JPH07210709A - Three-dimensional moving picture generation device - Google Patents

Abstract

PURPOSE:To add the easiness of modeling and gamelike factors, to efficiently decrease the arithmetic quantity for hidden-surface erasure at the time of scene generation at a specified view point position and the amount of unnecessary data when a three-dimensional scene viewed from the specified view point is generated, and to enable real-time animation generation. CONSTITUTION:This three-dimensional moving picture generation device is equipped with a shape modeling part 1 which generates and edits a shape in a way of stacking basic primitives on a screen when three-dimensional shape data are modeled, a scene generation part 3 which arrange components generated by it at optional positions in a scene, and a data management part 6 which subdivides the scene into plural subscenes, and manages and retrieve three- dimensional shape data on a three-dimensional body included in the respective subscenes and hierarchical image data viewed from plural view points, subscene by subscene, so as to render an optional generated scene.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a computer graphics (CG) field, various image processing fields, and a game field using them. The present invention relates to an interactive three-dimensional moving image generation device for generating and processing as.

[0002]

2. Description of the Related Art In recent CG, various algorithms have been developed for rapidly generating a high quality image faithful to a physical phenomenon. Considering future use of CG, it is important to improve the image quality and generate a moving image (animation) at high speed. In order to obtain a high-quality animation, it is necessary to improve the image quality (spatial frequency and brightness level) of an image per frame, and at the same time increase the resolution in the time domain, that is, the number of frames. In order to obtain a moving image equivalent to the current NTSC TV, the number of frames must be 30 frames / sec or more.
Also, at least 20 frames per second or more are required to obtain smooth movement.

On the other hand, in conventional CG animation, since it takes a considerable amount of time to generate one frame image, one frame image is generated over time, and each frame image is generated by a video or the like. It was animated by taking frames. In the field of image processing, a method called key frame interpolation is often used. This is a method of interpolating a basic frame (key frame) in the time axis direction by using a method such as detection or estimation of a motion vector from the key frame between the frame images.

On the other hand, in the case of games and hobby use in the home, it responds in real time to a user's instruction even if the image quality of the image is relatively lower than that of the high quality CG described above. Interactivity and prompt response are required. In such applications, by reducing the number of polygons (polygons) that make up a part existing in a scene per frame or by approximating the part with two-dimensional data (cells). , Reduces the amount of calculation and ensures real-time performance.

However, in the field of conventional game machines, in general, the scenes and shapes (including characters) developed in the game are almost always provided by the user. It cannot be changed as desired. To change or edit the shape, press C on WS.
It is necessary to return the data to an AD system or the like once and edit the three-dimensional data on CAD.

These are due to the hardware performance (calculation time and capacity) of the game machine, and existing game machines do not provide a function for interactively changing and drawing the three-dimensional shape data.

[0007]

However, in the frame-taking method as described above, an image of several hundreds of frames is required even when an animation of several tens of seconds is required, and several hours are required for each continuous scene. Is very inefficient. Also, even if you see the outline of the animation or make a partial change during the creation, it is impossible unless all the animations have been generated, and the practicality and interactivity were very low. In addition, since the capacity of the internal memory and the drawing performance are greatly limited in the game target device, the CG
However, there is no natural modeling function for 3D shape data, and a large amount of shape data cannot be processed in real time. In the future game market, high image quality as well as real-time property is required, and there is a problem that the image quality level cannot be satisfied only by simple data reduction such as frame interpolation.

In view of such a point, the present invention can add the easiness of modeling and the game-like factor, and, at the time of generating a three-dimensional scene viewed from a specified viewpoint, the specified scene can be specified. It is possible to efficiently reduce the amount of hidden surface removal calculation and unnecessary data amount when generating a scene at the viewpoint position, and to significantly reduce the amount of calculation, which enables real-time animation generation. An object is to provide a generator.

