toward an object.

In the initial light beam generation section 1, a light beam R having such information is generated corresponding to all pixels p(i,j) of the image P, and is transferred to the brightness calculation section 3.

FIG. 5 is an explanatory diagram showing object information set in the object information setting section 2. As shown in FIG. To simplify the explanation, the displayed object will be limited to a sphere, but the same procedure can be applied to displaying objects such as polyhedrons and free-form surfaces.

As shown in FIG. 5, object information is input from the information input section 6 and set in the object information setting section 2. As shown in FIG. The information on the object O _i to be set includes the object number n _i for distinguishing the object O _i , the center coordinates of the sphere as the object (x _i , y _i , z _i ), the radius r _i , and the diffusion value indicating the material of the object. coefficient
dif _i and reflection coefficient ref _i . Further, as information indicating the circumscribed area of the object O _i , the range of the smallest rectangular parallelepiped including the object is set.

FIG. 6 is an explanatory diagram showing a circumscribed region of a sphere as an object. As shown in Fig. 6, the circumscribed area of the object O _i is the range in the x direction (x _i- , x _i+ ), the range in the y direction (y _i- , y _i+ ), and the range in the z direction (z _i- , z _i+ ). These values are obtained using the following equations.

x _i- =x _i −r _i x _i+ =x _i +r _i y _i- =y _i −r _i y _i+ =y _i +r _i z _i- =z _i −r _i z _i+ =z _i +z _i 7th The figure is an explanatory diagram showing light source information set in the object information setting section 2. As shown in FIG. To simplify the explanation, the explanation will be limited to point light sources, but the same procedure can be applied to various types of illumination light such as parallel rays and spotlights. Information on a light source for illuminating an object is input from the information input section 6 and set in the object information setting section 2 as shown in FIG. The information of the light source to be set is point light source L _i
The position coordinates (x _i , y _i , z _i ) of , and the brightness I _Li of the light source.
This brightness I _Li is a real value between 0 and 1. The value of this brightness I _Li indicates the brightness of the light source, and is 1
When it is , it is the brightest light source, and when it is 0, it is a completely dark light source.

The object information and light source information set in this way are
The information is transmitted from the object information setting section 2 to the brightness calculation section 3 via the connection line 5.

FIG. 8 is an explanatory diagram showing a method of combining a plurality of brightness calculation units 3 into a three-dimensional array. As shown in FIG. 8, the brightness calculation units 3 are connected in a three-dimensional array via mutual communication means 31 provided within the brightness calculation units 3. That is, each brightness calculation unit 3
are the brightness calculation units 3 on both sides in the x, y, and z directions, respectively.
, and can mutually transmit information with those brightness calculation units 3. Here, x,
If the a, b, and c-th brightness calculation units 3 in the y and z directions are called the (a, b, c) brightness calculation unit 3,
(a, b, c) The brightness calculation unit 3 calculates (a-1, b,
c), (a+1, b, c), (a, b-1, c),
(a, b+1, c), (a, b, c-1), (a, b,
c+1) Connected to the brightness calculation section 3.

Further, A, B, and C brightness calculation units 3 are arranged in the x, y, and z directions in the figure, respectively, and the total number D of brightness calculation units 3 is as follows: D=A〓B〓C.

FIG. 9 is an explanatory diagram showing a method of dividing a space defining an object into areas S assigned to each brightness calculation unit 3. As shown in FIG. 9, x,
The object definition space is divided into a plurality of rectangular parallelepiped regions by planes perpendicular to one of the y and z coordinate axes. In this case, the number of regions in the x direction is A, y
The division is performed so that the number of regions in the direction is B and the number of regions in the z direction is C. And (a, b, c)
The brightness calculation unit 3 is assigned the a-th area in the x direction, the b-th area in the y-direction, and the c-th area in the z-direction. Thereby, all the divided areas can be assigned to one brightness calculation unit.

Moreover, by performing such allocation, the brightness calculation units 3 connected via the mutual communication means 31 are respectively in charge of adjacent areas.

10 and 11 show the area S in charge of the brightness calculation unit 3.
FIG. 1st
As shown in FIG. 0, the area information storage means 33 provided in the brightness calculation unit 3 stores area information indicating the area S in charge of the brightness calculation unit 3. This area information includes the values (a, b, c) indicating the number in the x, y, and z directions, and the range (x, y, and z directions) of the assigned area S as shown in FIG. _- , x ₊ ),
(y _- , y ₊ ), (z _- , z ₊ ) are stored.

