JP2001167289A - Device and method for processing image while including blending processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processor capable of efficiently performing rendering processing accompanied with blend processing of a translucent polygon. SOLUTION: In rendering processing of the translucent polygon, work data Xn as the cumulative multiplied values of transparency are found successively from the polygon of the front side in a display picture, and light source calculating processing for finding color data INTn from a light source or texture processing for finding color data TXCOn while utilizing a texture color or the like is parallel performed to plural polygons. Then, color synthesizing processing of plural polygons is performed at random for finding and accumulating image data FBn from the work data Xn-1, light source image data INTn, color data TXCOn and a opacity αn.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus and method for a plurality of polygons including a translucent polygon, and more particularly to a blending process utilizing transparency required for a translucent polygon and a general method for a plurality of overlapping polygons. The present invention relates to an image processing apparatus and method capable of performing efficient blending processing more efficiently and in a shorter time.

[0002]

2. Description of the Related Art In a game device or a simulation device, the positions of a plurality of polygons constituting a character or the like are determined in accordance with an operation input, and rendering processing is performed on the displayed polygons among the polygons to convert image data. Generating and displaying them is done.

[0003] Unlike a non-transparent polygon, a rendering process for a translucent polygon involves a blending process with another polygon located behind (rear side).
It is necessary to perform rendering processing including blending processing in order from the translucent polygon on the back side in the display screen.

On the other hand, rendering processing for opaque polygons is performed by performing hidden surface elimination processing using a Z value indicating the depth in the display screen, thereby imposing no restriction on the rendering order of a plurality of polygons, and thereby eliminating the overlapping part. Processing is performed. Further, when opaque polygons and translucent polygons are mixed, a method of processing opaque polygons in an arbitrary order by the hidden surface elimination processing using the Z value described above, and then processing the translucent polygons in order from the back side. General.

[0005]

However, when the number of polygons to be processed increases in the rendering processing,
The rendering process may not be completed within a predetermined time between frames. In that case, if emphasis is placed on the quality of the image processing, it is necessary to wait until the rendering processing of all the translucent polygons is completed, and the movement of the character being displayed is stopped, which is inconvenient in the game device or the simulation device. . If the movement of the character is emphasized, there are polygons that are not processed within the time between frames. In this case, since the rendering process of the translucent polygon is performed from the polygon on the far side, the process on the polygon on the near side, which is one of the most important characters in the image, is cut off.

Further, as described above, if importance is placed on the efficiency of rendering the opaque polygon and the translucent polygon, it is necessary to process the polygons separately and increase the burden of image processing. Such an increase in the load makes it difficult to complete the image processing within the time between frames as described above.

Further, the present invention is not limited to the blending process using the opacity of a semi-transparent polygon. In general blending process, the color data of the polygon on the far side of the display screen is obtained first, and the color data and the next color data are obtained. It has been proposed to perform a blending process on color data of a polygon to be processed by a predetermined source mixing count and a destination mixing count. Also in this case, it is required to perform processing in order from the polygon on the back side of the display screen, and there is the same problem as in the case of the translucent polygon.

Accordingly, the present applicant has solved the above-mentioned problem.
The invention was proposed. For example, Japanese Patent Application No. 11-225426, 1999
“Image processing equipment including blending processing” filed on August 9
And its method ". This case has not been completed at the time of filing the application.
It is public and therefore unknown. In the present invention, the semi-transparent
Rendering of light polygons in front of the display screen
Side, that is, from the viewpoint side polygon in the depth direction.
And features. Therefore, from the near side in the display screen
The color data T is assigned to the n-th polygon when arranged in order. _nAnd not
Transparency α_nIs given, the work data X_nage
And X_n= (1-α_n) ・ X_n-1 And further obtain image data D_nAs D_n= D_n-1+ T_n・ Α_n・ X_n-1 And work data X_nAnd image data D_nAsk for
The above processing is performed on all polygons. Soshi
And the last image data D_nAs image data for display
Use.

The present invention provides a novel image processing apparatus and image processing method capable of performing a rendering process at a higher speed by using the above-described image processing algorithm.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image processing apparatus and method capable of reliably performing rendering processing of a polygon on the front side within the time between frames in image processing of a translucent polygon.

Further, an object of the present invention is to provide an image processing apparatus which can perform rendering processing without discriminating both polygons in image processing of polygons in which opaque polygons and translucent polygons are mixed, and further improve processing efficiency. It is to provide the method.

Further, an object of the present invention is to provide an image processing apparatus having improved processing efficiency in general blending processing.
An object of the present invention is to provide an image processing method.

[0013]

SUMMARY OF THE INVENTION A first aspect of the present invention is as follows.
The rendering process of the translucent polygon includes light source calculation processing for obtaining work data Xn in order from the polygon on the near side in the display screen and color data INTn from the light source,
Texture processing for obtaining color data TXCOn using a texture color or the like is performed on a plurality of polygons in parallel. The work data Xn and the light source color data INT
n, color data TXCOn and opacity αn, and a color synthesis process for obtaining and accumulating image data FBn is performed for a plurality of polygons in random order.

For this purpose, pixel color data TXCO is added to the n-th polygon when arranged in order from the near side in the display screen.
_{If n} and opacity alpha _n and the light source color data INTn is given, as work data X _n, from the front side of the display screen with respect to _{X n = (1-α n} ) · X n-1 all polygons Ask in order. Furthermore, as image data FB _n, multiplication value INT _n
· TXCO _n · α _n · X _n-1 are calculated for all polygons and accumulated. That is, FB = Performing the calculation of _{Σ (INT n · TXCO n ·} α n · X n-1), which, since it is operation for adding the image data FBn of overlapping polygons, the image data F
Bn can be obtained in a random order for a plurality of polygons in the depth direction. Then, all the accumulated image data F
B is used as display image data.

In the above, the initial value X ₀ of the work data X _n is preferably 1, for example. Also, the image data FB _n
The initial value FB ₀ of, for example, 0 is preferable.

In the above image processing, the image data FB is
Pixel color data TXCOn, opacity αn, light source color data
Value INT of data INTn and work data Xn_n・ TXCO_n
・ Α _n・ X_n-1(N = 0 to j, j)
Is the number of overlapping polygons). And each polygo
Integrated value INT_n・ TXCO_n・ Α_n・ X_n-1Ask for
The operation is not affected by other polygons, so
As you can see, they can be performed in any order. And from the light source
Calculation of light source color data INTn showing the effect on rigone
Can be simple or complex depending on the type of polygon
Are mixed. Also, use a texture color, etc.
The calculation processing for cell color data TXCOn is also
In some cases, simple cases and complicated cases are mixed. Meanwhile, transparent
Work data X that is the cumulative multiplication value of degrees_n= (1-α_n)
・ X_n-1Is the work data X of another polygon_n-1Impact of
In order from the front of the screen in the depth direction
There is a need to do. However, the operation itself is simple.
And the number of processing steps is small.

[0017] In the present invention, the processing steps are relatively high light source calculation processing and texture processing is performed in parallel for a plurality of polygons, in order from the polygon these processing is completed, the accumulated value INT _n · TXCO _A color synthesizing process including a process for obtaining _n · α _n · X _n-1 and a process for accumulating them is performed. As a result, the light source calculation processing and the texture processing, which conventionally have a very heavy calculation load, can be efficiently performed in a short time.

The above work data X_nIs the near side (viewpoint
Side) multiplied by the transparency (1-α) of the n translucent polygons
Means the given value. Therefore, multiple overlapping polygons
Work data in order from the polygon on the front side
The processing for all polygons
Data X_nCan be obtained in order. Respectively
Pixel rendering data
TXCO_nIs the transparency of the n-1 polygons before it.
Work data value X multiplied by_n-1And the transparency α _nAnd the light
Source color data INT_nAccording to the calculation processing, the image data FBn
Ask for. And for each polygon
The accumulated image data FBn is accumulated (added) and
Image data FB is obtained.