[0009]

In order to attain the above object, the present invention achieves the above-mentioned object. When modeling three-dimensional shape data, shape modeling is performed in which basic primitives are simply created and edited by stacking them on a screen. Part, parts that are created using this part are placed at arbitrary positions in the scene, and in order to render the arbitrary scene created by the scene creation part, the scene is divided into multiple sub-scenes as data in advance. A data management unit that subdivides and manages and retrieves, for each sub-scene, the three-dimensional shape data of the three-dimensional object included in each sub-scene and the image data (hierarchical data) when viewed from a plurality of viewpoints. It is a configuration provided.

Further, when the viewpoint information is given, which hierarchical data of the above-mentioned component data is in the visible region is determined according to the positional relationship and the distance from the viewpoint, and is visible when observed from the viewpoint position. This is a configuration including a mechanism for switching image component data and transforming the viewpoint coordinates to draw in a common area.

[0011]

According to the present invention, with the above-described configuration, when modeling a shape or a scene, basic primitives are simply modeled in a manner of stacking blocks, and a scene is created in a manner of arranging parts in a garden. Also, by stacking basic primitives from primitives far from the viewpoint, polygon sorting time is reduced. Furthermore, when generating a three-dimensional image when a shape is viewed from an arbitrary viewpoint, image data of parts visible from the viewpoint is stored in a hierarchy according to the distance, and a plurality of image data are searched at high speed. Then, the image data is transformed into two-dimensional by creating the designated viewpoint coordinates. By providing these image data in a hierarchical structure, the data amount can be suppressed while maintaining the image quality.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A three-dimensional moving image generating apparatus according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a three-dimensional moving image generating apparatus in this embodiment.

In FIG. 1, reference numeral 1 denotes a shape modeling unit, 2
Is a shape registration unit, 3 is a scene creation unit, 4 is a viewpoint designating unit, 5
Is a motion setting unit, 6 is a data management unit, and 7 is a rendering unit. The operation of the three-dimensional moving image generating apparatus configured as described above will be described below with reference to FIG.

First, the shape modeling unit 1 creates a shape part to be placed in a scene to be generated. Generally, in modeling a shaped part, CAD is used to subdivide the three-dimensional shape of an object into polygon data, and the shape is defined by the three-dimensional coordinates of the vertices of the polygon. However, this process requires hidden surface removal to determine the front-back relation of polygons and the front and back of the surface, and the amount of calculation becomes enormous when the shape becomes complicated. In addition, to model complex shapes, very high computational performance and interactive GUI (Graphical User Int.
erface) is required.

The shape modeling unit 1 clarifies the difference from the conventional CAD system and presupposes interactive shape creation on a game machine or the like having a relatively low hardware performance. Therefore, in order to simplify the operation at the time of shape modeling as much as possible, a polyhedron (for example, a hexahedron or a cylinder), that is, a so-called primitive is defined as a basic unit forming the shape. The size of this basic primitive can be changed by an integral multiple of the most basic (smallest size).
In this way, a plurality of basic primitives defined in advance are used, and these are modeled by stacking blocks.

As shown in FIG. 2, considering virtual surfaces in the XYZ directions of the local coordinate system of the shape (cast),
The parts can be connected only from the positions on the grid points on each surface. This grid can be changed by an integer multiple of the size of the basic primitive. A shape is created while moving the primitive from the 6 directions in FIG. 2 toward the cast. When moving, the primitive allows rotation in all XYZ directions at 90 degree intervals as required. Furthermore, the point where each primitive touches the cast being created is determined as the connection position of that primitive.

Further, when it is desired to delete a certain primitive for the shape cast already modeled, the erasing primitive is moved in the same manner as described above, and the primitive at the position where the primitive comes into contact with the cast is deleted from the cast shape. The cast is modeled by repeating the above operations and stacking blocks while looking at the screen (see FIG. 3).

Further, in the method of FIG. 3, an example in which basic primitives are used one by one has been shown, but a plurality of primitives are defined in advance as a part (for example, a car door or body), These may be added to the cast in the manner described above. If the basic primitive is a cylinder or a concave shape, the boundary box is regarded as a basic primitive and the same processing is performed. The above is the embodiment of the basic operation of the shape modeling unit.

