Patent Application
20250218111
This application relates to the field of computer technologies, and in particular, to a light update method and apparatus for a virtual scene, an electronic device, a computer-readable storage medium, and a computer program product.
With development of multimedia technologies, types of games are increasingly diversified, and functions are increasingly enriched. To provide a player with more real game experience, a person skilled in the art focuses on improving fineness of a game picture. For example, a person skilled in the art improves overall fineness of a game picture by improving a degree of realism of light in the game picture.
In the related art, for high-frequency indirect light information restored by using a screen space occlusion technology, because a light update needs to be comprehensively performed on an entire virtual scene, light update efficiency for the virtual scene is low.
Embodiments of this application provide a light update method and apparatus for a virtual scene, an electronic device, a computer-readable storage medium, and a computer program product, to effectively improve light update efficiency for a virtual scene.
Technical solutions in embodiments of this application are implemented as follows:
Embodiments of this application provide a method for updating light in response to a content change in a virtual scene performed by an electronic device, the method including:
Embodiments of this application provide an electronic device, including:
Embodiments of this application provide a non-transitory computer-readable storage medium, having computer-executable instructions stored therein. The computer-executable instructions are configured for implementing the light update method for a virtual scene in response to a content change in the virtual scene provided in embodiments of this application when being executed by a processor of an electronic device.
Embodiments of this application have the following beneficial effects:
Voxelization processing is performed on a virtual scene to obtain a plurality of initial voxels corresponding to the virtual scene. At least one to-be-updated voxel is selected from the plurality of initial voxels in response to a content change in the virtual scene. Light information carried in each to-be-updated voxel is updated. Light information of each virtual viewpoint in a changed virtual scene is updated based on the updated voxel and an initial voxel among the plurality of initial voxels other than the to-be-updated voxel. In this way, at least one to-be-updated voxel is selected from the plurality of initial voxels in response to the content change in the virtual scene. Because a quantity of to-be-updated voxels is less than a quantity of initial voxels, the quantity of to-be-updated voxels is effectively reduced, so that update costs are effectively reduced, and light update efficiency for the virtual scene can be effectively improved.
To make the objectives, technical solutions, and advantages of this application clearer, the following describes this application in further detail with reference to the accompanying drawings. The described embodiments are not to be considered as a limitation to this application. All other embodiments obtained by a person of ordinary skill in the art without creative efforts shall fall within the protection scope of this application.
In the following descriptions, the term “some embodiments” describes subsets of all possible embodiments, but “some embodiments” may be the same subset or different subsets of all the possible embodiments, and can be combined with each other without conflict.
In the following descriptions, the terms “first”, “second”, and “third” are merely intended to distinguish between similar objects rather than describe a specific order of objects. The “first”, “second”, and “third” are interchangeable in order in proper circumstances, so that embodiments of this application described herein can be implemented in an order other than the order illustrated or described herein.
Unless otherwise defined, meanings of all technical and scientific terms used in this specification are the same as those usually understood by a person skilled in the art to which this application belongs. The terms used in this specification are merely intended to describe the objectives of embodiments of this application, but are not intended to limit this application.
Before embodiments of this application are further described in detail, terms in embodiments of this application are described, and the following explanations are applicable to the terms in embodiments of this application.
During implementation of embodiments of this application, the applicant finds that the related art has the following problems:
In the related art, for high-frequency indirect light information restored by using a screen space occlusion technology, because a light update needs to be comprehensively performed on an entire virtual scene, light update efficiency for the virtual scene is low.
Embodiments of this application provide a light update method and apparatus for a virtual scene, an electronic device, a computer-readable storage medium, and a computer program product, to effectively improve light update efficiency for a virtual scene. The following describes exemplary application of a light update system for a virtual scene provided in embodiments of this application.
The terminal 400 is configured to display a virtual scene on a graphical interface 410-1 (a graphical interface 410-1 is shown as an example) by a user by using a client 410. The terminal 400 and the server 200 are connected to each other through a wired or wireless network.
In some embodiments, the server 200 may be an independent physical server, or may be a server cluster or a distributed system that includes a plurality of physical servers, or may be a cloud server that provides basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a network service, cloud communication, a middleware service, a domain name service, a security service, a content delivery network (CDN), big data, and an AI platform. The terminal 400 may be a smartphone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart television, a smartwatch, an in-vehicle terminal, or the like, but is not limited thereto. The electronic device provided in embodiments of this application may be implemented by a terminal, or may be implemented by a server. The terminal and the server may be directly or indirectly connected through wired or wireless communication. This is not limited in embodiments of this application.
In some embodiments, the terminal 400 sends a virtual scene to the server 200; and the server 200 performs voxelization processing on the virtual scene to obtain a plurality of initial voxels, selects at least one to-be-updated voxel from the plurality of initial voxels in response to a content change in the virtual scene, updates light information carried in each to-be-updated voxel to obtain an updated voxel corresponding to each to-be-updated voxel, and updates light information of each virtual viewpoint in a changed virtual scene.
In some other embodiments, the server 200 performs voxelization processing on a virtual scene to obtain a plurality of initial voxels corresponding to the virtual scene, selects at least one to-be-updated voxel from the plurality of initial voxels in response to a content change in the virtual scene, and sends the to-be-updated voxel to the terminal 400; and the terminal 400 updates light information carried in each to-be-updated voxel to obtain an updated voxel corresponding to each to-be-updated voxel, and updates light information of each virtual viewpoint in a changed virtual scene.
In some embodiments, the terminal 400 sends a virtual scene to the server 200; and the server 200 performs voxelization processing on the virtual scene to obtain a plurality of initial voxels, selects at least one to-be-updated voxel from the plurality of initial voxels in response to a content change in the virtual scene, updates light information carried in each to-be-updated voxel to obtain an updated voxel corresponding to each to-be-updated voxel, and updates light information of each virtual viewpoint in a changed virtual scene.
In some other embodiments, embodiments of this application may be implemented by using a cloud technology. The cloud technology is a hosting technology that integrates a series of resources such as hardware, software, and network resources in a wide area network or a local area network to implement data computing, storage, processing, and sharing.
The cloud technology is a general term for a network technology, an information technology, an integration technology, a management platform technology, an application technology, and the like that are based on application of a cloud computing business model, and may constitute a resource pool for use on demand and therefore is flexible and convenient. A cloud computing technology is to become an important support. A background service of a technology network system requires a large number of computing and storage resources.
The processor 430 may be an integrated circuit chip with a signal processing capability, for example, a general-purpose processor, a digital signal processor (DSP), another programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The general-purpose processor may be a microprocessor, any conventional processor, or the like.
The memory 450 may be a removable memory, a non-removable memory, or a combination thereof. For example, a hardware device includes a solid-state memory, a hard disk drive, or an optical drive. In some embodiments, the memory 450 includes one or more storage devices physically located away from the processor 430.
The memory 450 includes a volatile memory or a non-volatile memory, or may include both a volatile memory and a non-volatile memory. The non-volatile memory may be a read-only memory (ROM). The volatile memory may be a random access memory (RAM). The memory 450 described in this embodiment of this application is intended to include any suitable types of memories.
In some embodiments, the memory 450 is capable of storing data to support various operations. Examples of the data include a program, a module, and a data structure or a subset or superset thereof. Examples are described below:
In some embodiments, the light update apparatus for a virtual scene provided in embodiments of this application may be implemented by using software.
In some other embodiments, the light update apparatus for a virtual scene provided in embodiments of this application may be implemented by using hardware. In an example, the light update apparatus for a virtual scene provided in embodiments of this application may be a processor in a form of a hardware decoding processor, and is programmed to perform the light update method for a virtual scene provided in embodiments of this application. For example, the processor in the form of the hardware decoding processor may be one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic elements.
In some embodiments, the terminal or the server may implement the light update method for a virtual scene provided in embodiments of this application by running a computer program or computer-executable instructions. For example, the computer program may be a native program (for example, a dedicated light update program) or a software module in an operating system, for example, a light update module that may be embedded in any program (for example, an instant messaging client, an album program, an electronic map client, or a navigation client). For example, the computer program may be a native application (APP), to be specific, a program that needs to be installed in an operating system to run. To sum up, the computer program may be an application, a module, or a plug-in in any form.
The light update method for a virtual scene provided in embodiments of this application is described with reference to the exemplary application and implementation of the server or the terminal provided in embodiments of this application.
Operation 101: Perform voxelization processing on a virtual scene to obtain a plurality of initial voxels corresponding to the virtual scene.
In some embodiments, different initial voxels have different positions in the virtual scene, and the initial voxel carries light information of a corresponding position in the virtual scene.
In some embodiments, the virtual scene is a virtual scene displayed or provided by an application when the application is run on a terminal. The virtual scene may be a simulated environment of a real world, or may be a semi-simulated semi-fictional virtual environment, or may be an entirely fictional virtual environment. The virtual scene may be any one of a two-dimensional virtual scene, a 2.5-dimensional virtual scene, or a three-dimensional virtual scene. A dimension of the virtual scene is not limited in this embodiment of this application. For example, the virtual scene may include the sky, the land, or the ocean. The land may include environmental elements such as the desert and a city. A user may control a virtual object to move in the virtual scene.
In some embodiments, the voxelization processing is a processing process of dividing the virtual scene into the plurality of initial voxels, the initial voxel carries light information of a corresponding position in the virtual scene, and different initial voxels have different positions in the virtual scene.
In some embodiments, operation 101 may be implemented in the following manner: The virtual scene is divided into a plurality of virtual scene blocks. Positions of the virtual scene blocks in the virtual scene are different, and sizes of the virtual scene blocks may be the same or different. Virtual scene information corresponding to each virtual scene block is obtained. The virtual scene information includes a virtual scene color, diffuse reflection information of the virtual scene, direct light information, and the like. Corresponding virtual scene information is assigned to each virtual scene block to obtain an initial voxel corresponding to the virtual scene block.
In some embodiments, a voxel carrying normal line information is configured for representing a surface element, to simplify description of a three-dimensional scene. First, cuboid space with a camera as a center is maintained, and a voxel is configured for simplifying description of three-dimensional scene information in the space. Content of the voxel includes a normal line, a diffuse reflection color, direct light information, and radiance information. To reduce memory consumption, in this embodiment of this application, voxels are sparsely stored in a hierarchical manner, and 4×4×4 voxels are considered as a group, which is referred to as a brick. A hierarchical index is represented by an n×m×k three-dimensional map. If a brick includes a voxel, an actual storage position of the voxel may be obtained based on the hierarchical index. If a value obtained based on the hierarchical index is 0, the brick does not include a voxel. In this way, memory space can be greatly saved.
Operation 102: Select at least one to-be-updated voxel from the plurality of initial voxels in response to a content change in the virtual scene.
In some embodiments, a quantity of to-be-updated voxels is less than a quantity of initial voxels. Because the quantity of to-be-updated voxels is less than the quantity of initial voxels, the quantity of to-be-updated voxels is effectively reduced, so that update costs are effectively reduced, and light update efficiency for the virtual scene can be effectively improved.
