Particle rendering method, system for simulating deep-sea sea snow settling process

Abstract

The application relates to a particle rendering method and system for simulating deep-sea sea snow sedimentation. The method comprises the following steps: dividing a three-dimensional scene space into a plurality of uniform grid units, constructing a sea snow particle model containing multi-attribute characteristics, synchronously driving a particle motion process and a biochemical process simulation containing aggregation, decomposition and feeding mechanism based on three-dimensional marine environment field data, updating and solving the particle state by using parallel computing, and performing three-dimensional scene rendering and simulating optical special effects in a deep-sea environment based on the updated particle state. The method can efficiently and accurately simulate the deep-sea sea snow sedimentation process, and significantly improves the accuracy and authenticity of sea snow sedimentation visualization.

Description

Technical Field

[0001] This application relates to the field of deep-sea environment simulation technology, and in particular to a particle rendering method and system for simulating the deep-sea snow deposition process. Background Technology

[0002] Deep-sea snow is a type of suspended particle mainly composed of organic suspended matter such as planktonic remains and excrement. As a key carrier of the "biological pump" of the marine carbon cycle, the simulation and visualization of its sedimentation process is of great significance for understanding the potential of marine carbon sinks, predicting climate change, and exploring the evolution of deep-sea life.

[0003] Current visualization techniques for marine snow particles suffer from the following problems: First, mainstream marine science visualization software primarily focuses on presenting static or quasi-static field data (such as the planar distribution of temperature, salinity, and chlorophyll concentration), typically failing to simulate key ecological processes like aggregation, decomposition, and feeding among marine snow particles. This results in overly simplified views with low scientific fidelity. Second, particle systems in computer graphics are usually just points with basic physical properties, lacking a range of biochemical attributes. This makes them unsuitable for calculating marine snow ecological processes and for coupling with marine environmental data, leading to simulation results that are disconnected from the real marine physical environment. Furthermore, when the particle scale reaches billions or more, current technologies struggle to achieve real-time interactive simulations while maintaining scientific accuracy due to limitations in computational architecture.

[0004] In summary, current simulation methods have technical bottlenecks in simultaneously achieving high-fidelity biochemical mechanism simulation and large-scale real-time rendering of deep-sea snow deposition processes, resulting in insufficient accuracy and real-time performance of visualization results. Summary of the Invention

[0005] Therefore, it is necessary to provide a particle rendering method and system that can accurately and efficiently simulate the deep-sea snow deposition process, addressing the aforementioned technical problems.

[0006] In a first aspect, this application provides a particle rendering method for simulating the settling process of deep-sea snow, comprising:

[0007] The three-dimensional scene space is divided into several uniform grid units, and several marine snow particle models are initialized in the three-dimensional scene space. The marine snow particle models include physical properties, biochemical properties and ecological properties.

[0008] Based on the input three-dimensional marine environmental field data, parallel computing is used to drive the marine snow particle model in each grid cell to simulate the motion process and biochemical process, and then update and solve the problem. The biochemical process simulation includes aggregation mechanism, decomposition mechanism and feeding mechanism.

[0009] Based on the updated state of the sea snow particle model, rendering is performed in a 3D scene, while simulating optical effects in a deep-sea environment.

[0010] In one embodiment, the process of driving the sea snow particle model to simulate motion based on the input three-dimensional marine environmental field data includes:

[0011] Determine the grid cell where the current position of the snow particle model is located, and obtain the flow velocity vector of each vertex of the grid cell;

[0012] Calculate the local coordinates of the sea snow particle model within the grid cells;

[0013] Based on the local coordinates, trilinear interpolation is performed on the velocity vectors of each vertex to obtain the velocity vector of the current position of the snow particle.

[0014] The position of the sea snow particle model is updated using kinematic equations based on the flow velocity vector.

[0015] In one embodiment, the physical, biochemical, and ecological properties of the marine snow particle model include:

[0016] The physical properties include position, velocity, mass, volume, density, and formation time;

[0017] The biochemical properties include organic carbon content, viscosity coefficient, and particle type identifier;

[0018] The ecological attributes include parameters used to control predation priority.

[0019] In one embodiment, the aggregation mechanism includes:

[0020] Determine the target grid cell where the current marine snow particle model resides;

[0021] Search for other sea snow particle models within the local search area formed by the target grid cell and its spatially adjacent grid cells;

[0022] When other sea snow particle models exist within the local search area, and the distance between the two sea snow particle models is less than a preset aggregation threshold, the aggregation probability is calculated based on the viscosity coefficient of the two sea snow particle models.

[0023] If the aggregation probability is greater than a preset threshold, the two marine snow particle models are merged into a new particle, and the physical and biochemical properties of the new particle are calculated and updated based on the physical and biochemical properties of the two marine snow particle models.

[0024] In one embodiment, the decomposition mechanism includes:

[0025] For the marine snow particle model within each grid cell, the particle mass decays over time according to the formation time of the marine snow particle model and the preset decomposition rate.

[0026] When the particle mass is lower than the decomposition threshold, the splitting probability is calculated according to a preset splitting probability model;

[0027] If the splitting probability meets the preset condition, the snow particle model is split into multiple sub-particle models, and the attributes of each sub-particle model are inherited from the original particle.

