A method and system for generating an animated video of a bird flock

Abstract

The application relates to a bird flock animation video generation method and system. The method comprises the following steps: constructing a group motion simulation model, and generating three-dimensional space motion trajectory data of each individual in a virtual bird flock changing over time according to the group motion simulation model; constructing a preset state conversion model, and determining a current visual performance state based on the preset state conversion model; for each individual at each time point, generating a two-dimensional image according to the three-dimensional space motion trajectory data and the visual performance state; and combining two-dimensional image frames generated at multiple time points into a video animation. The application closely combines high-fidelity biomechanical group trajectory simulation, individual dynamic visual state simulation based on a Markov process, and a high-quality visual element library, so that a large-scale bird flock animation video with rich details and dynamic reality can be efficiently generated.

Description

Technical Field

[0001] This invention relates to the field of group behavior simulation, and more specifically, to a method and system for generating animated bird flock videos. Background Technology

[0002] The demand for large-scale, realistic group animations (such as flocks of birds soaring, schools of fish swimming, and herds of animals galloping) is growing in numerous fields, including film and television production, game development, scientific research visualization, educational demonstrations, and virtual reality experiences. However, current technologies still face many challenges in generating such high-quality group animation videos: First, the macroscopic realism of group movement is difficult to guarantee. Many traditional animation methods rely on preset paths, simple particle systems, or scripted behavior lacking deep interactive logic. This makes simulated group movement often appear mechanical and repetitive, lacking the complex dynamics, self-organizing phenomena (such as vortex, ribbon, and spherical shape switching), and adaptive responses to the environment characteristic of natural groups. Although there are group behavior models based on multi-agent systems (such as the Boids and Vicsek models), they usually simplify the physical characteristics of biological individuals and complex environmental interactions. For example, the Boids model mainly focuses on geometric rules while ignoring the aerodynamic problems that flying organisms must face, resulting in simulated flight movements lacking physical basis. Second, the microscopic visual representation of individuals within a group is often monotonous. To reduce production costs and computational complexity, group members in animation are often given the same or a limited number of repetitive actions and appearances, making it difficult to showcase the rich postures, airfoil changes, and subtle movements that individuals should exhibit under different motion states (such as acceleration, deceleration, gliding, sharp turns, and hovering in flight) or different behavioral patterns (such as predation, evasion, and social interaction). For example, the shape and frequency of the wings of a bird accelerating at full speed and a bird gliding smoothly should be significantly different. Such details are often overlooked in traditional group animation, which greatly reduces the immersion and credibility of the animation.

[0003] Secondly, there are technical bottlenecks in acquiring and efficiently synthesizing high-quality visual materials. To give each dynamic individual in a group appropriate and detailed visual representation in every frame of a video, a large amount of high-quality visual materials (such as 2D sprites or 3D models) corresponding to various states are needed. Traditional methods of manually drawing, modeling, or extracting these materials from real images through manual or semi-automatic methods (such as chroma keying or rotating brushes) are not only time-consuming, labor-intensive, and costly, but also make it difficult to guarantee the diversity and segmentation accuracy of the materials. Especially when dealing with complex backgrounds, non-rigid objects, or targets with semi-transparent parts (such as wings with motion blur), edge defects or background residue are easily produced. Summary of the Invention

[0004] Therefore, it is necessary to provide a method and system for generating bird flock animation videos with better output to address the aforementioned technical problems.

[0005] One method for generating animated bird flock videos includes: S1: Construct a group motion simulation model, and generate three-dimensional spatial motion trajectory data of multiple individuals in the virtual flock of birds over time based on the group motion simulation model; S2: Construct a preset state transition model. For each individual, at multiple discrete time points on its motion trajectory, determine its current visual performance state based on the preset state transition model. S3: For each individual at each time point, generate a two-dimensional image based on the three-dimensional spatial motion trajectory data and visual performance status; S4: Combine two-dimensional image frames generated at multiple time points into a video animation.