According to another aspect of the present invention, in an image processing apparatus for generating image data for display by performing image processing on a plurality of overlapping polygons, the image processing apparatus is provided with a plurality of polygons arranged in order from the near side in the display screen in the depth direction. Pixel color data TXCOn for the nth polygon And opacity α _n or transparency (1
-Αn) and the light source color data INTn, and the image processing apparatus determines the transparency (1-α) of the overlapping pixel.
n) the light source color data INTn for each pixel in the depth direction according to the blend processing unit for obtaining the multiplied value _Xn in the depth direction from the near side and the light source data of the pixel.
Light source calculation unit to obtain the polygon color data PCOLn
And a texture processing unit for obtaining pixel color data TXCOn for each pixel in the depth direction according to the texture coordinates, and for each pixel in the depth direction,
A color combining unit that performs a color combining process that corrects and accumulates the pixel color data TXCOn of the pixel in accordance with the light source color data INTn, the opacity αn, and the multiplied value X _n−1 of the immediately preceding pixel. Having at least one of the light source calculation unit and the texture processing unit,
An image processing apparatus, wherein corresponding data generation processes are performed in parallel, and the color synthesis unit performs the color synthesis process in random order for pixels in a depth direction. Or, the image processing method.

Further, another aspect of the present invention is directed to overlapping overlapping
Performs image processing on a number of polygons and
In an image processing device that generates data,
N-th polygon when arranged in order from the front in the depth direction
And pixel color data TXCOn And light source color data INTn,
Destination mixing coefficient DE _nAnd source mixing coefficient
SR_nAnd the image processing apparatus
Is the source mixing coefficient SR of the overlapping pixels_nTo the power of
Arithmetic value X_nFrom the near side in the depth direction.
Command processing unit and the light source data of the pixel,
Obtain the light source color data INTn for each pixel in the depth direction
Light source calculation unit, polygon color data PCOLn and
According to the texture coordinates, each pixel in the depth direction is
Texture processing to find the pixel color data TXCOn
For each pixel in the depth direction,
Xel pixel color data TXCOn and light source color data INT
n, destination mixing coefficient DE_n, And one near side
Multiplied value X of the pixel_n-1Correct and accumulate according to
A color synthesis unit for performing color synthesis processing,
Light source calculation unit and texture processing unit
At least one performs the corresponding data generation processing in parallel,
The color synthesizing unit performs the color synthesizing process in a depth
The feature is that the order of pixels in the direction is performed in any order
An image processing device. The image processing method is also described.

[0021]

Embodiments of the present invention will be described below with reference to the drawings. However, the technical scope of the present invention is not limited to the embodiment.

First, a blending processing method proposed in Japanese Patent Application No. 11-225426 will be described.

FIG. 1 is a diagram for explaining an example of overlapping of translucent polygons. In this example, three translucent polygons P
1, P2, P3 and one opaque polygon P4 are sequentially overlapped from the near side in the depth direction (Z value) in the display screen. Therefore, the image data C ₄ of the area C ₄ is the color data T ₄ formed according to the environment light or the like from the texture data of the opaque polygon P ₄ .

In the present specification, the polygon color data T
Includes, for example, the color data of the polygon itself, the texture color data including the pattern, and the color data after shading the texture color data and the color data of the polygon itself. Any one of these color data will be referred to as polygon color data T. The color data obtained by the blending process or the like is image data C used for display. Therefore, the image data C is also a kind of color data.

That is, C ₄ = T ₄ (1) In the area C 3, since the translucent polygon P 3 overlaps the opaque polygon P 4, the image data is the opacity α ₃ (α ₃ = 1, opaque, α
₃ = 0 transparent, 1> α ₃ > 0 translucent) and color data T
_{According to 3} , the color data is obtained by blending the color data of the polygons P4 and P3. That is, according to the conventional general rendering method, color data of the area C3 _{is, C 3 = α 3 T 3} + (1-α 3) obtained in C ₄ (2).

Similarly, the image data C ₂ of the area C ₂ is also obtained from the opacity α ₂ and color data T _{2 of the} translucent polygon P ₂ and the color data C _{3 of the} area C ₃ as C ₂ = α ₂ T ₂ + (1 −α ₂ ) C ₃ (3), and the image data C ₁ of the area C ₁ is also obtained by C ₁ = α ₁ T ₁ + (1−α ₁ ) C ₂ (4).

[0027] As described above, the rendering process for the semi-transparent polygons, in order to obtain the image data C _n, in the region where the overlapping with another polygon, more image data C _{n + 1} on the back side with reference recursively Blending must be performed. Therefore, it is general to perform polygon rendering processing sequentially from the back side in the display screen.

Therefore, the invention proposed in the prior application is characterized by performing polygon rendering processing in order from the near side in the display screen in order to solve the above-mentioned conventional problems. According to the present invention, the above equations (2), (3),
From (4), when C ₁ is opened, C ₁ = α ₁ T ₁ + (1−α ₁ ) ｛α ₂ T ₂ + (1−
α ₂ ) [α ₃ T ₃ + (1−α ₃ ) C ₄ ]｝. Further, T _n can be summarized as follows: C ₁ = T ₁ α ₁ + T ₂ α ₂ (1−α ₁ ) + T ₃ α ₃ (1−α ₁ ) (1−α ₂ ) + C ₄ (1−α ₁ ) (1−α ₂ ) (1−α ₃ ) (5) Here, the polygon P4 is an opaque polygon, and the opacity data α ₄ = 1, and any work data X _n
As, and _{X n = (1-α 1} ) (1-α 2) .... (1-α n) = X n-1 (1-α n) defined as (6). Note that X ₀ = 1.

[0029] When inserting the above expression (6) into equation _{(5), C 1 = T} 1 α 1 X 0 + T 2 α 2 X 1 + T 3 α 3 X 2 + C 4 X 3 = T 1 α 1 X 0 + T _{2 α 2 X 1 + T 3} α 3 X 2 + T 4 α 4 X 3 next, focusing on each of _{T n α n X n-1} , FB n = FB n-1 + T n α n X n-1 ( 7) However, FB ₀ = 0. As a result, when n = 4, FB ₄ = C ₁ holds.

That is, the same processing as the conventional blending processing is performed by using the above equations (6) and (7), so that the color data _Tn and its color data Tn are sequentially converted into polygon data from the near side in the display screen. By giving the transparency α _n and performing arithmetic processing on all overlapping polygons, the finally obtained image data FB _n becomes the same as the image data C ₁ of the area C 1 in FIG. Thus, by utilizing the work data X _n newly defined, the image data FB _n = FB <sub>n-1
+ T n α n X n-</sub> 1 can be obtained. As is clear from the indices of these equations, it is not necessary to recursively call the data of the polygon on the far side in these operations.

FIG. 2 is a chart showing an arithmetic expression when the rendering processing of the present invention utilizing the above-described arithmetic operation is applied to the example of FIG. FIG. 2A shows the case of the invention of the prior application, and FIG. 2B shows the case of the conventional example.

As is clear from the table of FIG. 2, according to the conventional rendering processing, a semi-transparent polygon is
The image data C ₄ of the area C ₄ is obtained by using the color data T _{4 of the} back side polygon P _4, the opacity α ₄ and the initial value C ₅ , and further, the color data T ₄ of the immediately preceding polygon P ₃ is obtained. ₃ and using the opacity alpha ₃ and the initial value C _4, the image data C ₃ region C3 is determined. Further, the image data C ₁ of the image data C ₂ and the region C1 of the region C2 are determined sequentially.

On the other hand, according to the invention of the prior application, the color data T ₁ and the opacity α _{1 of} the polygon P1 located first in the foreground by using the above equations (6) and (7).
And the initial image values X ₀ and FB ₀ , and intermediate image data F
B ₁ is obtained, and further, the color data T ₂ and the opacity α ₂ of the polygon P _{2 on} the back side, and the intermediate image data FB
_1, utilizing the image data FB ₂ in the following way is required. Further, similarly, the next intermediate image data FB ₃ , F
B ₄ is required. Then, the final image data FB ₄ the processing for data of all polygons ends, the region C
It is used as _one image data C1.