In some embodiments, the content change in the virtual scene may be movement of a virtual item or movement of a virtual object in the virtual scene, a change in a color or a form of a virtual item or a virtual object in the virtual scene, or the like. Content changes may occur in the virtual scene at different time points. For example, a virtual object A moves from a position B to a position C in the virtual scene, and a content change occurs in the virtual scene during the movement from the position B to the position C.
In some embodiments, duration of the content change in the virtual scene may be determined based on an actual change situation. The duration of the content change in the virtual scene may be a range of a time interval from the start of the charge to the end of the change.
In an example, the virtual object A in the virtual scene starts to move at the twelfth minute and the tenth second. To be specific, a content change starts to occur in the virtual scene at the twelfth minute and the tenth second. During the movement, the virtual object A starts from the position B until the virtual object A moves to the position C. Time at which the virtual object A moves to the position C is the fourteenth minute and the twentieth second. To be specific, the content change stops in the virtual scene at the fourteenth minute and the twentieth second.
In some embodiments, in a process of selecting at least one to-be-updated voxel from the plurality of initial voxels, a quantity of selection actions is positively correlated with the duration of the content change in the virtual scene. For example, if the duration of the content change in the virtual scene is 10 seconds, a to-be-updated voxel is selected from the plurality of initial voxels at least once per second during the 10 seconds in which the content change occurs. Longer duration of the content change in the virtual scene indicates a larger quantity of selected to-be-updated voxels.
In some embodiments, the content change in the virtual scene may be movement of a virtual light source in the virtual scene. A light source position of the virtual light source and light intensity of the virtual light source may change in the virtual scene at different time points. For example, a virtual light source D moves from a position E to a position F in the virtual scene, and a content change occurs in the virtual scene during the movement from the position E to the position F. For another example, a content change occurs in the virtual scene in a process in which light intensity of the virtual light source becomes stronger or weaker.
In some embodiments, operation 102 is described below by using an example in which a voxel is selected once.
Operation 1021: In response to the content change in the virtual scene, obtain a camera position of a virtual camera in a virtual scene with changed content.
In some embodiments, the virtual camera is a “camera” built in computer animation software or a virtual engine. A function of the virtual camera for representing a viewpoint during animation production is equivalent to a camera in a conventional sense. An object photographed by the virtual camera is completely different from an object photographed by a physical camera, but functions of the cameras are quite similar. The physical camera photographs a real person or an actually established scene. The virtual camera photographs a model established in three-dimensional software, and can implement infinite possibilities. The virtual camera is presented in a virtual engine in a form of an icon, and also has parameters such as a lens, a focal length, a focus, an aperture, and a depth of field. The virtual camera can implement camera actions such as “pushing, pulling, shaking, moving, following, throwing, rising, lowering, and comprehensive movement”, and can implement photographing effects that are difficult or even impossible to be implemented by the physical camera, for example, passing through a wall, passing through a keyhole, and passing through an object. Parameters that need to be adjusted for the physical camera are distributed on a body of the physical camera, and manual operations are required. A camera parameter of the virtual camera is a button or a value input bar integrated in a panel. An operator only needs to enter a parameter or drag a mouse. Sometimes, a motion path of the virtual camera may be determined by using only several key frames. During actual photographing, the physical camera usually needs to have a stabilizer or a motion control system. However, even in this case, a picture still shakes.
In some embodiments, the camera position is configured for indicating position coordinates of the virtual camera in the virtual scene with changed content. When the virtual scene is a two-dimensional virtual scene, the position coordinates are two-dimensional position coordinates. When the virtual scene is a three-dimensional virtual scene, the position coordinates are three-dimensional position coordinates.
In an example, if a camera position of the virtual camera in the virtual scene is a position G and the camera position of the virtual camera in the virtual scene with changed content is a position H, the virtual camera has a position offset in the virtual scene before and after the position changes. If a camera position of the virtual camera in the virtual scene is a position G and the camera position of the virtual camera in the virtual scene with changed content is the position G, the virtual camera has no position offset in the virtual scene before and after the position changes.
Operation 1022: Obtain a voxel position of a voxel center point of each initial voxel in the virtual scene, and determine a voxel distance between the camera position and each voxel position.
In some embodiments, the voxel position of the voxel center point of the initial voxel in the virtual scene is configured for indicating position coordinates of the voxel center point in the virtual scene. When the virtual scene is a two-dimensional virtual scene, the position coordinates are two-dimensional position coordinates. When the virtual scene is a three-dimensional virtual scene, the position coordinates are three-dimensional position coordinates.
In some embodiments, the voxel distance between the camera position and each voxel position may be determined by using a distance calculation method such as a Euclidean distance, a Manhattan distance, or a Chebyshev distance. A specific calculation method for the voxel distance does not constitute a limitation on this application. The voxel distance is configured for indicating a distance between a voxel center point of an initial voxel and the virtual camera.
Operation 1023: Select at least one to-be-updated voxel from the plurality of initial voxels based on the voxel distance.
In some embodiments, an initial voxel with a smaller voxel distance has a higher probability of being selected as a to-be-updated voxel, and a value of the voxel distance is directly proportional to a value of a probability of being selected as a to-be-updated voxel.
In some embodiments, operation 1023 may be implemented in the following manner: determining the initial voxel as a first initial voxel when the voxel distance of the initial voxel is less than or equal to a voxel distance threshold; or determining the initial voxel as a second initial voxel when the voxel distance of the initial voxel is greater than the voxel distance threshold; and selecting a first quantity of first initial voxels and a second quantity of second initial voxels from the plurality of initial voxels, and determining the selected first initial voxel and second initial voxel as the to-be-updated voxel.
In some embodiments, the first quantity is greater than the second quantity, and the first quantity is at least one. Because the first quantity is greater than the second quantity and the first quantity is at least one, the second quantity is at least zero.
In some embodiments, the voxel distance threshold may be specifically set based on a real situation, and the voxel distance threshold is less than a maximum voxel distance. Initial voxels may be classified into the first initial voxel and the second initial voxel based on comparison between voxel distances and the voxel distance threshold. A voxel distance of the first initial voxel is less than a voxel distance of the second initial voxel.
In this way, the first quantity of first initial voxels and the second quantity of second initial voxels are selected from the plurality of initial voxels, and the selected first initial voxel and second initial voxel are determined as the to-be-updated voxel. Because the selected first quantity is greater than the second quantity, a quantity of selected second initial voxels farther away from the virtual camera is less than a quantity of first initial voxels closer to the virtual camera. Because a possibility that the first initial voxel closer to the virtual camera includes a virtual viewpoint is higher, selecting more first initial voxels can effectively ensure that light information of the virtual viewpoint is subsequently updated. This effectively ensures light update accuracy for the virtual viewpoint, and therefore effectively improves light update accuracy.
In an example, when the quantity of initial voxels is 10, a quantity of first initial voxels whose voxel distances are less than or equal to the voxel distance threshold may be 5, a quantity of second initial voxels whose voxel distances are greater than the voxel distance threshold may be 5, the selected first quantity may be 3, and the selected second quantity may be 1.
In an example, when the quantity of initial voxels is 10, a quantity of first initial voxels whose voxel distances are less than or equal to the voxel distance threshold may be 5, a quantity of second initial voxels whose voxel distances are greater than the voxel distance threshold may be 5, the selected first quantity may be 1, and the selected second quantity may be 0.
In this way, at least one to-be-updated voxel is selected from the plurality of initial voxels in response to the content change in the virtual scene, and the quantity of to-be-updated voxels is less than the quantity of initial voxels. Because the quantity of to-be-updated voxels is less than the quantity of initial voxels, the quantity of to-be-updated voxels is effectively reduced, so that update costs are effectively reduced, and light update efficiency for the virtual scene can be effectively improved.
Operation 103: Update light information carried in each to-be-updated voxel, to obtain an updated voxel corresponding to each to-be-updated voxel.
In some embodiments, operation 103 may be implemented in the following manner: performing the following processing on each to-be-updated voxel: updating light information, carried in the to-be-updated voxel, of a corresponding position in the virtual scene, to obtain an updated voxel corresponding to the to-be-updated voxel.
In some embodiments,
Operation 1031: Determine, from virtual scene points of the virtual scene, a plurality of target virtual scene points located in the to-be-updated voxel.
In some embodiments, a virtual scene point is a minimum constituent unit in the virtual scene, the virtual scene includes a plurality of virtual scene points, and different virtual scene points have different positions in the virtual scene.
In some embodiments, operation 1031 may be implemented in the following manner: performing the following processing on each virtual scene point in the virtual scene: comparing a scene point position of the virtual scene point in the virtual scene with a voxel position range of the to-be-updated voxel in the virtual scene, and determining the virtual scene point as a target virtual scene point when the scene point position falls within the voxel position range.
In an example, the scene point position of the virtual scene point in the virtual scene is (0.2, 5), a horizontal coordinate of the voxel position range of the to-be-updated voxel in the virtual scene ranges from 0 to 1, and a vertical coordinate ranges from 4 to 6. In this case, the scene point position falls within the voxel position range, and the virtual scene point is determined as a target virtual scene point.
Operation 1032: Obtain target light information of each target virtual scene point in the to-be-updated voxel.
In some embodiments, the target light information is configured for indicating light intensity or irradiance reflected at the target virtual scene point. The light intensity is a light flux of received visible light per unit area. The light intensity is referred to as radiance for short, and is measured in Lux (or lx). The light intensity is configured for indicating intensity of light and an amount of light in a surface area of an object. The irradiance, also referred to as radiance, is a radiant flux in a unit area of an irradiated surface and is measured in watt per square meter (W/m2).
In some embodiments, operation 1032 may be implemented in the following manner: performing the following processing on each target virtual scene point in the to-be-updated voxel: obtaining direct light information of the target virtual scene point; obtaining indirect light information of the target virtual scene point; and summing the direct light information and the indirect light information to obtain the target light information of the target virtual scene point.
In some embodiments, the direct light information is configured for indicating impact of a direct ray emitted by the virtual light source on light of the target virtual scene point, and the indirect light information is configured for indicating impact of a reflected ray corresponding to the direct ray on light of the target virtual scene point.
In some embodiments, when the direct light information is direct light intensity and the indirect light information is indirect light intensity, the direct light intensity and the indirect light intensity are summed to obtain a light intensity sum value, and the light intensity sum value is determined as the target light information of the target virtual scene point.
In an example, an expression of the target light information of the target virtual scene point may be as follows:
G is configured for indicating the target light information of the target virtual scene point, G1 is configured for indicating the direct light information of the target virtual scene point, and G2 is configured for indicating the indirect light information of the target virtual scene point.
In some embodiments, when the indirect light information is indirect light irradiance and the direct light information is direct light irradiance, the direct light irradiance and the indirect light irradiance are summed to obtain a light irradiance sum value, and the light irradiance sum value is determined as the target light information of the target virtual scene point.
