Patent Application
20240165511
The present specification is related generally to the field of video games and graphics processing. More specifically the present specification is related to offline automated generation of level of detail (LOD) assets, manual optimization of the auto-generated LOD assets as well as dynamic selection and rendering of the optimized LOD assets, during runtime.
Objects in a video game scene are typically modeled as three-dimensional meshes comprising geometric primitives such as, for example, triangles or other polygons whose coordinate points are connected by edges. For rendering and displaying such three-dimensional meshes or object models, large numbers of computations need to be performed in order to determine the locations of, and complex interactions with, those coordinate points and edges.
While the desired levels of geometric detail in games are increasing, gaming hardware (personal computer as well as console based) are not necessarily advancing as fast in order to manage the computationally intensive workload of rendering and displaying highly complex polygonal meshes. One way to address this technical challenge is to increase geometric detail when virtual objects are closer to a virtual camera position in a game and decrease geometric detail when virtual objects are further away from the virtual camera's position. This is a known technical approach, referred to as Level of Detail or LOD. There are many challenges in implementing an LOD approach, but, at the highest level, the challenges pertain to a) creating the right Level of Detail assets which should be used at various “distances” or “switch distances” and b) during the actual game (runtime), selecting the right LOD objects in the right context. If not done right, a player might experience “pixel pop”, where the switch from one LOD object to another becomes obvious. In other words, it is desirable to determine which level of detail to use when rendering a mesh, so that the mesh is detailed enough to provide a desired level of visual or perceptual fidelity without being so detailed as to overburden rendering resources.
Conventional LOD object generation is highly manual. However, when starting out with extremely high resolution (high-polygon) assets, doing everything manually is not feasible. At the same time, a purely automated approach does not result in optimized LOD assets every time, and therefore computer graphic artists do need to provide manual input. There are many variables which have to be chosen in order to effectively generate an optimal LOD and, conventionally, those variables have to be considered, and accounted for, manually. Those variables include, for example:
Once the LOD assets have been auto-generated and further optimized manually, during runtime a video game still needs to select the right LOD assets to use at the right time (when the virtual object, relative to a virtual camera position, is close, not-so-close, or far). Conventionally, LOD assets are chosen for rendering based on a fixed switch distance established offline. However, there are many considerations that must be taken in to account while selecting a LOD for rendering during gameplay situations that may be highly dynamic and intensive at least in terms of graphics pipeline processing workload.
Accordingly, there is a need for an automated process of generating LOD assets that also provides artists with an intuitive way to modify and optimize the generated LOD assets manually. There is also a need to determine optimal selection of LOD assets, during runtime, by integrating a plurality of variables into a LOD selection and rendering process. There is further a need for an improved way of generating robust LOD assets and selecting those LOD assets at the right switch distances.
The following embodiments and aspects thereof are described and illustrated in conjunction with systems, tools and methods, which are meant to be exemplary and illustrative, not limiting in scope. The present application discloses numerous embodiments.
In some embodiments, the present specification discloses a method of generating a plurality of level of detail (LOD) visual assets of an object model, for rendering in a video game, the method comprising: receiving into a LOD management module data indicative of a first LOD visual asset of the object model and a first switch distance associated with the first LOD visual asset; generating, using the LOD management module, a second LOD visual asset by iteratively reducing a polygon mesh complexity of the first LOD visual asset until a maximum geometric deviation of the second LOD visual asset approaches, but does not exceed, a first threshold distance value with respect to the first LOD visual asset; generating, using the LOD management module, a second switch distance at which the first threshold distance value, when represented on a display of a first resolution, comprises a predefined number of pixels; and storing the second LOD in association with the second switch distance if a difference between the first switch distance and second switch distance is greater than a second threshold distance value.
Optionally, if the difference between the first switch distance and the second switch distance is less than or equal to the second threshold distance value, then the second LOD visual asset and the second switch distance are provided as inputs to the LOD management module.
Optionally, the first LOD visual asset and the first switch distance correspond to the object model and an associated predefined switch distance at a beginning of an execution of the method.
Optionally, the first threshold distance value is a maximum distance between at least one vertex in the first LOD visual asset and a closest corresponding point on the second LOD visual asset.
Optionally, the first threshold distance value ranges from 0.001 to 24.0 inches.
Optionally, the predefined number of pixels ranges from 2.0 to 100.00 pixels.
Optionally, when measured using a scale native to the video game, the predefined threshold distance ranges from 10 to 100 feet.
Optionally, steps of the method are repeated until at least one of a first condition or a second condition is satisfied, wherein the first condition corresponds to a size of the second LOD visual asset being determined to be less than or equal to a predefined size, and wherein the second condition corresponds to a maximum number of the plurality of LOD visual assets being generated.
Optionally, the predefined size ranges from 4.0 to 1000.0 inches, and wherein the maximum number of the plurality of LOD visual assets is 6.
Optionally, the method of generating a plurality of level of detail (LOD) visual assets of an object model, for rendering in a video game further comprises generating, using the LOD management module, one or more graphical user interfaces for receiving user input to modify one or more of the generated plurality of LOD visual assets, wherein a portion of the one or more graphical user interfaces prompt for user inputs related to one or more of the following attributes: auto LOD visual asset generation, LOD visual asset count, a LOD visual asset index at which to start generating the plurality of LOD visual assets, distance scale, tri-coverage LOD visual asset threshold, LOD visual asset blink-out, reduction percentage, weight settings to modify a number of triangles, edges and points, importance settings to modify geometry, texture coordinates, normal, material ID, object ID, skinning and vertex color, and settings to modify a maximum deviation, weld threshold, vertex regularization, surface area and/or normal split tolerance.
In some embodiments, the present specification discloses a computer readable non-transitory medium comprising a plurality of executable programmatic instructions wherein, when said plurality of executable programmatic instructions are executed by a processor in a computing device, a process for generating a plurality of level of detail (LOD) of visual assets of an object model is performed for rendering in a video game, wherein the plurality of LOD visual assets comprises at least a first LOD visual asset and a second LOD visual asset, the plurality of executable programmatic instructions comprising: receiving, as input by a LOD management module, data indicative of the first LOD visual asset of the object model and a first switch distance associated with the first LOD visual asset; generating, as a first output by the LOD management module, the second LOD visual asset by iteratively reducing a polygon mesh complexity of the first LOD visual asset, wherein a geometric deviation of the second LOD visual asset reaches a predefined threshold value with respect to the first LOD visual asset; generating, as second output by the LOD management module, a second switch distance at which the predefined threshold value, when projected on a display of a default resolution, comprises a predefined number of pixels; storing, by the LOD management module, the first output in association with the second output if a difference between the first switch distance and the second switch distance is greater than a predefined threshold distance; and generating, by the LOD management module, one or more graphical user interfaces for receiving user input to modify one or more of the plurality of LOD visual assets, wherein a portion of the one or more graphical user interfaces prompt for user inputs related to one or more of a plurality of LOD visual asset related attributes.
Optionally, the computer readable non-transitory medium of claim 11, wherein the plurality of LOD visual asset related attributes include at least one of auto LOD visual asset generation, LOD visual asset count, a LOD visual asset index at which to start generating the plurality of LOD visual assets, distance scale, tri-coverage LOD visual asset threshold, LOD visual asset blink-out, reduction percentage, weight settings to modify a number of triangles, edges and points, importance settings to modify geometry, texture coordinate, normal, material ID, object ID, skinning and vertex color, or settings to modify a maximum deviation, weld threshold, vertex regularization, surface area and/or normal split tolerance.
Optionally, the first visual asset LOD and the first switch distance correspond to said object model and an associated predefined switch distance during a first execution of the process steps.
Optionally, the predefined threshold value is a maximum distance between vertices in the first LOD visual asset and the closest corresponding points on the second LOD visual asset.
Optionally, the predefined threshold value ranges from 0.001 to 24.0 inches.
Optionally, the predefined number of pixels ranges from 2.0 to 100.00 pixels.
Optionally, the predefined threshold distance ranges from 10 to 100 feet when evaluated using a scale native to the video game.
Optionally, the process is repeated until at least one of a first condition or a second condition is reached, wherein the first condition corresponds to a size of the second LOD visual asset being determined to be less than or equal to a predefined size, and wherein the second condition corresponds to a predefined maximum number of the plurality of LOD visual assets being generated.
Optionally, the predefined size ranges from 4.0 to 1000.0 inches, and wherein the predefined maximum number of the plurality of LOD visual assets is 6.