[0028] If the splitting probability does not meet the preset condition, the sea snow particle model is removed to simulate the degradation process.

[0029] In one embodiment, the feeding mechanism includes:

[0030] One or more specific cell grids in the three-dimensional scene space are defined as virtual predator regions;

[0031] When the sea snow particle model enters the grid cell corresponding to the virtual predator area through the movement process, the sea snow particle model is removed according to a preset probability to simulate the feeding process.

[0032] In one embodiment, the step of using parallel computing to update the state of the sea snow particle model includes:

[0033] The three-dimensional scene space is divided into multiple subdomains, and each subdomain corresponds to a computing node;

[0034] The snow particle models located in different subdomains are distributed and stored on the corresponding computing nodes;

[0035] Each computing node computes the state of the snow particle model in the corresponding subdomain in parallel.

[0036] Data exchange of the sea snow particle model located at the subdomain boundary is performed between adjacent computing nodes through the message passing interface.

[0037] In one embodiment, the optical effects simulating a deep-sea environment include:

[0038] A virtual camera and a simulated searchlight source are set in the three-dimensional scene space;

[0039] Based on the position of the virtual camera and the direction of the searchlight source, the light scattering intensity of the sea snow particle model is calculated according to the Tyndall effect model:

[0040] ;

[0041] in, Let be the angle between the position of the virtual camera and the direction of the searchlight source, d be the distance from the snow particle model to the virtual camera, and r be the radius of the snow particle. The scattering coefficient is related to the particle material and refractive index;

[0042] In one embodiment, fusing the light scattering intensity with the base color of the snow particle model to obtain the final pixel color and outputting the rendering result includes:

[0043] The rendered fragments of all sea snow particle models in the three-dimensional scene space are sorted by depth;

[0044] Based on the sorting results, and following the order from farthest to nearest, the color of the snow particle model after light scattering intensity fusion is mixed with the target color using the Alpha blending formula to output the final pixel color. The Alpha blending formula is as follows:

[0045] ;

[0046] in, The source color is calculated using the Yndall effect from the aforementioned sea snow particle model. For transparency parameters, For the target color that already exists on the screen, To output color.

[0047] Secondly, this application also provides a particle rendering system for simulating the settling process of deep-sea snow, comprising:

[0048] The model building module is used to divide the three-dimensional scene space into several uniform grid units and initialize several marine snow particle models in the three-dimensional scene space. The marine snow particle models include physical properties, biochemical properties and ecological properties.

[0049] The simulation module is used to drive the marine snow particle model in each grid cell to simulate the motion process and biochemical process based on the input three-dimensional marine environmental field data and to update and solve the problem using parallel computing. The biochemical process simulation includes aggregation mechanism, decomposition mechanism and feeding mechanism.

[0050] The real-time rendering module is used to render in a 3D scene based on the updated state of the sea snow particle model, while simulating optical effects in a deep-sea environment.

[0051] The aforementioned particle rendering method and system for simulating deep-sea snow deposition processes achieve accurate quantitative characterization of complex ecological behaviors of snow by constructing a snow particle model and introducing aggregation, decomposition, and feeding mechanisms. By introducing a parallel computing architecture, the efficiency of large-scale snow particle visualization is improved. Furthermore, by integrating optical effects simulation into 3D scene rendering, the realism of particle visual representation under deep-sea lighting conditions in 3D simulations is enhanced. These methods significantly improve the simulation of snow deposition processes in terms of behavioral realism, computational efficiency, and visual expressiveness. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is an application environment diagram of a particle rendering method used to simulate the deep-sea snow deposition process in one embodiment;

[0054] Figure 2 This is a flowchart illustrating a particle rendering method for simulating deep-sea snow deposition in one embodiment.

[0055] Figure 3 This is a flowchart illustrating the steps involved in updating the motion of a snow particle model using trilinear interpolation in one embodiment.

[0056] Figure 4 This is a flowchart illustrating a method for simulating optical effects in a deep-sea environment in another embodiment;

[0057] Figure 5 This is a structural block diagram of a particle rendering system used to simulate the settling process of deep-sea snow in one embodiment;

[0058] Figure 6 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0060] The particle rendering method for simulating deep-sea snow deposition provided in this application embodiment can be applied to, for example... Figure 1In the application environment shown, terminal 102 (e.g., a high-performance graphics workstation) communicates with server 104 (e.g., a database server storing marine environmental field data) via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located in the cloud. Specifically, terminal 102 can be a server for scientific computing, and the data storage system can be used to store three-dimensional marine environmental field data and simulation results. It should be noted that this embodiment only uses a graphics workstation as an example; in practical applications, terminal 102 can also be, but is not limited to, various personal computers, laptops, server clusters, etc. Server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0061] This embodiment provides a particle rendering method for simulating the settling process of deep-sea snow. Figure 2 This is a flowchart of a high-performance particle rendering method provided in an embodiment of this application. The execution subject of this method can be an electronic device. Optionally, the electronic device can be a server or a terminal device, but this application is not limited thereto. Specifically, as shown... Figure 2 As shown, the process includes the following steps:

[0062] Step S201: Divide the three-dimensional scene space into several uniform mesh units, and initialize several snow particle models in the three-dimensional scene space.