In some embodiments, the direct light information includes direct light intensity, and the obtaining direct light information of the target virtual scene point may be implemented in the following manner: determining a light source distance between the virtual light source in the virtual scene and the target virtual scene point; obtaining a camera distance between the target virtual scene point and the virtual camera, and adding up the camera distance and the light source distance to obtain a total distance; determining a loss value of light source light intensity of the virtual light source based on the total distance and the target virtual scene point; and subtracting the loss value from the light source light intensity to obtain the direct light intensity of the target virtual scene point.
In some embodiments, the loss value of the light source light intensity of the virtual light source is a loss of the light source light intensity during ray transmission of a virtual ray transmitted by the virtual light source from the virtual light source to the target virtual scene point and from the target virtual scene point to the virtual camera.
In some embodiments, the determining a loss value of light source light intensity of the virtual light source based on the total distance and the target virtual scene point may be implemented in the following manner: determining a first loss value of the virtual light source based on the total distance, determining a second loss value of the virtual light source based on the target virtual scene point, and adding up the first loss value and the second loss value to obtain the loss value of the light source light intensity of the virtual light source.
In some embodiments, the first loss value is configured for indicating a loss of the light source light intensity of the virtual light source in a ray path, and the second loss value is configured for indicating a loss of the light source light intensity of the virtual light source at a target virtual point.
In some embodiments, the ray path includes a first sub-path with the virtual light source as a start point and the target virtual scene point as an end point and a second sub-path with the target virtual scene point as a start point and the virtual camera as an end point in the virtual scene.
In some embodiments, the determining a first loss value of the virtual light source based on the total distance may be implemented in the following manner: obtaining a unit loss value of the light source light intensity of the virtual light source at a unit distance, and determining a product of the total distance and the unit loss value as the first loss value.
In some embodiments, the determining a second loss value of the virtual light source based on the target virtual scene point may be implemented in the following manner: obtaining an absorption rate of the target virtual scene point for the light source light intensity of the virtual light source, and determining a product of the absorption rate and the light source light intensity as the second loss value, the absorption rate being associated with a physical material indicated by the target virtual scene point.
In this way, the direct light intensity of the target virtual scene point is accurately determined by determining the loss value of the light source light intensity in the ray path, so that the target light information of the target virtual scene point is subsequently accurately determined with reference to the indirect light intensity. This effectively improves light accuracy of the virtual scene.
In some embodiments, the indirect light information includes indirect light intensity, and the obtaining indirect light information of the target virtual scene point may be implemented in the following manner: determining, from a plurality of light probes arranged in the virtual scene, at least one target light probe whose distance from the target virtual scene point is less than a distance threshold; and when one target light probe exists, determining light intensity stored in the target light probe as the indirect light intensity of the target virtual scene point; or when a plurality of target light probes exist, determining a weight of each target light probe based on a probe distance between each target light probe and the target virtual scene point; and performing, based on the weight, weighted summation on light intensity stored in all target light probes, to obtain the indirect light intensity of the target virtual scene point.
In some embodiments, the light probe is configured to store light intensity at a corresponding position in the virtual scene with changed content.
In some embodiments, the light probe stores “baking” information of light in the virtual scene, and a light map stores light information about a ray irradiating a surface of a virtual item in the virtual scene, but a light detector stores information about a ray passing through empty space in the virtual scene.
In some embodiments, a value of the probe distance is inversely proportional to a value of the weight of the target light probe. A larger probe distance corresponds to a smaller value of the weight of the target light probe. A smaller probe distance corresponds to a larger value of the weight of the target light probe. A sum of weights of all target light probes is equal to 1.
In an example, when a plurality of target light probes exist, for example, when a target light probe 1, a target light probe 2, and a target light probe 3 exist, a probe distance corresponding to the target light probe 1 is 3, a probe distance corresponding to the target light probe 2 is 4, a probe distance corresponding to the target light probe 3 is 5, a probe weight corresponding to the target light probe 1 is 0.2, a probe weight corresponding to the target light probe 2 is 0.3, and a probe weight corresponding to the target light probe 3 is 0.5.
In this way, at least one target light probe whose distance from the target virtual scene point is less than the distance threshold is determined from the plurality of light probes arranged in the virtual scene, a weight corresponding to each target light probe is determined, and weighted summation is performed, based on the weight, on light information stored in all target light probes, to obtain the indirect light intensity of the target virtual scene point. The light information stored in the target light probe can accurately indicate light information of a corresponding position in the virtual scene with changed content. In addition, the target light probe is close to the target virtual scene point, and therefore can more accurately indicate actual indirect light of the target virtual scene point. Moreover, weighted summation is performed on light information of a plurality of target light probes, to integrate the light information of the plurality of target light probes, so that the determined indirect light intensity can more accurately indicate indirect light in the virtual scene with changed content, and accuracy of the determined indirect light intensity is effectively improved.
In some embodiments, before the determining, from a plurality of light probes arranged in the virtual scene, at least one target light probe whose distance from the target virtual scene point is less than a distance threshold, light probes may be further arranged in the following manner: obtaining a camera position of the virtual camera in the virtual scene; determining a virtual scene region that is in the virtual scene and whose distance from the camera position is less than a distance threshold as a first virtual scene region, and determining a virtual scene region that is in the virtual scene and whose distance from the camera position is greater than or equal to the distance threshold as a second virtual scene region; and arranging a third quantity of light probes in the first virtual scene region, and arranging a fourth quantity of light probes in the second virtual scene region.
In some embodiments, the third quantity is greater than the fourth quantity, and the third quantity is an integer greater than or equal to 1.
In some embodiments, the distance threshold may be specifically set based on an actual situation, and the distance threshold is configured for dividing the virtual scene into at least two virtual scene regions.
In an example, 20 light probes are arranged in the first virtual scene region, and 15 light probes are arranged in the second virtual scene region.
In this way, because a quantity of virtual viewpoints (virtual scene points that can be captured by the virtual camera) in the first virtual scene region is greater than a quantity of virtual viewpoints in the second virtual scene region, arranging more light probes in the first virtual scene region can effectively ensure light accuracy of the first virtual scene region, and arranging a small quantity of light probes in the second virtual scene region can reduce a quantity of arranged light probes while ensuring light accuracy of the second virtual scene region. This effectively saves storage space, and effectively improves light update efficiency for the virtual scene.
Operation 1033: Perform weighted summation on all pieces of target light information to obtain updated light information.
In some embodiments, operation 1033 may be implemented in the following manner: performing the following processing on each piece of target light information: multiplying the target light information by a corresponding weight to obtain a light product result; and adding up light product results corresponding to all pieces of target light information, to obtain the updated light information.
Operation 1034: Update the light information carried in the to-be-updated voxel to the updated light information, to obtain the updated voxel corresponding to the to-be-updated voxel.
In some embodiments, operation 1034 may be implemented in the following manner: replacing the light information carried in the to-be-updated voxel with the updated light information, and determining a to-be-updated voxel obtained through replacement as the updated voxel corresponding to the to-be-updated voxel.
In this way, some initial voxels (to-be-updated voxels) of the virtual scene are updated to obtain updated voxels corresponding to the to-be-updated voxels, so that a quantity of to-be-updated voxels are effectively reduced, update costs are effectively reduced, and light update efficiency for the virtual scene can be effectively improved. In addition, indirect light information of the to-be-updated voxel can be accurately determined through arrangement of the light probes, direct light information of the to-be-updated voxel can be accurately determined through calculation of the loss value of the light source light intensity, and the direct light information and the indirect light information are summed, so that the obtained updated light information is more accurate, and accuracy of the determined updated light information is effectively improved.
Operation 104: Update light information of each virtual viewpoint in the virtual scene with changed content based on the updated voxel and an initial voxel among the plurality of initial voxels other than the to-be-updated voxel.
In some embodiments, the virtual viewpoint is a virtual scene point able to be captured by the virtual camera in the virtual scene with changed content.
In some embodiments, the updating light information of each virtual viewpoint in the virtual scene with changed content is a processing process of updating the light information of each virtual viewpoint in the virtual scene with changed content based on the updated light information of the updated voxel and light information of the initial voxel among the plurality of initial voxels other than the to-be-updated voxel.
In some embodiments,
Operation 1041: Determine, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located, and obtain light information of the target voxel.
In some embodiments, the determining, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located may be implemented in the following manner: comparing a viewpoint position of the virtual viewpoint in the virtual scene with position ranges respectively corresponding to the updated voxel and the initial voxel other than the to-be-updated voxel, and determining, as the target voxel, a voxel corresponding to a position range within which the viewpoint position falls.
Operation 1042: Determine the light information of the target voxel as updated light information of the virtual viewpoint.
In some embodiments, operation 1042 may be implemented in the following manner: when the target voxel is the updated voxel, determining the updated light information of the updated voxel as the updated light information of the virtual viewpoint; or when the target voxel is the initial voxel other than the to-be-updated voxel, determining initial light information of the initial voxel as the updated light information of the virtual viewpoint.
Operation 1043: Update the light information of the virtual viewpoint to the updated light information of the virtual viewpoint.
In this way, a target voxel in which each virtual viewpoint is located is determined, and light information of the target voxel is determined as updated light information of the virtual viewpoint. When the target voxel is the to-be-updated voxel, because of accuracy of the updated light information of the to-be-updated voxel, the updated light information of the to-be-updated voxel is determined as the updated light information of the virtual viewpoint, so that the updated light information of the virtual viewpoint is more accurate.
In some embodiments,
Operation 105: Perform spatial-domain correction on the updated light information to obtain first light information.
In some embodiments, the spatial-domain correction is a processing process of performing error correction on the updated light information of the target voxel based on a voxel other than the target voxel corresponding to the virtual viewpoint in scene space of the virtual scene. Accuracy of the first light information is higher than accuracy of the updated light information.
In some embodiments, operation 105 may be implemented in the following manner: determining, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located, and a plurality of adjacent voxels adjacent to the target voxel; selecting a target adjacent voxel from the plurality of adjacent voxels; obtaining light information of the target voxel and light information of each target adjacent voxel; and performing weighted summation on the light information of the target voxel and the light information of each target adjacent voxel, to obtain the first light information of the virtual viewpoint.
In some embodiments, a target voxel in which the virtual viewpoint is located and a plurality of adjacent voxels adjacent to the target voxel are determined from the updated voxel and the initial voxel other than the to-be-updated voxel.
In some embodiments, the determining, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located may be implemented in the following manner: comparing a viewpoint position of the virtual viewpoint in the virtual scene with position ranges respectively corresponding to the updated voxel and the initial voxel other than the to-be-updated voxel, and determining, as the target voxel, a voxel corresponding to a position range within which a voxel position falls.
In some embodiments, the adjacent voxel is a voxel that has an adjacency relationship with the target voxel.
In some embodiments, a target adjacent voxel is selected from the plurality of adjacent voxels.