In some embodiments, the present specification is directed towards a method of selecting a level of detail (LOD) visual asset, from a set of pre-generated generated LOD visual assets, for rendering during gameplay in a multiplayer game, said multiplayer game being executed in a multiplayer gaming network comprising at least one game server in data communication with a plurality of client devices located remote from each other, the method comprising: accessing data representative of the set of pre-generated LOD visual assets and associated switch distances; monitoring one or more of a plurality of variables related to the plurality of client devices and/or the gameplay being rendered on the plurality of client devices; determining if one or more of a plurality of corrective factors should be applied to the associated switch distances; and based on the determination, performing a first action or a second action, wherein the first action comprises applying one or more of the plurality of corrective factors to the associated switch distances in order to generate modulated switch distances and selecting a LOD from the set of pre-generated generated LOD visual assets based on one of the modulated switch distances, and wherein the second action comprises selecting a LOD visual asset from the set of pre-generated LOD visual assets based on the associated switch distances corresponding to the selected LOD visual asset.
Optionally, the plurality of variables comprise at least one of a field of view change, a resolution of display screen, a rate of vertex processing throughput, a processor performance or a memory budget.
Optionally, the plurality of corrective factors comprise at least one of a field of view scaling factor, a screen resolution scaling factor, a vertex processing efficiency scaling factor, a processor performance scaling factor or a memory budget-based biasing factor.
Optionally, if a field of view decreases then a switch distance associated with a pre-generated LOD visual asset is increased by said field of view scaling factor and wherein if the field of view increases then the switch distance associated with the pre-generated LOD visual asset is decreased by the field of view scaling factor.
Optionally, the associated switch distances are based on a display having a predefined resolution, and wherein the screen resolution scaling factor is applied to the associated switch distances if a resolution of display on one of said plurality of client devices is above or below the predefined resolution.
Optionally, said vertex processing efficiency scaling factor defines a vertex processing time that, across a plurality of computing platforms, does not vary beyond a range of 0.1 to 30 milliseconds.
Optionally, the processor performance scaling factor is applied to a switch distance associated with a pre-generated LOD visual asset if a processor workload on one of the plurality of client devices causes a maximum processor frame latency to exceed a predefined value.
Optionally, the maximum processor frame latency is 16.67 milliseconds.
Optionally, prior to or during rendering, performing one or more of the following interventions: preventing rendering of a LOD visual asset if a projected size of the LOD visual asset is smaller than a predefined size, switching off procedural vertex motion of a LOD visual asset being rendered at a distance ranging from 256 to 512 inches from a virtual camera, delaying or not allowing a static LOD visual asset to LOD visual asset out earlier than designated by an associated switch distance of the LOD visual asset, and recursively attempting to render a lower complexity LOD visual asset if a higher complexity cannot be rendered due to insufficient streaming bandwidth.
In some embodiments, the present specification discloses a computer readable non-transitory medium comprising a plurality of executable programmatic instructions wherein, when said plurality of executable programmatic instructions are executed by a processor in a computing device, a process for selecting a level of detail (LOD) visual asset, from a set of pre-generated LOD visual assets, is performed to render during gameplay in a multiplayer game, the multiplayer game being executed using a multiplayer gaming network comprising at least one game server in data communication with a plurality of client devices located remote from each other, the plurality of executable programmatic instructions comprising: accessing data representative of the set of pre-generated LOD visual assets and associated switch distances; monitoring one or more of a plurality of variables related to the plurality of client devices and/or the gameplay being rendered on the plurality of client devices; determining if one or more of a plurality of corrective factors should be applied to the associated switch distances; and based on the determination, performing a first action or a second action, wherein the first action comprises applying one or more of the plurality of corrective factors to the associated switch distances in order to generate modulated switch distances and selecting a LOD visual asset from the set of pre-generated LOD visual assets based on one of the modulated switch distances, and wherein the second action comprises selecting a LOD visual asset from the set of pre-generated LOD visual assets based on the associated switch distances corresponding to the selected LOD visual asset.
Optionally, the plurality of variables comprise at least one of a field of view change, a resolution of display screen, a rate of vertex processing throughput, a processor performance or a memory budget.
Optionally, the plurality of corrective factors comprise at least one of a field of view scaling factor, a screen resolution scaling factor, a vertex processing efficiency scaling factor, a processor performance scaling factor or a memory budget-based biasing factor.
Optionally, if a field of view decreases, then a switch distance associated with a pre-generated LOD visual asset is increased by said field of view scaling factor and wherein if the field of view increases then the switch distance associated with the pre-generated LOD visual asset is decreased by the field of view scaling factor.
Optionally, the associated switch distances are based on a display having a predefined resolution, and wherein the screen resolution scaling factor is applied to the associated switch distances if a resolution of a display on one of the plurality of client devices is above or below the predefined resolution.
Optionally, the vertex processing efficiency scaling factor establishes a vertex processing time that, across a plurality of target platforms, does not vary beyond a range of 0.1 to 30 milliseconds.
Optionally, the processor performance scaling factor is applied to a switch distance associated with a pre-generated LOD visual asset if a processor workload on one of said plurality of client devices causes a predefined maximum processor frame latency to exceed.
Optionally, said predefined maximum processor frame latency is 16.67 milliseconds.
Optionally, the computer readable non-transitory medium further comprises, prior to or during rendering, performing one or more of the following interventions: preventing rendering of a LOD visual asset if a projected size of the LOD visual asset is smaller than a predefined size, switching off procedural vertex motion of a LOD visual asset being rendered at a distance ranging from 256 to 512 inches from a virtual camera, delay or not allow a static LOD visual asset to LOD visual asset out earlier than designated by an associated switch distance of the LOD visual asset, and recursively attempting to render a lower complexity LOD visual asset if a higher complexity cannot be rendered due to insufficient streaming bandwidth.
In some embodiments, the present specification discloses a method for generating a plurality of proxy level of detail (LOD) visual assets for contents of a grid cell of a game map in a multi-player video game, wherein the game map is segmented into a plurality of grid cells, the method comprising: receiving, into a LOD management engine, data indicative of the contents of the grid cell; applying, by the LOD management engine, at least one of a plurality of functions to the data; generating, as first output by the LOD management engine, a base proxy LOD visual asset representative of the data and an associated base proxy switch distance as a result of the applying, wherein the base proxy LOD visual asset is stored in association with the base proxy switch distance; receiving, as input into a LOD management module, the base proxy LOD visual asset and the associated base proxy switch distance; generating, as second output by the LOD management module, a first proxy LOD visual asset by iteratively reducing a polygon mesh complexity of the base proxy LOD visual asset until a maximum geometric deviation of the first proxy LOD visual asset approaches, but does not exceeds, a predefined threshold value with respect to the base proxy LOD visual asset; generating, as third output by the LOD management module, a first proxy switch distance at which the predefined threshold value, when projected on a display of a default resolution, comprises a predefined number of pixels; and storing, by the LOD management module, the second output in association with the third output if a difference between the base proxy switch distance and the first proxy switch distance is greater than a predefined threshold distance.
Optionally, the contents include a plurality of static object models represented by their respective LOD visual assets.
Optionally, the plurality of functions include at least one of remeshing or material baking.
Optionally, if a difference between the base proxy switch distance and the first proxy switch distance is less than or equal to the predefined threshold distance, then the second output and the third output are provided as input to the LOD management module.
Optionally, the predefined threshold value is a maximum distance between vertices in the base proxy LOD visual asset and closest corresponding points on the first proxy LOD visual asset.
Optionally, the predefined threshold value ranges from 0.001 to 24.0 inches.
Optionally, the predefined number of pixels ranges from 2.0 to 100.00 inches.
Optionally, the predefined threshold distance ranges from 10 to 100 feet in a distance scale of the multi-player video game.
Optionally, the steps of receiving input, generating second and third outputs and storing second and third outputs are repeated until at least one of first or second conditions is satisfied, wherein the first condition corresponds to a size of the first proxy LOD visual asset being determined to be less than or equal to a predefined size, and wherein the second condition corresponds to a predefined maximum number of the plurality of proxy LOD visual assets being generated.
Optionally, the predefined size ranges from 4.0 to 1000.0 inches, and wherein a predefined maximum number of the plurality of proxy LOD visual assets is 6.
