A method for online interaction of metaverse digital assets based on a three-dimensional engine

Abstract

The present invention discloses an online interaction method for metaverse digital assets based on a three-dimensional engine, which relates to the technical field of metaverse. The present invention constructs an interaction data set, calculates the interaction weight of assets based on interaction frequency and interaction duration, and dynamically divides assets into high-priority interaction assets, medium-priority interaction assets and low-priority interaction assets according to a set threshold. The interaction data is periodically updated during the interaction process, and the asset priority is adjusted in real time, so that high-interaction assets always maintain high-precision rendering, while low-interaction assets are only loaded when necessary, thereby improving rendering efficiency. In the process of dynamic asset loading and scheduling, the user's interaction hotspots are predicted based on the user's perspective, movement direction and interaction trajectory, a visual cone area is established, and the future field of view is calculated in combination with the user's historical movement data, and assets that are about to enter the field of view are loaded in advance. In a multi-person online environment, the interaction hotspot distribution of multiple users is integrated, and high-interaction areas are loaded first, thereby reducing resource overhead.

Description

Technical Field

[0001] The present invention relates to the field of metaverse technology, and in particular to an online interaction method for metaverse digital assets based on a three-dimensional engine. Background Art

[0002] In various application scenarios of the metaverse, online interaction of digital assets has become the core of user experience; from virtual real estate transactions to NFT art exhibitions, to multi-person collaborative social spaces, three-dimensional assets not only constitute the basic elements of the virtual world, but also carry core interactive functions.

[0003] As the user base expands and scene complexity increases, high latency becomes a problem when it comes to the dynamic loading and rendering of large numbers of 3D assets. For example, in a multi-person collaborative virtual exhibition hall, users expect all exhibits to load quickly and be presented with high-definition and delicate visual effects upon entering the scene, while ensuring smooth operation. However, in real-world applications, 3D engines need to process massive amounts of data, such as architectural models, sculptures, and interactive installations. Every time the user's perspective changes or multiple people view the same exhibit at the same time, the system needs to perform a large amount of real-time calculations, resulting in excessive use of device resources, affecting frame rates, and even causing freezes. To further complicate matters, in multi-person collaborative scenarios, all clients must synchronize the asset status in their perspectives, and traditional rendering pipelines cannot provide low-latency interactive responses under high-load environments.

[0004] To alleviate this problem, some existing solutions adopt a step-by-step loading strategy, that is, the 3D assets are gradually loaded when the user approaches a certain area, and the LOD (Level of Detail) mechanism is used to reduce the details of distant objects to reduce rendering pressure. Frustum culling is used to render only objects within the currently visible range, and pre-computed lightmaps are used to reduce real-time lighting calculations. However, these solutions still have major limitations: each digital asset submits a rendering request separately, resulting in GPU computing bottlenecks; users only need to view some assets, but the 3D engine loads all visible objects by default; the model's multi-level LOD is not intelligently optimized for the user's perspective, resulting in unnecessary high-precision rendering.

[0005] In summary, it can be seen that some of the current traditional methods have improved rendering efficiency to a certain extent. However, since most of them rely on static rule settings, it is difficult to achieve true adaptive optimization when user behavior changes dynamically. When users move quickly or multiple people interact, LOD switching may cause abrupt visual jumps, and loading delays are still obvious. Frustum culling will still cause a large amount of unnecessary resource calculations in complex scenes. Therefore, a metaverse digital asset online interaction solution based on a 3D engine is urgently needed to solve such problems. Summary of the Invention

[0006] In view of the above existing problems, the present invention is proposed.

[0007] The present invention provides a method for online interaction of digital assets in the metaverse based on a three-dimensional engine to solve the problems of traditional solutions using step-by-step loading, LOD and culling optimization, but with high rendering overhead, loading lag and low interaction synchronization efficiency.

[0008] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0009] The embodiment of the present invention provides a method for online interaction of digital assets of the metaverse based on a three-dimensional engine, which includes:

[0010] Step S1: Acquire a set of 3D assets in the metaverse scene, establish a hierarchical management mechanism for 3D assets based on interaction requirements, dynamic changes in user perspectives, and computing resource allocation, and classify the 3D assets into high-priority interaction assets, medium-priority interaction assets, and low-priority interaction assets;

[0011] Step S2: Based on the user's perspective, movement direction, and interaction trajectory, predict the user's interaction hotspots in the metaverse scene, and dynamically load and schedule 3D assets accordingly;

[0012] Step S3: In the 3D engine, the high-priority interactive assets and the medium-priority interactive assets are rendered by combining occlusion culling and intelligent batch rendering optimization.

[0013] Before rendering, assets outside the current view are eliminated based on the frustum culling method;

[0014] Adopting an adaptive LOD multi-level of detail adjustment strategy, it automatically switches rendering precision for assets at different distances, maintaining high precision for close assets and rendering low precision for distant assets.

[0015] The texture cache reuse mechanism is introduced here;

[0016] Step S4: when the user operates the high-priority interactive asset in the metaverse scene, adjust its rendering strategy;

[0017] Step S5: After the user completes the interaction or exits the metaverse scene, dynamic resource management is performed on the three-dimensional asset set.

