A high-efficiency 3D animation production method based on artificial intelligence

Abstract

The present invention relates to the field of computer vision technology and discloses a high-efficiency 3D animation production method based on artificial intelligence, comprising the following steps: 1. Constructing a light propagation calculation model. Based on the propagation characteristics of photons in three-dimensional space, a light propagation calculation model is established to describe the distribution relationship of light in different media. The light propagation calculation model includes a description of the motion state of photons and calculates the light intensity distribution in combination with scene geometry information and material properties; 2. Calculating the light distribution based on the light propagation calculation model. By solving the photon motion state, the distribution of light in the three-dimensional scene is determined. Light path sampling is optimized using the Markov chain Monte Carlo method to achieve adaptive photon propagation path selection based on state transition probability, thereby reducing invalid light path calculations and improving the convergence speed of light calculations, thereby enabling efficient global illumination calculations in complex lighting environments.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular to a high-efficiency three-dimensional animation production method based on artificial intelligence. Background Art

[0002] 3D animation production technology has developed rapidly in recent years and is widely used in film, television, gaming, and virtual reality. Lighting calculation is a key factor influencing final image quality and rendering efficiency. High-quality lighting simulation enhances the realism of images, making the lighting and shadows of characters and scenes appear more natural. However, existing 3D animation production methods still face technical challenges in lighting calculation, including the following:

[0003] Traditional lighting calculation methods rely primarily on path tracing and photon mapping, whose core idea is to simulate global illumination by sampling light propagation paths. However, this method requires a large number of samples to approximate the actual lighting distribution, resulting in extremely high computational costs, especially in complex lighting environments.

[0004] In the production of 3D animation, lighting calculations often require handling global illumination for large-scale scenes as well as local illumination for complex geometric structures. Traditional global illumination methods, such as radiosity and path tracing, can effectively simulate multiple reflections of light, but their computational complexity increases dramatically in large-scale scenes, making real-time processing impossible.

[0005] Shadows are a crucial component of 3D animation rendering, enhancing the sense of depth and clarifying the spatial relationship between characters and the environment. Current shadow calculation methods primarily include shadow mapping and ray-traced shadows. Shadow mapping offers high computational efficiency but can easily produce jagged artifacts at the shadow edges, resulting in unnatural shadows. Ray-traced shadows provide high-quality dynamic shadows but are computationally expensive, especially under dynamic light sources and complex geometry, where shadow boundaries can become discontinuous or blurry.

[0006] In 3D animation production, materials within a scene often have varying optical properties. Traditional lighting calculation methods use a fixed reflectivity model to calculate a material's lighting response. This method is unable to dynamically adjust a material's adaptability to varying light sources, resulting in low lighting calculation accuracy and impacting the consistency and realism of the image.

[0007] High dynamic range rendering technology simulates the varying light intensity in the real world, making the highlights and shadows in 3D animation scenes more distinct. Traditional lighting calculation methods, however, can easily lead to overexposure of highlights and loss of detail in shadows in high dynamic range environments, resulting in a lack of depth in the image.

[0008] Therefore, those skilled in the art provide a high-efficiency 3D animation production method based on artificial intelligence to solve the above-mentioned problems. Summary of the Invention

[0009] In view of the deficiencies of the prior art, the present invention provides a high-efficiency three-dimensional animation production method based on artificial intelligence to solve the problems raised in the above background technology.

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions: a high-efficiency 3D animation production method based on artificial intelligence, comprising:

[0011] Step 1: Construct a light propagation calculation model. Based on the propagation characteristics of photons in three-dimensional space, a light propagation calculation model is established to describe the distribution relationship of light in different media. The light propagation calculation model includes a description of the motion state of photons and calculates the light intensity distribution by combining scene geometry information and material properties.

[0012] Step 2: Calculate the illumination distribution based on the light propagation calculation model. By solving the motion state of photons, determine the distribution of illumination in the 3D scene. Adjust the attenuation and scattering characteristics of illumination based on the material parameters of the scene to generate preliminary illumination data.