Optionally, the method for generating a plurality of proxy level of detail (LOD) visual assets for contents of a grid cell of a game map in a multi-player video game, wherein the game map is segmented into a plurality of grid cells, further comprises generating, by the LOD management module, one or more graphical user interfaces for receiving user input to modify one or more of the generated plurality of proxy LOD visual assets, wherein a portion of the one or more graphical user interfaces prompt for user inputs related to one or more of the following attributes: auto LOD visual asset generation, LOD visual asset count, a LOD visual asset index at which to start generating the plurality of LOD visual assets, distance scale, tri-coverage LOD visual asset threshold, LOD visual asset blink-out, reduction percentage, weight settings to modify a number of triangles, edges and points, importance settings to modify geometry, texture coordinate, normal, material ID, object ID, skinning and vertex color, and settings to modify a maximum deviation, weld threshold, vertex regularization, surface area and/or normal split tolerance.
Optionally, the present specification discloses a computer readable non-transitory medium comprising a plurality of executable programmatic instructions wherein, when said plurality of executable programmatic instructions are executed by a processor in a computing device, a process for generating a plurality of proxy level of detail (LOD) visual assets is performed for contents of a grid cell of a game map in a multi-player video game, wherein the game map is segmented into a plurality of grid cells, the plurality of executable programmatic instructions comprising: receiving, into a LOD management engine, data indicative of the contents of the grid cell; applying, by the LOD management engine, at least one of a plurality of functions to said data; generating, as first output by the LOD management engine, a base proxy LOD visual asset representative of said data and an associated base proxy switch distance as a result of said applying, wherein the base proxy LOD visual asset is stored in association with the base proxy switch distance; receiving, as input into a LOD management module, the base proxy LOD visual asset and the associated base proxy switch distance; generating, as second output by the LOD management module, a first proxy LOD visual asset by iteratively reducing a polygon mesh complexity of the base proxy LOD visual asset until a maximum geometric deviation of the first proxy LOD visual asset approaches, but does not exceeds, a predefined threshold value with respect to the base proxy LOD visual asset; generating, as third output by the LOD management module, a first proxy switch distance at which the predefined threshold value, when projected on a display of a default resolution, comprises a predefined number of pixels; and storing, by the LOD management module, the second output in association with the third output if a difference between the base proxy switch distance and the first proxy switch distance is greater than a predefined threshold distance.
Optionally, the contents include a plurality of static object models represented by their respective LOD visual assets.
Optionally, the plurality of functions include at least one of remeshing or material baking.
Optionally, if a difference between the base proxy switch distance and the first proxy switch distance is less than or equal to the predefined threshold distance, then the second output and the third output are not stored and instead provided as input to the LOD management module.
Optionally, the predefined threshold value is a maximum distance between vertices in the base proxy LOD visual asset and closest corresponding points on the first proxy LOD visual asset.
Optionally, the predefined threshold value ranges from 0.001 to 24.0 inches.
Optionally, the predefined number of pixels ranges from 2.0 to 100.00 inches.
Optionally, the predefined threshold distance ranges from 10 to 100 feet in a scale of the multi-player video game.
Optionally, the steps of receiving input, generating second and third outputs and storing second and third outputs are repeated until at least one of first or second conditions is satisfied, wherein the first condition corresponds to a size of the first proxy LOD visual asset being determined to be less than or equal to a predefined size, and wherein the second condition corresponds to a predefined maximum number of the plurality of proxy LOD visual assets being generated.
Optionally, the predefined size ranges from 4.0 to 1000.0 inches, and wherein a predefined maximum number of the plurality of proxy LOD visual assets is 6.
Optionally, the computer readable non-transitory medium comprising a plurality of executable programmatic instructions, wherein the plurality of executable programmatic instructions further comprise generating, by the LOD management module, one or more graphical user interfaces for receiving user input to modify one or more of the generated plurality of proxy LOD visual assets, wherein a portion of the one or more graphical user interfaces prompt for user inputs related to one or more of the following attributes: auto LOD visual asset generation, LOD visual asset count, a LOD visual asset index at which to start generating the plurality of LOD visual assets, distance scale, tri-coverage LOD visual asset threshold, LOD visual asset blink-out, reduction percentage, weight settings to modify a number of triangles, edges and points, importance settings to modify geometry, texture coordinate, normal, material ID, object ID, skinning and vertex color, and settings to modify a maximum deviation, weld threshold, vertex regularization, surface area and/or normal split tolerance.
The aforementioned and other embodiments of the present specification shall be described in greater depth in the drawings and detailed description provided below.
These and other features and advantages of the present specification will be appreciated, as they become better understood by reference to the following detailed description when considered in connection with the accompanying drawings, wherein:
The present specification is directed towards multiple embodiments. The following disclosure is provided in order to enable a person having ordinary skill in the art to practice the invention. Language used in this specification should not be interpreted as a general disavowal of any one specific embodiment or used to limit the claims beyond the meaning of the terms used therein. The general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the invention. Also, the terminology and phrascology used is for the purpose of describing exemplary embodiments and should not be considered limiting. Thus, the present invention is to be accorded the widest scope encompassing numerous alternatives, modifications and equivalents consistent with the principles and features disclosed. For purpose of clarity, details relating to technical material that is known in the technical fields related to the invention have not been described in detail so as not to unnecessarily obscure the present invention.
The term “a multi-player online gaming environment” or “massively multiplayer online game” may be construed to mean a specific hardware architecture in which one or more servers electronically communicate with, and concurrently support game interactions with, a plurality of client devices, thereby enabling each of the client devices to simultaneously play in the same instance of the same game. Preferably the plurality of client devices number in the dozens, preferably hundreds, preferably thousands. In one embodiment, the number of concurrently supported client devices ranges from 10 to 5,000,000 and every whole number increment or range therein. Accordingly, a multi-player gaming environment or massively multi-player online game is a computer-related technology, a non-generic technological environment, and should not be abstractly considered a generic method of organizing human activity divorced from its specific technology environment.
In various embodiments, the system includes at least one processor capable of processing programmatic instructions, has a memory capable of storing programmatic instructions, and employs software comprised of a plurality of programmatic instructions for performing the processes described herein. In embodiments, a computer-readable non-transitory medium comprises the plurality of executable programmatic instructions. In one embodiment, the at least one processor is a computing device capable of receiving, executing, and transmitting a plurality of programmatic instructions stored on a volatile or non-volatile computer readable medium.
In various embodiments, a computing device includes an input/output controller, at least one communications interface and system memory. The system memory includes at least one random access memory (RAM) and at least one read-only memory (ROM). These elements are in communication with a central processing unit (CPU) to enable operation of the computing device. In various embodiments, the computing device may be a conventional standalone computer or alternatively, the functions of the computing device may be distributed across multiple computer systems and architectures.
In some embodiments, execution of a plurality of sequences of programmatic instructions or code enable or cause the CPU of the computing device to perform various functions and processes. In alternate embodiments, hard-wired circuitry may be used in place of, or in combination with, software instructions for implementation of the processes of systems and methods described in this application. Thus, the systems and methods described are not limited to any specific combination of hardware and software.
The term “module”, “application” or “engine” used in this disclosure may refer to computer logic utilized to provide a desired functionality, service or operation by programming or controlling a general purpose processor. Stated differently, in some embodiments, a module, application or engine implements a plurality of instructions or programmatic code to cause a general purpose processor to perform one or more functions. In various embodiments, a module, application or engine can be implemented in hardware, firmware, software or any combination thereof. The module, application or engine may be interchangeably used with unit, logic, logical block, component, or circuit, for example. The module, application or engine may be the minimum unit, or part thereof, which performs one or more particular functions.
The term “platform” or “target gaming platform” used in this disclosure may refer to hardware and/or software specifications of a player's client device (which may be a PC or a gaming console, for example). In some embodiments, “platform” may refer to at least GPU (Graphics Processing Unit) specification, CPU specification, display screen resolution, RAM and hard disk space available and a type of operating system.
The term “offline” or “offline process” used in this disclosure refers to one or more programmatic instructions or code that may be implemented or executed while the game is not being played by any player (that is, while the one or more game servers are not rendering a game for playing).
The term “runtime” or “runtime process” used in this disclosure refers to one or more programmatic instructions or code that may be implemented or executed during gameplay (that is, while the one or more game servers are rendering a game for playing).
The term “switch distance” used in this disclosure refers to a distance of an object from an in-game virtual camera at which the game switches from one LOD to another. In some embodiments, for small objects (that is, objects having an on-screen size equal to or below a predefined threshold size) the switch distance refers to a “distance to pivot” that is, the distance of the virtual camera from a point of pivot on a small object (for example, for a small tree the point of pivot may lie at a base of a trunk of the small tree) and is defined as =length(
−
=length(max(
−
−
,
)).