Therefore, according to the invention of the prior application, the rendering process for the translucent polygon can be performed one after another by using the data (color data T and opacity data α) for the front polygon. Further, as will be apparent from a specific example described later, in the rendering processing for the polygon on the far side, the work data X _n-1 tends to be small, and the halfway obtained using the work data X _n-1 They tend to also change the image data FB _n decreases. Therefore, when the work data X _{n- 1} is reduced to some extent or becomes 0 to some extent, even if the subsequent rendering processing for the data of the back side polygon is omitted,
It is understood that there is almost no effect on the displayed image.

Further, in the above-described calculation of the rendering processing, if the opacity α _n is simply set to 1 (opaque) for the opaque polygon, the above-described calculation can be directly applied to the processing. That is, when the opacity α _n = 1, the work data X _n = 0, and there is no change in the rendering result of the polygons on the back side, so that rendering on those back side polygons is substantially unnecessary. Become. Therefore, by utilizing the rendering process of the invention of the prior application, even if translucent polygons and opaque polygons are mixed, translucent and opaque can be processed in order from the polygon on the nearer side.

The work data _Xn described above physically means a value obtained by cumulatively multiplying the transparency (1-α) when n polygons are overlapped. That is, as is apparent from the above equation (6), the work data _Xn is a value obtained by multiplying all the transparency (1−α) of each polygon. That is, when a polygon having high transparency, for example, a polygon in which (1-α) is close to 1, is superimposed, when adding the color data Tn of the _n-th polygon located on the back side thereof, The image data FB _n-1 is affected according to the cumulative multiplication value of the transparency (1-α) of the n-1 overlapping polygons. Then, desirably, color data T _n according to the opacity α _n of the _n-th polygon is added. This is because if the nth polygon itself is a transparent polygon (1−α _n = 1), the color data T _n should not be added.

As described above, the rendering processing by the above equation (7) FB _n = FB _n-1 + T _n α _n X _n-1 firstly involves the color data T of the polygon on the back side to be processed. _n is the previous n-
According to the value X _n-1 that is the accumulated transparency of one polygon,
Image data FB in which n-1 polygons in the foreground are superimposed
_This is processing to be added to _n-1 .

Second, in the rendering processing by the above equation (7), the color data _Tn is added according to the transparency of the polygon being processed. That is, the component of the color data T _n corresponding to the opacity alpha _n of the polygon being processed, it than in accordance with the accumulated value X _n-1 of the transparency of the front of the n-1 single polygon, n-1 This means that it is added to the image data FB _n−1 of the polygons that are overlapped.

Then, the accumulated value of transparency X_nInitial value X of₀
= 1 means that before the first polygon, another
There is no gon, meaning the transparency is maximum
I do. Therefore, the color data T of the first polygon₁Is that
As-is image data FB₁Used as However, the first
The opacity α of the polygon itself₁Color data T according to₁Success
Minute is image data FB₁Used as The image data
TA FB_nInitial value FB of ₀= 0 means that the luminance value of the color is 0
Is a black state without any color components
Means Initial value FB of image data₀Is not necessarily true
It doesn't have to be black.

The image processing method proposed in the prior application has been described above. This image processing complies with the following arithmetic expression.

FB _n = FB _n-1 + T _n α _n X _n-1 (7) X _n = X _n-1 (1-α _n ) (6) Therefore, by rewriting the above equation (7), FB _{n = Σ (T n α n} X n-1) may be a (8). This is a cumulative value obtained by adding the products of the opacity α _{n of} each polygon, the polygon color data T _n, and the work data X _n−1 in the depth direction in the display screen.

Here, the color data Tn of the polygon is color data that reflects the influence of the color of the light source such as ambient light on the color or texture color of the polygon itself. This will be described in more detail below. The image data is the polygon's own color PCOL given to the polygon's vertex data.
(Hereinafter referred to as polygon color data PCOL) and texture color data
Color data obtained from TX and TXCO (hereinafter referred to as pixel color data
TXCO) plus the effect of the color data INT from the light source.

For the color data of the light source, for example, yellow
The effect of diffused light from a
It can be represented by yellow color data INT with degrees. Some
Is affected by ambient light from the surrounding red wallpaper
Can be represented by red color data INT with a predetermined intensity.
it can. And the silver metal fabric of polygon
The effect of peculiar light is a silver color with a certain intensity
It can also be represented by data INT. Therefore, those light sources
Here, as the color data from
Collectively. As a result, the color data Tn of equation (7)
Tn = TXCO_n・ INT_n Can be replaced by That is, X_n= X_n-1・ (1-α_n(6) FB_n= FB_n-1+ INT_n・ TXCO_n・ Α_n・ X_n-1 (7) FB_n= Σ (INT_n・ TXCO_n・ Α_n・ X_n-1(8) As is clear from the above equation, the image data FBn is
Transparency for the previous polygon for each polygon
Work data Xn-1 which is the product of degrees (1-αn) and the
Rigon's opacity αn, light source color data INTn, pixel
Color data TXCOn in the frame buffer
By adding to the image data FB stored in the
Can be Moreover, INT in each polygon_n・
TXCO_n・ Α _n・ X_n-1The calculation for
If found, the result of the calculation of other overlapping polygons
Does not depend. Therefore, for overlapping polygons,
INT for each polygon above_n・ TXCO_n・ Α_n・ X_n-1Ask for
Can be performed in any order.

When real-time image processing is performed in a game or simulation, the light source color data IN
In the calculation processing for obtaining Tn, it is necessary to perform calculation processing for many light sources in some cases, and the calculation processing requires a very high number of steps. Similarly, in texture processing for obtaining pixel color data TXCO obtained from the polygon color data PCOL, which is the own color of the polygon, and the texture color data TX, a polygon having no texture, a polygon having a plurality of texture data, There are polygons and the like that require trilinear filter processing for obtaining a texture corresponding to an intermediate subdivision value from a plurality of textures provided corresponding to the subdivision values LOD. In some cases, an operation process that requires man-hours is required.

Therefore, in a preferred embodiment of the present invention, the light source calculation processing and the texture processing are performed on a plurality of polygons in parallel. For this purpose, a plurality of light source calculation circuits and texture processing circuits are provided in the image processing apparatus. The processing efficiency of a plurality of polygons is increased by sequentially distributing the processing to the empty processing circuits. Thereby, the entire image processing time can be shortened.

[Image Processing Apparatus] FIG. 3 is a configuration diagram of a game apparatus having a rendering processor according to an embodiment of the present invention. The rendering processor 20 is an image processing device, and is provided in a game device or a simulation device. The game device shown in FIG. 3 includes a main CPU 10, a program ROM storing a game program, a work RAM 14 used when executing the game program, and a rendering processor 20 via an internal bus 16. Connected.

The main CPU 10 executes a game program and performs necessary image processing in response to an operation input from an operator (not shown). In the example of FIG. 3, it is assumed that geometry processing such as matrix calculation and perspective transformation for moving an object is performed by the main CPU. Of course, this geometry processing may be performed by a dedicated processing processor provided separately.

The rendering processor 20 is a rendering command memory 22 for storing rendering commands issued by the main CPU 10 for instructing rendering.
And a setup block 24 for interpreting the drawing command. The setup block 24 gives the polygon data and the light source data in the display screen to the sorter unit 26.

The polygon data has vertex data. Each vertex data includes, for example, two-dimensional coordinate data (x, y) in the display screen, a Z value indicating depth (depth), and color data PCO.
L, opacity (or transparency) αp, normal vector data (Nx, Ny, Nz), texture coordinates (Tx, Ty), and the like.