In some embodiments, the selecting a target adjacent voxel from the plurality of adjacent voxels may be implemented in the following manner: performing the following processing on each adjacent voxel: obtaining first occlusion information of the adjacent voxel and second occlusion information of the target voxel; and determining the adjacent voxel as the target adjacent voxel when the first occlusion information indicates that no virtual item exists between the adjacent voxel and the virtual camera and the second occlusion information indicates that no virtual item exists between the target voxel and the virtual camera; or performing an occlusion check on the adjacent voxel and the target voxel to obtain an occlusion check result when the first occlusion information indicates that a virtual item exists between the adjacent voxel and the virtual camera and the second occlusion information indicates that a virtual item exists between the target voxel and the virtual camera; and determining the adjacent voxel as the target adjacent voxel when the occlusion check result indicates that the adjacent voxel and the target voxel pass the occlusion check.
In some embodiments, the first occlusion information of the adjacent voxel is configured for indicating whether a virtual item exists between the adjacent voxel and the virtual camera, and the second occlusion information of the target voxel is configured for indicating whether a virtual item exists between the target voxel and the virtual camera.
In some embodiments, when the first occlusion information of the adjacent voxel indicates that no virtual item exists between the adjacent voxel and the virtual camera, no virtual occlusion exists between the adjacent voxel and the virtual camera. In this case, a virtual scene point in the adjacent voxel is more likely to be captured by the virtual camera (provided that a plurality of virtual scene points in the adjacent voxel include at least one virtual viewpoint). When the second occlusion information indicates that no virtual item exists between the target voxel and the virtual camera, no virtual occlusion exists between the target voxel and the virtual camera. In this case, a virtual scene point in the target voxel is more likely to be captured by the virtual camera (provided that a plurality of virtual scene points in the target voxel include at least one virtual viewpoint, to be specific, provided that the plurality of virtual scene points in the target voxel include at least one virtual scene point that can be irradiated by a direct ray or an indirect ray). Therefore, when the first occlusion information indicates that no virtual item exists between the adjacent voxel and the virtual camera and the second occlusion information indicates that no virtual item exists between the target voxel and the virtual camera, both the target voxel and the adjacent voxel corresponding to the target voxel are more likely to be captured by the virtual camera. In this case, the adjacent voxel may be determined as the target adjacent voxel.
In some embodiments, when the first occlusion information indicates that a virtual item exists between the adjacent voxel and the virtual camera and the second occlusion information indicates that a virtual item exists between the target voxel and the virtual camera, both the target voxel and the adjacent voxel corresponding to the target voxel are less likely to be captured by the virtual camera. In this case, an occlusion check needs to be performed on the adjacent voxel and the target voxel, and the adjacent voxel is determined as the target adjacent voxel when an occlusion check result indicates that the adjacent voxel and the target voxel pass the occlusion check.
In some embodiments, when the first occlusion information indicates that a virtual item exists between the adjacent voxel and the virtual camera or the second occlusion information indicates that a virtual item exists between the target voxel and the virtual camera, the adjacent voxel is determined as a non-adjacent voxel.
In this way, the target adjacent voxel is selected from the plurality of adjacent voxels adjacent to the target voxel, and spatial-domain correction is performed on the updated light information of the target voxel based on the target adjacent voxel, so that the obtained first light information is more accurate, and the updated light information of the virtual viewpoint is more accurate.
In some embodiments, the performing an occlusion check on the adjacent voxel and the target voxel to obtain an occlusion check result may be implemented in the following manner: obtaining a first distance between the adjacent voxel and the corresponding virtual item, and a second distance between the target voxel and the corresponding virtual item; performing a first occlusion check on the adjacent voxel and the target voxel based on the first distance and the second distance, to obtain a first occlusion check result; and determining the occlusion check result as a first result when the first occlusion check result indicates that the adjacent voxel and the target voxel pass the first occlusion check; or when the first occlusion check result indicates that the adjacent voxel and the target voxel fail the first occlusion check, obtaining a third distance between the adjacent voxel and the target voxel, and performing a second occlusion check on the adjacent voxel and the target voxel based on the second distance and the third distance, to obtain a second occlusion check result; and determining the occlusion check result as a first result when the second occlusion check result indicates that the adjacent voxel and the target voxel pass the second occlusion check; or determining the occlusion check result as a second result when the second occlusion check result indicates that the adjacent voxel and the target voxel fail the second occlusion check.
In some embodiments, the first result is configured for indicating that the adjacent voxel and the target voxel pass the occlusion check, and the second result is configured for indicating that the adjacent voxel and the target voxel fail the occlusion check.
In some embodiments, when the first occlusion information indicates that a virtual item exists between the adjacent voxel and the virtual camera and the second occlusion information indicates that a virtual item exists between the target voxel and the virtual camera, two occlusion checks (the first occlusion check and the second occlusion check) are performed based on distances between the adjacent voxel, the virtual item corresponding to the adjacent voxel, the target voxel, and the virtual item corresponding to the target voxel, to accurately determine whether the adjacent voxel is to be determined as the target adjacent voxel.
In some embodiments, when the first occlusion check result indicates that the adjacent voxel and the target voxel fail the first occlusion check, the adjacent voxel cannot be determined as the target adjacent voxel. In this case, the second occlusion check further needs to be performed on the adjacent voxel and the target voxel, and the adjacent voxel can be determined as the target adjacent voxel only when the adjacent voxel and the target voxel pass the second occlusion check.
In some embodiments, the performing a first occlusion check on the adjacent voxel and the target voxel based on the first distance and the second distance, to obtain a first occlusion check result may be implemented in the following manner: comparing the first distance with the second distance to obtain a first comparison result; and determining the first occlusion check result as a third result when the first comparison result indicates that the first distance is less than or equal to the second distance and a dot product of a direction vector from the virtual viewpoint to a voxel center point of the target voxel and a direction vector from the target voxel to the corresponding virtual item is less than a dot product threshold.
In some embodiments, the third result is configured for indicating that the adjacent voxel and the target voxel pass the first occlusion check.
In some embodiments, the first occlusion check result is determined as a fifth result when the first comparison result indicates that the first distance is greater than the second distance or a dot product of a direction vector from the virtual viewpoint to a voxel center point of the target voxel and a direction vector from the target voxel to the corresponding virtual item is greater than or equal to a dot product threshold.
In some embodiments, the fifth result is configured for indicating that the adjacent voxel and the target voxel fail the first occlusion check.
In some embodiments, when the first comparison result indicates that the first distance is less than or equal to the second distance, a distance between the target voxel and the corresponding virtual item is shorter than a distance between the adjacent voxel and the corresponding virtual item. A shorter distance between a voxel and a corresponding virtual item indicates that more virtual scene points in the voxel are occluded by the virtual item, in other words, a proportion of virtual scene points in the voxel being occluded is higher. The proportion of virtual scene points in the voxel being occluded is directly proportional to a distance between the voxel and the corresponding virtual item. In this case, because a quantity of occluded virtual scene points in the target voxel is greater than a quantity of occluded virtual scene points in the adjacent voxel, light information of the adjacent voxel has great reference value for the updated light information of the target voxel, and spatial-domain correction may be performed on the updated light information of the target voxel based on the light information of the adjacent voxel, in other words, the adjacent voxel may be determined as the target adjacent voxel.
In some embodiments, when the first comparison result indicates that the first distance is less than or equal to the second distance, whether to determine the adjacent voxel as the target adjacent voxel may be further determined based on whether the dot product of the direction vector from the virtual viewpoint to the voxel center point of the target voxel and the direction vector from the target voxel to the corresponding virtual item is less than the dot product threshold.
In some embodiments, when the dot product of the direction vector from the virtual viewpoint to the voxel center point of the target voxel and the direction vector from the target voxel to the corresponding virtual item is less than the dot product threshold, an included angle between the two direction vectors is small, and occlusion is more likely to occur. In this case, the adjacent voxel may be determined as the target adjacent voxel.
In this way, the first occlusion check is performed on the adjacent voxel and the target voxel to obtain the first occlusion check result, and the adjacent voxel is determined as the target adjacent voxel when the first occlusion check result indicates that the target voxel and the adjacent voxel pass the first occlusion check, so that the target adjacent voxel is determined based on a contribution of the light information of the adjacent voxel to accuracy of the target voxel. Spatial-domain correction is performed on the updated light information of the target voxel based on the target adjacent voxel, so that the obtained first light information is more accurate, and the updated light information of the virtual viewpoint is more accurate.
In some embodiments, the performing a second occlusion check on the adjacent voxel and the target voxel based on the second distance and the third distance, to obtain a second occlusion check result may be implemented in the following manner: comparing the second distance with the third distance to obtain a second comparison result; and determining the second occlusion check result as a fourth result when the second comparison result indicates that the second distance is greater than the third distance.
In some embodiments, the fourth result is configured for indicating that the adjacent voxel and the target voxel pass the second occlusion check.
In some embodiments, the third distance between the adjacent voxel and the target voxel is configured for indicating a distance between a voxel center point of the adjacent voxel and the voxel center point of the target voxel.
In some embodiments, when the second distance between the target voxel and the corresponding virtual item is greater than the third distance between the adjacent voxel and the target voxel, the second distance between the target voxel and the corresponding virtual item is longer than the third distance between the adjacent voxel and the target voxel, and a contribution of the adjacent voxel to light of the target voxel is greater than a contribution of the virtual item to light of the target voxel. In this case, the second occlusion check result may be determined as the fourth result. The fourth result is configured for indicating that the adjacent voxel and the target voxel pass the second occlusion check, in other words, the adjacent voxel may be determined as the target adjacent voxel.
In this way, the first occlusion check is performed on the adjacent voxel and the target voxel to obtain the first occlusion check result, and the second occlusion check is further performed when the first occlusion check result indicates that the target voxel and the adjacent voxel fail the first occlusion check, so that the target adjacent voxel is determined based on a contribution of the light information of the adjacent voxel to accuracy of the target voxel. Spatial-domain correction is performed on the updated light information of the target voxel based on the target adjacent voxel, so that the obtained first light information is more accurate, and the updated light information of the virtual viewpoint is more accurate.
In some embodiments, light information of the target voxel and light information of each target adjacent voxel are obtained.
In an example, light information respectively corresponding to a target adjacent voxel 1, a target adjacent voxel 2, a target adjacent voxel 3, and a target adjacent voxel 4 of the target voxel is obtained.
In some embodiments, weighted summation is performed on the light information of the target voxel and the light information of each target adjacent voxel, to obtain the first light information of the virtual viewpoint.
In some embodiments, weights respectively corresponding to the light information of the target voxel and the light information of each target adjacent voxel may be set based on an actual situation. For example, a weight of the light information of the target voxel is greater than a weight of the light information of each target adjacent voxel, and weights of light information of all target adjacent voxels are equal.
Operation 106: Perform time-domain correction on the updated light information to obtain second light information.
In some embodiments, the time-domain correction is a processing process of performing error correction on the updated light information of the target voxel based on light information of a plurality of historical voxels of a pixel corresponding to the virtual viewpoint in a historical update process. Accuracy of the second light information is higher than accuracy of the updated light information.
In some embodiments,
Operation 1061: Determine, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located, and obtain light information of the target voxel.