Next, the order of joining the basic primitives at the time of shape modeling is defined as follows. First, when the position of the viewpoint for observing the cast is determined, the cast is moved in the order of primitives from a lattice point located at a large value in the Z direction in the viewpoint coordinate system, that is, a position farther than the viewpoint position. As a result, it is possible to omit the time for sorting for hidden surface removal, which is required at the time of drawing.

A method for registering the part modeled by the above method in the shape registration section 2 will be described below. When the shape modeling is completed and the shape is registered, unnecessary primitives are removed without being registered. In the modeling method described above, the primitives can be arranged only at the grid points on the XYZ plane of the cast local coordinate system. Therefore, when the primitives or their boundary boxes are joined, the size relationship of the joined surfaces is compared, If all the vertices are shared or all the vertices are included inside the other closed surface, delete the shared surface of the primitive that was joined later or the surface included in the closed surface from the data, or draw The shape registration unit 2 is provided with a mechanism for adding a flag that does not need to be drawn sometimes.

Further, in the shape modeled using the above-mentioned basic primitives, the outer shape may be angular, so that only the outermost primitives of the stacked primitives are searched and the shape corresponding to the ridgeline is searched. If necessary, chamfer. The outermost primitive is searched under the condition that two or more surfaces adjacent to the boundary are not shared with other primitives.

When this primitive is searched, (a) a plane passing through two planes sharing one side (slanted line in the figure) and (b) three planes including three sides sharing one vertex are shown in FIG. A new plane that passes through is generated. Primitives that require chamfering are newly registered in the shape registration unit 2 as a polyhedron composed of these surfaces.

Next, a method of creating a scene desired by the user by the scene creating unit 3 using the parts created by the shape modeling unit 1 will be described. Although FIG. 5 is an example of a scene, consider a virtual plane (scene A) that forms a scene as shown in the figure, and further subdivide this scene into some sub-scenes. The shape parts created by the shape modeling unit 1 and registered in the shape registration unit 2 are arranged for each sub-scene.

For each of the sub-scenes, the component position (X, Y) is determined from the direction perpendicular to the paper surface of FIG. 5, and the parts are arranged so as to be dropped from above the scene. Height when dropped (Z
The value) is specified by the distance from the virtual plane. In this sub-scene example, roads, buildings, etc. are all defined by the set of basic primitives described above. Also, when the scene is arranged, the position of the viewpoint for observing the scene later is dropped and arranged in the same manner as the parts. Viewpoint information (viewing angle, line of sight, etc.)
Is defined by the viewpoint setting unit 4 for each viewpoint. Further, the arrangement information (coordinate values in the sub-scene) of the parts in the created scenes is used as the data management unit 6 in sub-scene units.
Manage with.

Next, with respect to the parts and scenes set as described above, a method of rendering a three-dimensional image when observing this scene from a certain viewpoint position will be described. First, the motion setting unit 5 sets motion information (velocity, locus, etc.) of necessary parts placed in the scene for each part. Also, this example places some restrictions on the position of the viewpoint as it moves through the scene. For example, consider a case where the viewpoint moves on the road in the scene of FIG. Since the scene A is observed from the viewpoint while traveling on the road, it is necessary to search for a visible shape from the viewpoint centering on the road. As is done in a general CG system, the processing performance is not sufficient in a game machine or the like to have the three-dimensional data of all the shapes in the scene and render the three-dimensional shapes as they are.

Therefore, it is necessary to provide the data management unit 6 with a mechanism for efficiently providing the calculation amount and the data amount. The method will be described below. First, the data structure will be described with reference to FIG. Considering the case where the viewpoint moves from right to left on the road in the scene as shown in FIG. 6, first, the shape included in the sub-block A is very close to the viewpoint, so it is necessary to render it as a three-dimensional shape. .