The sorter unit 26 rearranges the polygons in the depth direction in accordance with the Z value (depth value), which is the distance from the viewpoint of the supplied polygon data, and records the polygons in a memory in the unit. Accordingly, data on a plurality of polygons in the frame is temporarily stored in the sorter unit 26. Then, the sorter unit 26 supplies the opacity data αp of the polygon to the blend processing unit 30 in the order of the depth direction for each pixel of the display screen.
In the blending unit, the texture opacity αt
In the case of a polygon using, the process of obtaining work data Xn is performed according to the pixel opacity αn obtained by combining the opacity αp obtained from the vertex data of the polygon and the opacity αt of the texture. In order to read the opacity αt of the texture from the texture memory 28, the texture coordinates of the pixel are stored in the sorter unit 26.
Need to be supplied from

In the case of a polygon that does not use the texture opacity αt, the work data Xn is obtained by using the opacity αp obtained from the vertex data of the polygon as the opacity αn of the pixel. The process for finding work data is
Specifically, X _n = X _n-1 (1-α _n ) in the above equation (6)
This is a process for performing an operation according to.

The texture memory 28 stores texture opacity αt and texture color data T. In the texture memory 28, the texture data is directly downloaded from the program ROM 12,
Will be saved. Further, the texture memory 28 can be provided in a storage area in the work RAM 14.

As described above, the blend processing unit 30 generates work data Xn and opacity αn for each pixel and in the depth direction and supplies them to the pixel work memory unit 36. Work data Xn and opacity αn
It is recorded in the pixel work memory in the pixel work memory unit 36.

The texture processing unit 32 outputs the texture coordinates (Tx, Tx) from the sorter unit 26 for each pixel.
y), polygon color data PCOL, and three-dimensional coordinates (X,
Y, Z) values, etc., and according to the texture coordinates,
The texture color data T is read from the texture memory 28, and the polygon color data PCOL, which is the color of the background of the polygon, and the texture color data T are combined, and the pixel color data TX
Perform processing to obtain COn. This process need not always be performed in the depth direction of the display screen. A plurality of processing circuits are provided in the texture processing unit.
Processing for a plurality of pixels is sequentially distributed to vacant processing circuits and performed. As a result, the pixel color data
TXCOn is generated in any order. This pixel color data T
XCOn is also recorded in the memory in the pixel work memory unit 36.

The light source calculation unit 34 is supplied with three-dimensional coordinate values (X, Y, Z) and normal vectors (Nx, Ny, Nz) for each pixel from the sorter unit 26 together with the light source data. Then, color data INTn from the light source is obtained.
This processing does not need to be performed in the depth direction of the display screen. A plurality of processing circuits are provided in the light source calculation unit, and the processing for the plurality of pixels is sequentially distributed to the vacant processing circuits and performed. As a result, the light source color data INTn of the pixel is generated in any order. The light source color data INTn is also stored in the pixel work memory unit 3.
6 is stored in the memory.

The light source color data is generated, for example, in the case of diffused light, by multiplying the color data of the light source by the inner product of the luminance vector of the light source and the normal vector of the pixel.

In the pixel work memory unit 36,
When the work data Xn, the opacity αn, the pixel color data TXCOn, and the light source color data INTn are all stored for the n-th pixel in the depth direction, those data are supplied to the color synthesis unit 38. The color synthesizing unit 38 multiplies the data, adds the data to the image data FB in the frame buffer 40, and overwrites the newly obtained image data FB in the frame buffer 40. This image data FB is color data obtained by subjecting pixel color data to light source processing and blend processing.

When the above-described color synthesizing process is performed for all the pixels, 1 is stored in the frame buffer 40.
The image data FB used for displaying the frame is stored. Then, the image data FB in the frame buffer 40 is converted into image data in a form required for the display device 44 and supplied to the display device 44.

The above is the description of the rendering processor 20.
It is a flow of processing inside. Here, only the blend processing unit 30 receives the pixel data in the order in the depth direction and sequentially performs processing for obtaining the work data Xn. The texture processing unit 32 and the light source calculation unit 3
In No. 4, since each arithmetic processing is very heavy, a plurality of processing circuits are provided and processed in parallel. In addition, the overall processing speed can be increased by sequentially distributing processing to unused processing circuits for a plurality of processing circuits. However, by performing such processing, the generation processing of the pixel color data TXCOn and the light source color data INTn is completed before and after in the depth direction. However, the color synthesizing unit 38 can perform the color synthesizing processing in order from the pixel on which the data generation processing is completed, so that there is no problem.

Next, the configurations of the sorter unit 26, the blend processing unit 30, the texture processing unit 32, the light source calculation unit 34, the pixel work memory unit 36, and the color composition unit 38 will be described.

FIG. 4 is a configuration diagram of the sorter unit 26. The sorter unit 26 is supplied with polygon data PG0 to PGj composed of vertex data, and generates a pixel data in each polygon by a pixel generation unit 50, and pixel data PX0,0 to PX generated by the pixel generation unit 50.
i and j are rearranged in the depth direction for each pixel,
It has pixel sorters 52, 54 and 56 for storing.

It is desirable that the number of pixel sorters be provided by the number of pixels on the display screen. However, since the number of pixels on the screen becomes enormous, for example, 640 × 480, it is practical to provide a pixel sorter for a part of the screen, for example, 8 × 8 = 64 pixels. In that case, i = 64 pixels. Alternatively, in an extreme example, a pixel sorter having only one pixel may be used.

As shown in FIG. 4, each pixel data PXm, n is obtained by interpolation from, for example, coordinate values (X, Y, Z), polygon color data PCOL, which is the color of the background of the polygon, and vertex data of the polygon. Obtained opacity αp, texture coordinates (Tx, Ty), and normal vector (Nx, Ny, Nz)
Having. Various methods have been proposed for an algorithm for generating such pixel data from polygon data composed of vertex data. Most commonly, a raster scan method is used. According to the raster scanning method, each pixel data scanned in the X-axis and Y-axis directions in a polygon is obtained by performing an interpolation operation on the vertex data of the polygon. Also, pixel data can be obtained by using the fractal method. The fractal method is disclosed in Japanese Patent Application Laid-Open No. H11-144074 filed separately by the present applicant.

According to the sorter unit of FIG. 4, the pixel data PXm, n for each pixel is arranged in order in the depth direction based on the Z value.

FIG. 5 is a configuration diagram of another sorter unit. This sorter unit 26 stores the polygon data P
G0 to PGj are sorted according to the Z value, rearranged in the depth direction, and stored in the polygon sorter 58. Then, the rearranged polygon data is supplied to the pixel generation unit 60 in order from the near side in the depth direction. The pixel generation unit 60 calculates
Pixel data PX0,0 to PXi, j in the polygon are generated by the above-described raster scan method or fractal method. Accordingly, the pixel data is supplied to the blend processing unit 30, the texture processing unit 32, and the light source calculation unit 34 in order from the near side in the depth direction.

FIG. 6 is a block diagram of the blend processing unit. The blend processing unit 30 shown here includes:
The opacity (transparency) αp obtained from the vertex data of the polygon and the texture coordinates are sequentially supplied along the depth direction. For example, the opacity data α of the pixel data PXm, 0 (m = 0 to i) sorted to the front side
p is supplied, and the opacity data αt of the texture corresponding to the pixel is read from the texture memory 28. The transparency processing circuit 301 obtains, for example, the opacity αn of the pixel synthesized by integrating these opacity data and other predetermined processing. If there is no texture opacity data, the polygon opacity data αp becomes the pixel opacity αn as it is.

The blend processing unit 30 further includes a multiplier 302, a subtractor 303, and an X data work memory 304.
And the work data Xn-1 of the immediately preceding pixel is read from the X data work memory 304,
It is stored in the nth area of the corresponding pixel in the pixel work memory unit 36. Then, from the read work data Xn-1 and the opacity αn, the work data Xn = Xn-1 (1-α) of the n-th pixel being processed.
n) is required. The obtained work data Xn is updated in the work data memory 304. Also, the opacity αn is stored in the nth area of the corresponding pixel in the pixel work memory unit 36.

FIG. 17 is a diagram showing a modification of the blend processing unit 30. This example is different from the blend processing unit 30 in FIG. 6 in that a multiplier 305 for multiplying the work data Xn−1 by the opacity αn is added. Therefore, the product Xn-1 ·
αn is stored in the pixel work memory unit 36 in the nth area of the corresponding pixel. By providing such a multiplier 305, the configurations of a pixel work memory unit and a color synthesizing unit described later can be simplified.