[0063] First, a uniform mesh space is used to divide the 3D scene space into several cubic mesh units. Specifically, the entire 3D simulation space can be divided into units of size [missing information]. A cubic mesh, in which Let be the maximum particle radius, and c be an adjustable parameter. It can be understood that after meshing, for a particle located in the grid (X, Y, Z), the search can be completed by searching and interacting with particles in its 26 neighboring grids. Preferably, the grid can be reduced to 8 grids. By meshing, the computational complexity can be reduced from O(N²) of the exhaustive search method to O(N), where N is the number of particles.

[0064] Then, in the memory of the computing device, the structure of the marine snow particle model is constructed, and several marine snow particle models are initialized.

[0065] It should be noted that, in addition to the particle model constructed during system initialization, a particle generation volume can be defined in the 0-100 meter depth region of the simulated water area to continuously generate new sea snow particles. For example, the generation of new particles can be based on two modes:

[0066] (1) Fixed rate: M new particles are generated at each time step. This method is suitable for maintaining the activity of the system.

[0067] (2) Dynamic rate coupled with external data: The generation rate G is calculated from surface chlorophyll concentration and net primary productivity data, and the formula is as follows: Where Input is the input data, and k is a fixed scaling factor. Using this method, it is possible to simulate the changes in the rate of sea snow formation caused by diurnal temperature variations, ocean currents, and monsoon events.

[0068] In some embodiments, the marine snow particle model includes physical properties, biochemical properties, and ecological properties. Physical properties include location, velocity, mass, volume, density, and formation time; biochemical properties include organic carbon content, viscosity coefficient, and particle type identifier, where particle type includes mayfly plant remains, fecal pellets, etc.; ecological properties include parameters used to control predation priority.

[0069] During initialization, the positions of the snowflake particles are randomly distributed within the generation area, with initial velocities close to zero and the direction of velocity set to the direction driven by gravity. Initial mass, volume, and viscosity coefficient are all randomly generated from a normal distribution to simulate the natural distribution characteristics of particles. The formation time is the time it takes for the particles to be generated. Upper and lower limits on the number of particles allowed by the system are set to prevent memory overflow or particle exhaustion. When the total number of particles exceeds the upper limit, some particles can be eliminated according to rules, such as prioritizing the removal of particles that have settled to the deepest part and contribute the least to the visual effect.

[0070] Step S202: Based on the input three-dimensional marine environmental field data, parallel computing is used to drive the marine snow particle model in each grid cell to simulate the motion process and biochemical process, and then update the solution.

[0071] Read three-dimensional marine environmental field data from the data file, including but not limited to three-dimensional current field, temperature field, geographic information (latitude, longitude, depth) data, and marine biological distribution data.

[0072] For each sea snow particle model, based on its current position coordinates, the environmental parameters at that position are obtained from the three-dimensional environment field through spatial interpolation, and the particle position and velocity are updated based on the physical kinematic equations, wherein the environmental parameters include at least flow velocity and temperature.

[0073] Based on this, biochemical process simulations are performed simultaneously. The aggregation mechanism calculates the aggregation probability based on the surface characteristic parameters of the particle models when the spatial distance between two particles is less than a preset aggregation threshold. When the aggregation conditions are met, the two particle models are merged into a new particle model, whose attributes are calculated by weighting or superimposing the attributes of the original particle models. The decomposition mechanism causes particle mass to decay over time according to a preset decomposition kinetic model based on the particle model's formation time and environmental parameters. When the particle mass falls below the decomposition threshold, the particle model splits or disappears. The feeding mechanism sets a virtual predator action area in three-dimensional space. When a particle enters this area, it is marked as consumed according to a preset feeding probability model, and corresponding metabolic product particles are optionally generated.

[0074] Understandably, due to the large number of particles and the interactions between them during the snowfall process, the computational complexity of simulating the aforementioned motion and biochemical processes is high. Therefore, a data-parallel distributed computing framework can be constructed to decompose the large-scale cluster state update task into multiple independently executable subtasks, and the parallel computing capabilities of the Graphics Processing Unit (GPU) can be used to achieve parallel solutions.

[0075] Specifically, all sea snow particle models are divided into multiple subsets and assigned to different computational units, each comprising multiple threads. Within each computational unit, synchronous state update calculations are performed on its assigned subset. These state updates include at least updates to the motion process and the detection and calculation of interactions between particle models, including neighbor particle model search and aggregation judgment in the aggregation mechanism, predation zone intrusion detection in the feeding mechanism, and particle attribute decay calculation and lifecycle state judgment in the decomposition mechanism.

[0076] Step S203: Based on the updated state of the sea snow particle model, render in the 3D scene while simulating optical effects in the deep sea environment.

[0077] Traditional methods for visualizing ocean particles often rely on general-purpose graphics engines, resulting in rendering effects that do not accurately reflect the real physical and optical characteristics of the deep-sea environment, particularly failing to accurately represent the unique light scattering phenomena of sea snow particles under artificial light sources such as searchlights. In this step, our approach couples particle data with a physically based optical model.