In the description and claims of the application, each of the words “comprise” “include” and “have”, and forms thereof, are not necessarily limited to members in a list with which the words may be associated. It should be noted herein that any feature or component described in association with a specific embodiment may be used and implemented with any other embodiment unless clearly indicated otherwise.
As used herein, the indefinite articles “a” and “an” mean “at least one” or “one or more” unless the context clearly dictates otherwise.
The one or more game servers 105 can be any computing device having one or more processors and one or more computer-readable storage media such as RAM, hard disk or any other optical or magnetic media. The one or more game servers 105 include a plurality of modules operating to provide or implement a plurality of functional, operational or service-oriented methods of the present specification. In some embodiments, the one or more game servers 105 include or are in communication with at least one database system 120. The database system 120 stores a plurality of game data including data representative of a plurality of LOD (Level of Detail) assets (along with switch distances corresponding to the plurality of LOD assets and LOD-related attributes, characteristics or parameters) associated with at least one game that is served or provided to the client devices 110 over the network 115. In some embodiments, the one or more game servers 105 may be implemented by a cloud of computing platforms operating together as game servers 105.
In accordance with aspects of the present specification, the one or more game servers 105 provide or implement a plurality of modules or engines such as, but not limited to, a master game module 130, a hybrid LOD management module 132, and a rendering module 134. In some embodiments, the one or more client devices 110 are configured to implement or execute one or more of a plurality of client-side modules some of which are same as or similar to the modules of the one or more game servers 105. For example, in some embodiments each of the player client devices 110 executes a client-side game module 130′ (also referred to as—client game module 130′) that integrates a client-side rendering module 134′ (also referred to as—client rendering module 134′) while at least one non-player client device 110g executes the client game module 130′ that integrates a client-side hybrid LOD management module 132′ (also referred to as—client LOD management module 132′) and the client rendering module 134′.
In some embodiments, the at least one non-player client device 110g does not include the client LOD management module 132′. Instead, the non-player client device 110g is used by the computer graphics artist or designer to log into the one or more game servers 105 and execute the hybrid LOD management module 132 on the server to auto-generate a plurality of LODs and their associated switch distances, offline. The auto-generated plurality of LODs and their associated switch distances are stored in the at least one database 120.
In some embodiments, the one or more game servers 105 do not implement the hybrid LOD management module 132. Instead, the hybrid LOD management module 132 is implemented on the non-player client device 110g wherein the computer graphics artist or designer execute the hybrid LOD management module 132 on the server to auto-generate a plurality of LODs and their associated switch distances, offline. The auto-generated plurality of LODs and their associated switch distances are then uploaded, via the network 115, and stored in the at least one database 120.
While various aspects of the present specification are being described with reference to functionalities or programming distributed across multiple modules or engines 132 and 134, it should be appreciated that, in some embodiments, some or all of the functionalities or programming associated with these modules or engines may be integrated within fewer modules or in a single module—such as, for example, in the master game module 130 itself on the server side and in the client gaming module 130′ on the client side.
In embodiments, the master game module 130 is configured to execute an instance of an online game to facilitate interaction of the players with the game. In embodiments, the instance of the game executed may be synchronous, asynchronous, and/or semi-synchronous. The master game module 130 controls aspects of the game for all players and receives and processes each player's input in the game. In other words, the master game module 130 hosts the online game for all players, receives game data from the client devices 110 and transmits updates to all client devices 110 based on the received game data so that the game, on each of the client devices 110, represents the most updated or current status with reference to interactions of all players with the game. Thus, the master game module 130 transmits game data over the network 115 to the client devices 110 for use and rendering by the game module 130′ to provide local versions and current status of the game to the players.
On the client-side, each of the one or more player client devices 110 implements the game module 130′ that operates as a gaming application to provide a player with an interface between the player and the game. The game module 130′ generates the interface to render a virtual environment, virtual space or virtual world associated with the game and enables the player to interact in the virtual environment to perform a plurality of game and other tasks and objectives. The game module 130′ accesses game data received from the game server 105 to provide an accurate representation of the game to the player. The game module 130′ captures and processes player inputs and interactions within the virtual world or environment and provides updates to the game server 110 over the network 115.
In embodiments, the game module 130′ (for each of the one or more player client devices 110) also integrates the client rendering module 134′ that, in data communication with the server-side rendering module 134, performs a plurality of tasks (during runtime or execution of gameplay) such as: a) determining which representation, version or LOD, out of a set, group or chain of LOD assets, to render for a given gameplay view or scene, b) assessing a player's client device configurations and platform specifications such as, but not limited to, display screen resolution, GPU capabilities and memory availability, c) monitoring network (for example, Internet) connectivity or streaming bandwidth fluctuations, GPU workload and performance parameters (such as, for example, frame latency), memory usage and the player's field of view (FOV) changes, d) dynamically applying one or more of a plurality of corrective factors to offline authored switch distances of various LOD assets, wherein the one or more of a plurality of corrective factors include factors such as FOV scaling, screen resolution scaling, vertex processing efficiency scaling, general processing unit (GPU) performance scaling and memory budget-based biasing, and c) implementing a plurality of case, context or scenario based interventions such as rejecting objects smaller than a predefined threshold, switching off procedural vertex motion or animation for more distant LODs, delaying or not allowing LODs corresponding to a ‘static’ object to switch to lower complexity LODs earlier than designated by the offline authored switch distances of the respective LODs, implementing LOD fallback plans, and rendering proxy LODs, when needed.
In embodiments, the at least one client device 110g is configured as a non-player computing device to enable a graphics artist or designer to interact with the one or more game servers 105. In embodiments, the at least one client device 110g also implements the client game module 130′ that, in some embodiments, further integrates the client LOD management module 132′ and the client rendering module 134′. In accordance with some aspects of the present specification, an offline execution of the client LOD management module 132′, in data communication with the server-side LOD management module 132, enables auto-generation of one or more chains of LOD assets with associated offline authored switch distances. The offline execution of the client LOD management module 132′, in data communication with the server-side LOD management module 132, further results in generating one or more GUIs (graphical user interfaces) to enable the graphics designer or artist to optimize and modify one or more of the auto-generated chains of LOD assets and associated offline switch distances on the basis of at least one of a plurality of LOD-related attributes, characteristics or parameters.
The database system 120 described herein may be, include, or interface to, for example, an Oracle™ relational database sold commercially by Oracle Corporation. Other databases, such as Informix™, DB2 (Database 2) or other data storage, including file-based, or query formats, platforms, or resources such as OLAP (On Line Analytical Processing), SQL (Structured Query Language), a SAN (storage area network), Microsoft Access™ or others may also be used, incorporated, or accessed. The database system 120 may comprise one or more such databases that reside in one or more physical devices and in one or more physical locations.
During an offline process, a hybrid LOD management module or engine 132 implements a plurality of instructions or programmatic code to a) enable automated generation of a plurality of representations or versions of an object model (that corresponds to a most detailed, complex or highest resolution object model). In embodiments, each of the plurality of representations or versions of the object model has a different level of detail (LOD) ranging from a most detailed or complex LOD representation to a least complex (or simplest) LOD representation, and b) generate one or more GUIs (graphical user interfaces) to enable a graphics designer or artist to modify one or more of the plurality of representations or versions of the object model on the basis of at least one of a plurality of LOD-related attributes, characteristics or parameters.
In embodiments, a representation, version or LOD of an object model is described using a plurality of geometric primitives such as, for example, triangles, quadrilaterals, polygons or non-uniform rational basis spline (NURBS). In embodiments, a base model is associated with a highest number of geometric primitives and therefore a highest complexity or resolution.
At step 202, the module 132 receives, as input, data indicative of a first representation or LOD of an object model and a first switch distance associated with the first representation or LOD. In some embodiments, the first representation or LOD and the associated first switch distance are stored in the at least one database 120 from which this data is accessible to the module 132.
At step 204, the module 132 iteratively reduces the polygon mesh complexity (that is, reduce the number of polygons) of the first representation or LOD to generate a second representation or LOD for which a maximum geometric deviation or error approaches a predefined threshold ‘D’, without exceeding it, compared to the first representation or LOD. Accordingly, in some embodiments, each vertex in the first representation diverges or deviates from a closest corresponding point on the second representation of the object by no more than ‘D’ units. Therefore, in some embodiments, the predefined threshold ‘D’ is a maximum distance between vertices in the first representation and the closest corresponding points on the second representation of the object. That is, the predefined threshold ‘D’ is a maximum distance between any point on the surface of the first representation and any closest corresponding point on the surface of the second representation of the object. In embodiments, the second representation corresponds to a less complex or simplified version of the object model compared to the first representation.