In some embodiments, the determining, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located may be implemented in the following manner: comparing a viewpoint position of the virtual viewpoint in the virtual scene with position ranges respectively corresponding to the updated voxel and the initial voxel other than the to-be-updated voxel, and determining, as the target voxel, a voxel corresponding to a position range within which a voxel position falls.
Operation 1062: Determine a pixel corresponding to the virtual viewpoint, query for a plurality of historical voxels of the pixel in a historical update process from the plurality of initial voxels, and obtain light information of each historical voxel.
In some embodiments, the pixel corresponding to the virtual viewpoint is a pixel corresponding to the virtual viewpoint on an imaging plane of the virtual camera, the virtual viewpoint is in a one-to-one correspondence with the pixel on the imaging plane of the virtual camera, and one virtual viewpoint in the virtual scene corresponds to one pixel on the imaging plane of the virtual camera in the virtual scene.
In some embodiments, a quantity of historical voxels is equal to a quantity of content updates in the virtual scene.
In an example, content of the virtual scene is updated five times within 10 seconds. Specifically, an update is performed at the first second, an update is performed at the third second, an update is performed at the fifth second, an update is performed at the seventh second, and an update is performed at the ninth second. A historical voxel corresponding to the pixel in a historical update (each update) process is queried, and light information of each historical voxel is obtained.
Operation 1063: Perform weighted summation on the light information of the target voxel and the light information of each historical voxel, to obtain the second light information of the virtual viewpoint.
In some embodiments, a content update in the virtual scene causes a change in a voxel corresponding to the pixel. One content update in the virtual scene causes one change in the voxel corresponding to the pixel. A content update in the virtual scene changes with time. For example, content of the virtual scene is updated five times within 10 seconds, and each content update causes one change in the voxel corresponding to the pixel. In this case, a plurality of initial voxels (historical voxels) corresponding to the pixel in a historical update (each update) process may be queried from the plurality of initial voxels, and weighted summation is performed on the light information of the target voxel and the light information of each historical voxel, to obtain the second light information of the virtual viewpoint. In this way, the updated light information is corrected within a time span (time domain) of a content update process.
In an example, content of the virtual scene is updated five times within 10 seconds. Specifically, an update is performed at the first second, an update is performed at the third second, an update is performed at the fifth second, an update is performed at the seventh second, and an update is performed at the ninth second. Weighted summation is performed on the light information of the target voxel and the light information of each historical voxel, to obtain the second light information of the virtual viewpoint.
In this way, the pixel corresponding to the virtual viewpoint is determined, the plurality of historical voxels of the pixel in the historical update process are queried from the plurality of initial voxels, and time-domain correction is performed on the updated light information based on a historical voxel corresponding to each update within a time-domain range of updates, to obtain the second light information, so that an error of the updated light information is effectively corrected, the obtained second light information is more accurate, and the updated light information of the virtual viewpoint is more accurate.
Operation 107: Perform error correction on the updated light information based on the first light information and the second light information.
In some embodiments, accuracy of the second light information is higher than accuracy of the updated light information, and accuracy of the first light information is higher than accuracy of the updated light information. The first light information mainly eliminates a spatial-domain error of the updated light information, and the second light information mainly eliminates a time-domain error of the updated light information. Error correction is performed on the updated light information based on the first light information and the second light information. In this way, obtained error-corrected updated light information not only can eliminate the spatial-domain error of the updated light information, but also can eliminate the time-domain error of the updated light information, so that the error-corrected updated light information is more accurate.
In some embodiments,
Operation 1071: Perform a validity check on the first light information to obtain a check result.
In some embodiments, the validity check is a processing process of checking validity of the first light information, and the check result is configured for indicating whether the first light information is valid.
In some embodiments, operation 1071 may be implemented in the following manner: determining, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located; constructing a check ray by using the target voxel as a start point and the virtual camera as an end point in the virtual scene with changed content; and determining the check result as a first check result when the check ray intersects with a virtual item in the virtual scene with changed content; or determining the check result as a second check result when the check ray does not intersect with a virtual item in the virtual scene with changed content.
In some embodiments, the first check result is configured for indicating that the first light information is invalid, and the second check result is configured for indicating that the first light information is valid.
In some embodiments, the determining, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located may be implemented in the following manner: comparing a viewpoint position of the virtual viewpoint in the virtual scene with position ranges respectively corresponding to the updated voxel and the initial voxel other than the to-be-updated voxel, and determining, as the target voxel, a voxel corresponding to a position range within which a voxel position falls.
In some embodiments, when the check ray intersects with the virtual item in the virtual scene with changed content, there is a possibility that the target voxel is occluded by the virtual item, and the first light information may be determined as invalid; or when the check ray does not intersect with the virtual item in the virtual scene with changed content, there is a possibility that the target voxel is not occluded by the virtual item, and the first light information may be determined as valid.
Operation 1072: Correct the updated light information to the second light information when the check result indicates that the first light information is invalid.
In some embodiments, the updated light information may be corrected to the second light information when the check result corresponding to the first light information indicates that the first light information is invalid.
Operation 1073: Correct the updated light information to the first light information when the check result indicates that the first light information is valid.
In some embodiments, the updated light information may be corrected to the first light information when the check result corresponding to the first light information indicates that the first light information is valid.
In this way, accuracy of the second light information is higher than accuracy of the updated light information, and accuracy of the first light information is higher than accuracy of the updated light information. The first light information mainly eliminates a spatial-domain error of the updated light information, and the second light information mainly eliminates a time-domain error of the updated light information. Error correction is performed on the updated light information based on the first light information and the second light information. In this way, obtained error-corrected updated light information not only can eliminate the spatial-domain error of the updated light information, but also can eliminate the time-domain error of the updated light information, so that the error-corrected updated light information is more accurate.
In some embodiments, after operation 104, target light information of the pixel may be further determined in the following manner: performing the following processing on each virtual viewpoint in the virtual scene with changed content: obtaining a pixel corresponding to the virtual viewpoint on an imaging plane of the virtual camera; determining a viewpoint distance between the virtual camera and the virtual viewpoint, and determining initial light information of the pixel based on the viewpoint distance and the updated light information of the virtual viewpoint; determining, from a plurality of screen probes arranged on the imaging plane, at least one target screen probe whose distance from the pixel is less than a distance threshold; and performing weighted summation on light information stored in each target screen probe and the initial light information of the pixel, to obtain target light information of the pixel.
In some embodiments, the target light information of the pixel is configured for performing picture rendering on the imaging plane of the virtual camera.
In some embodiments, the screen probe is configured to store light information of a corresponding position on the imaging plane.
In some embodiments, the viewpoint distance between the virtual camera and the virtual viewpoint may be determined by using a distance calculation method such as a Euclidean distance, a Manhattan distance, or a Chebyshev distance. A specific calculation method for the viewpoint distance is not limited. The viewpoint distance is configured for indicating a distance between the virtual camera and the virtual viewpoint.
In some embodiments, a Euclidean metric (also referred to as a Euclidean distance) is a commonly used distance definition, and is a true distance between two points in m-dimensional space, or a natural length of a vector (to be specific, a distance from the point to an origin). A Euclidean distance in two-dimensional space and three-dimensional space is an actual distance between two points. The Manhattan distance is a geometric term used in geometric metric space, and is configured for indicating a sum of absolute axial distances between two points in a standard coordinate system.
In some embodiments, weights respectively corresponding to the light information stored in each target screen probe and the initial light information of the pixel may be specifically set based on an actual situation, and a sum of weights respectively corresponding to light information stored in all target screen probes and the initial light information of the pixel is equal to 1.
In this way, voxelization processing is performed on a virtual scene to obtain a plurality of initial voxels corresponding to the virtual scene, at least one to-be-updated voxel is selected from the plurality of initial voxels in response to a content change in the virtual scene, light information carried in each to-be-updated voxel is updated, and light information of each virtual viewpoint in a changed virtual scene is updated based on the updated voxel and an initial voxel among the plurality of initial voxels other than the to-be-updated voxel. In this way, at least one to-be-updated voxel is selected from the plurality of initial voxels in response to the content change in the virtual scene, and the quantity of to-be-updated voxels is less than the quantity of initial voxels. Because the quantity of to-be-updated voxels is less than the quantity of initial voxels, the quantity of to-be-updated voxels is effectively reduced, so that update costs are effectively reduced, and light update efficiency for the virtual scene can be effectively improved.
The following describes exemplary application of embodiments of this application in an actual application scenario of an online game.
Embodiments of this application can be applied to any project and product that require real-time global illumination calculation, including games and three-dimensional visual designs, and can provide high-quality and realistic light rendering effects and significantly improve a visual manifestation capability of the product, to further improve user experience. In a typical application scenario such as a game, in embodiments of this application, an indirect light result of each pixel on a screen may be automatically calculated in real time based on placement positions and parameters of items and light in a game scene, to improve light details and vividness of a game picture.
In some embodiments,
In some embodiments,
Operation 201: Obtain three-dimensional scene information of a virtual scene.
Operation 202: Obtain a scene color and depth of a perspective.
In some embodiments, an obtained initial input is color and depth information obtained by rendering the virtual scene at a current perspective, and a voxel in world space is updated based on a three-dimensional structure of the virtual scene, such as an item posture and material, and a light source parameter. A voxel is used as a basic structure of a light buffer system, and voxel data is continuously updated to adapt to a dynamic change in the virtual scene and calculate indirect light transmission.
Operation 203: Update a scene voxel.
In some embodiments, operation 203 may be implemented in the following manner: updating the scene voxel based on the three-dimensional scene information of the virtual scene and the scene color and depth of the perspective.
In some embodiments, a voxel carrying normal line information is configured for representing a surface element in this embodiment of this application, to simplify description of a three-dimensional scene. First, cuboid space with a camera as a center is maintained, and a voxel is configured for simplifying description of three-dimensional scene information in the space. Content of the voxel includes a normal line, a diffuse reflection color, direct light information, radiance information, and the like. To reduce memory consumption, in this embodiment of this application, voxels are sparsely stored in a hierarchical manner, and 4×4×4 voxels are considered as a group, which is referred to as a brick. A hierarchical index is represented by an n×m×k three-dimensional map. If a brick includes a voxel, an actual storage position of the voxel may be obtained based on the hierarchical index. If a value obtained based on the hierarchical index is 0, the brick does not include a voxel. In this way, memory space can be greatly saved.
In this embodiment of this application, the scene voxel update (operation 203 shown in
In some embodiments, the screen voxel injection is usually performed by using a scene color map and a scene depth map. World coordinates of a pixel are calculated by using the depth map. Then information, such as a scene color and a diffuse reflection color, is recorded at a corresponding voxel position. Embodiments of this application provide an optimized screen voxel injection solution, to improve performance and accuracy of screen voxel injection. A process is divided into three parts:
A first part is group frustum culling: First, 8×8×8 bricks are considered as a group, and an intersection between each group and a view frustum is calculated in a GPU. A group that does not intersect with the view frustum is discarded. A group intersecting with the view frustum may be updated by frame based on a distance between the group and a camera. Update frequency of a group close to the camera is high, and update frequency of a group far away from the camera is low. A quantity of voxels that need to be updated is greatly reduced through group culling.