[0078] Specifically, the updated sea snow particle model state data is transferred to the graphics rendering domain. During the shading phase of the rendering process, the simulation of deep-sea optical effects is executed simultaneously. Based on the spatial positions of the virtual camera and the simulated light source, the set relationship of each particle model relative to the light source and the line of sight is calculated. Based on this geometric relationship, the light scattering intensity generated by the particle model is calculated using an optical scattering model. The calculated scattered light intensity is then blended with the particle's base color to generate the final output pixel color.

[0079] In the above embodiments, by constructing a marine snow particle model and introducing aggregation, decomposition, and feeding mechanisms, the problem that traditional particle models cannot accurately represent the complex ecological behavior of marine snow is solved; by synchronously driving the motion process and biochemical process simulation based on three-dimensional marine environmental field data, the technical defects of existing methods in the incomplete simulation of the unique ecological interaction mechanism of marine snow are overcome; by utilizing parallel computing, the efficiency of large-scale marine snow particle visualization is improved; and by integrating optical effects simulation into three-dimensional scene rendering, the realism of particle visual performance under deep-sea lighting conditions in three-dimensional simulation is improved.

[0080] In some exemplary embodiments, the sea snow particle model is driven to simulate motion processes based on input three-dimensional marine environmental field data, such as... Figure 3 As shown, it includes the following steps:

[0081] Step S301: Determine the grid cell where the current position of the snow particle model is located, and obtain the flow velocity vector of each vertex of the grid cell.

[0082] It is understandable that the velocity at a point is obtained from the environmental field data through trilinear interpolation, and the particle position is updated accordingly. In advection motion, the formula is as follows:

[0083] ;

[0084] in It is the position of particle i at time t. It is the velocity vector at the particle's location, obtained from the three-dimensional velocity field. This is the time step in the simulation. This formula describes the basic physical process by which snow particles are carried by ocean currents.

[0085] However, in computer simulations, physical fields (such as three-dimensional velocity fields and temperature fields) are discretely stored at the vertices of a three-dimensional mesh, while the positions of the particle model are continuous and may be located inside mesh cells. Therefore, this embodiment uses trilinear interpolation to simulate the motion process of the snow particle model, and the key lies in solving for the velocity vector at any point P(x, y, z). .

[0086] First, for a three-dimensional flow field, the mesh size is NX×NY×NZ, and each mesh point (i, j, k) stores a flow velocity vector value V(i, j, k). For any particle model position P(x, y, z), assuming the mesh cell spacing is 1, the mesh cells surrounding particle P(x, y, z) are:

[0087] ;

[0088] ;

[0089] ;

[0090] It should be noted that if the grid spacing is not 1, a normalization step is required before subsequent calculations.

[0091] Once the grid cells are determined, a particle can be considered to be surrounded by 8 vertices, which are: , , , , , , .

[0092] Step S302: Calculate the local coordinates of the sea snow particle model within the grid cells.

[0093] Then, the local coordinates of the particle within this unit grid cell are calculated, i.e., the weighting factor: Then (dx, dy, dz) are the interpolation weights in the three directions, with values ​​ranging from [0, 1).

[0094] Step S303: Based on the local coordinates, perform trilinear interpolation calculation on the velocity vector of each vertex to obtain the velocity vector of the current position of the snow particle model.

[0095] Next, a cubic linear interpolation step is performed. First, interpolation is performed along the X-axis (on four lines parallel to the X-axis). Layer, along The line is and Interpolation between ,along The line is and Interpolation between Similarly, in Layer, along The line is and Interpolation between; Layer, along The line is and Interpolation is performed between these values. This yields the particle's coordinates in the YZ plane. This gives the estimated velocity along the X-axis. Similarly, interpolation along the Y and Z axes yields the final velocity vector of the particle.

[0096] .

[0097] Step S304: Update the position of the sea snow particle model using kinematic equations based on the flow velocity vector.

[0098] Combining the above steps, we obtain the trilinear interpolation formula, which is then used to perform the same calculation independently for each component (Vx, Vy, Vz) of the velocity vector:

[0099] .

[0100] As can be understood, assuming the physical field changes continuously and linearly within a single grid cell, the above method estimates the field value at the particle model's location by weighted averaging the known values ​​at the eight corner points of the unit grid containing the particle model. The weights (dx, dy, dz) reflect the spatial proximity of the particle model to each corner point; the closer the particle model is to a corner point, the greater the weight of that corner point's value in the final result.

[0101] In the above embodiments, the motion process of each marine snow particle model is simulated by using a trilinear interpolation algorithm, which can ensure the accuracy of obtaining continuous spatial field values ​​from discrete grid data, thereby improving the accuracy of the description of the marine snow particle motion process.

[0102] Because marine snow particles collide during sedimentation due to turbulence and Brownian motion, and aggregate under certain conditions, this process affects their particle size distribution and sedimentation rate. To simulate this ecological process, in some exemplary embodiments, the aggregation mechanism can be implemented as follows:

[0103] Step S401: Determine the target mesh cell where the current marine snow particle model is located.

[0104] Step S402: Search for other sea snow particle models within the local search area formed by the target grid cell and its spatially adjacent grid cells.

