JP2007310682A - Transparent object rendering method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a transparent object rendering method that is suited to render many transparent objects in real time on hardware limited in performance by uncomplicated processing. <P>SOLUTION: The method for rendering a transparent object in which an opaque inner layer is covered in an optically transparent or translucent outer layer in a virtual three-dimensional space comprises the steps of: calculating a Fresnel term having the inner product of a normal vector and a view vector to vertices of the inner layer surface; pixel-processing the background; pixel-processing the inner layer with the calculated Fresnel term as transparency; pixel-processing the outer layer with fixed transparency; and combining the pixel-processed background, inner layer and outer layer. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a method for drawing a transparent object having a transparent or translucent light-transmitting outer layer in a virtual three-dimensional space, and more particularly, a large number of non-uniformities or thicknesses in real time on hardware with limited performance. The present invention relates to a CG (Computer Graphics) technique for drawing a transparent object having an outer layer having a light transmission property that changes.

  In recent years, CG technology such as game images has evolved at a very high speed. This is due to the evolution of hardware, especially the remarkable evolution of GPU (Graphics Processing Unit). Along with this, various shaders (drawing mechanisms) have been developed and improved in efficiency, but there are still many things that have an unrealistic amount of calculation to process in real time. Ray tracing drawing processing such as photon mapping is one of them, and although it is possible to draw very fine drawing, it is difficult to use it in a game due to the amount of calculation.

As the background of the invention of this time, we needed to create an object with an unprecedented fresh and grotesque texture as an enemy character creature (hill) that appeared in a game (shooting game in the jungle). As a condition required for that,
(1) Wet texture associated with the setting of living on wet and wet watersides (2) Visceral organs that move lively to produce unpleasantness. Therefore, we needed to express two different textures: the inner layer representing the internal organs and the outer layer surrounding it.

  First, as a result of investigating various things in order to search for actual materials with this kind of texture, we gained a hint of Japanese confectionery conflict. The appearance of the kuzu is smooth and glossy, and the inner eggplant is wrapped in a transparent or translucent kuzu. This was very close to the desired texture of the organism, so we tried a method to reproduce the equivalent texture.

The characteristics of conflict by observation are as follows.
(A) It has two different materials and textures, the outer layer of Kuzu and the inner layer of Isogo. This deserves the skin and internal organs of the organism.
(B) Since the thickness of the entire inner layer, particularly the outline, differs depending on the line-of-sight direction, it appears blurred due to non-uniform opacity depending on the angle of the line-of-sight direction. The volume of the inner layer, which is the contents of the outer layer, can be sensed by the bleeding of the contour.

  Hereinafter, processing in the case where a transparent object having an outer layer having such transparent or translucent light transmittance is drawn in real time will be described.

  FIG. 1 is a flowchart showing a conventional process for drawing a transparent object.

  In FIG. 1, when processing is performed in real time, an opaque portion including a background and an inner layer is rendered and stored in the frame buffer, for example, in the first pass in the GPU processing (step S1).

  In the second pass, the frame buffer is used as a refractive component of the light transmission portion of the outer layer, and rendering is performed by distribution ray-tracing or photon mapping (step S2).

  In addition, since the blur of the object to be refracted cannot be reproduced by the standard refraction process, the outline of the object to be blurred (inner layer) is further extracted and blurred by a filter such as Box or Gaussian. (Step S3).

  As described above, the conventional method requires at least two passes of rendering and blurring.

Related prior art documents are shown below.
“Reproduction of opal play-color phenomenon using a structural model” Takaaki Abe February 7, 2003 Nara Institute of Science and Technology (http://chihara.aist-nara.ac.jp/public/research/thesis/master /2003/abe0151003.pdf)

  As described above, when a transparent object is drawn by a conventional method, rendering and blurring processing of at least two passes are required, and a large amount of processing time is required for calculation processing for drawing, and a CPU accompanying an increase in calculation amount, There is a problem that real-time drawing is difficult due to a great load on the GPU. Therefore, it is necessary to perform a unique approximation based on observations that are not confined to conventional methods.

  The present invention has been proposed in view of the above-described conventional problems, and the object of the present invention is to provide a light transmission and transmission with many realistic drawings in real time on hardware with limited performance. An object is to provide a transparent object drawing method suitable for drawing an object.