[0018] As a preferred solution of the online interaction method of digital assets in the metaverse based on a three-dimensional engine described in the present invention, the high-priority interactive assets include directly operable virtual props and NFT assets, maintain high-precision rendering, and support real-time updates;

[0019] The medium-priority interactive assets include buildings and decorations that are about to enter the user's field of view and are preloaded with low precision;

[0020] The low-priority interactive assets include background and backdrop structures and are loaded only when necessary.

[0021] As a preferred solution of the online interaction method of metaverse digital assets based on a three-dimensional engine described in the present invention, the step of dividing the three-dimensional assets into high-priority interaction assets, medium-priority interaction assets, and low-priority interaction assets is as follows:

[0022] Construct interaction data set D:

[0023]

[0024] Where D represents the interaction data set, t i Indicates the timestamp of the i-th interaction, u i Indicates the user ID of the i-th interaction, a i represents the asset identifier involved in the i-th interaction, f i represents the type of interaction at the i-th time, p i Indicates the spatial location where the i-th interaction occurs, i represents the index number of the interaction data, and n represents the total number of interactions in the current time window.

[0025] Calculate the interaction frequency and define the calculation formula for interaction frequency as:

[0026]

[0027] Among them, F(a) represents the interaction frequency of asset a in the time window T, T represents the time window length of the interaction data statistics, δ(a i =a) is the indicator function, when a i =a, the value is 1, otherwise the value is 0.

[0028] Calculate the duration of a single interaction using the following formula:

[0029]

[0030] Where L(a) represents the average interaction duration of asset a, N represents the number of interactions of asset a within the time window T, represents the starting time of the i-th interaction, represents the end time of the i-th interaction,

[0031] Calculate the interaction weight based on the interaction frequency and interaction duration:

[0032] W(a)=αF(a)+βL(a),

[0033] Where W(a) represents the interaction weight of asset a, α and β are adjustment parameters, which are used to balance the influencing factors of interaction frequency and interaction duration respectively;

[0034] Dynamically adjust asset priorities based on interaction weights:

[0035] If W(a)>θ1, then P(a)=high priority,

[0036] If θ2≤W(a)≤θ1, then P(a)=medium priority,

[0037] If W(a)<θ2, then P(a)=low priority,

[0038] Where P(a) represents the priority of asset a, θ1 and θ2 are the interaction weight thresholds of high, medium and low priority respectively.

[0039] During the interaction process, the interaction data set D is updated periodically, and the asset priority is dynamically adjusted based on the latest data.

[0040] As a preferred solution of the online interactive method of metaverse digital assets based on a three-dimensional engine described in the present invention, the method of dynamically loading and scheduling three-dimensional assets is as follows:

[0041] Calculate the field of view that the user is about to enter, and add the medium-priority interactive assets to the preloading queue;

[0042] In a multi-person online interactive environment, the overlapping interaction areas of multiple users are calculated, and 3D assets that multiple people are interested in are rendered first.

[0043] As a preferred solution of the online interaction method of metaverse digital assets based on a three-dimensional engine described in the present invention, the steps of predicting the user's interaction hotspots in the metaverse scene based on the user's perspective, movement direction and interaction trajectory, and performing dynamic loading and scheduling of three-dimensional assets accordingly are as follows:

[0044] Define the user's current viewing direction vector V u :

[0045] V u =(x u ,y u ,z u ),

[0046] Among them, V u Represents the user's viewing direction vector, x u Indicates the component of the user's viewing direction on the x-axis, y u Indicates the component of the user's viewing direction on the y-axis, z u Indicates the component of the user's viewing direction on the z-axis.

[0047] Define the user viewing cone area:

[0048] Ω u ={p|cos(θ p )≥cos(θ c )},in,

[0049] Ω u represents the user's viewing cone area, p represents any point in the three-dimensional space, θ p Represents point p and viewing direction vector V u The angle between them, θ c Indicates the maximum viewing angle of the viewing cone,

[0050] Calculate the user's displacement per unit time using the following formula:

[0051]

[0052] Among them, M u Represents the user's moving direction vector, Represent the three-dimensional coordinates of the user at time t, Represent the three-dimensional coordinates of the user at time t+1,

[0053] Based on the user's perspective u and moving direction M u Predicting the future horizon, the prediction formula is:

[0054] in,

[0055] represents the possible viewing cone area of ​​the user at time t+1, Δt represents the prediction time step,

[0056] The probability of interactive hotspots is calculated based on historical user interaction data. The calculation formula is:

[0057]

[0058] Among them, H(p) represents the probability of point p being an interaction hotspot, K(pp i ) represents the kernel density estimation function, f i represents the type weight of the i-th interaction, p i Indicates the spatial location where the i-th interaction occurs, i represents the index number of the interaction data, and n represents the total number of interactions in the time window.

[0059] Dynamically adjust the loading order of 3D assets based on the distribution of interactive hotspots:

[0060] If H(p a )>λ1, then L(a)=priority loading,

[0061] If λ2≤H(p a )≤λ1, then L(a)=delayed loading,

[0062] If H(p a )<λ2, then L(a)=on-demand loading,

[0063] Where L(a) represents the loading scheduling strategy of asset a, p a represents the location of asset a, λ1 and λ2 are the thresholds for loading scheduling;

[0064] In a multi-person online environment, calculate the joint distribution of interaction hotspots:

[0065]

[0066] Among them, H multi (p) represents the joint probability of multi-person interaction hotspots, m represents the total number of online users, and w j represents the weight of user j, H j (p) represents the interaction hotspot probability calculated by user j, j represents the user index number,

[0067] When the probability of multi-person interaction hotspot H multi (p) When the value is higher than the set threshold, prioritize loading and rendering 3D assets that are of interest to multiple people.