[0105] Step S403: When there are other marine snow particle models in the local search area, and the distance between the two marine snow particle models is less than the preset aggregation threshold, the aggregation probability is calculated based on the viscosity coefficient of the two marine snow particle models.

[0106] Specifically, each snowflake particle model is traversed, and its neighboring particles are searched for each particle. The aggregation threshold is related to the physical size of the particle. In one embodiment, for original particles j and k, where r is the particle's equivalent radius, the determination method is as follows:

[0107] ;

[0108] in, This is the threshold coefficient, used to simulate the effective range of the particle model.

[0109] When the distance condition is met, a probabilistic aggregation decision is made based on the surface characteristics of the particle model. Specifically, the aggregation probability is calculated based on the viscosity coefficients of the two sea snow particle models:

[0110] ;

[0111] in, These are the viscosity coefficients of the two particle models, and A is the rate constant related to the particle collision cross section and the intensity of environmental turbulence. The specific value can be calibrated by comparing with field observation data or configured according to the physical characteristics of the target simulation area.

[0112] In one embodiment, the order of magnitude of the rate constant A can be expressed by the formula An estimate is made, in which, and These are the equivalent radii of the interacting particle models j and k, respectively. This approximates the product of the collision cross-section and the relative velocity. Larger particles have a larger collision target area and a higher relative velocity in turbulence. The turbulent dissipation rate of seawater, and the polymerization efficiency coefficient. Calibration can be achieved through experimental data. to Within a reasonable range. This is the time step, indicating that the probability is related to the duration of action. This formula macroscopically quantifies microscopic adhesion as a random event.

[0113] In step S404, if the aggregation probability is greater than a preset threshold, the two sea snow particle models are merged into a new particle, and the physical and biochemical properties of the new particle model are calculated and updated based on the physical and biochemical properties of the two sea snow particle models.

[0114] The properties of the merged new particle model can be obtained by calculating the properties of the two original particle models. For the original particle models j and k, the mass of the merged new particle model is... The center of mass is The position is the center of mass of the two-particle model, and its density and other physical properties are calculated similarly. The viscosity coefficient of the new particle model i is output by weighting the two particle models according to their surface areas. The velocity of the new particle model i is determined by the law of conservation of momentum. Calculations are performed, with the formation time set at age=0, and the formation position at the average of the displacement vectors of particles j and k before their aggregation. The particle type can be determined using a preset inheritance rule, which identifies the new particle type.

[0115] It can be understood that the decomposition mechanism simulates the material decay and morphological evolution process of marine snow particles during sedimentation due to microbial degradation, hydrolysis, and other processes. In some exemplary embodiments, the decomposition mechanism is implemented according to the following steps:

[0116] Step S501: For the marine snow particle model in each grid cell, the particle mass is made to decay over time according to the formation time of the marine snow particle model and the preset decomposition rate.

[0117] Specifically, for each snowflake particle model, its mass is reduced based on its travel time and a set decomposition rate. When the mass falls below a threshold, the particle model is marked for deletion or split into multiple smaller particle models. For example, an exponential decay model is used to simulate this process:

[0118] ;

[0119] in, Let i be the mass of particle model i at the current time step. The decomposition rate constant is The time step is the decomposition rate constant. It can be correlated with environmental parameters, such as decreasing with increasing water depth or decreasing temperature. A decomposition event is triggered when the particle model mass is detected to be below a preset decomposition threshold. The decomposition threshold can be set as a fixed percentage of the initial particle model mass, or dynamically adjusted according to the particle model type.

[0120] Step S502: When the particle mass is lower than the decomposition threshold, the splitting probability is calculated according to the preset splitting probability model.

[0121] When the decomposition condition is met, the particle model randomly decomposes into two parts or disappears. Preferably, in one implementation, the system determines the decomposition based on a preset probability. This determines whether the particle model disappears or splits.

[0122] When generating random numbers Or the particle mass is already extremely small (e.g. The particle model was removed directly from the system to simulate situations where organic matter is completely degraded into a dissolved state, or where the particles are too small to be resolved at the simulation scale.

[0123] Step S503: If the splitting probability meets the preset condition, the snow particle model is split into multiple sub-particle models, and the attributes of each sub-particle model are inherited from the original particle.

[0124] when When that happens, decomposition is performed, and the definition is made. For example, a random allocation factor can be taken as... ~U(0.3,0.7) means that it follows a random distribution between 0.3 and 0.7. After the split, the new particle inherits the matter density of the original particle model and obeys the law of conservation of momentum:

[0125] ;

[0126] ;

[0127] in, This is a random velocity vector, representing the tiny impulse generated during the split; its direction is random, and its magnitude can be set to... The positions of the new particle models are obtained as follows:

[0128]

[0129] ;

[0130] Here, offset is a displacement vector with a random direction and a magnitude proportional to the merging radius of the two new particle models. This is to avoid generating two new particle models in exactly the same location, which would lead to visual overlap and physical calculation anomalies.

[0131] The new particle model has a formation time of 0, inherits the type of the original particle, and its viscosity coefficient can be fine-tuned based on the surface area or volume ratio of the new particle model, for example... This simulates situations where the viscosity of a fresh fracture surface may differ.