In the blend processing unit 30, as described above, the work data Xn for the foremost pixel is defined by the opacity αn and the previous work data Xn−
It is obtained in order from 1. Since this arithmetic processing is performed by a multiplier, a subtractor, and the like, and is relatively light processing, the arithmetic operation for all the pixels in the depth direction can be completed in a relatively short time.

If necessary, a plurality of blend processing units 30 can be provided, and each unit can be parallelized in correspondence with display screen (screen) coordinates. Even in the case of such a configuration, each processing unit performs an operation for sequentially obtaining work data for the corresponding pixel from the near side in the depth direction.

FIG. 7 is a configuration diagram of the light source calculation unit. The light source calculation unit 34 includes a plurality of light source calculation circuits 342, 343, and 344 that perform light source calculation. In the example of FIG. 7, three light source calculation circuits are provided, and the light source calculation can be processed in parallel. The light source calculation unit 34 is further supplied with the light source data of the pixels from the sorter unit, and outputs the light source calculation circuits 342, 343, and 34 in an empty state.
4 is a calculation instruction distribution circuit 34 for sequentially distributing the light source calculation processing.
1 is provided. Further, the light source calculation unit 34 is provided with an operation result output circuit 345 that supplies the light source color data INT generated by the light source calculation circuit to the pixel work memory unit 36 in the order of generation.

FIG. 8 is an operation timing chart of the light source calculation circuit. Here, six pixel data PX1
The timing of the operation of the light source calculation circuit for .about.PX6 is shown. In the example of FIG. 8, it is assumed that the light source calculation processing for the pixel data PX1 requires six clocks. Similarly, the number of processing clocks of the pixel data PX2 to PX6 is 2, 3, 7,
Assume 4,2. Then, from the sorter unit 26, light source data, coordinate values (X, Y, Z) and normal vectors (Nx, Ny, Nz) are obtained as pixel data necessary for light source calculation.
Are supplied in order.

In the above case, the operation instruction distribution circuit 341
The processing is sequentially distributed to the empty light source calculation circuits. As shown in FIG. 8, first, in the cycle of clock 1, the arithmetic processing on the pixel data PX1 is performed by the calculation circuit 1
Assigned to. Here, it is assumed that one clock cycle is required since the pixel data is supplied via the bus 346 in the unit in order to allocate the processing.
The calculation for the pixel data PX1 is completed in the calculation circuit 1 in six clock cycles.

Next, the processing of the supplied pixel data PX 2 is assigned to the light source calculation circuit 2. Since the allocation process using the bus 346 cannot overlap the pixel data PX1, the allocation is performed in the cycle of the clock 2, and the calculation circuit 2 requires two cycles to complete the calculation. The next pixel data PX3 is calculated by the light source calculation circuit 3
Assigned to. Then, the process is completed in three clock cycles.

The processing of the pixel data PX4 supplied thereafter is assigned to the light source calculation circuit 2 in an empty state at the clock 4. Then, the process is completed in seven clock cycles. Similarly, the pixel data PX5 is supplied to the light source calculation circuit 3 which becomes empty at the clock 6. The pixel data PX6 is supplied to the light source calculation circuit 1 which becomes empty at the clock 7.

As is clear from FIG. 8, all calculations for the light source calculation for six pixel data are completed at clock 10. When the light source calculation for this pixel data is executed serially, a total of 6 + 2 + 3 + 7 + 4 + 2 = 24 clock cycles are required. In comparison, the processing can be completed twice or more times. This has been achieved by providing a plurality of light source calculation circuits and distributing the light source calculation for the plurality of pixel data in parallel and sequentially to the empty calculation circuits. However, the calculation results for each pixel data are completed in any order.

FIG. 9 is a block diagram of the texture processing unit. This configuration is equivalent to the light source calculation unit,
A plurality of texture processing circuits 322, 323, and 324 are provided, and texture processing is sequentially assigned to processing circuits in an empty state, as in the case of the light source calculation unit.
The calculation instruction distribution circuit 321 distributes pixel data necessary for texture processing to three processing circuits. Each texture processing circuit reads out the texture color data T from the texture memory 28 in accordance with the texture coordinates (Tx, Ty), and supplies the background polygon color data PCOL of the supplied polygon.
To generate pixel color data TXCOn.

In this texture processing, the processing cycle varies from pixel to pixel.
Of the 2,323,324, an empty circuit performs texture processing on a plurality of pixels. The operation timing is the same as the algorithm described in FIG. 8 in the light source calculation. Therefore, the pixel color data TXCOn is also generated in any order and stored in the pixel work memory unit 36.

It is desirable that pixel data corresponding to a common polygon be continuously supplied to the light source calculation unit 34 and the texture processing unit 32. This is because if pixel data is for the same polygon, common processing may be included, and the processing efficiency of each unit can be increased.

FIG. 10 is a configuration diagram of the pixel work memory unit. Pixel work memory unit 36
The work data Xn supplied from the blend processing unit 30, the opacity αn, the pixel color data TXCOn in the polygon supplied from the texture processing unit 32, and the light source color data INTn supplied from the light source calculation unit 34 Pixel work memory 362 that records each time
Having. FIG. 10 shows the pixel work memory 362
Is configured by a matrix-shaped memory area for storing the calculated pixel data EPX0,0 to EPXi, j. As shown in FIG. 10, the pixel data EPXm, n includes the work data Xn-1, the opacity αn, the pixel color data TXCOn, the light source color data INTn, and four pieces of data indicating that those data are stored. It has flags F1 to F4. In the pixel data EPXm, n, m distinguishes pixels on the display screen, and n distinguishes pixels in the depth direction of the display screen. However, n does not necessarily have to be in the order in the depth direction.

The pixel work memory unit 36 further has a pixel work memory controller 361 for recording the supplied calculated pixel data in each memory area. When the calculated pixel data is stored, the pixel work memory controller 361 sets flags F1 to F4 corresponding to the stored data,
The state is changed from 0 in the unstored state to 1 in the stored state. Furthermore,
The pixel work memory controller 361 sequentially supplies the pixel data in which all of these four flags have become 1 to the color synthesizing unit 38.

The blend processing unit 30 obtains the work data Xn and the opacity αn in the depth direction from the near side and supplies them to the pixel work memory unit 36. Therefore, the flags F1 and F2 corresponding to the above two pieces of pixel data are arranged in the order of depth from the foreground, that is, EPXm, 0, EPXm, 1, EPXm, 2. . . The storage state is changed in the order of EPXm, j. In addition, the pixel color data TXCOn and the light source color data INTn are obtained by the texture processing unit 32 and the light source calculation unit 34, respectively. The timing is not always the order in the depth direction. As a result, the timing at which all the four data are stored in the pixel work memory 362 is also random, depending on the pixel.

Therefore, the pixel work memory controller 361 monitors the four flags of each pixel data, and determines from the pixel data EPX that has been all stored,
It is supplied to the color synthesizing unit 38.

FIG. 11 is a configuration diagram of the color synthesizing unit. The color synthesizer 38 includes a multiplier 381 and an adder 38.
And 2. The multiplier 381 outputs the pixel data EPX supplied from the pixel work memory controller 361.
m, n work data Xn-1, opacity αn, pixel color data TXCOn, and light source color data INTn are multiplied. Adder 38
2 adds the image data FB corresponding to the pixel being processed in the frame buffer 40 and the result of the multiplication, and adds the newly generated image data FB to the frame buffer 4.
0 is stored in the same pixel area. This arithmetic processing is not always performed in the depth direction in an order, but is performed in any order. The color synthesizing unit 38 performs the above-described operation on all pixel data EPXm, n in the depth direction in the same pixel, thereby displaying the display image data FB.
Can be stored in the frame buffer 40.

When the modification of the blend processing unit shown in FIG. 16 is used, a multiplier 381 of the color synthesizing unit 38 is supplied with a product value of the work data Xn and the opacity αn. Therefore, the configuration of the multiplier 381 can be simplified accordingly.