[0006] Further, in step S1, the generation of three-dimensional spatial trajectory data of multiple individuals in the virtual flock of birds over time based on the group motion simulation model is specifically as follows: The group motion simulation model generates three-dimensional spatial motion trajectory data of multiple individuals in a virtual flock of birds over time, based on the interaction rules of local topological neighborhoods between individuals and biomechanics related to individual motion characteristics.

[0007] Furthermore, the interaction rules based on the local topological neighborhood between individuals include separation forces, alignment forces, and cohesion forces, encompassing social forces, physical forces, and random perturbations; among the physical forces are aerodynamics, gravity, and individual thrust. Aerodynamics includes lift and drag.

[0008] Furthermore, the interaction rules based on local topological neighborhoods between individuals specifically reflect the forces of local social interactions between individuals. They employ topological neighborhood-based interaction, meaning each individual interacts only with a fixed number of its k nearest neighbors; including: Separation force A short-range repulsive force designed to prevent individuals from getting too close to their neighbors to avoid collisions; its force is calculated based on the distance between the individual and its neighbors; assuming the distance between individual i and neighbor j is... The strength of the separating force is directly proportional to the reciprocal of the distance or decreases exponentially; its mathematical expression is:

[0009] in For individuals The neighborhood group, and Individuals and neighbors The position vector.

[0010] Alignment Force A medium-range force ensures that an individual's direction of motion aligns with the average direction of motion of its neighbors; this is crucial for achieving coordinated movement within a group. Its mathematical expression is:

[0011] in and Individuals and neighbors The velocity vector.

[0012] Cohesion ( ( ): A long-range attraction that drives individuals to move toward the geometric center of their neighboring groups to maintain group cohesion. Its mathematical expression is:

[0013] Comprehensive social force ( An individual's total social power is composed of the weighted sum of the above three types of forces:

[0014] in These are weighting coefficients that control the strength of separation, alignment, and cohesion, respectively.

[0015] Furthermore, biomechanics related to individual movement characteristics specifically refers to physical forces ( )include: Lift and drag, as well as gravity and individual active thrust; assuming the flight speed of each individual is... Its size is Then the components of each physical force are:

[0016]

[0017] in It is the lift coefficient. It is the drag coefficient. It is the wing area. It refers to air density. The resultant force of physical forces also includes gravity. And the active thrust generated by the individual through flapping its wings .

[0018] random perturbation ( To simulate uncertainties in the natural environment, a random perturbation force is introduced. The expression for this force is:

[0019] in Individual The intensity of the disturbance. It is a unit vector with a random direction.

[0020] Model Update: At each time step, the individual's motion state is updated by the combined effects of social forces, physical forces, and random perturbations. The individual's acceleration is given by Newton's second law:

[0021] in It is the individual mass. Its velocity and position are then updated using numerical integration.

[0022] Further, step S2 specifically includes: Define the visual state set: Predefine a set of discrete states that can fully describe the typical appearance or movement cycle of the target individual; =“Full-force upward flapping phase” =“Flapping the wings to the middle position” =“Wings fully extended, ready to flap” =“Full-force downward flapping phase” =“Wings flapping down to the middle position” =“Gliding posture - wingspan” =“Gliding posture - wings slightly tucked in to adjust direction”; Construct a pre-defined state transition model: Define the probability of transitioning between visual states in the visual state set per unit time; the model is dynamic, where the transition probability is affected by the individual's current kinematic parameters, which are obtained from trajectory data.

[0023] Further, in step S2, the probability of transitioning between visual states in the visual state set per unit time is defined as follows: When individual speed High and acceleration When smaller, it is more likely to transition to and maintain various gliding attitudes.

[0024] When an individual needs to accelerate, climb, or make a sharp turn, i.e., accelerates... When the value is large, it will enter and rapidly cycle through flapping-related states with a high probability. to ; In the simulation model Flapping frequency parameters are used to directly guide the switching rate of visual states related to flapping. Angular velocity: calculated by the velocity vector The angular velocity of an individual is obtained by changing direction in a short period of time; when the angular velocity exceeds a certain threshold, the probability of the individual entering a "tilting turn" state increases.