[0068] As a preferred solution of the online interaction method of metaverse digital assets based on a three-dimensional engine described in the present invention, the step of rendering the high-priority interactive assets and the medium-priority interactive assets in the three-dimensional engine in combination with occlusion culling and intelligent batch rendering optimization is as follows:

[0069] Based on the frustum culling method, assets outside the current viewing angle are culled. The culling process is expressed as:

[0070] V c ={a|a∈A,p a ∈Ω u}, where V c Represents the asset collection within the viewing cone, A represents the collection of all three-dimensional assets in the scene, p a represents the location of asset a,

[0071] Ω u Represents the user's viewing cone area,

[0072] Based on the depth buffer and occlusion query technology, assets blocked by other objects are culled. The culling process is expressed as follows:

[0073] V o ={a|a∈V c ,D a >Dthreshold},

[0074] Among them, V o represents the set of visible assets, D a represents the depth value of asset a relative to the user, and D threshold is the depth threshold for occlusion culling;

[0075] An adaptive LOD adjustment strategy is adopted to dynamically adjust the asset rendering precision according to the user's distance:

[0076] If d a < d1, then R(a) = high precision,

[0077] If d1 ≤ d a ≤ d2, then R(a) = high precision,

[0078] If d a > d2, then R(a) = high precision,

[0079] Among them, R(a) represents the rendering precision of asset a, and d a represents the distance from asset a to the user, and d1 and d2 are the distance thresholds for high, medium, and low precision rendering respectively;

[0080] Based on the caching strategy, the rendering efficiency is improved, and the reuse process is expressed as:

[0081] T c = {t a | a ∈ V o , t a ∈ C cache},

[0082] Among them, T c represents the set of reusable textures, t a represents the texture of asset a, and C cache represents the set of textures in the current cache.

[0083] As a preferred solution of the online interaction method for metaverse digital assets based on a 3D engine described in this invention, among them: the rendering strategy adjustment method in step S4 includes:

[0084] When the user performs a pick-up, rotation or editing operation, the rendering priority of the asset is increased;

[0085] Incremental rendering is adopted, and only the changed area is rendered and updated;

[0086] In a multi-person collaboration scenario, independent interaction areas are set for different users.

[0087] As a preferred solution of the online interactive method of metaverse digital assets based on a three-dimensional engine described in the present invention, the step of adjusting the rendering strategy during the operation is as follows:

[0088] When the user performs an interactive operation, the rendering priority of the interactive asset is adjusted using the following formula:

[0089] P′(a)=P(a)+γf a ,

[0090] Among them, P'(a) represents the new rendering priority after interaction, P(a) represents the rendering priority before interaction, γ is the interaction impact factor, and f a Represents the interaction type weighted value of asset a,

[0091] When an asset changes, only the changed areas are updated. The update formula is:

[0092]

[0093] Among them, R' represents the set of areas that need to be re-rendered, p represents a point in three-dimensional space, and A mod Represents the set of assets that have changed, A static Represents a collection of assets that have not changed.

[0094] As a preferred solution of the online interactive method of metaverse digital assets based on a three-dimensional engine described in the present invention, the dynamic resource management method is:

[0095] For interactive assets, lower their rendering priority;

[0096] For assets that have not been accessed for a long time, a progressive unloading strategy is implemented to release video memory and retain low-precision placeholder data.

[0097] In multi-person collaboration scenarios, automatic degradation is performed on areas where no one is interacting.

[0098] As a preferred solution of the online interactive method of metaverse digital assets based on a three-dimensional engine described in the present invention, the steps of dynamic resource management are:

[0099] Lower the rendering priority for interactive assets:

[0100] P′(a)=P(a)-δ,

[0101] Among them, P'(a) represents the asset rendering priority after adjustment, P(a) represents the asset rendering priority before adjustment, and δ represents the priority attenuation factor.

[0102] For assets that have not been accessed for a long time, execute the uninstallation strategy:

[0103] If T a >T max , then U(a)=complete unloading,

[0104] If T min ≤T a ≤T max , then U(a)=partial unloading,

[0105] If T a <T min , then U(a)=Retain T a >T max ,

[0106] Among them, U(a) represents the unloading strategy of asset a, T a Indicates the duration of time that asset a has not been accessed, T max Indicates the time threshold for complete uninstallation, T min Indicates the time threshold for partial unloading;

[0107] In a multi-person collaborative environment, reduce the resource usage of the unmanned interaction area. The resource adjustment formula is:

[0108] P′(A idle )=P(A idle )-∈,

[0109] Among them, P'(A idle ) represents the adjusted asset priority in the unmanned interaction area, P(A idle ) represents the asset priority of the unmanned interaction area before adjustment, A idle represents the set of asset areas without human interaction, and ∈ is the priority reduction factor.