Furthermore, this is a case where the brightness calculation units 3 located on the outer periphery of the three-dimensional array are not connected to each other by the mutual communication means 31. Therefore, in each direction connected by the mutual communication means 31,
Whether or not the brightness calculation section 3 is actually connected is stored in the area information storage means 33 as presence/absence information.

These area information are input from the information input means 6 and transmitted to each brightness calculation section 3.

FIG. 12 is an explanatory diagram showing the process of transferring information on the light ray R from the initial light generation section 1 to the brightness calculation section 3. Information on all the light rays R generated by the initial light generation section 1 is transferred to all the brightness calculation sections 3 via the connection line 5. Then, as shown in FIG. 12, the information on the light ray R transferred from the initial light generation section 1 is inputted to the light ray information determination means 37 provided in each brightness calculation section 3 via the communication means 32. Ru. At the same time, the information of the assigned area S stored in the area information storage section 33 is transferred to the ray information determining means 3.
7. The starting point position coordinates (S _x , S _y , S _z ) of the input light ray R are set to the position where the light ray R first enters the object definition space. Therefore, if the starting point position coordinates (S _x , S _y , S _z ) are included in the assigned area S, the light ray R will pass through the assigned area S first.

Therefore, the ray information determining means 37 determines the starting point position coordinates (S _x , S _y , S _z ) of the ray R and the range (x _- , x ₊ ), (y _- , y ₊ ), (z _- , z ₊ ) are compared.

If all three conditional expressions x _- ≦S _x ＜x ₊ y _- ≦S _y ＜y ₊ z _- ≦S _z ＜z ₊ are satisfied, the starting point position coordinates (S _x , S _y , S _z ) are It is included in the assigned area S, and the light ray R passes through the assigned area S first. In this case, information on the ray R is transferred from the ray information determination means 37 to the ray information storage means 34,
be remembered.

Further, if any of these conditional expressions does not hold, it means that the light ray R is first incident on another assigned area S'. Therefore, information on the light ray R is not stored in this brightness calculation section 3.

For the above reasons, information on the light ray R generated by the initial light generation section 1 is stored in the light ray information storage means 34 of the brightness calculation section 3 which is in charge of the area where the light ray first enters.

FIG. 13 is an explanatory diagram showing a process of storing object information and light source information from the object information setting section 2 in the object information storage means 35 and light source information storage means 36 provided in the brightness calculation section 3. All object information and light source information stored in the object information setting section 2 are
It is transferred to all the brightness calculation units 3 at once via the connection line 5. As shown in FIG. 13, the light source information storage means 36 provided in the brightness calculation section 3 stores all the light source information transferred via the communication means 32.

Further, the object information setting means 38 provided in the brightness calculation section 3 first receives object information transferred via the communication means 32. Next, the range of the area S in charge of the brightness calculation section 3 is read out from the area information storage means 33. The range of this responsible area S (x _- , x ₊ ), (y _- ,
y ₊ ), (z _- , z ₊ ) and the range of the circumscribed area of the object information (x _i- , x _i+ ), (y _i- , y _i+ ), (z _i- , z _i+ ) are compared. Ru.

FIGS. 14a and 14b show the area S in charge of the brightness calculation unit 3.
FIG. 4 is an explanatory diagram showing a comparison process between the object information and the circumscribed area of the object information. As shown in FIGS. 14a and 14b, when the assigned area S and the circumscribed area have no common part, at least one of the following conditional expressions holds true.

_{x - ≧ x i + _ _ _ _ _ _} , it is determined whether the assigned area S and the circumscribed area have a common part. As a result, if there is a common part, it is determined that the object O _i is included in the assigned area S, and all information about the object O _i is stored in the object information storage means 35.
If they do not have a common part, they are not stored.

Through the above processing, among the object information stored in the object information setting section 2, only the object information included within the assigned area S is stored in the object information storage means 35 provided in the brightness calculation section 3.

FIG. 15 is an explanatory diagram showing the process of enlarging the area S in its duty by the brightness calculation unit 3. This enlargement area processing of the assigned area S is performed by moving the boundary surface of the assigned area S perpendicular to a specific coordinate axis in a direction parallel to this specific coordinate axis. This specific coordinate axis is determined by information input from the information input section 6. Here, a case will be described in which the z coordinate axis is a specific coordinate axis, but generality is not impaired.

As shown in FIG. 15, the load determination means 43 provided in the brightness calculation section 3
The number of light rays R stored in the brightness calculation unit 3
Find it as the load F. Note that although this method of determining the load is used here for the sake of simplicity, various other methods of determining the load are possible. Even if a method other than this is used, processing can be performed in substantially the same way. For example, load determining means 4
3, the time during which each brightness calculation unit 3 actually performs processing and the time during which it has information on the ray R may be measured, and the load on each brightness calculation unit 3 may be determined from these values. .