[0013] Step 3: Perform neural network optimization on the preliminary illumination data. Based on the constructed light propagation calculation model and illumination distribution, a neural network is used to learn the characteristics of the illumination data. The calculated illumination distribution is adjusted to adapt the illumination data to the characteristics of different scene light sources. The neural network input is the illumination distribution data, and the output is the optimized illumination distribution information.

[0014] Step 4: Perform light path optimization based on the optimized light distribution. Based on the light distribution information obtained by neural network optimization, a path optimization method is used to adjust the light propagation path in three-dimensional space. The directionality of the light is optimized in combination with the photon propagation state, so that the light propagation on the surface of different media conforms to the optical characteristics.

[0015] Step 5: Calculate global illumination based on the light path optimization results. Based on the propagation information obtained from the light path optimization and the geometric structure of the 3D scene, use the illumination calculation method to solve the global illumination and generate global illumination data. The global illumination data serves as the input information for the final illumination rendering.

[0016] In step 5, the global illumination calculation uses the finite element method to solve the illumination distribution, and the calculation formula is as follows:

[0017]

[0018] Among them, K lmis a finite element matrix, which represents the mutual influence of light in different spatial units;

[0019] I m is the light intensity of the mth unit;

[0020] F l Contribution to the external light source of unit l;

[0021] Q is the total number of elements divided by the finite element;

[0022] Step 6: Render the animation scene based on the global illumination data. Based on the calculated global illumination data and combined with the material properties and lighting direction in the 3D animation scene, perform a light mapping calculation on the objects in the scene to generate a 3D animation image that conforms to the lighting characteristics. The light mapping calculation combines the lighting distribution information in the global illumination data to perform light and shadow adjustments to ensure that the rendering result conforms to the lighting changes in the 3D scene.

[0023] In step 6, the animation scene rendering uses the boundary element method to optimize the local illumination calculation, and the calculation formula is as follows:

[0024] K·I=F,

[0025] Among them, K is the boundary element matrix, which represents the propagation relationship of boundary illumination;

[0026] I is the boundary light intensity distribution vector;

[0027] F is the boundary light source contribution vector.

[0028] Preferably, the light propagation calculation model in step 1 uses a light propagation formula based on wave optics to calculate the light intensity distribution, and the calculation formula is as follows:

[0029] E(r, t)=E0e i(k·r-ωt) ,

[0030] Where E(r, t) represents the amplitude of the light wave at position r in three-dimensional space;

[0031] E0 is the initial amplitude of light, which indicates the energy intensity of the light source;

[0032] e i(k·r-ωt) is the phase factor of the light wave, which describes the propagation of the light wave in space and time;

[0033] k is the wave vector of light, which describes the propagation direction of light in different media;

[0034] ω is the angular frequency of light, reflecting the time evolution characteristics of light waves;

[0035] t is the time variable.

[0036] Preferably, the calculation of the illumination distribution in step 2 uses the Monte Carlo integration method to calculate the photon state, and the calculation formula is as follows:

[0037]

[0038] Where I(r) is the light intensity at position r in three-dimensional space;

[0039] N is the number of Monte Carlo sampling times, which indicates the number of samples used when calculating lighting;

[0040] f(r i ) is the photon at position r i Contribution factor at

[0041] P(r i ) is the photon at position r i The probability distribution at .

[0042] Preferably, in step 3, the neural network optimization of the illumination data uses a convolutional neural network to extract illumination features and calculate the optimization error of the illumination data. The loss function is as follows:

[0043]

[0044] Among them, L cnn Optimizing the loss function for lighting calculations for neural networks;

[0045] M is the number of training samples;

[0046] I cnn (s j ) is the light intensity calculated by the neural network;

[0047] I gt (s j ) is the actual light intensity;

[0048] s j is the spatial location of the training sample.