A second part is brick collection: Brick collection is performed on a group that needs to be updated in a current frame. In this process, whether a brick intersects with a screen depth is determined. If the brick intersects with the screen depth, the brick is collected for subsequent voxel injection; otherwise, the brick is not collected. In addition, all bricks in front of a scene depth are collected every 10 frames to complete a voxel clearing operation. Embodiments of this application provide a new “eight-vertex test method” herein. Whether a cuboid intersects with a scene depth is tested by determining, through comparison, a front-back relationship between eight vertexes of the cuboid and the scene depth. If some of the eight vertexes are located in front of the scene depth and some vertexes are located after the scene depth, it is determined that the cuboid intersects with the scene depth. In addition, for ease of determining, if any one of the eight vertexes is quite close to the scene depth, it is determined that the cuboid intersects with the scene depth.
A third part is voxel injection: A thread is created for each voxel of a collected brick. First, a three-dimensional point is randomly selected from the voxel. A screen depth is obtained based on a projection of the point in a camera coordinate system. If world coordinates corresponding to the screen depth are quite close to the point, the voxel is updated by using a screen color corresponding to the point. If the world coordinates are far away from the point, the sampling point is invalid, and enters a clearing stage. Two conditions need to be met for clearing a voxel. If it is determined, by using the “eight-vertex test method”, that a voxel does not intersect with the scene depth, and a voxel depth is in front of the screen depth, the corresponding voxel is cleared.
In some embodiments, in single-model voxelization, a three-dimensional model in a scene is transmitted to a GPU, and model voxelization is completed through hardware rasterization. A model projection solution needs to be set to orthogonal projection. In addition, a depth test is closed, and instance drawing is simultaneously performed on an x-axis, a y-axis, and a z-axis in one drawing call pass. Because a PC solution includes a subsequent voxel light update solution, only a diffuse reflection color and a normal line of a voxel are written in the single-model voxelization stage. However, in a mobile terminal solution, for consideration of performance, no light transmission is performed for a voxel; instead, light for a voxel is updated through screen voxel injection, to be specific, operations shown in
In some embodiments,
Operation 2031: Perform spatial-temporal division on a virtual scene.
In some embodiments, a voxel light transmission update is mainly configured for updating and calculating direct light and indirect light in each voxel, so that a light buffer structure of a system can adapt to a dynamic changes in a scene. However, because a large quantity of voxels exist in a scene, it is difficult to complete a light update within limited time. Therefore, an update operation is performed only on some voxels in each frame.
In some embodiments, space division and time division first need to be performed on all voxel brick structures in a scene. In space, a scene is divided into the following two parts based on a specific threshold by using a game player or a camera as an origin: a short distance and a long distance. For example, space within a length of 300 meters, a width of 300 meters, and a height of 150 meters is the short distance, and the rest is the long distance.
Operation 2023: Collect to-be-updated non-empty voxels in a current frame.
In some embodiments, all voxel brick structures in the short-distance space and the long-distance space are separately allocated to a plurality of game frames for light updates. The short-distance space is divided into nine game frames, and the long-distance space is divided into 121 game frames. Then non-empty voxel brick structures are collected for all voxel brick structures allocated to the current frame. To be specific, light of voxels in the voxel brick structures needs to be calculated. Then non-empty voxels in each voxel brick structure are marked. For example, 64 bits may be used to mark whether 64 voxels are in an empty state. Finally, uniform sampling and random extraction are performed on non-empty voxels in a voxel brick structure, and only some non-empty voxels are updated. For example, at most 16 non-empty voxels may be randomly updated. Due to impact of the random policy, all non-empty voxels in the voxel brick structure are updated after a plurality of frames elapse.
Operation 2033: Mark a non-empty voxel in a non-empty voxel brick.
Operation 2034: Randomly select a non-empty voxel.
Operation 2035: Calculate a starting point in the voxel.
In some embodiments,
Operation 2036: Calculate direct light at a position of the starting point in the voxel.
Operation 2037: Calculate indirect light at the position of the starting point in the voxel.
In some embodiments, the indirect light at the position of the starting point in the voxel further needs to be calculated, a specific quantity of rays are randomly sampled on a hemispherical surface in a normal direction of the starting point, and current indirect light of the voxel may be obtained by using an importance sampling or resampling technology.
Operation 2038: Perform an interpolation update on the indirect light.
In some embodiments, interpolation and blending are performed on a current indirect light calculation result and a result of a previous frame. For example, three rays may be uniformly sampled on a hemispherical surface in a normal direction of a voxel starting point in the current voxel, and light of the three rays is separately calculated through ray tracing. Then current indirect light is calculated through resampling. Nine frames are accumulated in time domain. To be specific, an interpolation weight is 1/9.
Operation 204: Generate a light probe system for the scene.
In some embodiments, the light probe system for the scene is generated based on a voxel structure, and is configured to perform subsequent indirect light calculation.
In this embodiment of this application, indirect light information of the scene is buffered by using three-dimensional probes in world space. The three-dimensional probes in this embodiment of this application also store two parts: indirect light information and visibility information. In this embodiment of this application, the indirect light information is stored in different formats based on platforms: On a PC, radiance information is stored by using an octahedral map. On a mobile terminal, radiance information is stored by using an ambient cube. For the visibility information, in this embodiment of this application, a variance shadow map (VSM) distance, to be specific, an average distance from closest items in all surrounding directions and a square of the average distance, is stored by using an octahedral map, and the VSM is configured for subsequently calculating visibility. The three-dimensional probes are uniformly distributed in a cuboid centered on a camera and are implemented in a nested hierarchy. Some probes close to the camera have high density, and some probes far away from the camera have low density.
In this embodiment of this application, a layer is selected in each frame to update a three-dimensional probe. In an update process, a probe that needs to be updated is first collected, and a determining condition is that a valid voxel exists within a specific range around the probe. This collection solution avoids a full probe update, and can reduce a large amount of performance consumption.
For each to-be-updated three-dimensional probe, a direction of a ray is first randomly selected from a spherical surface, and a probe surface that needs to be updated is determined based on the direction of the ray. If a mobile terminal is used, a surface that needs to be updated is first randomly selected, and then a direction of a ray is randomly selected by using cosine of an included angle of a normal line as an update probability. A start point of the ray is a center of the probe, and a voxel at an intersection position may be obtained by using a ray tracing algorithm. In a mobile terminal solution, the ray tracing algorithm is implemented through software-based voxel tracing. An implementation solution is to store occupation information of a next-level node in a hierarchical manner. A voxel is located at a bottom layer. Information indicating whether 4×4×4 next-level nodes are occupied is stored by using 64 bits in each previous-level node. Then the ray tracing algorithm is implemented by using a hierarchical digital differential analyzer (HDDA). In a PC solution, different solutions such as hardware-based ray tracing may alternatively be selected. If no voxel intersects with the probe within a quite long range, probe surface content, such as irradiance and unit radiant intensity, is updated by using daylight, and a VSM distance is updated. If a voxel intersects with a front surface, a probe surface is updated by using a voxel color, and a VSM distance is updated. If a voxel intersects with a rear surface, only a VSM distance is updated, and no color is updated. To avoid light leakage or shadow leakage caused by a probe in an item structure or a voxel, this embodiment of this application also provides a probe relocation function. Based on this, a position is quantized, and the position and transition information are stored, to ensure a smooth and natural relocation process. After the relocation, a color of the probe is reset, to prevent light leakage or shadow leakage.
Operation 205: Perform ray tracing.
In some embodiments, the ray tracing is configured for: for a world space position corresponding to each pixel on a screen, emitting a ray on a hemispherical surface of the pixel, obtaining intersection information, and solving an initial indirect light result by using the intersection information.
In some embodiments, an indirect light result at each pixel position may be obtained through calculation by a ray tracing module, and the indirect light result is a noisy indirect light result before being processed by a noise reduction module. For example, when RTX hardware-based ray tracing is used, for each pixel on the screen, sampling is performed based on a material of the pixel at a world space position corresponding to the pixel, to generate an initial ray, then ray tracing is performed to obtain intersection information, and then a light result is obtained from a corresponding scene voxel structure.
Operation 206: Perform noise reduction.
In some embodiments, an initial light result is full of noise, and enters a noise reduction module. The module performs noise elimination by using a related spatial-temporal algorithm, to generate a smooth and stable indirect light calculation result. Light rendering quality of an application project or product may be greatly improved by using the result, to provide a realistic visual effect.
In some embodiments,
Operation 2061: Obtain a noisy light result.
Operation 2062: Perform ReSTIR time-domain multiplexing.
In some embodiments, for an inputted noisy indirect light result, general ReSTIR time-domain multiplexing and spatial-domain multiplexing are first performed to preliminarily reduce noise. To reduce a calculation amount to meet a requirement for real-time operation, in this embodiment of this application, no visibility check operation is performed for ray tracing in these two operations. If no visibility check operation is performed, a large quantity of deviations are introduced, and compared with a physically correct reference standard, an outputted indirect light result is specifically manifested as a loss of high-frequency details of indirect light, a loss of a shadow of indirect light, a reduction in overall contrast of a picture, and a loss of a three-dimensional sense of light. In addition, a small deviation amount is introduced by the time-domain multiplexing, and a large deviation amount is introduced by the spatial-domain multiplexing.
A simple exemplary process is as follows: A noisy light result has been calculated for each pixel on the screen, and an initial sampling point So obtained by the ray tracing module for the pixel is stored. After a time-domain multiplexing operation is performed, a sampling point St having a maximum contribution to a current pixel is selected based on sampling information of a historical frame before the pixel. After a spatial-domain multiplexing operation is performed, a sampling point Ss having a maximum contribution to the current pixel is selected based on sampling information of other pixels around the pixel. An indirect light result of the pixel may be solved based on the sampling points.
Operation 2063: Perform ReSTIR spatial-domain multiplexing.
In some embodiments, a noisy light result has been calculated for each pixel on the screen, and an initial sampling point So obtained by the ray tracing module for the pixel is stored. After a time-domain multiplexing operation is performed, a sampling point St having a maximum contribution to a current pixel is selected based on sampling information of a historical frame before the pixel. After a spatial-domain multiplexing operation is performed, a sampling point Ss having a maximum contribution to the current pixel is selected based on sampling information of other pixels around the pixel. An indirect light result of the pixel may be solved based on the sampling points.
Operation 2064: Perform visibility detection.
In some embodiments,
Operation 20641: Obtain a spatial-domain multiplexing result Ss.
In some embodiments, an obtained input is a result sampling point Ss, obtained through spatial-domain multiplexing, of a current pixel.
Operation 20642: Emit a ray to the sampling point Ss.
In some embodiments, to determine whether Ss is visible to the current pixel, a ray is emitted from a world space position corresponding to the current pixel to the sampling point Ss, and whether the visible ray is occluded is determined through ray tracing.
Operation 20643: Determine whether the visible ray is occluded.