[0110] The beneficial effects of the present invention are as follows: the present invention constructs an interaction data set, calculates the interaction weight of assets based on interaction frequency and interaction duration, dynamically divides assets into high-priority interaction assets, medium-priority interaction assets and low-priority interaction assets according to set thresholds, and periodically updates interaction data during the interaction process, adjusts asset priorities in real time, so that high-interaction assets always maintain high-precision rendering, and low-interaction assets are only loaded when necessary, thereby optimizing computing resource allocation and improving rendering efficiency.

[0111] During the dynamic asset loading and scheduling process, the present invention predicts the user's interaction hotspots based on the user's perspective, movement direction, and interaction trajectory, establishes a visual cone area, and calculates the future field of view in combination with the user's historical movement data. Assets that are about to enter the field of view are loaded in advance, and the probability of interaction hotspots is calculated through kernel density estimation. In a multi-person online environment, the interaction hotspot distribution of multiple users is integrated, and high-interaction areas are loaded first, reducing unnecessary resource overhead.

[0112] This invention adopts occlusion culling combined with intelligent batch rendering to remove assets outside the field of view based on the view frustum culling method before rendering, and reduces the computational complexity of occluded assets through depth buffer occlusion culling. At the same time, it introduces adaptive LOD adjustment to dynamically adjust the rendering accuracy according to the distance between the asset and the user, so that close assets maintain high precision and distant assets are rendered with low precision. In addition, the texture cache reuse mechanism is used to avoid repeated rendering of the same resources, further improving rendering efficiency. During the interaction process, a dynamic adjustment of the asset's rendering strategy is adopted. When the user performs operations such as picking, rotating, and editing, the rendering priority of the asset is temporarily increased, and an incremental rendering mechanism is used to update only the changed area instead of redrawing the entire area, reducing the computational complexity and improving the response speed.

[0113] In a multi-person collaborative environment, the present invention sets independent interaction areas for different users to avoid image quality degradation caused by resource competition. After the user completes the interaction or exits the scene, a dynamic resource management strategy is implemented to reduce the rendering priority of the assets where the interaction is completed. For assets that have not been accessed for a long time, progressive unloading is performed, first reducing the accuracy and then completely removing them, balancing resource usage and rendering efficiency. In addition, automatic degradation is implemented in areas where no one is interacting, reducing unnecessary rendering calculations and optimizing video memory management.

[0114] In summary, the method of the present invention effectively alleviates the problems existing in traditional solutions, such as GPU computing bottlenecks, redundant resource loading, inefficient LOD management, multi-person interaction synchronization delays, and uneven resource allocation. It improves the rendering efficiency and interactive response speed of three-dimensional assets, and enables the metaverse scene to maintain a smooth and stable visual experience even in highly complex and multi-user interactive situations. BRIEF DESCRIPTION OF THE DRAWINGS

[0115] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0116] Figure 1 Schematic diagram of the process of the online interaction method of metaverse digital assets based on a three-dimensional engine of the present invention. DETAILED DESCRIPTION

[0117] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0118] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0119] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0120] Example 1, with reference to Figure 1 This embodiment provides a method for online interaction of digital assets in the metaverse based on a three-dimensional engine, including:

[0121] Step S1: Acquire a set of 3D assets in the metaverse scene, establish a hierarchical management mechanism for 3D assets based on interaction requirements, dynamic changes in user perspectives, and computing resource allocation, and classify the 3D assets into high-priority interaction assets, medium-priority interaction assets, and low-priority interaction assets;

[0122] The high-priority interactive assets include directly operable virtual props and NFT assets, maintaining high-precision rendering and supporting real-time updates;

[0123] The medium-priority interactive assets include buildings and decorations that are about to enter the user's field of view and are preloaded with low precision;

[0124] The low-priority interactive assets include background and background structures, which are loaded only when necessary;

[0125] The step of dividing the three-dimensional assets into high-priority interactive assets, medium-priority interactive assets and low-priority interactive assets is:

[0126] Construct interaction data set D:

[0127]

[0128] Where D represents the interaction data set, t i Indicates the timestamp of the i-th interaction, u i Indicates the user ID of the i-th interaction, a i represents the asset identifier involved in the i-th interaction, f i represents the type of interaction at the i-th time, p i Indicates the spatial location where the i-th interaction occurs, i represents the index number of the interaction data, and n represents the total number of interactions in the current time window.

[0129] Calculate the interaction frequency and define the calculation formula for interaction frequency as:

[0130]

[0131] Among them, F(a) represents the interaction frequency of asset a in the time window T, T represents the time window length of the interaction data statistics, δ(a i =a) is the indicator function, when a i =a, the value is 1, otherwise the value is 0.

[0132] Calculate the duration of a single interaction using the following formula:

[0133]

[0134] Where L(a) represents the average interaction duration of asset a, N represents the number of interactions of asset a within the time window T, represents the starting time of the i-th interaction, represents the end time of the i-th interaction,

[0135] Calculate the interaction weight based on the interaction frequency and interaction duration:

[0136] W(a)=αF(a)+βL(a),

[0137] Where W(a) represents the interaction weight of asset a, α and β are adjustment parameters, which are used to balance the influencing factors of interaction frequency and interaction duration respectively;

[0138] Dynamically adjust asset priorities based on interaction weights:

[0139] If W(a)>θ1, then P(a)=high priority,

[0140] If θ2≤W(a)≤θ1, then P(a)=medium priority,

[0141] If W(a)<θ2, then P(a)=low priority,

[0142] Where P(a) represents the priority of asset a, θ1 and θ2 are the interaction weight thresholds of high, medium and low priority respectively.