[0049] Preferably, the light path optimization in step 4 adopts the Markov chain Monte Carlo method to adjust the light propagation path, and the probability of the photon path is calculated as follows:

[0050] P(x k+1 |x k )=A(x k , x k+1 )P(x k+1 ),

[0051] Among them, P(x k+1 |x k ) is the photon from x k Propagate to x k+1probability;

[0052] x k is the current photon position, indicating the propagation point of the photon in space;

[0053] x k+1 The next photon propagation bit for the candidate;

[0054] A(x k , x k+1 ) is the acceptance probability, which is calculated as follows:

[0055]

[0056] Among them, P(x k+1 ) is the candidate new path point x k+1 probability;

[0057] P(x k ) is the current path point x k probability.

[0058] Preferably, in step 6, when rendering the animation scene, an adaptive light mapping method is used to adjust the light reflectivity of the material, and the calculation formula is as follows:

[0059] R mat (λ)=R0+α(λ-λ0),

[0060] Among them, R mat (λ) is the material light reflectance at wavelength λ;

[0061] R0 is the reflectivity at the reference wavelength λ0;

[0062] α is the reflectivity variation coefficient of the material;

[0063] λ is the wavelength of the current light; λ0 is the reference wavelength.

[0064] Preferably, in step 6, during the animation scene rendering process, the shadow distribution of the object is adjusted based on the illumination direction, and the shadow generation function is calculated as follows:

[0065] S(p,θ)=max(0,cosθ),

[0066] Where S(p, θ) is the shadow intensity at position p in three-dimensional space;

[0067] p is the spatial position in the three-dimensional scene, indicating the coordinates of the point where the shadow needs to be calculated;

[0068] θ is the incident angle of light; cosθ is the angle between the object surface and the light source direction.

[0069] Preferably, in step 6, in order to enhance light and shadow details during the animation scene rendering process, the light brightness is adjusted using a high dynamic range, and the calculation formula is as follows:

[0070]

[0071] Among them, L hdr (x) is the brightness of the light after high dynamic range adjustment;

[0072] L raw (x) is the original calculated light intensity.

[0073] The present invention provides a high-efficiency 3D animation production method based on artificial intelligence. It has the following beneficial effects:

[0074] 1. The present invention optimizes light path sampling through the Markov chain Monte Carlo method, realizes adaptive photon propagation path selection based on state transition probability, reduces invalid light path calculations, improves the convergence speed of illumination calculations, and enables efficient global illumination calculations in complex illumination environments.

[0075] 2. The present invention optimizes illumination calculation by combining the finite element method with the boundary element method, and implements a layered calculation strategy for global illumination and local illumination, so that global illumination is used for large-scale scene calculations, and local illumination optimizes the light and shadow transition of complex geometric structures, thereby improving rendering accuracy, reducing illumination boundary errors, and enhancing the sense of light and shadow layering in three-dimensional animation images.

[0076] 3. The present invention performs dynamic shadow calculation through an angle-based light and shadow calculation model, realizes the calculation of light and shadow distribution according to the light source direction and the object normal vector, and enables the shadow to be dynamically adjusted according to the lighting direction, thereby avoiding the problems of boundary discontinuity, blur and unreality in traditional shadow calculation methods, and making the shadows in three-dimensional animation conform to the real physical lighting characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0078] To help those skilled in the art understand the present invention, the following will provide a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments are only partial embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0079] The present invention is described in detail below with reference to the accompanying drawings:

[0080] Example:

[0081] Please see the attached Figure 1 The embodiment of the present invention provides a high-efficiency 3D animation production method based on artificial intelligence, comprising:

[0082] Step 1: Construct a light propagation calculation model. Based on the propagation characteristics of photons in three-dimensional space, a light propagation calculation model is established to describe the distribution relationship of light in different media. The light propagation calculation model includes a description of the motion state of photons and calculates the light intensity distribution by combining scene geometry information and material properties.

[0083] Step 2: Calculate the illumination distribution based on the light propagation calculation model. By solving the motion state of photons, determine the distribution of illumination in the 3D scene. Adjust the attenuation and scattering characteristics of illumination based on the material parameters of the scene to generate preliminary illumination data.