In some embodiments, whether the visible ray is occluded may be determined through ray tracing.
Operation 20644: Mark the sampling point Ss as invalid when the visible ray is occluded.
In some embodiments, if the visible ray is occluded, a ray emitted by the sampling point Ss cannot reach the current pixel, and the sampling point Ss is marked as invalid. This indicates that the spatial-domain multiplexing result has a quite large deviation. To reduce the deviation, the spatial-domain multiplexing result Ss is restored to a time-domain multiplexing result St, and the time-domain multiplexing result St has a small deviation. In addition, an occlusion distance Ot and an occlusion direction Od of Ss are recorded.
Operation 20645: Mark the sampling point Ss as valid when the visible ray is not occluded.
In some embodiments, if the visible ray is not occluded, the sampling point Ss is marked as valid. This indicates that the spatial-domain multiplexing result does not have a large deviation.
Operation 20646: Restore the time-domain multiplexing result St.
In some embodiments, to reduce the deviation, the spatial-domain multiplexing result Ss is restored to the time-domain multiplexing result St, and the time-domain multiplexing result St has a small deviation.
Operation 20647: Store the occlusion distance Ot and the occlusion direction Od.
In some embodiments, the occlusion distance Ot and the occlusion direction Od of Ss may be recorded.
Operation 2065: Perform heuristic ReSTIR spatial-domain multiplexing.
In some embodiments, a spatial-domain multiplexing result with a large deviation is detected and is restored to a time-domain multiplexing result in this embodiment of this application. However, compared with spatial-domain multiplexing, the time-domain multiplexing result has large noise. To further eliminate the noise, in this embodiment of this application, heuristic spatial-domain multiplexing is performed again based on a result of a visibility check module. This operation does not include an additional visibility check operation for ray tracing. Compared with the simple spatial-domain multiplexing in 2063, high-frequency details of indirect light can be retained, to greatly improve quality of the indirect light and make the indirect light closer to a physically correct reference result. In an original spatial-domain multiplexing process, filtering is performed based on a difference between a current pixel and a neighborhood pixel in terms of geometrical parameters such as a depth value and a normal value, but it is still difficult to retain high-frequency details of indirect light.
In some embodiments, the first occlusion information described above is occlusion determining 1 described below.
In some embodiments, the second occlusion information described above is occlusion determining 2 described below.
In some embodiments,
Operation 20651: Determine whether P is occluded.
Operation 20652: When P is not occluded, determine whether Q is occluded.
Operation 20653: When P is occluded, determine whether Q is occluded.
Operation 20654: Perform occlusion determining 1 when P is occluded and Q is occluded.
Operation 20655: Accept Q when P is not occluded and Q is not occluded, or when the occlusion determining 1 (namely, the first occlusion check described above) succeeds.
Operation 20656: Reject Q when P is occluded and Q is not occluded, or when P is not occluded and Q is occluded.
Operation 20657: Perform occlusion determining 2 (namely, the second occlusion check described above) when the occlusion determining 1 fails.
In some embodiments, if the point P is not occluded and the point Q is not occluded either, the neighborhood Q is accepted, to be specific, a spatial-domain multiplexing operation is performed between P and Q. If one of P and Q is occluded and the other is not occluded, the neighborhood Q is rejected, to be specific, no spatial-domain multiplexing operation is performed between P and Q. When both P and Q are occluded, occlusion determining further needs to be performed. An occlusion distance calculated by the visibility check module for the point P is denoted as Pot, an occlusion direction of the point P is Pod, an occlusion distance of the point Q is Qot, an occlusion direction of the point Q is Qod, a world space position corresponding to the point P is Wp, and a light sampling point position corresponding to the point Q is Sq. A distance from Wp to Sq is D, and a direction is L. The occlusion determining includes two operations. First, the occlusion determining 1 is performed. If Qot≤Pot and a dot product of Pod and L is less than a threshold a, where a may range from 0 to 1, the neighborhood Q is accepted through occlusion determining. Otherwise, the occlusion determining 2 is further performed. If Pot>bD, to be specific, if the occlusion distance Pot of the point P is greater than b times of D, the neighborhood Q is accepted through occlusion determining. A value of b is greater than 0. Usually, b may be equal to 1. If the occlusion determining 2 fails either, the neighborhood Q is rejected. This heuristic neighborhood filtering policy can be directly applied to a basic spatial-domain multiplexing process to form a heuristic spatial-domain multiplexing method in embodiments of this application, to efficiently retain shadow details of indirect light while reducing noise.
Operation 2066: Perform multi-sampling shading.
In some embodiments, after processing in all the foregoing processes is completed, each pixel has an optimal light sampling point after being processed by using a related algorithm. However, a final indirect light result calculated by using only one light sampling point obtained through resampling is full of a large amount of color noise, and color spectrum information of light is considered in none of the foregoing noise reduction processes. In this embodiment of this application, multi-sampling shading calculation is performed to resolve this problem. For a pixel whose shading information needs to be calculated, light calculation is performed by aligning sampling results of other pixels within a small neighborhood range of the pixel. For a calculation process, refer to a standard algorithm. The neighborhood range may be 3×3, 5×5, 7×7, or another size. Ato-be-shaded pixel is denoted as P, and a neighborhood pixel is denoted as Q. Assuming that nine neighborhood pixels are selected, nine standard spatial-domain multiplexing operations are separately performed. During each time of multiplexing, light sampling information of P is set to null and is then reused with light sampling information of the point Q. Indirect light R is calculated based on a result obtained through reusing. Weighted averaging is performed on R obtained through nine times of calculation, to obtain a final indirect light calculation result of the pixel P. A weighted averaging policy is not specifically limited in this embodiment of this application. For example, a uniform weight may be selected. To be specific, if nine neighborhood pixels are selected, a weight of indirect light R obtained through calculation each time is 1/9. In addition, a multi-sampling shading module may also be configured to increase a resolution. All noise reduction processing before the module may be performed at a low resolution, for example, half resolution, provided that resolutions of initial noisy light results are kept consistent. In the multi-sampling shading module, shading calculation may be directly performed on high-resolution screen pixel information by using an existing noise reduction result, to increase a resolution and retain material details and geometrical details of a high screen resolution.
Operation 2067: Perform time-domain filtering.
Operation 2068: Perform spatial-domain filtering.
In some embodiments, filtering is mainly intended to eliminate remaining noise, to provide a smooth and stable noise-reduced indirect light result.
Operation 2069: Generate a noiseless light result.
Operation 207: Generate an indirect light result.
In some embodiments,
In some embodiments, to implement a global illumination solution on a mobile terminal, the system captures indirect light by using a screen probe, and operation 305 may be implemented in the following manner: Lightweight simplified deferred pipeline: In this system, a forward pipeline commonly used on a conventional mobile terminal is reconstructed into a lightweight simplified deferred pipeline. To be specific, one R8G8B8A8 map is output in addition to a screen color and a screen depth in a base drawing call (BasePass) stage. RGB channels store diffuse reflection colors, and an alpha channel stores compressed normal line information. Because the normal line information including only eight bits has insufficient precision, a grid-like defect occurs on a surface of an arc model. This system provides noise perturbation within a small range based on a normal line of the screen, to resolve the problem of the grid defect. During actual implementation, one bit may be reserved in the B channel to store information indicating whether a three-dimensional model is static or dynamic. This manner can resolve a problem of dynamic object smearing that occurs when a time super-sampling algorithm is performed without a speed map. Generation of screen probes: In this system, screen space is divided into 64×64 pixel grids. One screen space probe is placed in each grid. The screen space probe is generated through interpolation on a three-dimensional space probe. The probe only needs to store indirect light information of a hemispherical surface facing a camera. Time super-sampling may be performed through jitter of a screen probe generation position. Collection of screen probes: In this system, sampling and interpolation are performed on screen probes pixel by pixel at half resolution, and bilateral filtering with depth detection and normal line detection is performed to obtain half-resolution indirect light of the screen. If a speed map is generated in a rendering project, the speed map may be superposed for time super-sampling. If no speed map is generated in a rendering project, three-dimensional probes may be directly collected based on a dynamic object identifier while skipping collection of screen probes. Indirect light upsampling: Upsampling is performed on a half-resolution indirect light result through bilateral filtering. Optimization of a smearing problem: Because a screen probe has a limited resolution, if none of four probes around a rendering position passes depth detection or normal line detection, the system records the rendering position, and generates one more screen probe at the position in a next frame, to avoid a smearing problem caused by incorrect collection of probes. A smearing problem of a moving object is resolved by generating a screen probe before the moving object is drawn.
Operation 306: Generate an indirect light result.
In this way, in terms of a mobile terminal, embodiments of this application provide an innovative screen space injection solution and a screen probe system, to control performance consumption of dynamic global illumination within 4 ms on a high-end mobile device, and implement application of dynamic global illumination for a big world on the mobile terminal. In terms of a PC, embodiments of this application provide an innovative voxel update solution, and provide a new noise reduction process, including an innovative visibility check method, heuristic spatial-domain multiplexing, multi-sampling shading, and the like. This can provide a stable global illumination effect for a highly dynamic rendering animation, and provide a high-precision indirect light shadow.
Data related to a virtual scene or the like is involved in embodiments of this application. When embodiments of this application are applied to a specific product or technology, user permission or consent is required, and collection, use, and processing of related data need to comply with related laws, regulations, and standards in related countries and regions.
The following further describes an exemplary structure of the light update apparatus 455 for a virtual scene provided in embodiments of this application when the apparatus is implemented as software modules. In some embodiments, as shown in
In some embodiments, the selection module is further configured to: in response to the content change in the virtual scene, obtain a camera position of the virtual camera in the virtual scene with changed content; obtain a voxel position of a voxel center point of each initial voxel in the virtual scene, and determine a voxel distance between the camera position and each voxel position; and select at least one to-be-updated voxel from the plurality of initial voxels based on the voxel distance.
In some embodiments, the selection module is further configured to: determine the initial voxel as a first initial voxel when the voxel distance of the initial voxel is less than or equal to a voxel distance threshold; or determine the initial voxel as a second initial voxel when the voxel distance of the initial voxel is greater than the voxel distance threshold; and select a first quantity of first initial voxels and a second quantity of second initial voxels from the plurality of initial voxels, and determine the selected first initial voxel and second initial voxel as the to-be-updated voxel, the first quantity being greater than the second quantity, and the first quantity being at least one.
In some embodiments, the voxel update module is further configured to perform the following processing on each to-be-updated voxel: determining, from virtual scene points of the virtual scene, a plurality of target virtual scene points located in the to-be-updated voxel; obtaining target light information of each target virtual scene point in the to-be-updated voxel, and performing weighted summation on all pieces of target light information to obtain updated light information; and updating the light information carried in the to-be-updated voxel to the updated light information, to obtain the updated voxel corresponding to the to-be-updated voxel.