  In order to solve the above problems, according to the present invention, as described in claim 1, in the virtual three-dimensional space, the opaque inner layer is covered with a transparent or translucent light-transmitting outer layer. A method of drawing an object, the step of calculating the Fresnel term by the inner product of the normal vector of the vertex of the inner layer surface and the line-of-sight vector, the step of performing background pixel processing, and using the calculated Fresnel term as transparency And performing the pixel processing of the inner layer, the pixel processing of the outer layer using fixed transparency, and the step of combining the background, the inner layer and the outer layer that have been pixel processed. The gist of the object drawing method.

  Further, as described in claim 2, in the transparent object drawing method according to claim 1, the line-of-sight vector used for the calculation of the Fresnel term can be constant regardless of the vertex.

  Further, as described in claim 3, in the transparent object drawing method according to claim 1, the sign of the line-of-sight vector used for calculation of the Fresnel term can be inverted in advance.

  Further, as described in claim 4, in the transparent object drawing method according to claim 1, the calculation of the Fresnel term can be performed in vertex processing.

  Further, according to the fifth aspect of the present invention, in the virtual three-dimensional space, a transparent object drawing device in which an opaque inner layer is covered with a transparent or translucent light-transmitting outer layer, the vertex of the inner layer surface Means for calculating the Fresnel term by the inner product of the normal vector and the line-of-sight vector, means for performing background pixel processing, means for performing pixel processing of the inner layer using the calculated Fresnel term as transparency, and fixed It is possible to configure as an object drawing apparatus including means for performing pixel processing of the outer layer using specific transparency, and means for combining the background subjected to pixel processing, the inner layer, and the outer layer.

  Further, according to a sixth aspect of the present invention, there is provided a drawing program for a transparent object in which an opaque inner layer is covered with a transparent or translucent light-transmitting outer layer in a virtual three-dimensional space, the computer comprising the inner layer Means for calculating the Fresnel term by the inner product of the normal vector and the line-of-sight vector of the surface vertex, means for performing background pixel processing, means for performing pixel processing of the inner layer using the calculated Fresnel term as transparency, fixed It can be configured as an object drawing program that functions as means for performing pixel processing of the outer layer using specific transparency, the background subjected to pixel processing, the inner layer, and the outer layer.

  In the transparent object drawing method of the present invention, it is possible to draw a large number of transparent objects on which real processing is performed in real time on hardware with limited performance and light processing.

  Hereinafter, preferred embodiments of the present invention will be described.

<Overview>
The transparent object drawing method according to the present invention is approximated by simplifying the calculation in the expression of bleeding caused by refraction / diffusion when light passes through a light-transmitting substance that requires a very heavy processing in the conventional method. It is a technique to get an appearance. The conventional method referred to here is distributed ray tracing or photon mapping. Since these require enormous calculations, the real-time property that we demanded here was completely absent, and a new approximation method had to be developed. To express the inner layer (internal organs), the general method is to simply treat the inner layer as opaque and express the outer layer by transparent or semi-transparent processing. However, a new method for obtaining a more expressive quality A transparent processing expression method was developed.

  Here, it is an object representation that has a transparent or translucent outer layer and an opaque inner layer, and the inner layer (internal) model is shaded so that the contents bleed from the inside to the outside, and the inner layer model is transparent or translucent. It is possible to overlap the outer layer model with light processing. Thereby, a substance in a volume (for example, muddy water, jelly, etc.) containing impurities such as bubbles and solids can be drawn naturally.

<System configuration>
FIG. 2 is a diagram illustrating a configuration example of an object drawing apparatus according to an embodiment of the present invention.

  In FIG. 2, the object drawing apparatus includes a 3D application 1 including drawing commands for 3D (3 Dimension) objects such as game applications, and 3D-API (Application Program) such as OpenGL and Direct3D that receive drawing commands and the like from the 3D application 1. Interface) 2 and a GPU 3 that executes drawing processing.

  The GPU 3 receives a GPU command and a data stream from the 3D-API 2, and performs coordinate transformation (matrix transposition) by receiving 3D vertex data from the GPU front end 31 and projecting it to the 2D screen space. A programmable vertex processor (VS) 32 and a basic assembly unit 33 that assembles a vertex index stream supplied from the GPU front-end unit 31 and vertex data coordinate-converted by the programmable vertex processor 32 are provided. Further, the GPU 3 performs rasterization / interpolation unit 34 that performs rasterization and interpolation from polygon, line, and point data assembled by the basic assembly unit 33, and rasterized pixel (fragment) data from the rasterization / interpolation unit 34. A raster processing that performs a raster operation from a programmable pixel processor (PS) 35 that receives and performs texture mapping and the like, a pixel position stream provided from the rasterization / interpolation unit 34 and a texture-mapped data provided from the programmable pixel processor 35 And a frame buffer 37 into which drawing contents are written / updated.