[0025] Through the above process, each individual is given a visual state that best represents its current movement intention and behavior at each moment on the trajectory.

[0026] Further, step S3 specifically includes: A visual element library is constructed, which contains multiple two-dimensional image sprites or three-dimensional models. Each sprite or group of sprites / three-dimensional models corresponds to a visual representation state. Furthermore, the two-dimensional image sprites are preferably processed by using an image segmentation model to process the original image containing the target individual in order to extract the precise contour of the target individual and generate visual elements with transparent background channels. Based on the individual's position in 3D space and camera parameters, project it onto a 2D image plane; determine the rotation angle of the visual element on the image plane based on the individual's 3D velocity vector or body orientation; adjust the display size of the visual element based on the individual's 3D distance from the camera to achieve a perspective effect; and process the rendering order between multiple individuals to achieve the correct occlusion relationship.

[0027] Further, step S4 specifically involves synthesizing the final animated video using a video encoding tool.

[0028] Furthermore, the present invention also provides a bird flock animation video generation system, comprising: Trajectory Module: Constructs a group motion simulation model and generates three-dimensional spatial motion trajectory data of multiple individuals in the virtual flock of birds over time based on the group motion simulation model; Visual performance module: Constructs a preset state transition model. For each individual, at multiple discrete time points on its movement trajectory, the current visual performance state is determined based on the preset state transition model. Two-dimensional image module: For each individual at each time point, a two-dimensional image is generated based on the three-dimensional spatial motion trajectory data and visual performance status; Video generation module: Combines two-dimensional image frames generated at multiple time points into a video animation.

[0029] This invention combines highly realistic biomechanical group trajectory simulation, Markov process-based individual dynamic visual state simulation, and a high-quality visual element library by constructing a group motion simulation model and a preset state transition model. This enables the efficient generation of large-scale bird flock animation videos with rich details and dynamic realism.

[0030] This embodiment combines highly realistic biomechanical group trajectory simulation, Markov process-based individual dynamic visual state simulation, and a high-quality visual element library by constructing a group motion simulation model and a preset state transition model. This enables the efficient generation of large-scale bird flock animation videos with rich details and dynamic realism. Attached Figure Description

[0031] Figure 1 Here is a flowchart of a method for generating animated bird flock videos in one embodiment; Figure 2 This is a block diagram of a bird flock animation video generation system in one embodiment; Figure 3 This is a flowchart illustrating the overall architecture of a method for generating animated bird flock videos in one embodiment; Figure 4 This is a detailed logic block diagram of the starling flock trajectory generation submodule in one embodiment; Figure 5 This is a schematic diagram of a Markov process model used to determine the visual state of an individual starling in one embodiment; Figure 6 This is a schematic diagram illustrating the local social interactions (social forces) among individual birds in one embodiment; Figure 7 This is an example diagram illustrating the rendering process of processing individual 2D bionic bird materials in a single video frame, as shown in one embodiment. Figure 8 This is an example image of a single frame from a video of a flock of starlings in flight, as shown in one embodiment. Detailed Implementation 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.

[0032] Example 1: This embodiment provides, as follows: Figure 1 A method for generating animated bird flock videos is shown, comprising: S1: Construct a group motion simulation model, and generate three-dimensional spatial motion trajectory data of multiple individuals in the virtual flock of birds over time based on the group motion simulation model; S2: Construct a preset state transition model. For each individual, at multiple discrete time points on its motion trajectory, determine its current visual performance state based on the preset state transition model. S3: For each individual at each time point, generate a two-dimensional image based on the three-dimensional spatial motion trajectory data and visual performance status; S4: Combine two-dimensional image frames generated at multiple time points into a video animation.