[0143] During the interaction process, the interaction data set D is updated periodically, and the asset priority is dynamically adjusted based on the latest data;

[0144] Specifically, it dynamically adjusts the priority of 3D assets through user interaction data, optimizes computing resource allocation, and improves rendering efficiency;

[0145] Here, we construct a data set containing interaction time, user ID, asset ID, interaction type, and interaction location. Based on this data, we calculate interaction frequency and duration to quantify the user's attention to different assets. We use a weight calculation formula to integrate interaction frequency and duration, and set priority thresholds to dynamically classify assets, effectively balancing rendering quality and performance consumption.

[0146] Step S2: Based on the user's perspective, movement direction, and interaction trajectory, predict the user's interaction hotspots in the metaverse scene, and dynamically load and schedule 3D assets accordingly;

[0147] The method for dynamic loading and scheduling of three-dimensional assets is as follows:

[0148] Calculate the field of view that the user is about to enter, and add the medium-priority interactive assets to the preloading queue;

[0149] In a multi-person online interactive environment, the system calculates the overlapping interaction areas of multiple users and prioritizes rendering of 3D assets that multiple people are interested in.

[0150] The steps of predicting the user's interaction hotspots in the metaverse scene based on the user's perspective, movement direction and interaction trajectory, and dynamically loading and scheduling three-dimensional assets accordingly are as follows:

[0151] Define the user's current viewing direction vector V u :

[0152] V u =(x u ,y u ,z u ),

[0153] Among them, V u Represents the user's viewing direction vector, x u Indicates the component of the user's viewing direction on the x-axis, y u Indicates the component of the user's viewing direction on the y-axis, z u Indicates the component of the user's viewing direction on the z-axis.

[0154] Define the user viewing cone area:

[0155] Ω u ={p|cos(θ p )≥cos(θ c )},in,

[0156] Ω u represents the user's viewing cone area, p represents any point in the three-dimensional space, θ p Represents point p and viewing direction vector V u The angle between them, θ c Indicates the maximum viewing angle of the viewing cone,

[0157] Calculate the user's displacement per unit time using the following formula:

[0158]

[0159] Among them, M u Represents the user's moving direction vector, Represent the three-dimensional coordinates of the user at time t, Represent the three-dimensional coordinates of the user at time t+1,

[0160] Based on the user's perspective u and moving direction M u Predicting the future horizon, the prediction formula is:

[0161] in,

[0162] represents the possible viewing cone area of ​​the user at time t+1, Δt represents the prediction time step,

[0163] The probability of interactive hotspots is calculated based on historical user interaction data. The calculation formula is:

[0164]

[0165] Among them, H(p) represents the probability of point p being an interaction hotspot, K(pp i ) represents the kernel density estimation function, f i represents the type weight of the i-th interaction, p i Indicates the spatial location where the i-th interaction occurs, i represents the index number of the interaction data, and n represents the total number of interactions in the time window.

[0166] Dynamically adjust the loading order of 3D assets based on the distribution of interactive hotspots:

[0167] If H(p a )>λ1, then L(a)=priority loading,

[0168] If λ2≤H(p a )≤λ1, then L(a)=delayed loading,

[0169] If H(p a )<λ2, then L(a)=on-demand loading,

[0170] Where L(a) represents the loading scheduling strategy of asset a, p a represents the location of asset a, λ1 and λ2 are the thresholds for loading scheduling;

[0171] In a multi-person online environment, calculate the joint distribution of interaction hotspots:

[0172]

[0173] Among them, H multi (p) represents the joint probability of multi-person interaction hotspots, m represents the total number of online users, and w j represents the weight of user j, H j (p) represents the interaction hotspot probability calculated by user j, j represents the user index number,

[0174] When the probability of multi-person interaction hotspot H multi (p) When the threshold is higher than the set threshold, prioritize loading and rendering 3D assets that are of interest to multiple people;

[0175] Specifically, we predict interactive hotspots based on the user's perspective, movement direction, and interaction history, and dynamically load and optimize 3D assets:

[0176] Define the user's viewing direction vector and establish a viewing cone to determine the spatial range of the user's current focus. Calculate the future viewing cone based on the user's historical movement data to pre-load assets that will soon enter the field of view. Use a kernel density estimation algorithm to identify high-interaction hotspots. In a multiplayer online environment, comprehensively calculate the hotspot distribution of multiple users to increase the rendering priority of key areas.

[0177] Step S3: In the 3D engine, the high-priority interactive assets and the medium-priority interactive assets are rendered by combining occlusion culling and intelligent batch rendering optimization.

[0178] Before rendering, assets outside the current view are eliminated based on the frustum culling method;

[0179] Adopting an adaptive LOD multi-level of detail adjustment strategy, it automatically switches rendering precision for assets at different distances, maintaining high precision for close assets and rendering low precision for distant assets.