Next, at this viewpoint position, the sub-block
The shape included in B is farther than the sub-block A.
Image data of all the parts in the sub-block B projected on the surface A and the surface B, respectively, and further, the parts included in the sub-block C and the sub-block D are stored in the virtual surface C. Images of all shapes (FIG. 7C) in the projected sub-blocks C and D are kept as they are.

When the road is curved like the sub-block F, the road is divided into three areas indicated by dotted lines in the figure, and images of the surfaces A and B facing the road are stored. . Thus, for each sub-block, three layers of (1) three-dimensional shape parts, (2) image data projected on a virtual surface facing the road, and (c) image data projected on a surface perpendicular to the road As the data, the data management unit 6 has a mechanism for storing the data. However, considering the direction of running on the road, if it is a straight line, it is necessary to have two image data for each of surface A and surface B.

Next, at the time of rendering, according to the distance from the viewpoint, the hierarchical data corresponding to the distance is selected from the above hierarchical data and selected as shown in FIG. 7, and the shape of the closest sub-block is 3 3D rendering, the shapes in the next adjacent sub-blocks are drawn by converting the viewpoint coordinates of the images projected on the surfaces A and B, and the shapes in the other visible sub-blocks are shown in FIG. ) Image is drawn as it is, and these are combined to create one three-dimensional image.
This operation is sequentially repeated according to the distance from the viewpoint as the viewpoint moves, so that a continuous animation is generated.

[0030]

As described above, according to the present invention, by using the basic primitives when modeling a three-dimensional shape and creating a scene, the degree of freedom at the time of modification is limited, but the ease of modeling and game-like performance are reduced. Factors can be added.

When generating a three-dimensional scene viewed from a specified viewpoint, two-dimensional image data when the scene is viewed from a plurality of viewpoint positions is managed and searched according to the distance from the viewpoint. Thus, it is possible to efficiently reduce the amount of hidden surface removal calculation and the amount of unnecessary data when generating a scene at a specified viewpoint position.

Further, when a visual point is given, a visible image is searched, and only the image data of the visual image is transformed into the visual point coordinates, whereby the amount of calculation can be greatly reduced and real-time animation can be generated.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a three-dimensional moving image generating apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing a modeling method during shape editing.

FIG. 3 is a diagram showing a cast shape modeling method.

FIG. 4 is a diagram showing chamfering of a basic shape (hexahedral).

FIG. 5 is a diagram showing the arrangement of parts when creating a scene.

FIG. 6 is a diagram showing how to provide a data structure of the present invention.

FIG. 7 is a diagram showing a rendering method during scene playback.

[Explanation of symbols]

 1 Shape Modeling Section 2 Shape Registration Section 3 Scene Creation Section 4 Viewpoint Designation Section 5 Motion Setting Section 6 Data Management Section 7 Rendering Section

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yoshimori Nakase 1006 Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (8)