[0132] Step S504: If the splitting probability does not meet the preset condition, the sea snow particle model is removed to simulate the degradation process.

[0133] In the above embodiments, by introducing a decomposition mechanism, the shortcomings of traditional particle simulation methods in reflecting the biodegradation process of marine snow particles are solved.

[0134] In some exemplary embodiments, since the distribution of predators and their feeding behavior are spatially heterogeneous and random, a feeding mechanism is used to simulate the ecological characteristics of marine snow particles: one or more specific cell grids in the three-dimensional scene space are defined as virtual predator regions; when the marine snow particle model enters the cell corresponding to the virtual predator region through the movement process, the marine snow particle model is removed according to a preset probability to simulate the feeding process.

[0135] Specifically, virtual predator agents with ecological attributes are set up in a three-dimensional space. For each agent, a marine snow particle model within the predator's activity range is detected. When a particle model enters any virtual predator's area, a predation probability model is used to determine whether predation has occurred.

[0136] In one possible implementation, a fixed probability value is used for judgment; in another, more complex implementation, the predation probability is related to one or more factors such as particle type, particle size, and predator density in the region. If predation is determined to have occurred, the sea snow particle model is removed from the system.

[0137] In a preferred embodiment, the feeding mechanism preferably includes a step of metabolite production. When the sea snow particle model is removed to simulate the feeding process, new particles representing predator excrement are generated at the feeding location.

[0138] Specifically, calculate the total mass of the consumed material. One of them (AE stands for assimilation efficiency) is assimilated and permanently removed from the system. The remaining portion... Accumulation occurs. When the accumulated mass reaches a threshold T, a new excrement particle is generated near the agent location. The new particle type has two labels, "fecal clump" and "new type," and its mass is equal to the accumulated excrement mass. Its density and viscosity coefficient can be set to specific values ​​based on biological knowledge, and its age is reset to 0.

[0139] In some embodiments, step S202, which utilizes parallel computing to drive the marine snow particle model within each grid cell to simulate motion and biochemical processes and update the solution, can be implemented using the parallel computing capabilities of a graphics processing unit (GPU) cluster. Specifically, it may include the following steps:

[0140] Step S601: Divide the three-dimensional scene space into multiple subdomains, with each subdomain corresponding to a computing node.

[0141] Initializing the high-performance computing cluster environment typically involves multi-node communication based on a Message Passing Interface (MPI). The 3D scene space is logically divided into multiple non-overlapping or partially overlapping subdomains. The entire 3D ocean simulation space is then decomposed into a 3D spatial domain, further dividing the global computational domain into multiple subdomains. Each subdomain is assigned to a computing node in the cluster (each node can contain multiple GPUs). On each computing node, its local GPU's computing environment is further initialized.

[0142] The spatial partitioning strategy can be either uniform grid partitioning or non-uniform partitioning based on particle density distribution, in order to ensure load balancing across computing nodes.

[0143] Step S602: Distribute and store the snow particle models located in different subdomains on the corresponding computing nodes.

[0144] Based on the particle model's position in 3D space, it is allocated and stored in the corresponding subdomain computing node. Each computing node maintains and processes only the particle data set within its local subdomain. Within each computing node, the particle data allocated to that node is transferred from its host memory to the global video memory of one or more GPUs on that node.

[0145] Step S603: Each computing node computes the state of the snow particle model in the corresponding subdomain in parallel.

[0146] On each node, the dimensions of the thread grid and thread blocks are configured for its GPU, and kernel functions are launched in parallel to process all particle models within the local subdomain. This update process fully includes the simulation logic of motion processes as well as all biochemical processes such as aggregation, decomposition, and feeding. Each node utilizes its local computing resources to perform synchronous calculations on its respective particle model, achieving task-level parallelism.

[0147] Step S604: Data exchange of the snow particle model located at the subdomain boundary is performed between adjacent computing nodes through the message passing interface.

[0148] MPI communication steps are inserted before or after the kernel function execution. Specifically, each computing node identifies the particles in its subdomain boundary region, packages the state data of these particles and sends them to neighboring nodes, and receives boundary particle data from neighboring nodes.

[0149] Through the aforementioned distributed parallel computing, large-scale particle update calculations can be distributed across multiple computing units, thereby significantly improving simulation speed.

[0150] In some of these embodiments, such as Figure 4 As shown, the optical effects simulating a deep-sea environment are achieved through the following methods:

[0151] Step S701: Set up a virtual camera and a simulated searchlight source in the three-dimensional scene space.

[0152] The virtual camera is used to define the observer's viewpoint position, angle, and orientation, while the simulated searchlight source simulates lighting equipment used in the deep-sea environment and has specific spatial position, illumination direction, and light intensity attributes.

[0153] Step S702: Based on the position of the virtual camera and the direction of the searchlight source, calculate the light scattering intensity of the sea snow particle model according to the Tyndall effect model.