[0033] This embodiment combines highly realistic biomechanical group trajectory simulation, Markov process-based individual dynamic visual state simulation, and a high-quality visual element library by constructing a group motion simulation model and a preset state transition model. This enables the efficient generation of large-scale bird flock animation videos with rich details and dynamic realism.

[0034] Example 2: This embodiment further discloses information based on Embodiment 1: Further, in step S1, the generation of three-dimensional spatial trajectory data of multiple individuals in the virtual flock of birds over time based on the group motion simulation model is specifically as follows: The group motion simulation model generates three-dimensional spatial motion trajectory data of multiple individuals in a virtual flock of birds over time, based on the interaction rules of local topological neighborhoods between individuals and biomechanics related to individual motion characteristics.

[0035] Furthermore, the interaction rules based on the local topological neighborhood between individuals include separation forces, alignment forces, and cohesion forces, encompassing social forces, physical forces, and random perturbations; among the physical forces are aerodynamics, gravity, and individual thrust. Aerodynamics includes lift and drag.

[0036] Furthermore, the interaction rules based on local topological neighborhoods between individuals specifically reflect the forces of local social interactions between individuals. They employ topological neighborhood-based interaction, meaning each individual interacts only with a fixed number of its k nearest neighbors; including: Separation force A short-range repulsive force designed to prevent individuals from getting too close to their neighbors to avoid collisions; its force is calculated based on the distance between the individual and its neighbors; assuming the distance between individual i and neighbor j is... The strength of the separating force is directly proportional to the reciprocal of the distance or decreases exponentially; its mathematical expression is:

[0037] in For individuals The neighborhood group, and Individuals and neighbors The position vector.

[0038] Alignment Force A medium-range force ensures that an individual's direction of motion aligns with the average direction of motion of its neighbors; this is crucial for achieving coordinated movement within a group. Its mathematical expression is:

[0039] in and Individuals and neighbors The velocity vector.

[0040] Cohesion ( ( ): A long-range attraction that drives individuals to move toward the geometric center of their neighboring groups to maintain group cohesion. Its mathematical expression is:

[0041] Comprehensive social force ( An individual's total social power is composed of the weighted sum of the above three types of forces:

[0042] in These are weighting coefficients that control the strength of separation, alignment, and cohesion, respectively.

[0043] Furthermore, biomechanics related to individual movement characteristics specifically refers to physical forces ( )include: Lift and drag, as well as gravity and individual active thrust; assuming the flight speed of each individual is... Its size is Then the components of each physical force are:

[0044]

[0045] in It is the lift coefficient. It is the drag coefficient. It is the wing area. It refers to air density. The resultant force of physical forces also includes gravity. And the active thrust generated by the individual through flapping its wings .

[0046] random perturbation ( To simulate uncertainties in the natural environment, a random perturbation force is introduced. The expression for this force is:

[0047] in Individual The intensity of the disturbance. It is a unit vector with a random direction.

[0048] Model Update: At each time step, the individual's motion state is updated by the combined effects of social forces, physical forces, and random perturbations. The individual's acceleration is given by Newton's second law:

[0049] in It is the individual mass. Its velocity and position are then updated using numerical integration.

[0050] Further, step S2 specifically includes: Define the visual state set: Predefine a set of discrete states that can fully describe the typical appearance or movement cycle of the target individual; =“Full-force upward flapping phase” =“Flapping the wings to the middle position” =“Wings fully extended, ready to flap” =“Full-force downward flapping phase” =“Wings flapping down to the middle position” =“Gliding posture - wingspan” =“Gliding posture - wings slightly tucked in to adjust direction”; Construct a pre-defined state transition model: Define the probability of transitioning between visual states in the visual state set per unit time; the model is dynamic, where the transition probability is affected by the individual's current kinematic parameters, which are obtained from trajectory data.

[0051] Further, in step S2, the probability of transitioning between visual states in the visual state set per unit time is defined as follows: When individual speed High and acceleration When smaller, it is more likely to transition to and maintain various gliding attitudes.