[Claims] 1. A shape modeling unit that defines a three-dimensional shape as a set of basic primitives, and a shape registration unit that registers the shape parts created by the shape modeling unit.
A scene creation unit that arranges the shape parts registered in the shape registration unit in a scene to create a scene, a viewpoint designation unit that sets viewpoint data for observing the scene, and a scene created by the scene creation unit On the other hand, with regard to the scene data defined by the scene setting unit, the motion setting unit that sets motion information of a specific part in the scene, the viewpoint data defined by the viewpoint setting unit, and the motion setting unit. Using the defined motion information, a data management unit that selects a shape that is visible from a viewpoint at an arbitrary time, and a scene data provided from the data management unit are used to draw a three-dimensional scene at an arbitrary time. When modeling three-dimensional shape data, if the shape primitive is a complicated shape, consider the boundary box of the shape primitive and The active or boundary boxes are arranged or deleted only at the positions of the variable grid points on the plane perpendicular to the XYZ directions of the three-dimensional local coordinates of the part in the manner that the blocks are piled up. A three-dimensional moving image generation apparatus characterized in that a shape is defined with a time point as a joining position. 2. A shape modeling unit that defines a three-dimensional shape as a set of basic primitives, and a shape registration unit that registers the shape parts created by the shape modeling unit.
A scene creation unit that arranges the shape parts registered in the shape registration unit in a scene to create a scene, a viewpoint designation unit that sets viewpoint data for observing the scene, and a scene created by the scene creation unit On the other hand, with regard to the scene data defined by the scene setting unit, the motion setting unit that sets motion information of a specific part in the scene, the viewpoint data defined by the viewpoint setting unit, and the motion setting unit. Using the defined motion information, a data management unit that selects a shape that is visible from a viewpoint at an arbitrary time, and a scene data provided from the data management unit are used to draw a three-dimensional scene at an arbitrary time. And a rendering unit for performing three-dimensional shape data modeling, the viewpoint coordinate system is used to model the three-dimensional shape data from the viewpoint position in the XY plane. A three-dimensional moving image generation device characterized in that basic primitives located at distant grid points are sequentially arranged. 3. A shape modeling unit that defines a three-dimensional shape as a set of basic primitives, and a shape registration unit that registers the shape part created by the shape modeling unit,
A scene creation unit that arranges the shape parts registered in the shape registration unit in a scene to create a scene, a viewpoint designation unit that sets viewpoint data for observing the scene, and a scene created by the scene creation unit On the other hand, with regard to the scene data defined by the scene setting unit, the motion setting unit that sets motion information of a specific part in the scene, the viewpoint data defined by the viewpoint setting unit, and the motion setting unit. Using the defined motion information, a data management unit that selects a shape that is visible from a viewpoint at an arbitrary time, and a scene data provided from the data management unit are used to draw a three-dimensional scene at an arbitrary time. When modeling three-dimensional shape data, the face primitives of the basic primitive or boundary box share all vertices. Characterized by removing from the data structure the face of a primitive or boundary box that was touched later, or the face contained in one closed face, if all vertices of one face are contained in the other closed face A three-dimensional moving image generation device. 4. A shape modeling unit that defines a three-dimensional shape as a set of basic primitives, and a shape registration unit that registers the shape parts created by the shape modeling unit.
A scene creation unit that arranges the shape parts registered in the shape registration unit in a scene to create a scene, a viewpoint designation unit that sets viewpoint data for observing the scene, and a scene created by the scene creation unit On the other hand, with regard to the scene data defined by the scene setting unit, the motion setting unit that sets motion information of a specific part in the scene, the viewpoint data defined by the viewpoint setting unit, and the motion setting unit. Using the defined motion information, a data management unit that selects a shape that is visible from a viewpoint at an arbitrary time, and a scene data provided from the data management unit are used to draw a three-dimensional scene at an arbitrary time. When modeling three-dimensional shape data, one of the edges or vertices of the basic primitive located at the outermost side of the shape is provided. In a plane that passes through multiple vertices of a face that shares the edges of, or through multiple vertices of the face that shares multiple edges,