<Operation>
FIG. 3 is a flowchart showing an example of processing for drawing a transparent object, and shows processing by one pass. Note that the actual processing in the GPU 3 is automatically optimized, so the actual processing order may be changed.

  In FIG. 3, when processing is started (step S100), vertex processing is performed by the programmable vertex processor 32 and the basic assembly unit 33 of the GPU 3 (step S101). In this vertex processing, background vertex processing (step S102), inner layer vertex processing (step S103), and outer layer vertex processing (step S105) are performed in parallel.

  Further, following (or in parallel with) the vertex processing of the inner layer (step S103), the Fresnel term used for transparency when expressing the blur of the inner layer contour by the transparent or translucent outer layer (ordinary Fresnel term) Is calculated) (step S104).

  The inner layer has a texture that is less glossy than the outer layer. This is because the light is dispersed by the turbidity of the outer layer until the light reaches the inner layer and when the light comes out of the inner layer. This means that the light that is transmitted and refracted is also dispersed, resulting in contour blurring. It is most important to reproduce this blurred outline. In order to correctly calculate bleeding, it is generally necessary to use a rendering method that uses probability such as dispersion ray tracing, or rendering that considers media involved in light within the material, but the processing becomes complicated, so real-time performance is completely Absent.

  Bleeding is a phenomenon caused by light scattering when passing through an outer layer having optical transparency, so it is necessary to obtain how much thickness (distance) is from the inner layer to the outer layer. In order to obtain numerical information on the thickness, the distance between arbitrary vertices of the inner layer and the outer layer is measured, so that vertex coordinate processing is required at least twice. In order to obtain the distance accurately, it is necessary to determine the collision with each other by sending a line-of-sight vector (ray) from the inner layer side to the outer layer or from the outer layer side to the inner layer from the virtual camera position, and to obtain the distance based on the intersection . The time required for this processing and the computation load on the CPU and GPU are too large. Therefore, in the present invention, the angle corresponding to the degree of bleeding is calculated from the inner product of both vectors by using the angles of the line-of-sight vector and the normal vector. As a result, it is possible to employ a method similar to the normal calculation of Fresnel terms (considering the line of sight and the inclination of the geometry).

FIG. 4 is a diagram showing how the degree of blurring of the inner layer contour by the outer layer having light transmittance depends on the inner product of the line-of-sight vector and the normal vector, and I ₁ and I ₂ are the inner layer IL from the viewpoint (camera) E. The line-of-sight vectors N ₁ and N ₂ facing the upper vertices P ₁ and P ₂ are normal vectors facing the direction of the geometry normal at the vertices P ₁ and P ₂ of the inner layer IL. In addition, OL is an outer layer.

4, compared to the case sight vector I ₂ and the normal vector N ₂ as vertices P ₂ is collinear (angle of 0 degrees), line-of-sight vector I ₁ and the normal vector as vertices P ₁ It can be considered that the thickness of the outer layer OL (broken line portion) increases as the angle of N ₁ approaches 90 degrees (0 in the inner product). The thickness of the outer layer OL is the degree of light diffusion, and the contour of the inner layer IL appears to be blurred according to the thickness.

  A general Fresnel term Fresnel is calculated based on the following equation.

Fresnel = max (0, min (1, Bias + Scale
* (1 + pow (dot (I, N), Shininess))))
Here, max (x, y) is a function for obtaining the maximum value of x and y, min (x, y) is a function for obtaining the minimum value of x and y, pow (x, y) is a function for obtaining ^xy , dot (x, y) is a function for calculating the inner product of the vector x and the vector y, and Bias, Scale, and Shininess are various parameters.

Also, in general Fresnel term Fresnel calculation, at each vertex,
I = normalize (Position-EyePosition)
And the line-of-sight vector I for each vertex is obtained. Here, normalize (x) is a function that returns a vector of length 1 pointing in the same direction as the vector x, Position is the vertex coordinates, and EyePosition is the viewpoint coordinates.