[0052] When an individual needs to accelerate, climb, or make a sharp turn, i.e., accelerates... When the value is large, it will enter and rapidly cycle through flapping-related states with a high probability. to ; In the simulation model Flapping frequency parameters are used to directly guide the switching rate of visual states related to flapping. Angular velocity: calculated by the velocity vector The angular velocity of an individual is obtained by changing direction in a short period of time; when the angular velocity exceeds a certain threshold, the probability of the individual entering a "tilting turn" state increases.

[0053] Through the above process, each individual is given a visual state that best represents its current movement intention and behavior at each moment on the trajectory.

[0054] Further, step S3 specifically includes: A visual element library is constructed, which contains multiple two-dimensional image sprites or three-dimensional models. Each sprite or group of sprites / three-dimensional models corresponds to a visual representation state. Furthermore, the two-dimensional image sprites are preferably processed by using an image segmentation model to process the original image containing the target individual in order to extract the precise contour of the target individual and generate visual elements with transparent background channels. Based on the individual's position in 3D space and camera parameters, project it onto a 2D image plane; determine the rotation angle of the visual element on the image plane based on the individual's 3D velocity vector or body orientation; adjust the display size of the visual element based on the individual's 3D distance from the camera to achieve a perspective effect; and process the rendering order between multiple individuals to achieve the correct occlusion relationship.

[0055] Further, step S4 specifically involves synthesizing the final animated video using a video encoding tool.

[0056] This embodiment combines highly realistic biomechanical group trajectory simulation, Markov process-based individual dynamic visual state simulation, and a high-quality visual element library by constructing a group motion simulation model and a preset state transition model. This enables the efficient generation of large-scale bird flock animation videos with rich details and dynamic realism.

[0057] Example 3: This embodiment also provides, for example Figure 2 The system shown is a bird flock animation video generation system, comprising: Trajectory Module: Constructs a group motion simulation model and generates three-dimensional spatial motion trajectory data of multiple individuals in the virtual flock of birds over time based on the group motion simulation model; Visual performance module: Constructs a preset state transition model. For each individual, at multiple discrete time points on its movement trajectory, the current visual performance state is determined based on the preset state transition model. Two-dimensional image module: For each individual at each time point, a two-dimensional image is generated based on the three-dimensional spatial motion trajectory data and visual performance status; Video generation module: Combines two-dimensional image frames generated at multiple time points into a video animation.

[0058] This embodiment combines highly realistic biomechanical group trajectory simulation, Markov process-based individual dynamic visual state simulation, and a high-quality visual element library by constructing a group motion simulation model and a preset state transition model. This enables the efficient generation of large-scale bird flock animation videos with rich details and dynamic realism.

[0059] Example 4: This embodiment provides, as follows: Figure 3 The method for generating a flock of bird animation videos includes generating an animation video of a flock of starlings in flight.

[0060] The three-dimensional trajectory generation stage of starling flocks, as follows Figure 4 As shown, its core principle is to combine social forces, physical forces, and random perturbations to update the motion state of each individual. In practical implementation, simulation parameters will be set according to this model description, and simulation loops will be executed to generate trajectory data.

[0061] The stages of determining the dynamic visual state of an individual starling, such as Figure 5 As shown, for each individual generated in the first stage The trajectory at each point in time (or selected keyframe): A schematic diagram of local social interactions (social forces) among individual birds, as shown below. Figure 6 As shown.

[0062] Define a visual state set S: for example .

[0063] Initialize individual state: In At that time, for each individual Give an initial visual state .

[0064] State transition based on Markov process: Construct a state transition probability matrix , where matrix elements Indicates from state Transition to state The probability. This probability can be dynamically adjusted based on the individual's kinematic parameters (velocity, acceleration, turning rate). At each time point According to the individual Current visual state And the adjusted transition probability, select its next visual state .

[0065] Visual material preparation and rendering / embedding stage Building a visual resource library: Collect / create raw materials: Obtain multiple high-resolution, realistic photos or high-quality 3D models of birds from various visual perspectives.