[0180] The texture cache reuse mechanism is introduced here;

[0181] The steps of rendering the high-priority interactive assets and the medium-priority interactive assets in the 3D engine by combining occlusion culling and intelligent batch rendering optimization are as follows:

[0182] Based on the frustum culling method, assets outside the current viewing angle are culled. The culling process is expressed as:

[0183] V c ={a|a∈A,p a ∈Ω u}, where V c Represents the asset collection within the viewing cone, A represents the collection of all three-dimensional assets in the scene, p aIndicates the location of asset a,

[0184] Ω u Indicates the frustum region of the user,

[0185] Based on depth buffering and occlusion query techniques, assets occluded by other objects are removed. The removal process is expressed as:

[0186] V o ={a|a∈V c ,D a >D threshold},

[0187] where V o represents the set of visible assets, D a represents the depth value of asset a relative to the user, and D threshold is the depth threshold for occlusion removal;

[0188] An adaptive LOD adjustment strategy is adopted to dynamically adjust the asset rendering precision according to the user's distance:

[0189] If d a <d1, then R(a)=high precision,

[0190] If d1≤d a ≤d2, then R(a)=medium precision,

[0191] If d a >d2, then R(a)=low precision,

[0192] where R(a) represents the rendering precision of asset a, d a represents the distance from asset a to the user, and d1 and d2 are the distance thresholds for high, medium, and low precision rendering respectively;

[0193] Based on the caching strategy, the rendering efficiency is improved. The reuse process is expressed as:

[0194] T c ={t a |a∈V o ,t a ∈C cache [[ID=,67]]},

[0195] where T c represents the set of reusable textures, t a represents the texture of asset a, and C cache represents the set of textures in the current cache;

[0196] Specifically, frustum culling technology removes assets outside the user's field of view to avoid useless calculations; a depth-buffered occlusion culling method reduces the need to render assets obscured by other objects; and an adaptive LOD multi-level of detail strategy adjusts rendering accuracy based on the distance between the user and the asset, rendering close assets with high precision and distant assets with low precision to reduce the computational burden.

[0197] Step S4: when the user operates the high-priority interactive asset in the metaverse scene, adjust its rendering strategy;

[0198] The rendering strategy adjustment method in step S4 includes:

[0199] When the user performs a pick, rotate, or edit operation, the rendering priority of the asset is increased;

[0200] Use incremental rendering to only update the rendering of the changed areas;

[0201] In multi-person collaboration scenarios, set up independent interaction areas for different users;

[0202] When performing the operation, the steps for adjusting the rendering strategy are:

[0203] When the user performs an interactive operation, the rendering priority of the interactive asset is adjusted using the following formula:

[0204] P′(a)=P(a)+γf a ,

[0205] Among them, P'(a) represents the new rendering priority after interaction, P(a) represents the rendering priority before interaction, γ is the interaction impact factor, and f a Represents the interaction type weighted value of asset a,

[0206] When an asset changes, only the changed areas are updated. The update formula is:

[0207]

[0208] Among them, R' represents the set of areas that need to be re-rendered, p represents a point in three-dimensional space, and A mod Represents the set of assets that have changed, A static Represents a collection of assets that have not changed;

[0209] Specifically, when a user picks, rotates, or edits an asset, the rendering priority of the asset is dynamically increased, allowing it to receive higher rendering resource allocation. At the same time, an incremental rendering strategy is adopted, updating only the changed areas rather than a global redraw, thereby reducing computational effort and improving response speed.

[0210] Furthermore, in multiplayer online environments, independent interaction areas are set for each user to avoid image quality degradation caused by resource competition. This effectively improves the visual quality of highly interactive assets while reducing global rendering pressure, achieving a balance between performance and user experience.

[0211] Step S5: After the user completes the interaction or exits the metaverse scene, dynamic resource management is performed on the three-dimensional asset set;

[0212] The dynamic resource management method is:

[0213] For interactive assets, lower their rendering priority;

[0214] For assets that have not been accessed for a long time, a progressive unloading strategy is implemented to release video memory and retain low-precision placeholder data.

[0215] In multi-person collaboration scenarios, automatic degradation is performed on unmanned areas;

[0216] The steps of dynamic resource management are:

[0217] Lower the rendering priority for interactive assets:

[0218] P′(a)=P(a)-δ,

[0219] Among them, P'(a) represents the asset rendering priority after adjustment, P(a) represents the asset rendering priority before adjustment, and δ represents the priority attenuation factor.

[0220] For assets that have not been accessed for a long time, execute the uninstallation strategy:

[0221] If T a >T max , then U(a)=complete unloading,

[0222] If T min ≤T a ≤T max , then U(a)=partial unloading,

[0223] If T a <T min , then U(a)=Retain T a >T max ,

[0224] Among them, U(a) represents the unloading strategy of asset a, T a Indicates the duration of time that asset a has not been accessed, T max Indicates the time threshold for complete uninstallation, T min Indicates the time threshold for partial unloading;

[0225] In a multi-person collaborative environment, reduce the resource usage of the unmanned interaction area. The resource adjustment formula is:

[0226] P′(A idle )=P(A idle )-∈,

[0227] Among them, P'(A idle ) represents the adjusted asset priority in the unmanned interaction area, P(A idle ) represents the asset priority of the unmanned interaction area before adjustment, A idle represents the set of asset areas with no human interaction, ∈ is the priority reduction factor;

[0228] Specifically, after the user completes the interaction or exits the scene, the rendering priority of the assets that have completed the interaction is gradually lowered, thereby freeing up computing resources; for assets that have not been accessed for a long time, a progressive unloading strategy is adopted, first reducing their accuracy and then completely removing them, so that resource management can transition smoothly; in addition, in a multiplayer online environment, automatic degradation processing is implemented for areas without human interaction to reduce unnecessary rendering calculations, effectively reduce video memory usage, and optimize system response speed.