  Since it is computationally expensive to perform such calculation for each vertex, in the present invention, as shown in FIG. 5, a single line-of-sight vector I is used regardless of the vertex. FIG. 6 is a diagram illustrating an example of a difference in drawing result between the case where a line-of-sight vector is calculated for each vertex (a) and the case where a single line-of-sight vector is used (b), where the line-of-sight vector is calculated for each vertex. In (a), the rendering quality is not good and the appearance is broken, but in the case of using a single line-of-sight vector (b), the calculation cost is not increased and the rendering quality is improved and the appearance is improved. This is considered to cause deterioration in quality because the interpolation between vertices is automatically performed when calculating the Fresnel term Fresnel in the vertex processing. Although the Fresnel term Fresnel can be calculated in the pixel processing, the calculation cost is higher than that in the vertex processing. Note that the cost here refers to the processing time for computation and the load on the CPU, GPU, and the like.

  Thus, as a result of considering whether there is a simpler and better-looking method, it is very easy to use the same gaze vector at all the vertices on the inner layer surface, and how much the normal is directed to the viewpoint. It can be obtained as a value that attenuates from the center of the inner layer to the outer side. As a result, a much smoother attenuation value can be obtained than in the normal method, and the appearance does not break down. Furthermore, since the line-of-sight vector only needs to be calculated when each drawing frame or line of sight changes, the calculation cost is reduced.

  Further, in order to simplify the calculation, the sign of the line-of-sight vector is inverted. As a result, a necessary correct calculation result can be obtained without executing the function min () on the inner product of the line-of-sight vector and the normal vector.

  The following is a summary of the calculation formula.

I = normalize (-1 * EyeVector)
Fresnel = pow (max (0, dot (N, I)), a) * Scale
Here, a is a parameter equivalent to the Fresnel term Shininess, and a and Scale adjust the degree of attenuation.

  Returning to FIG. 3, after the vertex processing (step S <b> 101), pixel processing (step S <b> 106) is performed by the rasterizing / interpolating unit 34, the programmable pixel processor 35, and the raster computing unit 36. In this pixel processing, background pixel processing (step S107), inner layer pixel processing (step S108), and outer layer pixel processing (step S109) are performed in parallel.

  Here, in the pixel processing of the inner layer (step S108), texture mapping is performed using the Fresnel term already calculated in the calculation of the Fresnel term (step S104) in the vertex processing (step S101) as the transparency.

  In the outer layer pixel processing (step S109), texture mapping that expresses gloss using a fixed transparency is performed.

In other words, the surface of the outer layer is a wet texture,
(B) It has two points of light transmission (b) gloss (specular reflection, gloss reflection, etc.).

  The inner layer can be displayed by being transparent or translucent, and the presence and texture of the outer layer can be emphasized by having gloss. Since the type of gloss is not particularly limited, a shading model that can reproduce gloss, for example, an existing reflection model such as Phong or Blinn may be used.

Since it is necessary to leave gloss, the calculation of transparency Transparency is
Transparency = Alpha + Specular + Reflection + Diffuse
It is said. Alpha indicates the texture alpha and vertex color alpha, Specular indicates refraction, Reflection indicates reflection, and Diffuse indicates diffusion.

actually
Transparency = saturate (Speccular * Scale1
+ Reflection * Scale2 + Diffuse * Scale3) * Alpha
Is used. Saturate (x) is a function corresponding to min (x, max (x, 1), 0), and Scale1 to Scale3 are parameters (Scale1 = 1.2, Scale2 = 0.2, Scale3 = 1.0).

  Next, in FIG. 3, the result of each pixel process (steps S107 to S109) is written into the frame buffer 37 and synthesized (step S110), and the process ends (step S111).

  FIG. 7 and FIG. 8 are diagrams showing a drawing example, FIG. 7A is a drawing result of the inner layer (representing bleeding on the outline), FIG. 7B is a drawing result of the outer layer (representing gloss on the surface), FIG. 8A shows the synthesis result of the inner layer and the outer layer, and FIG. 8B shows the synthesis result of the inner layer, the outer layer, and the background. In this manner, high-quality contour bleeding can be reproduced for the inner layer included in the glossy outer layer. Even if the inner layer and the outer layer are arranged separately, it does not feel that the contents are clogged. You must always render the two as one.

  FIG. 9 is a diagram showing another example of a transparent object, which is an example in which a transparent object of a hill whose internal organs are covered with a transparent or semi-transparent skin on a background assuming a jungle or the like.