Returning to FIG. 3, when the final image data FB for all pixels is stored in the frame buffer 40, the image data FB is supplied to the display device 44 via the display control unit 42, Is displayed.

FIG. 12 is a processing flowchart of the above-mentioned rendering processor. The setup block 24 decodes the drawing command and supplies polygon data to the sorter unit 26 (S10). The sorter unit 26 decomposes the polygon into pixels,
It is supplied to the pixel sorters 52, 54, 56 (S1
2). The pixel sorter is used to determine the distance from the viewpoint (depth, Z
The pixel data is rearranged according to the value (S1).
4). After that, the sorter unit 26 sorts the pixel data into the blend processing unit 3
0, texture processing unit 32, light source calculation unit 3
4 (S16).

In each processing unit, as described above, the generation of the work data Xn (S18), the pixel color data TXCO
n (S20) and light source color data INTn (S20).
Perform 22). The blend processing unit 30 needs to obtain work data in order from the previous pixel,
Texture processing unit 32 and light source calculation unit 34
Is to sequentially use empty circuits of a plurality of processing circuits,
Generate data in any order. Blend processing unit 30
Obtains work data in order from the pixel in front of it. When work data is obtained in the depth direction up to an opaque pixel having an opacity αn of 1, the subsequent work data Xn is 0. finish. Therefore, the blend processing unit does not always repeat the processing for all pixels in the depth direction.

[0089] When the generation of the data for each pixel is completed, the color synthesis unit, the operation already pixel data, color data _{INT n · TXCO n · α n} · X n , which is a light source processing and blending processing of each pixel _-1 is calculated, added to the image data in the frame buffer, and stored again (S24).

The above steps S16 to S24 are repeated until all the pixel data in the depth direction of each pixel are completed and all the pixels are completed. When the processing is completed for all the pixels, display is performed using the image data stored in the frame buffer (S26). Steps S10 to S26
Is repeated for each frame.

[Application to General-purpose Blending Process] In the above-described embodiment, when the color of a semi-transparent polygon and the color of a polygon located on the back side thereof are mixed and blended with an opacity α, a front-end blending process is performed. The example in which processing is performed in order from the polygon on the side has been described. However, a general computer-based graphic library can perform various blending processes as standardized by, for example, an open GL (Graphic Library). In the generalized blending process, the color data FBm-1 in the frame buffer and the color data PIXm of the pixel being processed this time, which have been obtained by the previous processing, are combined with the source mixed count SRm and the destination mixed count Dm, respectively.
Em and Blending. That is, when expressed by an arithmetic expression, the color data FBm obtained in the current processing and stored in the frame buffer is FBm = PIXm * DEm + FBm-1 * SRm (8) Here, m is an integer, and the smaller the value, the earlier the processing is performed (* indicates a product; the same applies hereinafter).

In the above blending operation equation, the source mixing count SRm and the destination mixing count D
Em is appropriately selected from the following 11 types.

0, FBn-1, PIXn, PIXn * α, FBn-1 * α,
1, 1-FBn-1, 1-PIXn, 1-PIXn * α, 1-FBn-1 *
α, (FB1-n * α, 1−PIXn * α), whichever is smaller As understood from the above equation (8), general blending processing requires color data of a frame buffer before drawing. And the calculation result also depends on the drawing order.
Therefore, in general computer graphics, polygons that require blending processing are usually drawn in order from the back side.

FIG. 13 is a diagram showing the blending processing of translucent polygons. Translucent polygon P on the near side
Gm and the polygon PGm-1 on the back side partially overlap,
This is a case where the opacity of the front-side semi-transparent polygon PGm is α. In this case, as described in the conventional example, the color data FBm-1 of the polygon PGm-1 on the far side is obtained first, and the color data of the semi-transparent polygon PGm on the near side is calculated based on the opacity αm and the transparency (1 -Αm) to perform a blending process. That is, the arithmetic expression is as shown in the figure: FBm = PIXm * αm + FBm-1 * (1-αm) (9) That is, the above equation (9) is obtained by adding the destination mixing coefficient DEm = αm, source mixing coefficient
This corresponds to the case where SRm = (1−αm).

FIG. 14 is a diagram showing a specific example of the auxiliary filter processing. A general-purpose graphic library includes auxiliary filter processing as blend processing. This process is different from the translucent polygon blending process of FIG. 13 in that the color data of the back polygon PGm-1 is left without being attenuated regardless of the blend coefficient αm of the front polygon PGm. It is.

For example, it is assumed that the polygon PGm on the near side is a polygon corresponding to a flame and the polygon PGm-1 on the far side is a predetermined opaque polygon. In this case, from the viewpoint, the opaque polygon PG passes through the flame polygon PGm.
You are looking at m-1. In that case, the flame polygon P
Although the color of Gm is added, the color data of the polygon PGm-1 on the far side is not attenuated.

In that case, the color data of the flame polygon PGm
Destination mixing coefficient DEm = α for PIXm
By setting m and the source mixing coefficient SRm = 1, the above general formula (8) can be used. As a result, the arithmetic expression of the auxiliary filter processing is FBm = PIXm * αm + FBm-1 * 1 (10) as shown in the figure. That is, the color data FBm-1 of the polygon PGm-1 obtained in the previous process is left as it is, and the color data PIXm of the polygon PGm currently being processed is added by the mixing coefficient αm.

In this auxiliary filter processing, for example, in addition to the above-described flame polygon, light from a red light source
The same processing is applied when m-1 is irradiated.

FIG. 15 is a diagram showing the color filter processing. In the color filter processing, for example, in order to extract only a specific color component of the polygon PGm-1 in the background, a color filter polygon PGm having no color component itself is arranged on the near side. In this blending operation, the color data FBm-1 of the background polygon PGm-1 is multiplied by the color extraction data PIXm of the color filter polygon PGm in order to extract only a specific color component.

That is, the arithmetic expression is FBm = PIXm * 0 + FBm-1 * PIXm (11) as shown. This equation is obtained by comparing the general equation (8) with the destination mixing coefficient DEm = 0 for the color extraction data PIXm of the color filter polygon PGm, and the source mixing coefficient SRm = for the previously obtained color data FBm-1. It is obtained by setting it to PIXm.

As described above, in addition to the blending of semi-transparent polygons, various blending processes such as a complementary color filter process and a color filter process are performed by adding the appropriately selected destination mixing coefficients to the general formula (8).
The arithmetic processing can be performed by applying DEm and the source mixing coefficient SRm. A general computer graphic processing device is configured to be applicable to the general formula (8),
As a result, various blending processes can be performed by appropriately selecting the destination mixing coefficient DEm and the source mixing coefficient SRm.

Therefore, a method and a processing device which can sequentially process polygons from the viewpoint side (front side) in the general blending processing will be described below.

As in the case of FIG. 1, it is assumed that four polygons overlap from the near side. In this case, the following four arithmetic expressions are derived from the general expression (8) by m = 1 to 4.

FB ₁ = PIX ₁ * DE ₁ + FB ₀ * SR ₁ (21) FB ₂ = PIX ₂ * DE ₂ + FB ₁ * SR ₂ (22) FB ₃ = PIX ₃ * DE ₃ + FB ₂ * SR ₃ (23 ) FB ₄ = PIX ₄ * DE ₄ + FB ₃ * SR ₄ (24) Here, it should be noted that the blending process is
Since the back side polygon is processed first and its color data is recorded in the frame buffer, a larger m means a front side polygon, and a smaller m means a back side polygon.