As shown in
In embodiments, the value of the predefined threshold ‘D’ is manually customizable and may vary at least across different types of object models such as those corresponding to, for example, a hero object, simple props, and peripheral objects. In some embodiments, the predefined threshold ‘D’ ranges from 0.001 to 24.0 inches depending upon the type of object model. In alternate embodiments, the aforementioned range for the predefined threshold ‘D’ may be expressed in other units such as, for example, millimeters, centimeters and feet.
It should be appreciated that the polygon mesh simplification is directed towards reducing the complexity of a mesh (number of polygons) while optimizing visual fidelity. In various embodiments, polygon mesh simplification is achieved using geometric reduction methods known to persons of ordinary skill in the art such as, for example, edge collapse, vertex pair collapse, triangle collapse, cell collapse, vertex removal, polygon merging, volumetric processing, and simplification envelopes.
At step 206, the module 132 calculates a second switch distance (corresponding to the second representation or LOD) at which the threshold ‘D’, when projected on a display of a default resolution, consumes or occupies a predefined number of pixels ‘P’. As shown in
In some embodiments, the default resolution corresponds to a 1080p display. In some embodiments, the predefined number of pixels ‘P’ ranges from 2.0 to 100.00 pixels for a 1080p display. In some embodiments, the module 132 calculates the second switch distance by translating the geometric deviation or error ‘D’ to a pixel distance on-screen by projecting a sphere of a certain diameter onto a screen of the default resolution and determining a size of the resultant circle. Accordingly, in some embodiments, the second switch distance (switchDist) is calculated using the following equations:
where, P is a predefined number of pixels, D is a predefined geometric deviation; modelRadius (in inches) is the smallest 3D sphere that tightly encompasses a given model; screenRatio is a different representation of pixelSize ‘P’, normalized against ‘screenSize’; screenSize is a fixed size of a reference display screen (and that in some embodiments is 1080.0 pixels); nearClipDist is a distance to the near clipping plane of a normalized (unit) view frustum defined by “fovRadians”; fovRadians is a field of view of a reference display screen in radians (which, in some embodiments, is 1.13446401 (65.0 degrees)); and bsphereAngle is an angle, in radians, of a given model's bounding sphere (defined by its center and “modelRadius”) projected on the reference display.
At step 208, the module 132 compares the second switch distance with the first switch distance to determine if the first and second switch distances are too close to each other. Stated differently, if a difference between the first and second switch distances is determined to be less than or equal to a predefined threshold ‘d’ then, at step 210, the module 132 does not store the second representation or LOD in the at least one database 120. That is, if the difference between the first and second switch distances is not greater than the predefined threshold ‘d’ then the two switch distances are considered to be too close to each other. Subsequently, the flow control moves back to step 202 with the second representation or LOD and the second switch distance being received as input by the module 132 in order to generate a subsequent LOD and corresponding switch distance.
However, if the difference between the first and second switch distances is determined to be greater than the predefined threshold ‘d’ then, at step 212, the module 132 associates the calculated second switch distance with the second representation or LOD and stores the second switch distance in association with the second representation in the at least one database 120. In some embodiments, the predefined threshold ‘d’ ranges from 10 to 100 feet for a 1080p display.
It should be appreciated that the comparison of two subsequent switch distances, such as the first and second switch distances, is directed towards preventing wastage of memory space (for storing various object model representations or LODs) and LOD assets. As an illustration, if there are two LODs that switch within, say, 1 inch of each other, it indicates that one of the two LODs will hardly ever be seen during runtime or gameplay. Therefore, an objective is to ensure that any two LODs have switch distances that are at least some significant distance apart so that each of the two LODs would be actually seen during gameplay (otherwise these two LODs would just take up memory space). In some embodiments, the predefined threshold ‘d’ is at least equal to the sum of all previous values of ‘d’. For example, when determining ‘d’ for LOD3, it will at least be equal to a first ‘d’ for LOD1 (i.e. the maximum geometric deviation between LOD0 and LOD1) plus a second ‘d’ for LOD2 (i.e. the maximum geometric deviation between LOD1 and LOD2.
During a first pass of the steps 202, 204, 206, 208, 210 and 212, in some embodiments, the first representation or LOD corresponds to a full object model. The full object model is associated with a highest number of geometric primitives and therefore a highest complexity or resolution. Accordingly, the first switch distance, in some embodiments, corresponds to a predefined switch distance associated with the full object model. Thus, in the first pass of the steps 202, 204, 206, 208, 210 and 212, the module 132 iteratively reduces the polygon mesh complexity of the first representation or LOD (the full object model) to generate the second representation or LOD (the base representation or LOD0) for which a geometric deviation or error reaches the predefined threshold ‘D’ compared to the first representation or LOD. Thereafter, the module 132 calculates a second switch distance (corresponding to the second representation or LOD) at which the threshold ‘D’, when projected on the display of the default resolution, consumes or occupies the predefined number of pixels ‘P’.
Steps 202, 204, 206, 208, 210 and 212 are repeated by the module 132 to generate subsequent representations or LODs. For example, to generate a third representation or LOD, the previously generated second representation is received as input in the module 132, at a repeat of step 202. At a repeat of step 204, the module 132 iteratively reduces the polygon mesh complexity of the second representation or LOD to generate the third representation or LOD for which a geometric deviation or error reaches the predefined threshold ‘D’ compared to the second representation or LOD. Thereafter, at a repeat of step 206, the module 132 calculates a third switch distance (corresponding to the third representation or LOD) at which the threshold ‘D’, when projected on the display of the default resolution, consumes or occupies the predefined number of pixels ‘P’. Now, at step 208, the model 132 compares the second switch distance with the third switch distance and, based on the comparison, either rejects the third representation, at a repeat of step 210, or stores the third switch distance in association with the third representation in the at least one database 120, at a repeat of step 212.
Similarly, the third representation or LOD is taken as input by the module 132 (at another repeat of step 202) in order to generate a fourth representation or LOD, and so on. Thus, the module 132 iterates through the steps 202, 204, 206, 208, 210 and 212 to generate a plurality of representations or LODs of the object model such that a generated representation or LOD is characterized by increasing level of simplicity or reduced number of geometric primitives compared to an immediate prior generated representation or LOD.
In some embodiments, the module 132 repeats the steps 202, 204, 206, 208, 210 and 212 until a first condition is met, wherein the first condition corresponds to a size of a generated representation or LOD being determined to be less than or equal to a predefined size of ‘s’ inches at a correspondingly calculated switch distance. That is, the module 132 stops generating further representations or LODs of the object model when the size of a generated representation approaches the predefined size ‘s’ indicative of the size of the generated representation being too small. In some embodiments, the predefined size ‘s’ ranges from 4.0 to 1000.0 inches on a 1080p display. In alternate embodiments, the aforementioned range for the predefined size ‘s’ may be expressed in other units such as, for example, millimeters, centimeters and feet.
In some embodiments, the module 132 repeats the steps 202, 204, 206, 208, 210 and 212 until a second condition is met, wherein the second condition corresponds to a predefined maximum number of representations or LODs being generated. In some embodiments, the predefined maximum number of representations or LODs is 5. In some embodiments, when including the base representation or LOD, the predefined maximum number of representations or LODs is 6.
In some embodiments, the module 132 repeats the steps 202, 204, 206, 208, 210 and 212 until at least one of the first or second condition is met.
In embodiments, as part of an offline process, a computer graphics artist or designer exports an object model (represented by plurality of geometric primitives) from a Digital Content Creation (DCC) application such as, for example, 3ds Max or Maya. The exported object model is provided as an input to the hybrid LOD management module or engine 132 that executes or implements the method 200 of
In accordance with aspects of the present specification, the module 132 enables the computer graphics artist or designer to further optimize the auto generated plurality of geometric representations, versions or LODs, if required.
Accordingly, in some embodiments, at step 222, the module 132 generates one or more GUIs (Graphical User Interfaces) and seeds or populates the one or more GUIs with data representative of the plurality of auto-generated geometric representations, versions or LODs (that were generated by the module 132 using the method 200 of
The module 132 generates the one or more GUIs to prompt the computer graphics artist or designer for inputs related to modulation of one or more of the plurality of LOD-related attributes, characteristics or parameters in order to modify one or more of the plurality of auto-generated geometric representations, versions or LODs.