The load F determined by the load determining means 43 in this way
of the brightness calculation unit 3 via the mutual communication means 31.
The brightness calculation unit 3 connected in the z ₊ direction (hereinafter referred to as the brightness calculation unit 3 (z ₊ )) and the brightness calculation unit 3 connected in the z _- direction (hereinafter referred to as the brightness calculation unit 3)
(referred to as (z _- )). Then, the expansion determining means 44 calculates the load F transmitted from the brightness calculation units 3 (z _- ) and (z ( ₊ ) via the mutual communication means 31.
(z _- ), F(z( ₊ ) (the load on the brightness calculation unit 3 (z _- ) is F
(z _- ), the load on the brightness calculation unit 3 (z ₊ ) is written as F (z ₊ )
and the load F determined by the load determining means 43. This expansion determining means 44 is
Based on the load comparison, it is determined whether to expand the assigned area S as follows, and the results dF _- and dF ₊ are output.

When F (z _- ) - F > TH dF _- = 1 When F - F (z _- ) > TH dF _- = -1 Otherwise dF _- = 0 When F (z ₊ ) - F > TH dF ₊ = 1 When F-F(z ₊ ) > TH dF ₊ = -1 Otherwise dF ₊ = 0 TH: Positive threshold value set in advance through the information input section 6 As a result, dF _- 1, the responsible area S is z _-
Expand in the direction. Furthermore, when dF _- is -1 as a result, the area S in charge is not expanded, but the area in charge of the brightness calculation section 3 connected in the z _- direction is expanded.

Similarly, when dF ₊ is 1, the responsible area S is z ₊
Expand in the direction. Furthermore, when dF ₊ is −1 as a result, the area S in charge is not expanded, but the area in charge of the brightness calculation unit 3 connected in the z ₊ direction is expanded.

Expansion direction of the assigned area S determined in this way
Given dF _- and dF ₊ , if these values are 1, then
The area expanding means 45 expands the range of the area S in charge. The area enlarging means 45 first reads out and stores the range (z _- , z ₊ ) of the area S in its duty in the z direction from the area information storage means 33 . Next, find a new range (z' _- , z' ₊ ) as shown below.

When dF _- = 1, z' _- = z _- -dS When dF ₊ = 1, z' ₊ = z ₊ + dS dS: Positive enlargement amount set in advance through the information input section 6 Obtained in this way The range (z' _- , z' ₊ ) is written in the area information storage means 33 to expand the range of the area S in charge.

By performing such enlargement processing of the assigned area S, the number of objects included in the assigned area S and the number of light rays passing through can be increased, and the amount of calculation by the brightness calculation unit 3 can be increased. can. At the same time, the load on the adjacent brightness calculation units 3(z ₊ ) and 3(z _- ) can be reduced. Therefore, it is possible to average out the total amount of calculations and redistribute the load appropriately.

When the assigned area S is expanded in this way, information on objects included in the expanded area must be received from the adjacent brightness calculation section 3 and added to the object information recording means 35.

For this purpose, when dF _- is -1, the region enlarging means 45 outputs the value of dF _- to the object information storage means 35 as control information. The object information storage means 35 that receives this transfers the information of each object O _i to the brightness calculation section 3 (z _- ) via the mutual communication 31 .

Further, when dF _- is 1, the area expanding means 45
outputs the value of dF _- to the object information determining means 38.
Upon receiving this, the object information determining means 38 reads out the expanded range of the area S in its duty from the area information storage means 33. Furthermore, the object O _j transferred from the brightness calculation unit 3 (z _- ) via the mutual communication means 31
Read out the information. Then, a process is performed to determine whether or not the circumscribed region of this object O _j and the assigned region S have a common part. The determination process at this time is performed in the same manner as already shown in FIG. As a result of this determination process, if the circumscribed area of object O _j and the assigned area S have a common part, it is determined that the object O _j is included in the assigned area S, and the information of this object O _j is used as the object information. The information is stored in the storage means 35.

However, information about the same object as this object O _j may already be stored in the object information storage means 35. Therefore, in order to avoid storing the information of object O _j repeatedly, the object number n j of this object O _j and the object number n _j stored in the object information storage means 35 are used.
Compare O _i with object number n _i . Based on this comparison result, if the same object O _i as the object O _j is not stored, information about the object O _j is stored in the object information storage means 35 .

Furthermore, similar processing is performed for dF ₊ to change object information.

When the assigned area S is expanded through the above processing, the object information stored in the object information storage means 35 is added.