FIG. 10 is a diagram showing another pattern example of the transparent object. FIG. 10A is a pattern in which one inner layer IL is included in one outer layer OL described so far, and FIG. 10B is a diagram in one outer layer OL. plural pattern that includes an inner layer _IL 1, IL ₂ of (two in the illustrated example), (c) is included one inner layer IL ₁ in one of the outer layer OL, the inner layer IL ₂ further inner layer IL ₁ The pattern which contains is shown.

  FIG. 11 is a diagram showing a drawing example in the case where two inner layers are arranged side by side in the outer layer (pattern (b) in FIG. 10), and (a) and (b) show different viewpoints.

<Summary>
With the method described above, an unprecedented soft permeation process that we originally requested can be realized with sufficient real-time characteristics. In this embodiment, since the outer periphery of the bleeding portion is transparent, aliasing does not occur and anti-aliasing is not necessary.

  The present invention has been described above by the preferred embodiments of the present invention. While the invention has been described with reference to specific embodiments, various modifications and changes may be made to the embodiments without departing from the broad spirit and scope of the invention as defined in the claims. Obviously you can. In other words, the present invention should not be construed as being limited by the details of the specific examples and the accompanying drawings.

It is a flowchart which shows the process for drawing of the transparent object in the past. It is a figure which shows the structural example of the object drawing apparatus concerning one Embodiment of this invention. It is a flowchart which shows the process example for drawing of a transparent object. It is a figure which shows a mode that the degree of the bleeding of the outline of the inner layer by the outer layer which has a light transmittance is dependent on the inner product of a gaze vector and a normal vector. It is a figure using a single gaze vector. It is a figure which shows the example of the difference in the drawing result at the time of using the gaze vector for every vertex, and using a single gaze vector. FIG. 6 is a diagram (part 1) illustrating a drawing example. It is a figure (the 2) which shows a drawing example. It is a figure which shows the other example of a transparent object. It is a figure which shows another example of a pattern of a transparent object. It is a figure which shows the example of a drawing in case two inner layers exist side by side in an outer layer.

Explanation of symbols

1 3D application 2 3D-API
3 GPU
31 GPU front-end unit 32 Programmable vertex processor 33 Basic assembly unit 34 Rasterization / interpolation unit 35 Programmable pixel processor 36 Raster operation unit 37 Frame buffer

Claims (6)

In a virtual three-dimensional space, a transparent object drawing method in which an opaque inner layer is covered with a transparent or semi-transparent outer layer,
Calculating a Fresnel term by the inner product of the normal vector and the line-of-sight vector of the vertex of the inner layer surface;
Performing background pixel processing;
Performing the pixel processing of the inner layer using the calculated Fresnel term as transparency;
Performing pixel processing of the outer layer using a fixed transparency;
A transparent object drawing method comprising: combining the pixel-processed background, the inner layer, and the outer layer. The transparent object drawing method according to claim 1,
A transparent object drawing method, wherein a line-of-sight vector used for calculating the Fresnel term is constant regardless of a vertex. The transparent object drawing method according to claim 1,
A transparent object drawing method, wherein a sign of a line-of-sight vector used for calculating the Fresnel term is inverted in advance. The transparent object drawing method according to claim 1,
A transparent object drawing method, wherein the calculation of the Fresnel term is performed in vertex processing. In a virtual three-dimensional space, a transparent object drawing device in which an opaque inner layer is covered with a transparent or semi-transparent outer layer,
Means for calculating the Fresnel term by the inner product of the normal vector and the line-of-sight vector of the vertex of the inner layer surface;
Means for performing background pixel processing;
Means for pixel processing of the inner layer using the calculated Fresnel term as transparency;
Means for pixel processing of the outer layer using fixed transparency;
An object drawing apparatus, comprising: the pixel-processed background, the inner layer, and the outer layer. In a virtual three-dimensional space, a transparent object drawing program in which an opaque inner layer is covered by a transparent or translucent light-transmitting outer layer,
Computer
Means for calculating the Fresnel term by the inner product of the normal vector and the line-of-sight vector of the vertex of the inner layer surface;
Means for pixel processing of the background,
Means for performing pixel processing of the inner layer using the calculated Fresnel term as transparency,
Means for pixel processing of the outer layer using a fixed transparency;
An object drawing program that functions as means for synthesizing the pixel-processed background, the inner layer, and the outer layer.