[0066] SAM2-assisted sprite extraction: For 2D image materials, the SAM2 model is used to process each original image. The bird subject is segmented through interactive prompts (click or selection), generating PNG format sprites with transparent backgrounds, and storing them in the library according to their corresponding visual states.

[0067] Single-frame image rendering (for each frame of the video, corresponding to the simulation time) ): Configure the virtual camera: Define the camera's position, orientation, field of view, etc.

[0068] Iterate through all individuals: For each individual To obtain its trajectory data , and visual state Select the corresponding sprite from the library, perform 3D projection, rotation, and scaling transformations, and finally utilize... The blending technique renders it onto the background while handling occlusion based on depth values. The rendering process for individual 2D bionic bird footage within a single video frame is as follows: Figure 7 As shown.

[0069] The video sequence generation stage combines each frame of the image sequence rendered in time sequence in stage III into a standard video file (MP4 format) using a video encoding tool (FFmpeg).

[0070] Example of a single frame image from a generated video of a flock of starlings in flight: Figure 8 As shown in the figure. This embodiment combines highly realistic biomechanical group trajectory simulation, Markov process-based individual dynamic visual state simulation, and a high-quality visual element library by constructing a group motion simulation model and a preset state transition model, thereby efficiently generating large-scale bird flock animation videos with rich details and dynamic realism.

Claims

1. A method for generating animated videos of bird flocks, characterized in that, include: S1: Construct a group motion simulation model, and generate three-dimensional spatial motion trajectory data of multiple individuals in the virtual flock of birds over time based on the group motion simulation model; S2: Construct a preset state transition model. For each individual, at multiple discrete time points on its motion trajectory, determine its current visual performance state based on the preset state transition model. S3: For each individual at each time point, generate a two-dimensional image based on the three-dimensional spatial motion trajectory data and visual performance status; S4: Combine two-dimensional image frames generated at multiple time points into a video animation.

2. The method for generating animated bird flock videos according to claim 1, characterized in that, In step S1, the generation of three-dimensional spatial trajectory data of multiple individuals in the virtual flock of birds over time, based on the group motion simulation model, is specifically as follows: The group motion simulation model generates three-dimensional spatial motion trajectory data of multiple individuals in a virtual flock of birds over time, based on the interaction rules of local topological neighborhoods between individuals and biomechanics related to individual motion characteristics.

3. The method for generating animated bird flock videos according to claim 2, characterized in that, Interaction rules based on local topological neighborhoods between individuals include separation forces, alignment forces, and cohesion forces. Biomechanical constraints include social forces, physical forces, and random perturbations. Physical forces include aerodynamics, gravity, and individual thrust. Aerodynamics includes lift and drag.

4. The method for generating animated bird flock videos according to claim 2, characterized in that, The interaction rules based on local topological neighborhoods between individuals specifically reflect the forces of local social interactions between individuals. The interaction is based on topological neighborhoods, meaning that each individual only interacts with its nearest fixed number of k neighbors. include: Separation force A short-range repulsive force designed to prevent individuals from getting too close to their neighbors to avoid collisions; its force is calculated based on the distance between the individual and its neighbors; assuming the distance between individual i and neighbor j is... The strength of the separating force is directly proportional to the reciprocal of the distance or decreases exponentially; its mathematical expression is: in For individuals The neighborhood group, and Individuals and neighbors The position vector; Alignment Force A medium-range force causes an individual's direction of motion to align with the average direction of motion of its neighbors; this is key to achieving coordinated movement within a group. Its mathematical expression is: in and Individuals and neighbors The velocity vector; Cohesion ( ): A long-range attraction that drives individuals to move toward the geometric center of their neighboring groups to maintain group cohesion; its mathematical expression is: Comprehensive social force ( An individual's total social power is composed of the weighted sum of the above three types of forces: in These are weighting coefficients that control the strength of separation, alignment, and cohesion, respectively.