A three-dimensional moving image generating apparatus, which chamfers an edge or a vertex of the basic primitive. 5. A shape modeling unit that defines a three-dimensional shape in units of basic primitives, a shape registration unit that registers the shape part created by the shape modeling unit, and a shape part registered in the shape registration unit. Is placed in the scene to create a scene, a viewpoint designating unit that sets viewpoint data for observing the scene, and a specific part in the scene for the scene created by the scene creating unit. The motion setting unit that sets the motion information of the scene, and the scene data defined by the scene creation unit, using the viewpoint data defined by the viewpoint setting unit and the motion information defined by the motion setting unit. A data management unit that selects a shape that is visible from the viewpoint at any time, and a scene data provided from the data management unit is used to draw a three-dimensional scene at any time. A scene creating method for arranging a shape part created by the shape modeling section in a scene, the virtual two-dimensional plane constituting the scene is considered, and the scene is divided into a plurality of sub-scenes. It subdivides and arranges and manages parts for each sub-scene, and when parts and viewpoints are arranged in the sub-scene, parts or viewpoints are dropped from a direction perpendicular to the virtual two-dimensional plane and arranged in the sub-scene. A three-dimensional moving image generation apparatus characterized in that a scene is created by determining coordinate positions and further defining heights of parts and viewpoints by distances from a plane. 6. A shape modeling unit that defines a three-dimensional shape as a set of basic primitives, and a shape registration unit that registers the shape parts created by the shape modeling unit.
A scene creation unit that arranges the shape parts registered in the shape registration unit in a scene to create a scene, a viewpoint designation unit that sets viewpoint data for observing the scene, and a scene created by the scene creation unit On the other hand, with regard to the scene data defined by the scene setting unit, the motion setting unit that sets motion information of a specific part in the scene, the viewpoint data defined by the viewpoint setting unit, and the motion setting unit. Using the defined motion information, a data management unit that selects a shape that is visible from a viewpoint at an arbitrary time, and a scene data provided from the data management unit are used to draw a three-dimensional scene at an arbitrary time. And a rendering unit for performing, when generating an image of the scene observed from the set viewpoint position for the created shape part and scene,
An arbitrary part in the scene is managed for each sub-scene obtained by subdividing the scene, and the shape included in the sub-scene including the viewpoint and the shape included in the visible sub-scene closest to the sub-scene. Further, it is characterized in that it is separated into a hierarchical structure composed of shapes included in a visible sub-scene other than the above, and the shape parts are hierarchically managed by three-dimensional data and a two-dimensional image projected on a virtual surface in the sub-scene. And a three-dimensional moving image generation device. 7. A shape modeling unit that defines a three-dimensional shape as a set of basic primitives, and a shape registration unit that registers the shape component created by the shape modeling unit.
A scene creation unit that arranges the shape parts registered in the shape registration unit in a scene to create a scene, a viewpoint designation unit that sets viewpoint data for observing the scene, and a scene created by the scene creation unit On the other hand, with regard to the scene data defined by the scene setting unit, the motion setting unit that sets motion information of a specific part in the scene, the viewpoint data defined by the viewpoint setting unit, and the motion setting unit. Using the defined motion information, a data management unit that selects a shape that is visible from a viewpoint at an arbitrary time, and a scene data provided from the data management unit are used to draw a three-dimensional scene at an arbitrary time. When rendering a three-dimensional image from the viewpoint at the arbitrary time, the shape data is managed by three hierarchical data, and the sub-scene When deciding the virtual projection plane of, the projection image in which all the parts in the sub-scene are projected on the virtual plane parallel to the viewpoint trajectory at any time and the vertical virtual plane is held as image data, and at the time of rendering. ,
A three-dimensional moving image generation apparatus, which draws while converting the viewpoint coordinates using the three-dimensional coordinates of the four vertices of the image data according to the position of the viewpoint. 8. A shape modeling unit that defines a three-dimensional shape as a set of basic primitives, and a shape registration unit that registers the shape component created by the shape modeling unit.
A scene creation unit that arranges the shape parts registered in the shape registration unit in a scene to create a scene, a viewpoint designation unit that sets viewpoint data for observing the scene, and a scene created by the scene creation unit On the other hand, with regard to the scene data defined by the scene setting unit, the motion setting unit that sets motion information of a specific part in the scene, the viewpoint data defined by the viewpoint setting unit, and the motion setting unit. Using the defined motion information, a data management unit that selects a shape that is visible from a viewpoint at an arbitrary time, and a scene data provided from the data management unit are used to draw a three-dimensional scene at an arbitrary time. When rendering a three-dimensional image from the viewpoint at the arbitrary time, the viewpoint position at the arbitrary time and all of the scenes to be drawn are included. Rendering data is selected from three hierarchical data according to the distance to each sub-scene, and three-dimensional data is used for the shape in the sub-scene including the viewpoint, and the visible sub-scene adjacent to the sub-scene is selected. The three-dimensional moving image generating apparatus according to claim 7, wherein the included shape is drawn using image data while changing the complexity of the shape according to the distance from the viewpoint.