Referring now to
It should be appreciated that both distance scale 408 and tri-coverage LOD threshold 410 are used to modulate the distance used for LOD switching, albeit with different metrics. Scaling distance directly, using the distance scale 408, is more intuitive (for example, a value of “2” means double the distance) but disregards runtime performance or visual implications. Scaling via tri coverage 410 has more deterministic runtime visual and performance implications, but since that final scale value is highly dependent on source model topology, its results are not intuitive.
Also, each LOD (425, 426, 427 and 428) has a group of associated parameters such as vertices count 418—indicative of the number of vertices in a polygon mesh of a LOD, LOD distance 420—indicative of a distance at which an associated LOD is switched, surfaces count 422—indicative of a number of surface in a polygon mesh of a LOD and bone count 424. Persons of ordinary skill in the art would appreciate that all models have at least one bone. Articulated models, such as animated characters, can have a plurality of bones and upwards of hundreds. The more bones in a given LOD, the more expensive it is to animate and render in-game. As such, it is prudent to reduce the number of bones required to represent a given model LOD. A surface is the smallest representation of model geometry understood by a renderer, typically representing all triangles shared by a single material/bone combination. Any model contains one or more surfaces. As such it is also prudent to minimize their counts with subsequent model LOD's as well.
As can be seen in the first GUI 400a, the vertices count 418 reduces from LOD0 425 to LOD1 427 to LOD2 429 to LOD3 430 in accordance with the reduction % 416 with reference to the base LOD0 425. A LOD distance override parameter 435 when selected/checked enables the graphic artist to modulate the auto-generated switch distance associated with a LOD. For example, as shown in the first GUI 400a, the graphic artist has selected the distance override parameter 435 for LOD1 and LOD2 and modified the corresponding switch distances, auto-generated by the module 132, to the values of 380 and 1000 respectively. The auto-generated switch distance for LOD3 has been retained by the graphics artist.
The second GUI 400b of
These values are used to help the mesh reduction algorithm or method (such as the method 200 of
Referring back to
In some embodiments, the module 132 also generates and provides the graphic artist with one or more visual, intuitive guidance tools to enable the artist to gauge the runtime efficiency (such as in terms of, for example, overall memory usage and GPU workload) of a chain of LOD assets as a result of the graphic artist making modifications to the plurality of LOD-related attributes, characteristics or parameters (such as, but not limited to, those illustrated in the first and second GUIs 400a, 400b) of the auto-generated plurality of LODs.
For example, in some embodiments, the module 132 may be configured to auto-generate LODs based on a predefined target reduction of ‘r %’ such that each auto-generated LOD has ‘r %’ less number of vertices than a prior LOD in a chain/group of LOD assets. In some embodiments, the predefined target reduction of r % is 50%. Similarly, the module 132 determines the switch distances (for each of the LODs in the chain/group of LOD assets) based on a predefined geometric deviation or error threshold ‘D’ as described earlier in the method 200 of
However, the graphics artist may end up modifying one or more of the plurality of LOD-related attributes, characteristics or parameters (such as, but not limited to, those illustrated in the first and second GUIs 400a, 400b) of the auto-generated plurality of LODs. As a non-limiting example, the graphics artist may use the first and second GUIs 400a, 400b to modify the reduction % and/or the auto-generated switch distances due to various gameplay considerations. Modification of the plurality of LOD-related attributes, characteristics or parameters may, however, penalize the runtime efficiency of the LODs.
Therefore, an objective of the first tool 500a is to visually guide the graphics artist in terms of a plurality of predefined target ranges of distances (of an object from the camera) that may be adhered to for using each LOD, in a chain of LODs, so as to minimize incurring costs in terms of the runtime efficiency of the LODs. The tool 500a broadly emphasizes that the highest quality LODs need to be used for short distances of the object from the camera. For example, in accordance with some embodiments, a base LOD0 525 may be used within a first predefined target distance range of 0 to M units. In some embodiments, M is 250 inches. A first representative model LOD1 527 may be deployed within a second predefined target distance range of M and N units. In some embodiments, N is 500 inches. Most of the additional LODs 529 (example, LOD2, LOD3, etc.) may be deployed within a third predefined target distance range of N and O units. In some embodiments, O is 1500 inches. A last LOD 532 may be used for distances ranging between O and P units. In some embodiments, P is 5000 inches. In some embodiments, beyond P units the last LOD blinks-out (that is, is not rendered).
A second curve 542 is indicative of a current decimation or reduction curve, for the chain of LODs, based at least on the graphic artist's modification of the predefined target reduction of ‘r %’. As can be observed, the second curve 542 is steeper than the first curve 540 indicating that the graphic artist's tweaking of at least the predefined target reduction of ‘r %’ has resulted in sharper or more aggressive reduction in the number of vertices for each LOD in the chain of LODs. This is desirable since, in some embodiments, the second curve 542 is indicative of improvement in the runtime efficiency of the LODs. However, if the graphic artist's tweaking of at least the predefined target reduction of ‘r %’ results in a third curve (not shown) that has a slope less steep than the first curve 540 it may indicate to the artist that the chain of LODs, corresponding to the third curve, would have lower runtime efficiency compared to those represented by the first curve 540. Accordingly, the graphic artist may take corrective measures and re-modify the LODs to ensure cutback in the undesired deviation of the third curve with respect to the first curve 540.
In accordance with some aspects of the present specification, the hybrid LOD management module 132 performs LOD packing which, in some embodiments, is an offline process by which all vertex attributes (color, normal, texture coordinates, skinning weights, material ID) are stored in separate, compact streams of data for each LOD of a model.
In accordance with further aspects of the present specification, the module 132 further performs quantization that, in general, refers to mapping from a large set of values to a smaller set. In some embodiments, vertex position values (x, y, z) are mapped, with each of the three coordinates starting as 32-bit floating point values, to 75-bits total. By using the BoundsRadius (or modelRadius) it is ensured that a smaller target set of this mapping is within the model's bounds, which is much smaller than the original set of all real numbers. This ensures a minimum loss of vertex position fidelity, as it is a lossy form of data compression.
In some embodiments, object-space positional quantization is defined as follows:
At runtime, the rendering module or engine 134 (in data communication with a client-side rendering module or engine 134′) implements a plurality of instructions or programmatic code to determine which representation, version or LOD, out of a set, group or chain of LOD assets, to render for a given gameplay view or scene. Stated differently, once a chain of LOD assets along with associated switch distances (hereinafter referred to as ‘offline authored switch distances’) have been auto-generated offline for an object and, if required, further modulated by a graphics artist, the module 134 needs to select an appropriate LOD at an appropriate switch distance so as to enable consistent visuals and performance for all gameplay scenes across platforms while ensuring optimal runtime efficiency in terms of, at least, predefined target GPU performance and memory usage.
In accordance with aspects of the present specification, at runtime or gameplay, the module 134 applies at least one of a plurality of corrective factors to an offline authored switch distance, associated with each LOD of a chain of LOD assets, in order to determine a modulated switch distance for each LOD.
As an illustration, let us assume that the offline authored switch distances are s0, s1, s2, s3 and so on for a chain of LOD assets LOD0, LOD1, LOD2, LOD3 and so on. These offline authored switch distances are determined based on the method 200 of
In some embodiments, the plurality of corrective factors correspond to the following variables—FOV (Field Of View) change, resolution of screen (or scene), rate of vertex processing throughput (that is, how many vertices are processed per second), general GPU performance (that is, dynamically reducing load on GPU to accommodate a predefined target workload), and available memory (that is, memory budget for a rendered gameplay scene). Accordingly, the plurality of corrective factors comprises:
In some embodiments, the FOV scaling factor is an inverse scaling factor. For example, if the FOV change corresponds to an FOV decrease of, say, 50% then the offline authored switch distances are increased by the FOV scaling factor (and vice versa) such that LODs corresponding to higher level details are retained in a gameplay scene for a longer duration.
Accordingly, in some embodiments, the FOV scaling factor is calculated using the following equations:
where, HalfFovY refers to half of a FOV angle along a vertical y-axis; and HalfFovX refers to half of a FOV angle along a horizontal x-axis. In some embodiments, the FovY (that is, the FOV angle along the vertical axis) is 80.0 degrees.