Therefore, for the above (21) to (24), the color data of the foremost polygon is changed from FB ₄ to D ₁ and the farthest side is obtained, as performed by the equations (5), (6) and (7). replacing color data of polygons respectively replaced from FB ₁ to D _4, as in the case of semi-transparent polygons, color data PIX of being processed pixel to the multiplier value TxCON * INTn pixel color data TxCON and the light source color data INTn Is as follows.
That is, D ₄ = FB _{1 to} D ₁ = FB ₄ and PIX ₄ = TXCO ₁ * I
_{NT 1 ~PIX 1 = TXCO 4 *} INT 4 _{Then, D 4 = TXCO 4 * INT} 4 * DE 4 + D 5 * SR 4 (31) D 3 = TXCO 3 * INT 3 * DE 3 + D 4 * SR 3 (32 _{) D 2 = TXCO 2 * INT} 2 * DE 2 + D 3 * SR 2 (33) D 1 = TXCO 1 * INT 1 * DE 1 + D 2 * SR 1 (34) Therefore, the work data Xn (where X ₀ = 1 ) Is _expressed as X _n = SR _n * X _n-1 = SR _n * SR _n-1 * SR _n-2 *.. * SR ₁ (35), the equation (34) is expanded as follows.

D ₁ = TXCO ₁ * INT ₁ * DE ₁ * X ₀ + TXCO ₂ * INT
₂ * DE ₂ * X ₁ + TXCO ₃ * INT ₃ * DE ₃ * X ₂ + TXCO ₄ * INT ₄ * D
Because it is E ₄ * X _3, when attention is paid to the _{TXCO n * INT n * DE n} * X n-1, Dn = D n-1 + TXCO n * INT n * DE n * X n-1 = Σ (TXCO _n * INT _n * DE _n * X _n-1 ) (36) In this case, as a result of the above replacement, the smaller the value of n, the closer the polygon is, so that the blending process can be performed in order from the more recent polygon. That is, using the processing result Dn-1 of the polygon located on the near side, the data Dn of the polygon located on the next deeper side can be obtained. Then, the above equations (35) and (36)
, SRn = (1−αn), DEn = αn, and the color data Dn of the polygon is further replaced by the data FBn of the frame buffer, the result becomes the same as the equations (6) and (7) shown in the processing of the translucent polygon. It will be understood that

According to the above equation (35), the work data Xn is the accumulated value of the source mixing coefficient SRn. According to the above equation (36), the color data Dn (FBn) obtained by processing the n-th polygon from the viewpoint is the color data TXCO _n of the _n-th polygon and the light source color data INT _n , Processing is performed in accordance with the destination mixing coefficient DE _n and the cumulative multiplication value (work data) X _n−1 of the source mixing coefficients SR _{1 to} SR _n−1 of the _n−1 polygons located in front of the destination mixing coefficient DE _n . It is obtained by a rendering process for adding the image data Dn. In addition, if the work data Xn is smaller than the predetermined minimum value, even if the processing of the subsequent polygons is omitted, the displayed color data will not be greatly affected, as in the case of the translucent polygon blending processing.

FIG. 16 is a configuration diagram of a blend processing unit provided in the rendering processor when performing the above-described general-purpose rendering processing. As above,
When applied to general-purpose rendering processing, the blend processing unit 30 needs to sequentially obtain work data Xn multiplied by the source coefficient SRn from the viewpoint side along the depth direction. In this case, the source coefficient SRn needs to be processed differently when using the opacity αn or when using the texture color data Tn, depending on the type of blending processing.

The blend processing unit 30 shown in FIG.
Is a source / destination coefficient generation circuit 301
, A multiplier 302, and an X data work memory 304. The coefficient generation circuit 301 stores the texture memory 28
The texture opacity αt and the texture color data T can be read from A and 28B. Further, the opacity αp of the pixel and the texture coordinates (Tx, Ty) are supplied to the coefficient generation circuit 301 from the sorter unit.

In accordance with the type of the blending process, the coefficient generation circuit 301 calculates 1-α
n, 1 and Tn are generated and supplied to the multiplier 302. The multiplier 302 calculates the source mixing coefficient SRn by using the work data X read from the X data work memory 304.
Multiply by n-1. The coefficient generation circuit 301 also outputs the destination mixing coefficient DEn at the same time, supplies the same to the pixel work memory 36, and stores the same. Further, the work data Xn− read from the X data work memory 304 is read.
1 is also supplied to the pixel work memory 36 and stored. In the above three blending processes, the destination mixing coefficient DEn is either the opacity αn or 0.

That is, the blending unit 30 can perform the above three processes by selecting a mixing coefficient as follows.

Mixed blending processing of translucent polygon: DEn = αn, SRn = 1−αn Auxiliary filter processing: DEn = αn, SRn = 1 Color filter processing: DEn = 0, SRn = Tn Can be selected. For example, 0, Tn * αn, 1, 1-Tn, 1-T
n * αn or the like can be appropriately selected. However, the data FBn-1, that is, the color data Dn + 1 obtained by processing later.
Cannot be selected.

FIG. 18 is a diagram showing a modification of the blend processing unit. Compared to FIG. 16, the modification of FIG. 18 further includes a multiplier 305 for multiplying the destination coefficient DEn by the X work data Xn-1. And
The multiplied value Xn-1 * DEn of the multiplier 305 is stored in the pixel work memory unit 36. By using such a blend processing unit, the configuration of the multiplier 381 of the color synthesis unit can be simplified.

The configurations of the texture processing unit 32 and the light source calculation unit 34 are the same as those described above even in the case of general-purpose blending processing. That is, the generation (texture processing) of the pixel color data TXCOn requiring a relatively large number of processing steps and the light source calculation for obtaining the light source color data are obtained in a random order by a parallel circuit, and are sequentially stored in the pixel work memory unit 36. Then, a synthesis process is performed by the color synthesis unit 38 from the pixel for which the processing of the pixel data is completed.

In the color synthesizing unit 38, the immediately preceding work data Xn-1 and the destination mixing coefficient DE
n, pixel color data TXCOn and light source color data INTn,
The data is multiplied by the multiplier 381, added to the image data FB in the frame buffer 40, and rewritten in the frame buffer 40.

The blend processing unit 30 shown in FIG.
Is the source mixing coefficient SRn
The color data T of the texture was used as it is.
However, in order to more precisely obtain the color data of the color filter, the light source processing is performed by multiplying the pixel color data TXCOn generated by the texture processing unit 32 and the light source color data INTn generated by the light source calculation unit 34. There is a need. Therefore, it is necessary to supply the pixel color data subjected to the light source processing to the blend processing unit 30 and use it as a source mixing coefficient. For that,
In the color filter processing, the blend processing unit 3
It is necessary to stop the process of generating work data by 0 for a while. Then, the work data Xn is generated while the color data TXCOn and the light source color data INTn of the pixel generated and recorded in the pixel work memory are supplied to the blending processing unit 30.

Since the above processing is limited to the case of the color filter processing, the influence on the whole rendering processing is not so large.

In the above embodiment, the blend processing unit 30 in the rendering processor 20
Texture processing unit 32 and light source calculation unit 3
4 may be each realized by a microcomputer. When the rendering processor 20 is realized by a one-chip processor, a macro circuit having a configuration of a microcomputer is embedded as the processing unit. In this case, the texture processing unit 32 and the light source calculation unit 34 have a plurality of microcomputers, and perform each process by executing a program. In the microcomputer provided with the plurality of units, the texture processing and the light source calculation are distributed as soon as the microcomputer becomes empty. Therefore, even if the processing time of each pixel is different, the entire processing time can be shortened.

As described above, the protection scope of the present invention is not limited to the above embodiments, but extends to the inventions described in the claims and their equivalents.

[0120]

As described above, according to the present invention, in the image processing of a semi-transparent polygon, an algorithm for sequentially processing the polygon from the viewpoint side of the display screen in the depth direction is employed, so that the depth direction can be reduced. Color synthesis processing can be performed in any order. As a result, a rendering processor capable of processing at higher speed can be realized.

Further, according to the present invention, a general-purpose rendering processor capable of performing general blending processing can also realize a configuration that performs processing at a higher speed.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating an example of overlapping semi-transparent polygons.

FIG. 2 shows a rendering process according to the present invention and a conventional example.
9 is a table showing an arithmetic expression when applied to the example of FIG.

FIG. 3 is a configuration diagram of a game device having a rendering processor according to the embodiment of the present invention.