[0154] According to the Tyndall effect model, the light scattering intensity coefficient can be calculated as follows:

[0155] ;

[0156] in, The angle between the direction of observation and the direction of the light source. This is a simplified approximation of the anisotropy term of the Rayleigh scattering function, indicating that forward and backward scattering is stronger than side scattering, where d is the distance from the particle to the camera. This indicates that light intensity decreases with the square of the distance. This represents the cross-sectional area of ​​the particle's scattering surface; the larger the area, the stronger the scattered light. The scattering coefficient is related to the particle material and refractive index. It should be noted that the specific mathematical expression of the Tyndall effect model can have multiple forms, and this application does not impose any limitations on it.

[0157] Step S703: The light scattering intensity is fused with the base color of the snow particle model to obtain the final pixel color and output the rendering result.

[0158] During the rendering and shading stage, the light scattering intensity is blended with the base color of the snow particle model. For example, the blending method is as follows:

[0159] ;

[0160] Where Color_final is the current particle color, Color_particle is the particle color at the previous time step, and the initial particle color is randomly generated according to the particle type value. The attenuation coefficient of the water body, The depth of the particle.

[0161] In the above embodiments, by incorporating deep-sea optical physical properties into the rendering process of deep-sea snow particle models, the accuracy and realism of the visualization results can be improved, thereby supporting researchers in conducting accurate visual analysis and judgment.

[0162] It is understandable that, since the snow particles are semi-transparent objects, they need to be blended from back to front during the final rendering to achieve the correct visual effect. In one embodiment, a parallel depth-sorting synthesis algorithm is used to render the snow particle model, specifically including the following steps:

[0163] First, the rendering fragments of all the sea snow particle models in the three-dimensional scene space are sorted by depth.

[0164] For example, when using parallel computing, the master node divides the image into multiple regions and assigns them to different nodes. Each node sends the fragment data of its assigned image region to the corresponding node. Each node receives all the fragments belonging to its own image region and sorts the locally rendered fragments by depth value z.

[0165] Then, based on the sorting results, in order from far to near, the color of the snow particle model after light scattering intensity fusion is mixed with the target color according to the Alpha mixing formula to output the final pixel color.

[0166] The Alpha mixing formula is:

[0167] ;

[0168] in, The source color represents the color calculated by the Tyndall effect from a single sea snow particle model; it is a vector containing RGB components. For source transparency, exemplarily, A value of 1 indicates complete opacity. A value of 0 indicates complete transparency. The target color represents the existing color on the screen. When drawing the first particle, this value is the background color; when drawing subsequent particles, it is the color that is already displayed on the screen after all previous particles have been blended. To output color.

[0169] For example, during parallel computing, each node sends the synthesized local image region back to the master node, which then stitches all regions together to form the final image.

[0170] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0171] Based on the same inventive concept, this application also provides a particle rendering system for simulating deep-sea snow deposition processes to implement the methods described above. The solution provided by this system is similar to the implementation schemes described in the above methods. Therefore, the specific limitations of one or more particle rendering system embodiments for simulating deep-sea snow deposition processes provided below can be found in the limitations of the corresponding methods above, and will not be repeated here.

[0172] In one exemplary embodiment, such as Figure 5 As shown, a particle rendering system 800 for simulating deep-sea snow deposition processes is provided, including: a model building module 801, a simulation module 802, and a real-time rendering module 803, wherein:

[0173] The model building module 801 is used to divide the three-dimensional scene space into several uniform grid units and initialize several marine snow particle models in the three-dimensional scene space. The marine snow particle models include physical properties, biochemical properties and ecological properties.

[0174] The simulation module 802 is used to drive the marine snow particle model in each grid cell to simulate the motion process and biochemical process based on the input three-dimensional marine environmental field data and to update and solve the problem using parallel computing. The biochemical process simulation includes aggregation mechanism, decomposition mechanism and feeding mechanism.

[0175] The real-time rendering module 803 is used to render in a three-dimensional scene based on the updated state of the sea snow particle model, while simulating optical effects in a deep-sea environment.

[0176] The modules in the particle rendering system described above for simulating deep-sea snow deposition can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0177] In one embodiment, a computer device is provided, which may be a server or a graphics workstation, and its internal structure diagram may be as follows: Figure 6 As shown, the computer device includes a processor, a graphics processing unit (GPU), memory, and a network interface connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media (such as hard disks) and main memory (such as random access memory). The non-volatile storage media stores the operating system, computer programs, and a database (for storing marine environmental field data). The main memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with an external data server via a network connection. When the computer program is processed and executed, it implements the aforementioned high-performance particle rendering method.

[0178] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0179] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the steps in the above-described method embodiments.

[0180] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0181] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0182] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A particle rendering method for simulating the settling process of deep-sea snow, characterized in that, The method includes: The three-dimensional scene space is divided into several uniform grid units, and several marine snow particle models are initialized in the three-dimensional scene space. The marine snow particle models include physical properties, biochemical properties and ecological properties. Based on the input three-dimensional marine environmental field data, parallel computing is used to drive the marine snow particle model in each grid cell to simulate the motion process and biochemical process, and then update and solve the problem. The biochemical process simulation includes aggregation mechanism, decomposition mechanism and feeding mechanism. Based on the updated state of the sea snow particle model, rendering is performed in a 3D scene, while simulating optical effects in a deep-sea environment. The aggregation mechanism includes: Determine the target grid cell where the current marine snow particle model resides; Search for other sea snow particle models within the local search area formed by the target grid cell and its spatially adjacent grid cells; When other sea snow particle models exist within the local search area, and the distance between the two sea snow particle models is less than a preset aggregation threshold, the aggregation probability is calculated based on the viscosity coefficient of the two sea snow particle models. If the aggregation probability is greater than a preset threshold, the two snow particle models will be merged into a new particle. Based on the physical and biochemical properties of the two marine snow particle models, the physical and biochemical properties of the new particle are calculated and updated.