5. The method for generating animated bird flock videos according to claim 3, characterized in that, Biomechanics related to individual motion characteristics includes social forces, physical forces, and random disturbances; physical forces include aerodynamics, gravity, and individual thrust; aerodynamics includes lift and drag. Physical forces include: Lift and drag, as well as gravity and individual active thrust; assuming the flight speed of each individual is... Its size is Then the components of each physical force are: in It is the lift coefficient. It is the drag coefficient. It is the wing area. It refers to air density; the resultant force of physical forces also includes gravity. And the active thrust generated by the individual through flapping its wings ; random perturbation ( To simulate uncertainties in the natural environment, a random perturbation force is introduced; the expression for this force is: in Individual The intensity of the disturbance. It is a unit vector with a random direction; Model update: At each time step, the individual's motion state is determined by the combined effects of social forces, physical forces, and random perturbations; the individual's acceleration is given by Newton's second law. in It is the individual mass; then its velocity and position are updated by numerical integration.

6. The method for generating animated bird flock videos according to claim 4, characterized in that, Step S2, specifically: Define the visual state set: Predefine a set of discrete states that can fully describe the typical appearance or movement cycle of the target individual; = "Wings flapping upwards at full power" =="Flap from the wing to the middle position", =="Wings fully extended, ready to unleash the fan". = "Wings flapping down at full power" =="wing flapping down to the middle position", =Gliding posture - wings spread out", =Gliding posture - wings slightly tucked in to adjust direction; Construct a pre-defined state transition model: Define the probability of transitioning between visual states in the visual state set per unit time; the model is dynamic, where the transition probability is affected by the individual's current kinematic parameters, which are obtained from trajectory data.

7. The method for generating animated bird flock videos according to claim 6, characterized in that, Step S2, define the probability of transitioning between visual states in the visual state set per unit time as follows: When individual speed High and acceleration When smaller, it is more likely to transition to and maintain various gliding attitudes; When an individual needs to accelerate, climb, or make a sharp turn, i.e., accelerates... When the value is large, it will enter and rapidly cycle through flapping-related states with a high probability. to ; In the simulation model Flapping frequency parameters are used to directly guide the switching rate of visual states related to flapping. Angular velocity: calculated by the velocity vector The angular velocity of an individual is obtained by changing direction in a short period of time; when the angular velocity exceeds a certain threshold, the probability of the individual entering a "tilting turn" state increases. Through the above process, each individual is given a visual state that best represents its current movement intention and behavior at each moment on the trajectory.

8. The method for generating animated bird flock videos according to claim 7, characterized in that, Step S3 is as follows: A visual element library is constructed, which contains multiple two-dimensional image sprites or three-dimensional models. Each sprite or group of sprites / three-dimensional models corresponds to a visual representation state. Furthermore, the two-dimensional image sprites are preferably processed by using an image segmentation model to process the original image containing the target individual in order to extract the precise contour of the target individual and generate visual elements with transparent background channels. Based on the individual's position in 3D space and camera parameters, project it onto a 2D image plane; determine the rotation angle of the visual element on the image plane based on the individual's 3D velocity vector or body orientation; adjust the display size of the visual element based on the individual's 3D distance from the camera to achieve a perspective effect; and process the rendering order between multiple individuals to achieve the correct occlusion relationship.

9. A method for generating animated bird flock videos according to claim 8, characterized in that, Step S4 specifically involves synthesizing the final animated video using a video encoding tool.

10. A system for generating animated bird flock videos, characterized in that, include: Trajectory Module: Constructs a group motion simulation model and generates three-dimensional spatial motion trajectory data of multiple individuals in the virtual flock of birds over time based on the group motion simulation model; Visual performance module: Constructs a preset state transition model. For each individual, at multiple discrete time points on its movement trajectory, the current visual performance state is determined based on the preset state transition model. Two-dimensional image module: For each individual at each time point, a two-dimensional image is generated based on the three-dimensional spatial motion trajectory data and visual performance status; Video generation module: Combines two-dimensional image frames generated at multiple time points into a video animation.

Images