The aforementioned equations are based on an assumption that all runtime FOV changes occur only in a horizontal axis. Also, assuming a 16:9 aspect ratio, the module 134 determines how much of a FOV scale factor should be applied to maintain the same visual behavior at a fixed vertical FOV and aspect ratio. In embodiments, vertical FOV is used to ensure same results regardless of display aspect ratio (for example, 4:3, 16:9, and 16:10).
For example, since one of the objectives is to minimize pixel popping, and a 4k resolution display has a higher pixel density compared to the default 1080p display, the screen resolution scaling factor ensures that a relative severity of all pixel pops is the same at the 4k display as at the 1080p display. This means scaling all offline authored switch distances so that the LODs are rendered at a farther/closer distance to match the player's target screen resolution.
Accordingly, in some embodiments, the screen resolution scaling factor is calculated using the following equations:
The aforementioned equations are based on an assumption that all offline authored switch distances were based on the geometric deviation or error projected on the default 1920×1080 display. This means that the “defaultScreenHeight” is always 1080 and the “defaultFOV” is always 65.0. The equations are solved for the screen resolution scaling factor that ensures a “targetScreenHeight” of 2160 produces the same projected geometric deviation or error. In some embodiments, the module 134 calculates the screen resolution scaling factor once when the player uses his client device 110 (
In embodiments, the vertex processing efficiency is a combination of at least GPU power, specifications or capabilities, memory bandwidth or space and shader compiler differences. Taking these variables into account, a plurality of empirical measurements, in vertex processing efficiency, are taken across platforms to determine an approximate vertex processing efficiency scaling factor for each of the target platforms (and stored in the at least one database 120 for access by the module 134). In an embodiment, a suite of reference meshes of varying complexities were rendered a large number (for example, on the order of hundreds) of times on all target platforms. Detailed GPU statistics were gathered from each step of the vertex processing pipeline for each platform. Those were then averaged and normalized against the platform used for most model LOD tuning —such as, for example, a PS4, running at 1080p.
Accordingly, the module 134 applies the vertex processing efficiency scaling factor, associated with a gaming platform, to offline authored switch distances (for LODs) to generate modulated switch distances optimized for the gaming platform.
Accordingly, in some embodiments, the GPU performance scaling factor is calculated using the following equation:
gpuPerfScale=mapToLogCurveWithHysteresis(gpuFrameTime)
If while rendering a gameplay scene, the GPU workload causes the predefined maximum GPU frame latency to be reached or exceeded then, in accordance with the aforementioned equation, the GPU frame time is mapped to a log curve in order to determine the GPU performance scaling factor. Thus, if the GPU workload, at the player's client device, results in the predefined maximum GPU frame latency to be reached or exceeded, the module 134 applies the GPU performance scaling factor to the offline authored switch distances (for LODs) to generate modulated switch distances as a result of which the LODs switch (to less complex LODs or to LODs with fewer number of geometric primitives) sooner (that is, at a nearer distance to the virtual camera) thereby reducing GPU workload.
To avoid visual artifacts (such as flickering) while skinning all visible LODs, the module 134 is configured to monitor and determine if a first predetermined memory space threshold, for the UAV, is being exceeded while rendering a gameplay scene. If the first predetermined memory space threshold is exceeded, the module 134 applies a first budge-based biasing factor to the offline authored switch distances in order to generate modulated switch distances for skinned LODs. In other words, the first budge-based biasing factor results in dynamically scaling down of the LODs and reducing vertex counts.
In embodiments, the geometry pipeline buffers, that handle vertex data, also have a fixed size or budget. Accordingly, the module 134 is configured to monitor and determine if a second predetermined memory space threshold, for the geometry pipeline buffers, is being exceeded while rendering a gameplay scene. If the second predetermined memory space threshold is exceeded, the module 134 applies a second budge-based biasing factor across all content to ensure that the total vertex count for the gameplay scene remains within budget.
Accordingly, the second budge-based biasing factor is determined based on the following steps:
The aforementioned steps of determining the second budget-based biasing factor may signify that: a) for each frame, monitoring how close a current rendered scene memory size is to a maximum geometry pipeline buffer limit; b) if the second predetermined memory space threshold (which, in some embodiments, is 85% of the maximum geometry pipeline buffer limit) is passed/exceeded for a frame then a predefined non-zero distance bias value (that is, the second budge-based biasing factor) is established; c) this non-zero bias value or the second budge-based biasing factor is added to all runtime LOD distance calculations; d) if the second predetermined memory space threshold is exceeded again in the very next frame, then the distance bias value or the second budge-based biasing factor is doubled (with reference to the predefined non-zero distance bias value at step b); and e) if for a frame the memory size of a rendered scene falls below the second predetermined memory space threshold, a decay factor is applied to the second budge-based biasing factor until it reaches 0 again.
In some embodiments, each of the steps a, b, c, d, and e above are repeated for global skinning UAV limit. The only difference is that the skinning UAV limit bias (or, in different terms, the first budge-based biasing factor) is only added to switch distances calculated for models that are animated/skinned (for example, characters, and vehicles). In some embodiments, the first and second budge-based biasing factor can be combined (or added).
In various embodiments, the module 134 determines modulated switch distance SDmodulated for a LOD as a function of a corresponding offline authored switch distance SDoffline and the plurality of corrective factors. Specifically, in some embodiments, a modulated switch distance for a LOD is calculated using the following equation:
where, SDmodulated is a modulated switch distance for a LOD, SDoffline is an offline determined switch distance for the LOD, invFovScale is a FOV scaling factor, scenResScale is a scene resolution scaling factor, vertexProcessingScale is a vertex processing efficiency scaling factor, gpuPerfScale is a GPU performance scaling factor, and geoBias is a memory budge-based biasing factor (including first and/or second biasing factors corresponding to memory limits for UAVs and/or geometry pipeline buffers, respectively).
Thus, during runtime, the module 134 is configured to select and render one or more LOD assets on the basis of the associated modulated switch distances that are determined by modifying the associated offline authored switch distances for the one or more LOD assets based on at least one of the plurality of corrective factors. In some embodiments, the modulated switch distances are determined by modifying the offline authored switch distances for the one or more LOD assets based on at least two of the plurality of corrective factors.
In accordance with further aspects of the present specification, at runtime, the module 134 implements a plurality of instructions or programmatic code to monitor gameplay scenes and apply one or more case, context or scenario based interventions aimed at striking a balance between visual or perceptual quality and GPU performance. In embodiments, the one or more case, context or scenario based interventions are those related to a size of the LOD being rendered, procedural vertex motion or animation for the LOD, whether the LOD represents a static object, and streaming bandwidth deterioration between the player's client device and the one or more game servers. Accordingly, in some embodiments, the following case, context or scenario based interventions are implemented by the module 134:
In some embodiments, the module 134 dynamically applies a size-based scaling factor to the predefined size threshold of ‘L’ units based on parameters that tend to affect general GPU performance. In some embodiments, with each drop in scene resolution (up to, say, 15), the size-based scaling factor is increased to match the new scene resolution using the same equation as that of the “screen resolution scaling factor” described earlier in this specification. For example, if screen or scene resolution dynamically drops by 2%, the predefined size threshold is scaled accordingly by applying the sized-based scaling factor. Thus, if screen or scene resolution dynamically drops then the predefined size threshold also drops and vice versa.
However, in accordance with aspects of the present specification, the module 134 is configured to either delay or not allow LODs, corresponding to a ‘static’ object, to LOD-out earlier than designated by the offline authored switch distances of the LODs. This prevents players using lower-end platforms (or client devices) from gaining unfair competitive advantage vis-à-vis players using higher-end platforms. For example, during a gameplay scene a player may be using a static object, such as a wall, boulder or a barrel, as cover. Now, if the scene causes a GPU usage spike then, on a higher-end platform, as a result of an application of screen resolution scaling and/or GPU performance scaling factors the LODs corresponding to the static object may switch to less complex LODs or the last LOD much faster thereby causing the static object to disappear from the scene earlier compared to the same scene being rendered on a lower-end platform. In other words, the static object LOD as cover would be visible for a longer duration in the scene being rendered on the lower-end platform compared to the duration of visibility of the static object LOD in the same scene being rendered on the higher-end platform due to runtime scaling.
Referring now to
At step 604, the rendering module 134 monitors one or more of a first plurality of variables related to a player's gameplay and/or client device 110. In some embodiments, the first plurality of variables comprise FOV change, resolution of screen (or scene), rate of vertex processing throughput, general GPU performance, and available memory (that is, memory budget for a rendered gameplay scene).