2. The method according to claim 1, characterized in that, The simulation of the motion process of the marine snow particle model based on the input three-dimensional marine environmental field data includes: Determine the grid cell where the current position of the snow particle model is located, and obtain the flow velocity vector of each vertex of the grid cell; Calculate the local coordinates of the sea snow particle model within the grid cells; Based on the local coordinates, trilinear interpolation is performed on the velocity vectors of each vertex to obtain the velocity vector of the current position of the snow particle. The position of the sea snow particle model is updated using kinematic equations based on the flow velocity vector.

3. The method according to claim 1, characterized in that, The physical, biochemical, and ecological properties of the marine snow particle model include: The physical properties include position, velocity, mass, volume, density, and formation time; The biochemical properties include organic carbon content, viscosity coefficient, and particle type identifier; The ecological attributes include parameters used to control predation priority.

4. The method according to claim 3, characterized in that, The decomposition mechanism includes: For the marine snow particle model within each grid cell, the particle mass decays over time according to the formation time of the marine snow particle model and the preset decomposition rate. When the particle mass is lower than the decomposition threshold, the splitting probability is calculated according to a preset splitting probability model; If the splitting probability meets the preset condition, the snow particle model is split into multiple sub-particle models, and the attributes of each sub-particle model are inherited from the original particle. If the splitting probability does not meet the preset condition, the sea snow particle model is removed to simulate the degradation process.

5. The method according to claim 1, characterized in that, The feeding mechanism includes: One or more specific cell grids in the three-dimensional scene space are defined as virtual predator regions; When the sea snow particle model enters the grid cell corresponding to the virtual predator area through the movement process, the sea snow particle model is removed according to a preset probability to simulate the feeding process.

6. The method according to claim 1, characterized in that, The process of updating and solving the state of the sea snow particle model using parallel computing includes: The three-dimensional scene space is divided into multiple subdomains, and each subdomain corresponds to a computing node; The snow particle models located in different subdomains are distributed and stored on the corresponding computing nodes; Each computing node computes the state of the snow particle model in the corresponding subdomain in parallel. Data exchange of the sea snow particle model located at the subdomain boundary is performed between adjacent computing nodes through the message passing interface.

7. The method according to claim 1, characterized in that, The optical effects simulated in a deep-sea environment include: A virtual camera and a simulated searchlight source are set in the three-dimensional scene space; Based on the position of the virtual camera and the direction of the searchlight source, the light scattering intensity of the sea snow particle model is calculated according to the Tyndall effect model: ； in, Let be the angle between the position of the virtual camera and the direction of the searchlight source, d be the distance from the snow particle model to the virtual camera, and r be the radius of the snow particle. The scattering coefficient is related to the particle material and refractive index; The light scattering intensity is fused with the base color of the snow particle model to obtain the final pixel color and output the rendering result.

8. The method according to claim 7, characterized in that, The process of fusing the light scattering intensity with the base color of the snow particle model to obtain the final pixel color and output the rendering result includes: The rendered fragments of all sea snow particle models in the three-dimensional scene space are sorted by depth; Based on the sorting results, and following the order from farthest to nearest, the color of the snow particle model after light scattering intensity fusion is mixed with the target color using the Alpha blending formula to output the final pixel color. The Alpha blending formula is as follows: ； in, The source color is calculated using the Tyndall effect from the aforementioned snow particle model. For transparency parameters, For the target color that already exists on the screen, To output color.

9. A particle rendering system for simulating the settling process of deep-sea snow, characterized in that, The system includes: The model building module is used to divide the three-dimensional scene space into several uniform grid units and initialize several marine snow particle models in the three-dimensional scene space. The marine snow particle models include physical properties, biochemical properties and ecological properties. The simulation module is used to drive the marine snow particle model within each grid cell to simulate motion and biochemical processes based on input three-dimensional marine environmental field data using parallel computing, and to update and solve the results. The biochemical process simulation includes aggregation, decomposition, and feeding mechanisms. The aggregation mechanism includes: determining the target grid cell where the current marine snow particle model is located; searching for other marine snow particle models within a local search area consisting of the target grid cell and its spatially adjacent grid cells; when other marine snow particle models exist within the local search area and the distance between two marine snow particle models is less than a preset aggregation threshold, calculating the aggregation probability based on the viscosity coefficient of the two marine snow particle models; if the aggregation probability is greater than the preset threshold, merging the two marine snow particle models into a new particle; and calculating and updating the physical and biochemical properties of the new particle based on the physical and biochemical properties of the two marine snow particle models. The real-time rendering module is used to render in a 3D scene based on the updated state of the sea snow particle model, while simulating optical effects in a deep-sea environment.

Images