FIG. 4 is a configuration diagram of a sorter unit.

FIG. 5 is a configuration diagram of another sorter unit.

FIG. 6 is a configuration diagram of a blend processing unit.

FIG. 7 is a configuration diagram of a light source calculation unit.

FIG. 8 is an operation timing chart of the light source calculation circuit.

FIG. 9 is a configuration diagram of a texture processing unit.

FIG. 10 is a configuration diagram of a pixel work memory unit.

FIG. 11 is a configuration diagram of a color synthesis unit.

FIG. 12 is a processing flowchart of a rendering processor.

FIG. 13 is a diagram showing a blending process of translucent polygons.

FIG. 14 is a diagram illustrating auxiliary filter processing.

FIG. 15 is a diagram illustrating color filter processing.

FIG. 16 is a configuration diagram of a blend processing unit when performing a general-purpose rendering process.

FIG. 17 is a configuration diagram of a modified example of the blend processing unit.

FIG. 18 is a configuration diagram of a modified example of a blend processing unit when performing a general-purpose rendering process.

[Explanation of symbols]

34 renderer, rendering processing unit 342 X buffer, work data buffer 343 D buffer, image data buffer T _n color data, texture data D _n image data α _n opacity data P1 to P4 polygon

Claims (12)

[Claims] An image processing apparatus for generating image data for display by performing image processing on a plurality of overlapping polygons, wherein the n-th polygon when arranged in the depth direction from the near side in the display screen is: Pixel color data TXCOn If, opacity alpha _n or transparency and (1-αn), the light source color data INTn is given, the image processing apparatus, the transparency of the overlapping pixels multiplied value X _n of (1-αn), from the front side A blend processing unit that determines in the order of depth direction, a light source calculation unit that determines light source color data INTn for each pixel in the depth direction according to the light source data of the pixel, and a polygon color data PCOLn and texture coordinates,
Pixel color data TX for each pixel in the depth direction
A texture processing unit for determining COn; for each pixel in the depth direction, pixel color data TXCOn of the pixel; light source color data INTn, opacity αn, and the multiplied value X _{n of the} immediately preceding pixel _A color synthesis unit that performs a color synthesis process that corrects and accumulates according to -1.At least one of the light source calculation unit and the texture processing unit performs a corresponding data generation process in parallel, and the color synthesis unit performs An image processing apparatus for performing the color synthesizing process on pixels in a depth direction in any order. 2. A sorter unit according to claim 1, further comprising: a sorter unit for rearranging a plurality of polygon data or pixel data in the frame in order from a near side in the display screen in a depth direction. Supplies the pixel data to a light source calculation unit and a texture processing unit, and supplies the pixel data to the blend processing unit in the depth direction from the near side. 3. The color synthesizing process according to claim 1, wherein the color synthesis processing includes pixel color data TXCOn, light source color data INTn, opacity αn of the pixel being processed, and the multiplication value X _{n−1 of the} immediately preceding pixel. An image processing apparatus for obtaining a multiplied value of と and accumulating them. 4. The image processing apparatus according to claim 3, wherein a cumulative value obtained by the color synthesis processing is used as the display image data. 5. The pixel work memory unit according to claim 1, further comprising a pixel work memory unit for recording data of the multiplied value Xn, pixel color data TXCOn, light source color data INTn, and opacity αn of a pixel in a depth direction. An image processing apparatus, wherein the color synthesizing unit performs the color synthesizing process on a pixel in which all of the data is recorded in the pixel work memory unit. 6. The method according to claim 1, wherein the blend processing unit further determines the opacity αn or transparency (1) of the pixel from the opacity or transparency of the polygon supplied from the sorter unit and the opacity or transparency of the texture. -Αn) is generated. 7. The texture processing unit according to claim 1, further comprising a plurality of texture processing circuits for obtaining the pixel color data TXCOn according to polygon color data PCOLn and color data T of a texture corresponding to the texture coordinates. An image processing apparatus, comprising: a texture processing circuit in an empty state sequentially generating the pixel color data. 8. The light source calculation unit according to claim 1, wherein the light source calculation unit has a plurality of light source calculation circuits for obtaining light source color data INTn, and a light source processing circuit in an empty state sequentially generates the light source color data. Image processing apparatus. 9. An image processing method for a plurality of overlapping polygons.
Processing device that performs image processing and generates display image data
In the display screen, the number n when arranged in order from the near side to the depth direction in the display screen
Pixel color data TXCOn in the eye polygon And the light source color
Data INTn and destination mixing coefficient DE _nAnd so
Source mixing coefficient SR_nGiven by the image processing apparatus, wherein the image processing device comprises:_nMultiplied value of
X_nFrom the near side in the depth direction.
Each pixel in the depth direction according to the processing unit and the light source data of the pixel
Light source calculation unit that calculates the light source color data INTn for
According to the polygon color data PCOLn and the texture coordinates,
Pixel color data TX for each pixel in the depth direction
A texture processing unit for obtaining COn; and for each pixel in the depth direction,
Xel color data TXCOn, light source color data INTn,
Bination mixing coefficient DE_n, And the pixel on the near side
The multiplied value X of_n-1Color composition corrected and accumulated according to
A color synthesizing unit for performing processing, wherein the number of light source calculating units and
At least one performs the corresponding data generation processing in parallel.
The color synthesizing unit performs the color synthesizing process.
The feature is that the order is performed in random order for pixels in the depth direction.
Image processing device. 10. The color synthesizing process includes the pixel color data TXCOn, the light source color data INTn, the destination mixing coefficient DE _n of the pixel under processing, and the multiplication value X _{n-1 of the} immediately preceding pixel. An image processing apparatus for obtaining a multiplied value of, and accumulating the multiplied values, wherein the accumulated value is used as the display image data. 11. An image processing method for generating image data for display by performing image processing on a plurality of overlapping polygons, wherein an n-th polygon arranged in order from a near side in a display screen in a depth direction includes: Pixel color data TXCOn If, opacity alpha _n or transparency and (1-αn), given the source color data INTn, the image processing method, a multiplication value X _n of the transparency of the overlapping pixels (1-αn), the front side And a light source calculation process for obtaining light source color data INTn for each pixel in the depth direction according to the light source data of the pixel, and a polygon color data PCOLn and texture coordinates according to the pixel light source data.
Pixel color data TX for each pixel in the depth direction
A texture processing step of obtaining COn; and for each pixel in the depth direction, pixel color data TXCOn of the pixel, light source color data INTn, opacity αn, and the multiplied value X _{n of the} immediately preceding pixel. _A color synthesis step of performing a color synthesis processing that corrects and accumulates according to -1.At least one of the light source calculation step and the texture processing step, a corresponding data generation process is performed in parallel, and in the color synthesis step, An image processing method, wherein the color synthesis processing is performed in random order for pixels in a depth direction. 12. An image for a plurality of overlapping polygons.
Image processing method that performs processing and generates display image data
In the method, the number n when arranged in the depth direction from the near side in the display screen
Pixel color data TXCOn in the eye polygon And the light source color
Data INTn and destination mixing coefficient DE _nAnd so
Source mixing coefficient SR_nAnd the image processing method comprises: providing the source mixing coefficient SR of overlapping pixels_nMultiplied value of
X_nFrom the near side in the depth direction.
Each pixel in the depth direction according to the processing steps and the light source data of the pixel
A light source calculation step of obtaining light source color data INTn for the following: polygon color data PCOLn and texture coordinates
Pixel color data TXCO for the pixels in the depth direction
a texture processing step for determining n, and for each pixel in the depth direction,
Xel color data TXCOn, light source color data INTn,
Bination mixing coefficient DE_n, And the pixel on the near side
The multiplied value X of_n-1Color composition corrected and accumulated according to
A color synthesizing step of performing processing, and at least the light source calculating step and the texture processing step.
On the other hand, the corresponding data generation processes are performed in parallel,
In the color synthesis process, the color synthesis processing is performed in the depth direction.
An image processing method characterized by performing cells in any order.
Law.