FIGS. 16a and 16b are explanatory diagrams showing a method of transferring information on the light ray R between the brightness calculation units 3. FIG. Figure 15b
The respective areas in charge of the brightness calculation units 3a to 3d shown in FIG. 15A are the areas in charge S _a to _{S d} shown in FIG. As shown in FIG. 16a, changing the assigned area S causes the boundary surfaces of the areas to overlap. Moreover, as shown in FIG. 16b,
The connection between the brightness calculation units 3 via the mutual communication means is as follows:
Fixed. Therefore, in the case shown in FIGS. 16a and 16b, the information about the light ray R that enters the responsible area S _a from the responsible area S _a is directly calculated from the brightness calculation unit 3 a via the mutual communication means 31. It cannot be transferred to section 3b.

Therefore, such information on the light ray R is first transferred from the brightness calculation section 3a to the brightness calculation section 3c via the mutual communication means 31. The information on the transferred light ray R is stored in the light ray information storage means 34 of the brightness calculation section 3c.

The information on the ray R stored in the ray information storage means 34 in this way is first read out by the ray information determination means 37. This light ray information determining means 37 determines that the light ray R
The z-coordinate value S _z of the starting point position coordinate is compared with the range (z _- , z ₊ ) of the assigned area S read out from the area information storage means 33 in the z direction. If S _z ≧z ₊ , information on the light ray R is transferred to the brightness calculation unit 3d connected in the z ₊ direction via the mutual communication means 31. Further, when S _z <z _- , information on the light ray R is transferred to the brightness calculation unit 3 b connected in the z _- direction via the mutual communication means 31 .

In other words, the range in the z direction of the responsible area S (z _- ,
z ₊ ) is changed, so if the information on the light ray R is transferred to the brightness calculation unit 3 connected in the z direction, the information on the light ray R can be processed by the correct brightness calculation unit 3.

FIG. 17 is an explanatory diagram showing a method of calculating the brightness of each pixel by the process of determining the intersection between a light ray and an object in the brightness calculation unit 3. As shown in FIG. 17, the information on the ray R stored in the ray information storage means 34 is as follows:
First, the light beam information is read out by the light beam information determining means 37. As a result of the determination process in this light beam information determination means 37,
Information about the ray R that has not been transferred to the brightness calculation section 3 (z _- ) or the brightness calculation section 3 (z ₊ ) is sent to the intersection determination means 39 . This intersection determination means 39 performs intersection determination processing between the light ray R and the object stored in the object information storage means 35.

Furthermore, for the ray R that does not intersect with the object, the intersection of the ray R and the boundary surface of the assigned area S is first determined. From this point of intersection, the adjacent region S' into which the light ray R will next enter is determined, and the direction in which the information of the light ray R should be transferred is determined. Information on the light ray R is transferred via the mutual communication means 31 to the brightness calculation unit 3 in the determined direction.

At this time, the starting point position coordinates of the light ray R are changed to the position coordinates of the obtained intersection point. As a result, the starting point position coordinates of the light ray R will be included in the area to which the light ray R should enter next.

The information on the light beam R transferred in this way is read out from the mutual communication means 31 and is read out from the mutual communication means 31 and
4 is stored.

Furthermore, if the ray R intersects the object, the ray R
When new light rays R' and light rays R _L are generated, information on these light rays R' and light rays R _L is stored in the light ray information storage means 34.

Such intersection determination processing is shown in each of the aforementioned papers and FIG. 2.

When the brightness I of the light ray R is determined by this process, the brightness I is added to the pixel p (i, j) of the image storage unit 4 indicated by the information about the light ray R via the communication means 32. Ru.

Through the above-described intersection determination process, each brightness calculation unit 3
The composition of the image P is completed when all the information on the light ray R stored in the light ray information storage means 34 has been processed.

Note that here, the brightness calculation unit 3 adjacent in the z direction
The explanation has been made assuming that load redistribution processing is performed only between. If the brightness calculation unit 3 adjacent in the x and y directions
When performing load redistribution processing between servers, the same process can be used to redistribute the load.

In addition, the object information storage means 35 of each brightness calculation section 3
It is also possible to store information on all objects in the area, select only the objects included in the area in charge, and perform the intersection determination process. By doing so, there is no need to transfer object information between the brightness calculation units 3 during load redistribution processing.

〔Effect of the invention〕

In the object image synthesis device of the present invention, the shape of the area in charge of each brightness calculation section is a rectangular parallelepiped, and each surface of the rectangular parallelepiped is perpendicular to the coordinate axis. Therefore, it is very easy to determine the intersection between the boundary surface of the assigned area and the light beam. Therefore, when a light beam passes through the assigned area, it is possible to determine the area to which the information of this light beam should be transferred with a much smaller amount of processing than in the past.