[0229] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for online interaction of digital assets in the Metaverse based on a three-dimensional engine, characterized by: include, Step S1: Acquire a set of 3D assets in the metaverse scene, establish a hierarchical management mechanism for 3D assets based on interaction requirements, dynamic changes in user perspectives, and computing resource allocation, and classify the 3D assets into high-priority interaction assets, medium-priority interaction assets, and low-priority interaction assets; Step S2: Based on the user's perspective, movement direction, and interaction trajectory, predict the user's interaction hotspots in the metaverse scene, and dynamically load and schedule 3D assets accordingly; Step S3: In the 3D engine, the high-priority interactive assets and the medium-priority interactive assets are rendered by combining occlusion culling and intelligent batch rendering optimization. Before rendering, assets outside the current view are eliminated based on the frustum culling method; Adopting an adaptive LOD multi-level of detail adjustment strategy, it automatically switches rendering precision for assets at different distances, maintaining high precision for close assets and rendering low precision for distant assets. The texture cache reuse mechanism is introduced here; Step S4: when the user operates the high-priority interactive asset in the metaverse scene, adjust its rendering strategy; Step S5: After the user completes the interaction or exits the metaverse scene, dynamic resource management is performed on the three-dimensional asset set; The high-priority interactive assets include directly operable virtual props and NFT assets, maintaining high-precision rendering and supporting real-time updates; The medium-priority interactive assets include buildings and decorations that are about to enter the user's field of view and are preloaded with low precision; The low-priority interactive assets include background and background structures, which are loaded only when necessary; The step of dividing the three-dimensional assets into high-priority interactive assets, medium-priority interactive assets and low-priority interactive assets is: Construct interaction data set D: Where D represents the interaction data set, t i Indicates the timestamp of the i-th interaction, u i Indicates the user ID of the i-th interaction, a i represents the asset identifier involved in the i-th interaction, f i represents the type of interaction at the i-th time, p i Indicates the spatial location where the i-th interaction occurs, i represents the index number of the interaction data, and n represents the total number of interactions in the current time window. Calculate the interaction frequency and define the calculation formula for interaction frequency as: Among them, F(a) represents the interaction frequency of asset a in the time window T, T represents the time window length of the interaction data statistics, δ(a i =a) is the indicator function, when a i =a, the value is 1, otherwise the value is 0. Calculate the duration of a single interaction using the following formula: Where L(a) represents the average interaction duration of asset a, N represents the number of interactions of asset a within the time window T, represents the starting time of the i-th interaction, represents the end time of the i-th interaction, Calculate the interaction weight based on the interaction frequency and interaction duration: W(a)=αF(a)+βL(a), Where W(a) represents the interaction weight of asset a, α and β are adjustment parameters, which are used to balance the influencing factors of interaction frequency and interaction duration respectively; Dynamically adjust asset priorities based on interaction weights: If W(a)>θ1, then P(a)=high priority, If θ2≤W(a)≤θ1, then P(a)=medium priority, If W(a)<θ2, then P(a)=low priority, Where P(a) represents the priority of asset a, θ1 and θ2 are the interaction weight thresholds of high, medium and low priority respectively. During the interaction process, the interaction data set D is updated periodically, and the asset priority is dynamically adjusted based on the latest data.

2. The method for online interaction of digital assets in the Metaverse based on a three-dimensional engine according to claim 1, characterized in that: The method for dynamic loading and scheduling of three-dimensional assets is as follows: Calculate the field of view that the user is about to enter, and add the medium-priority interactive assets to the preloading queue; In a multi-person online interactive environment, the overlapping interaction areas of multiple users are calculated, and 3D assets that multiple people are interested in are rendered first.

3. The method for online interaction of digital assets in the Metaverse based on a three-dimensional engine according to claim 2, characterized in that: The steps of predicting the user's interaction hotspots in the metaverse scene based on the user's perspective, movement direction and interaction trajectory, and dynamically loading and scheduling three-dimensional assets accordingly are as follows: Define the user's current viewing direction vector V u : V u =(x u ,y u ,z u ), Among them, V u Represents the user's viewing direction vector, x u Indicates the component of the user's viewing direction on the x-axis, y u Indicates the component of the user's viewing direction on the y-axis, z u Indicates the component of the user's viewing direction on the z-axis. Define the user viewing cone area: Oh u ={p|cos(θ p )≥cos(θ c )}, among them, Ω u represents the user's viewing cone area, p represents any point in the three-dimensional space, θ p Represents point p and viewing direction vector V u The angle between them, θ c Indicates the maximum viewing angle of the viewing cone, Calculate the user's displacement per unit time using the following formula: Among them, M u Represents the user's moving direction vector, Represent the three-dimensional coordinates of the user at time t, Represent the three-dimensional coordinates of the user at time t+1, Based on the user's perspective u and moving direction M u Predicting the future horizon, the prediction formula is: in, represents the possible viewing cone area of ​​the user at time t+1, Δt represents the prediction time step, The probability of interactive hotspots is calculated based on historical user interaction data. The calculation formula is: Among them, H(p) represents the probability of point p being an interaction hotspot, K(pp i ) represents the kernel density estimation function, f i represents the type weight of the i-th interaction, p i Indicates the spatial location where the i-th interaction occurs, i represents the index number of the interaction data, and n represents the total number of interactions in the time window. Dynamically adjust the loading order of 3D assets based on the distribution of interactive hotspots: If H(p a )>λ1, then L(a)=priority loading, If λ2≤H(p a )≤λ1, then L(a)=delayed loading, If H(p a )<λ2, then L(a)=on-demand loading, Where L(a) represents the loading scheduling strategy of asset a, p a represents the location of asset a, λ1 and λ2 are the thresholds for loading scheduling; In a multi-person online environment, calculate the joint distribution of interaction hotspots: Among them, H multi (p) represents the joint probability of multi-person interaction hotspots, m represents the total number of online users, and w j represents the weight of user j, H j (p) represents the interaction hotspot probability calculated by user j, j represents the user index number, When the probability of multi-person interaction hotspot H multi (p) When the value is higher than the set threshold, prioritize loading and rendering 3D assets that are of interest to multiple people.