In some embodiments, the voxel update module is further configured to perform the following processing on each target virtual scene point in the to-be-updated voxel: obtaining direct light information of the target virtual scene point, the direct light information being configured for indicating impact of a direct ray emitted by a virtual light source on light of the target virtual scene point; obtaining indirect light information of the target virtual scene point, the indirect light information being configured for indicating impact of a reflected ray corresponding to the direct ray on light of the target virtual scene point; and summing the direct light information and the indirect light information to obtain the target light information of the target virtual scene point.
In some embodiments, the direct light information includes direct light intensity, and the voxel update module is further configured to: determine a light source distance between the virtual light source in the virtual scene and the target virtual scene point; obtain a camera distance between the target virtual scene point and the virtual camera, and add up the camera distance and the light source distance to obtain a total distance; determine a loss value of light source light intensity of the virtual light source based on the total distance and the target virtual scene point; and subtract the loss value from the light source light intensity to obtain the direct light intensity of the target virtual scene point.
In some embodiments, the voxel update module is further configured to: determine, from a plurality of light probes arranged in the virtual scene, at least one target light probe whose distance from the target virtual scene point is less than a distance threshold, the light probe being configured to store light information of a corresponding position in the virtual scene with changed content; and when one target light probe exists, determine light information stored in the target light probe as the indirect light information of the target virtual scene point; or when a plurality of target light probes exist, determine a weight of each target light probe based on a probe distance between each target light probe and the target virtual scene point; and perform, based on the weight, weighted summation on light information stored in all target light probes, to obtain the indirect light information of the target virtual scene point.
In some embodiments, the light update apparatus for a virtual scene further includes an arrangement module, configured to: obtain a camera position of the virtual camera in the virtual scene; determine a virtual scene region that is in the virtual scene and whose distance from the camera position is less than a distance threshold as a first virtual scene region, and determine a virtual scene region that is in the virtual scene and whose distance from the camera position is greater than or equal to the distance threshold as a second virtual scene region; and arrange a third quantity of light probes in the first virtual scene region, and arrange a fourth quantity of light probes in the second virtual scene region, the third quantity being greater than the fourth quantity.
In some embodiments, the light update module is configured to perform the following processing on each virtual viewpoint in the virtual scene with changed content: determining, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located, and obtaining light information of the target voxel; determining the light information of the target voxel as updated light information of the virtual viewpoint; and updating the light information of the virtual viewpoint to the updated light information of the virtual viewpoint.
In some embodiments, the light update apparatus for a virtual scene further includes a correction module, configured to: perform the following processing on updated light information obtained by updating each virtual viewpoint: performing spatial-domain correction on the updated light information to obtain first light information, and performing time-domain correction on the updated light information to obtain second light information; and performing error correction on the updated light information based on the first light information and the second light information.
In some embodiments, the correction module is further configured to: determine, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located, and a plurality of adjacent voxels adjacent to the target voxel; select a target adjacent voxel from the plurality of adjacent voxels; and obtain light information of the target voxel and light information of each target adjacent voxel, and perform weighted summation on the light information of the target voxel and the light information of each target adjacent voxel, to obtain the first light information of the virtual viewpoint.
In some embodiments, the correction module is further configured to perform the following processing on each adjacent voxel: obtaining first occlusion information of the adjacent voxel and second occlusion information of the target voxel; and determining the adjacent voxel as the target adjacent voxel when the first occlusion information indicates that no virtual item exists between the adjacent voxel and the virtual camera and the second occlusion information indicates that no virtual item exists between the target voxel and the virtual camera; or performing an occlusion check on the adjacent voxel and the target voxel to obtain an occlusion check result when the first occlusion information indicates that a virtual item exists between the adjacent voxel and the virtual camera and the second occlusion information indicates that a virtual item exists between the target voxel and the virtual camera; and determining the adjacent voxel as the target adjacent voxel when the occlusion check result indicates that the adjacent voxel and the target voxel pass the occlusion check; or determining the adjacent voxel as a non-adjacent voxel when the first occlusion information indicates that a virtual item exists between the adjacent voxel and the virtual camera or the second occlusion information indicates that a virtual item exists between the target voxel and the virtual camera.
In some embodiments, the correction module is further configured to: obtain a first distance between the adjacent voxel and the corresponding virtual item, and a second distance between the target voxel and the corresponding virtual item; perform a first occlusion check on the adjacent voxel and the target voxel based on the first distance and the second distance, to obtain a first occlusion check result; and determine the occlusion check result as a first result when the first occlusion check result indicates that the adjacent voxel and the target voxel pass the first occlusion check; or when the first occlusion check result indicates that the adjacent voxel and the target voxel fail the first occlusion check, obtain a third distance between the adjacent voxel and the target voxel, and perform a second occlusion check on the adjacent voxel and the target voxel based on the second distance and the third distance, to obtain a second occlusion check result; and determine the occlusion check result as a first result when the second occlusion check result indicates that the adjacent voxel and the target voxel pass the second occlusion check; or determine the occlusion check result as a second result when the second occlusion check result indicates that the adjacent voxel and the target voxel fail the second occlusion check, the first result being configured for indicating that the adjacent voxel and the target voxel pass the occlusion check, and the second result being configured for indicating that the adjacent voxel and the target voxel fail the occlusion check.
In some embodiments, the correction module is further configured to: compare the first distance with the second distance to obtain a first comparison result; and determine the first occlusion check result as a third result when the first comparison result indicates that the first distance is less than or equal to the second distance and a dot product of a direction vector from the virtual viewpoint to a voxel center point of the target voxel and a direction vector from the target voxel to the corresponding virtual item is less than a dot product threshold, the third result being configured for indicating that the adjacent voxel and the target voxel pass the first occlusion check.
In some embodiments, the correction module is further configured to: compare the second distance with the third distance to obtain a second comparison result; and determine the second occlusion check result as a fourth result when the second comparison result indicates that the second distance is greater than the third distance, the fourth result being configured for indicating that the adjacent voxel and the target voxel pass the second occlusion check.
In some embodiments, the correction module is further configured to: determine, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located, and obtain light information of the target voxel; determine a pixel corresponding to the virtual viewpoint, query for a plurality of historical voxels of the pixel in a historical update process from the plurality of initial voxels, and obtain light information of each historical voxel; and perform weighted summation on the light information of the target voxel and the light information of each historical voxel, to obtain the second light information of the virtual viewpoint.
In some embodiments, the correction module is further configured to: perform a validity check on the first light information to obtain a check result; and correct the updated light information to the second light information when the check result indicates that the first light information is invalid; or correct the updated light information to the first light information when the check result indicates that the first light information is valid.
In some embodiments, the correction module is further configured to: determine, from the updated voxel and the initial voxel other than the to-be-updated voxel, a target voxel in which the virtual viewpoint is located; construct a check ray by using the target voxel as a start point and the virtual camera as an end point in the virtual scene with changed content; and determine the check result as a first check result when the check ray intersects with a virtual item in the virtual scene with changed content, the first check result being configured for indicating that the first light information is invalid; or determine the check result as a second check result when the check ray does not intersect with a virtual item in the virtual scene with changed content, the second check result being configured for indicating that the first light information is valid.
In some embodiments, the light update apparatus for a virtual scene further includes: performing the following processing on each virtual viewpoint in the virtual scene with changed content: obtaining a pixel corresponding to the virtual viewpoint on an imaging plane of the virtual camera; determining a viewpoint distance between the virtual camera and the virtual viewpoint, and determining initial light information of the pixel based on the viewpoint distance and the updated light information of the virtual viewpoint; determining, from a plurality of screen probes arranged on the imaging plane, at least one target screen probe whose distance from the pixel is less than a distance threshold; and performing weighted summation on light information stored in each target screen probe and the initial light information of the pixel, to obtain target light information of the pixel, the target light information of the pixel being configured for performing picture rendering on the imaging plane of the virtual camera.
Embodiments of this application provide a computer program product. The computer program product includes a computer program or computer-executable instructions. The computer program or the computer-executable instructions are stored in a computer-readable storage medium. A processor of an electronic device reads the computer-executable instructions from the computer-readable storage medium, and the processor executes the computer-executable instructions, so that the electronic device performs the light update method for a virtual scene in embodiments of this application.
Embodiments of this application provide a non-transitory computer-readable storage medium, having computer-executable instructions stored therein. When the computer-executable instructions are executed by a processor, the processor is enabled to perform the light update method for a virtual scene provided in embodiments of this application, for example, the light update method for a virtual scene shown in
In some embodiments, the computer-readable storage medium may be a memory such as an FRAM, a ROM, a PROM, an EPROM, an EEPROM, a flash memory, a magnetic memory, a compact disc, or a CD-ROM; or may be various electronic devices including one of or any combination of the foregoing memories.
In some embodiments, the computer-executable instructions may be written in the form of a program, software, a software module, a script, or code and according to a programming language (including a compiler or interpreter language or a declarative or procedural language) in any form, and may be deployed in any form, including an independent program or a module, a component, a subroutine, or another unit suitable for use in a computing environment.
In an example, the computer-executable instructions may, but do not necessarily, correspond to a file in a file system, and may be stored in a part of a file that saves another program or other data, for example, be stored in one or more scripts in a hypertext markup language (HTML) file, stored in a file that is specially used for a program in discussion, or stored in a plurality of collaborative files (for example, be stored in files of one or more modules, subprograms, or code parts).
In an example, the computer-executable instructions may be deployed on one electronic device for execution, or may be executed on a plurality of electronic devices at one location, or may be executed on a plurality of electronic devices that are distributed at a plurality of locations and that are interconnected through a communication network.
To sum up, embodiments of this application have the following beneficial effects:
In this application, the term “module” in this application refers to a computer program or part of the computer program that has a predefined function and works together with other related parts to achieve a predefined goal and may be all or partially implemented by using software, hardware (e.g., processing circuitry and/or memory configured to perform the predefined functions), or a combination thereof. Each module can be implemented using one or more processors (or processors and memory). Likewise, a processor (or processors and memory) can be used to implement one or more modules. Moreover, each module can be part of an overall module that includes the functionalities of the module. The foregoing descriptions are merely embodiments of this application and are not intended to limit the protection scope of this application. Any modification, equivalent replacement, or improvement made without departing from the spirit and scope of this application shall fall within the protection scope of this application.
| Number | Date | Country | Kind |
|---|---|---|---|
| 202310318681.5 | Mar 2023 | CN | national |
This application is a continuation application of PCT Patent Application No. PCT/CN2024/073461, entitled “LIGHT UPDATE METHOD AND APPARATUS FOR VIRTUAL SCENE, ELECTRONIC DEVICE, COMPUTER-READABLE STORAGE MEDIUM, AND COMPUTER PROGRAM” filed on Jan. 22, 2024, which claims priority to Chinese Patent Application No. 2023103186815, entitled “LIGHT UPDATE METHOD AND APPARATUS FOR VIRTUAL SCENE, ELECTRONIC DEVICE, COMPUTER-READABLE STORAGE MEDIUM, AND COMPUTER PROGRAM” filed on Mar. 21, 2023, all of which is incorporated herein by reference in their entirety.
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/CN2024/073461 | Jan 2024 | WO |
| Child | 19083189 | US |