Further, the shape of the assigned area is changed by moving a boundary surface perpendicular to a specific coordinate axis in a direction parallel to that coordinate axis. Therefore, even if the shape of the area in charge is changed, the area in charge is kept in the shape of a rectangular parallelepiped, so the above-mentioned effects are not impaired.

Furthermore, when the shape of the assigned area is changed, information on objects included in this assigned area must be changed. Even in this case, since the area in charge is kept in the form of a rectangular parallelepiped, object information can be easily changed.

In this way, since the amount of processing required when changing the shape of the assigned area is small, it is possible to sufficiently obtain the effect of redistributing the load of each brightness calculation unit by changing the shape of the assigned area.

[Brief explanation of the drawing]

FIGS. 1a and 1b are block diagrams showing the overall configuration of an object image synthesis device according to an embodiment of the present invention, and a block diagram showing the detailed configuration of the brightness calculation section. FIG. ), FIG. 3 is a diagram showing information on the ray R that should be set when the ray R is generated, and FIG. 4 is a diagram for explaining the operation of the initial ray generating section 1. 5 is a diagram showing the object information set in the object information setting section 2, FIG. 6 is an explanatory diagram showing the circumscribed area of a sphere as an object, and FIG. 7 is a diagram showing the information set in the object information setting section 2. FIG. 8 is an explanatory diagram showing a method of combining a plurality of brightness calculation units 3 into a three-dimensional array, and FIG. 9 is a diagram showing how to allocate a space defining an object to each brightness calculation unit 3. FIGS. 10 and 11 are diagrams showing area information and a three-dimensional explanatory diagram showing the area S in charge of the brightness calculation unit 3, respectively. FIG. FIG. 13 is a block diagram showing the process of transferring information on the light ray R from the brightness calculation unit 3 to the brightness calculation unit 3. 14. Block diagram showing the process of storing object information and light source information from
Figures a and b are explanatory diagrams illustrating the comparison process of the brightness calculation unit 3 between the assigned area S and the circumscribed area of the object information, and show the cases where the assigned area S and the circumscribed area have a common part and the case where they do not, respectively. FIG. 15 is a block diagram showing the process of changing the area S in charge of the brightness calculation unit 3;
FIGS. 16a and 16b are explanatory diagrams showing a method of transferring information on the light ray R between the brightness calculation units 3, and FIG. 17 shows a method of calculating the brightness of each pixel by the intersection determination process between a light ray and an object in the brightness calculation unit 3. FIG. 1... Initial light generation section, 2... Object information setting section, 3, 3a, 3b, 3c, 3d... Brightness calculation section, 4... Image storage section, 5... Connection line, 6... Information input section , 31... Mutual communication means, 32... Communication means, 33... Area information storage means, 34... Ray information storage means, 35... Object information storage means, 36
. . . Light source information storage means, 37 . . . Light ray information determination means, 38 . . . Object information determination means, 39 . . . Intersection determination means, 43 . .

Claims (1)

[Claims] 1. Reverse the path of the light ray from the light source to the viewpoint, perform intersection determination processing between the object and the light ray, and calculate the brightness of each pixel constituting the image that should display the object. An object image synthesis device based on a ray tracing method, comprising: an initial ray generation section that generates information on a plurality of rays passing through each pixel from the viewpoint; and an object information setting section that sets information on the object; Intersection determination processing is performed between a light ray passing through a responsible region that is one of a plurality of regions generated by dividing a space defined by a plane perpendicular to the coordinate axis and the object included in the responsible region. a plurality of brightness calculation units that calculate the brightness of the pixel by calculating the brightness of the pixel, and a pixel storage unit that stores the brightness calculated by the brightness calculation unit as the pixel, and each of the brightness calculation units has a range of the area in its duty. an area information storage means for storing, a load determination means for determining respective loads, and an adjacent brightness calculation unit which is the brightness calculation unit responsible for the area adjacent to the responsible area in directions parallel to each of the coordinate axes. mutual communication means for mutual communication;
Comparing the load of the adjacent brightness calculation unit obtained through this mutual communication means with its own load, and determining whether to expand the area in charge of the adjacent brightness calculation unit in the direction of the area in charge of the expansion. The present invention is characterized by comprising a determining means, and an area expanding means for expanding the range of the responsible area stored in the area information storage means in order to expand the responsible area based on the determination by the expansion determining means. Object image synthesis device.