4. The method for online interaction of digital assets in the Metaverse based on a three-dimensional engine according to claim 3, characterized in that: The steps of rendering the high-priority interactive assets and the medium-priority interactive assets in the 3D engine by combining occlusion culling and intelligent batch rendering optimization are as follows: Based on the frustum culling method, assets outside the current viewing angle are culled. The culling process is expressed as: V c ={a|a∈A,p a ∈Ω u }, where V c Represents the asset collection within the viewing cone, A represents the collection of all three-dimensional assets in the scene, p a represents the location of asset a, Ω u Represents the user's viewing cone area, Based on the depth buffer and occlusion query technology, assets blocked by other objects are culled. The culling process is expressed as follows: V o ={a|a∈V c ,D a >D threshold }, Among them, V o Represents the visible asset set, D a Indicates the depth value of asset a relative to the user, D threshold is the depth threshold for occlusion culling; Adopting an adaptive LOD adjustment strategy, the asset rendering accuracy is dynamically adjusted according to the user distance: If d a <d1, then R(a) = high precision, If d1≤d a ≤d2, then R(a)=high precision, If d a >d2, then R(a)=high precision, Among them, R(a) represents the rendering accuracy of asset a, d a Indicates the distance from asset a to the user, d1 and d2 are the distance thresholds for high, medium, and low precision rendering respectively; Based on the cache strategy, rendering efficiency is improved, and the reuse process is expressed as: T c ={t a |a∈V o ,t a ∈C cache }, Among them, T c Represents a reusable texture set, t a Represents the texture of asset a, C cache Represents the collection of textures currently in the cache.

5. The method for online interaction of digital assets in the metaverse based on a three-dimensional engine according to claim 4, characterized in that: The rendering strategy adjustment method in step S4 includes: When the user performs a pick, rotate, or edit operation, the rendering priority of the asset is increased; Use incremental rendering to only update the rendering of the changed areas; In multi-person collaboration scenarios, set up independent interaction areas for different users.

6. The method for online interaction of digital assets in the metaverse based on a three-dimensional engine according to claim 5, characterized in that: When performing the operation, the steps for adjusting the rendering strategy are: When the user performs an interactive operation, the rendering priority of the interactive asset is adjusted using the following formula: P′(a)P(a)+γf a , Among them, P'(a) represents the new rendering priority after the interaction, P(a) represents the rendering priority before the interaction, γ is the interaction impact factor, and f a Represents the interaction type weighted value of asset a, When an asset changes, only the changed areas are updated. The update formula is: Among them, R' represents the set of areas that need to be re-rendered, p represents a point in three-dimensional space, and A mod Represents the set of assets that have changed, A static Represents a collection of assets that have not changed.

7. The method for online interaction of digital assets in the metaverse based on a three-dimensional engine according to claim 6, characterized in that: The dynamic resource management method is: For interactive assets, lower their rendering priority; For assets that have not been accessed for a long time, a progressive unloading strategy is implemented to release video memory and retain low-precision placeholder data. In multi-person collaboration scenarios, automatic degradation is performed on areas where no one is interacting.

8. The method for online interaction of digital assets in the metaverse based on a three-dimensional engine according to claim 7, characterized in that: The steps of dynamic resource management are: Lower the rendering priority for interactive assets: P′(a)=P(a)-δ, Among them, P'(a) represents the asset rendering priority after adjustment, P(a) represents the asset rendering priority before adjustment, and δ represents the priority attenuation factor. For assets that have not been accessed for a long time, execute the uninstallation strategy: If T a >T max , then U(a)=complete unloading, If T min ≤T a ≤T max , then U(a)=partial unloading, If T a <T min , then U(a)=Retain T a >T max , Among them, U(a) represents the unloading strategy of asset a, T a Indicates the duration of time that asset a has not been accessed, T max Indicates the time threshold for complete uninstallation, T min Indicates the time threshold for partial unloading; In a multi-person collaborative environment, reduce the resource usage of the unmanned interaction area. The resource adjustment formula is: P′(A idle )=P(A idle )-∈, Among them, P'(A idle ) represents the adjusted asset priority in the unmanned interaction area, P(A idle ) represents the asset priority of the unmanned interaction area before adjustment, A idle represents the set of asset areas without human interaction, and ∈ is the priority reduction factor.

Images