At step 606, based on monitoring of one or more of the first plurality of variables, the rendering module 134 determines if one or more of a plurality of corrective factors should be applied to the offline authored switch distances associated with each LOD of the chain of LOD assets. If yes, then at step 608, the module 134 applies one or more of the plurality of corrective factors to the offline authored switch distances associated with each LOD of the chain of LOD assets, in order to generate modulated switch distances corresponding to each of the LODs. In various embodiments, the plurality of corrective factors comprise FOV scaling factor, screen resolution scaling factor, vertex processing efficiency scaling factor, GPU performance scaling factor and memory budget-based biasing factor. Thereafter, at step 610, the rendering module 134 selects a LOD, from the chain of LOD assets, on the basis of its corresponding modulated switch distance.
However, if it is determined that none of the plurality of corrective factors need to be applied, then, at step 612, the rendering module 134 selects a LOD, from the chain of LOD assets, on the basis of its corresponding offline authored switch distance.
At step 614, the module 134 proceeds to render the selected LOD from the chain of LOD assets. Finally, prior to or during the process of rendering the selected LOD, the module 134, at step 616, applies one or more case, context or scenario based interventions, if required. In embodiments, the one or more case, context or scenario based interventions are those related to a size of the LOD being rendered, procedural vertex motion or animation for the LOD, whether the LOD represents a static object, and streaming bandwidth deterioration between the player's client device 110 and the one or more game servers 105.
Some game modes are characterized by large, highly complex topographical maps. As a non-limiting illustration, a portion of a large game map may include static content such as a city block containing a bank building with numerous amounts of detailing comprising, say, 10,000 object models, 5 million vertices and 30,000 materials. Such large, high object density, complex game maps pose challenges in terms of memory budgets and streaming bandwidth pressures. In some cases these maps cannot be efficiently rendered at a frame rate of 60FPS (for the default 1080p display) even if all distant static object models were represented by their respective simplest LODs. This is so, because rendering such high object density, complex game maps requires just too many draw calls and too many sub-pixel triangles (or geometric primitives) for a GPU to efficiently deal with without breaching the predefined maximum GPU frame latency.
To address this issue and as part of an offline process, in some embodiments, a large, high object density, complex game map is broken down or segmented into a plurality of grid cells or areas. In some embodiments, a size of each of the plurality of grid cells or areas is not uniform and varies depending upon a density or complexity of static content encompassed within each cell and based on streaming and/or memory budgets or limits. In some embodiments, by default, the game map is automatically divided into uniformly-sized (for example, 10,000×10,000 units) grid cells or areas. Designers then have the power to further divide these grid cells into smaller ones of arbitrary dimensions, for gameplay or runtime budgeting refinement.
At step 702, the module 132 receives, as input, data indicative of the contents of the grid cell. In various embodiments, the contents include a plurality of static object models represented by their respective LODs.
At step 704, the module 132 applies one or more of a plurality of functions to the data in order to auto-generate a single base model or base proxy LOD representative of the contents of the grid cell (or representative of all constituent LODs of the grid cell). The module 132 also auto-generates or assigns a base proxy switch distance corresponding to the base proxy LOD. In some embodiments, the plurality of functions includes:
At step 706, the module 132 iteratively reduces the polygon mesh complexity (that is, reduce the number of polygons) of the base proxy LOD to generate a chain or set of successive proxy LODs. The successive proxy LODs, in the chain of proxy LODs, being characterized by a descending order of complexity. In some embodiments, a proxy LOD is generated based on its geometric deviation or error approaching, but not exceeding, a predefined threshold ‘D’ compared to an immediate prior proxy LOD. Also, a proxy switch distance for each proxy LOD is calculated (by the module 132) at which the threshold ‘D’, when projected on a display of a default resolution, consumes or occupies a predefined number of pixels ‘P’. In some embodiments, the default resolution corresponds to a 1080p display. In some embodiments, the predefined number of pixels ‘P’ ranges from 2.0 to 100.0 pixels for a 1080p display.
Thus, in some embodiments, the step 706 of auto-generating the chain or set of successive proxy LODs along with their associated proxy switch distances, comprises implementing the method 200 of
Specifically, the step 706 includes the following sub-steps: a) receiving, as input into a LOD management module, the base proxy LOD and the associated base proxy switch distance; b) generating, as second output by the LOD management module, a first proxy LOD by iteratively reducing a polygon mesh complexity of the base proxy LOD until a maximum geometric deviation of the first proxy LOD approaches, but does not exceeds, a predefined threshold value with respect to the base proxy LOD. In some embodiments, the predefined threshold value ranges from 0.001 to 24.0 inches; c) generating, as third output by the LOD management module, a first proxy switch distance at which the predefined threshold value, when projected on a display of a default resolution, comprises a predefined number of pixels. In some embodiments, the predefined number of pixels ranges from 2.0 to 100.00 inches; and d) storing, by the LOD management module, the second output in association with the third output if a difference between the base proxy switch distance and the first proxy switch distance is greater than a predefined threshold distance. However, wherein if a difference between the base proxy switch distance and the first proxy switch distance is less than or equal to the predefined threshold distance, then the second output and the third output are not stored and instead provided as input to the LOD management module. In some embodiments, the predefined threshold distance ranges from 0.001 to 24.0 inches. Sub-steps a), b), c) and d) are repeated until at least one of first or second conditions is satisfied, wherein the first condition corresponds to a size of the first proxy LOD being determined to be less than or equal to a predefined size, and wherein the second condition corresponds to a predefined maximum number of the plurality of proxy LODs being generated. In some embodiments, the predefined size ranges from 4.0 to 1000.0 inches, and wherein a predefined maximum number of the plurality of proxy LODs is 6.
Consequently, each proxy LOD has an associated proxy switch distance at which the proxy LOD is rendered for the grid cell.
Finally, at step 708, the module 132 enables a computer graphic artist or designer to optimize and modulate one or more of the auto-generated proxy LODs and their associated proxy switch distances, if needed. In embodiments, the module 132 may generate one or more GUIs to prompt the graphics artist and receive inputs related to various proxy LOD related attributes or parameters described with reference to
It should be appreciated that since the proxy switch distances for each proxy LOD is known, the complexity of remeshing inputs can be reduced and visual coherence can be ensured by choosing (as inputs at step 702) only static model LODs which would be visible at corresponding proxy LODs. For example, if a proxy switch distance was 5000 units, it is ensured that all models used as inputs to the corresponding proxy LOD use their LOD representations that would be otherwise selected at 5000 units. This is advantageous since a large number (sometimes in 1000's) of small models can be omitted from proxy LOD generation as a result of them LOD'ing out of visibility prior to the proxy switch distance they would otherwise be included in.
Also, when any modification is made to the static geometry of a game level, all proxy LODs which are sourced from these bits of static geometry are automatically regenerated. This is done by establishing cell residency for all geometry in the game map. This cell residency is then used to rebuild only proxy LODs which represent those grid cells. In embodiments, every piece of static geometry has a bounding box. The x-y coordinates of the center point of the bounding box exists in exactly one world grid cell. The piece of static geometry associated with that bounding box is then assigned to that grid cell.
In accordance with some aspects of the present specification, there are multiple levels of caching in place to ensure that proxy LOD creation is as efficient as possible. When a new proxy LOD is generated, the geometry and material associated with that model are added to a distributed cache, shared across all designers and/or developers. When another designer tries to build that same proxy LOD model, the module 132 detects that all inputs match a proxy LOD model that has already been added to the distributed cache. In such a case, the cached assets are pulled directly instead of remeshing, thus saving on map iteration times.
In some embodiments, the geometry and material associated with the new proxy LOD are stored using a distributed caching system such as, for example, developed by ATVI called “Bill”. The newly generated proxy LOD data are first cached locally on standard HDD (Hard Disk Drive) storage, then distributed to standard server storage, and thereafter copied to the local caches (HDD storage) of various developers. In embodiments, this process occurs automatically in the background.
The above examples are merely illustrative of the many applications of the system of present specification. Although only a few embodiments of the present specification have been described herein, it should be understood that the present specification might be embodied in many other specific forms without departing from the spirit or scope of the specification. Therefore, the present examples and embodiments are to be considered as illustrative and not restrictive, and the specification may be modified within the scope of the appended claims.
The present application is a continuation application of U.S. patent application No. 17/037,262, titled “Methods and Systems for Selecting a Level of Detail Visual Asset in a Video Game” and filed on Sep. 29, 2020, which is incorporated by reference herein in its entirety.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 17037262 | Sep 2020 | US |
| Child | 18528239 | US |