[0084] Step 3: Perform neural network optimization on the preliminary illumination data. Based on the constructed light propagation calculation model and illumination distribution, a neural network is used to learn the characteristics of the illumination data. The calculated illumination distribution is adjusted to adapt the illumination data to the characteristics of different scene light sources. The neural network input is the illumination distribution data, and the output is the optimized illumination distribution information.

[0085] Step 4: Perform light path optimization based on the optimized light distribution. Based on the light distribution information obtained by neural network optimization, a path optimization method is used to adjust the light propagation path in three-dimensional space. The directionality of the light is optimized in combination with the photon propagation state, so that the light propagation on the surface of different media conforms to the optical characteristics.

[0086] Step 5: Calculate global illumination based on the light path optimization results. Based on the propagation information obtained from the light path optimization and the geometric structure of the 3D scene, use the illumination calculation method to solve the global illumination and generate global illumination data. The global illumination data serves as the input information for the final illumination rendering.

[0087] Step 6: Render the animation scene based on the global illumination data. Based on the calculated global illumination data, combined with the material properties and lighting direction in the three-dimensional animation scene, perform illumination mapping calculations on the objects in the scene to generate a three-dimensional animation image that conforms to the lighting characteristics. The illumination mapping calculation is combined with the illumination distribution information in the global illumination data to perform light and shadow adjustments to ensure that the rendering results conform to the lighting changes in the three-dimensional scene.

[0088] Benefits of Step 1: By modeling the propagation of photons in three-dimensional space, lighting calculations can accurately simulate the reflection, refraction, and scattering of light in different media, enhancing the physical realism of lighting calculations. Combining scene geometry and material properties to calculate the spatial distribution of light intensity allows for precise representation of lighting in complex three-dimensional environments, reducing the distortion often seen in traditional methods.

[0089] Benefits of Step 2: By solving the photon motion state, we optimize the propagation trajectory of light in the 3D scene, ensuring that lighting calculations conform to actual optical laws and reducing non-physical lighting effects. Combined with scene material parameters, we dynamically adjust light attenuation and scattering properties to ensure natural lighting distribution on different material surfaces, enhancing the realism of lighting and shadows in animation scenes.

[0090] Benefits of Step 3: Using a neural network to learn the characteristics of lighting data allows lighting calculations to automatically adapt to different light conditions, improving the generalization of lighting calculations. Deep learning methods are used to optimize lighting data, making lighting calculations smoother under low sampling conditions. This effectively reduces high-frequency noise issues in Monte Carlo sampling methods and improves rendering quality.

[0091] Benefits of Step 4: Based on the neural network-optimized illumination data, path optimization methods are used to adjust photon propagation trajectories, improving ray tracing efficiency, reducing redundant ray calculations, and lowering computing resource consumption. Light directionality is optimized based on the photon propagation state, ensuring that light propagation on different media surfaces conforms to optical properties. This avoids the inconsistent illumination directionality found in traditional methods and improves light and shadow quality.

[0092] Benefits of Step 5: Utilizing optimized light path information through global illumination calculation methods, the lighting calculations for 3D scenes are more efficient, reducing sampling errors in global illumination calculations and improving the accuracy of lighting distribution. By integrating the scene geometry into the lighting solution, the lighting calculations fully account for environmental factors, improving the layering and realism of the lighting, and enhancing the visual expressiveness of 3D animation scenes.

[0093] Benefits of Step 6: Combining global illumination data with the material information of the 3D animation scene allows for precise lighting mapping calculations on the object surface, ensuring uniform lighting distribution and improving the lighting and shadow realism of the final rendering. By combining the lighting distribution information from the global illumination data with lighting adjustments, the rendering result conforms to the scene's lighting changes, achieving dynamic lighting effects and enhancing the visual impact of the 3D animation.

[0094] The light propagation calculation model in step 1 uses the light propagation formula based on wave optics to calculate the light intensity distribution. The calculation formula is as follows:

[0095] E(r, t)=E0e i(k·r-ωt) ,

[0096] Where E(r, t) represents the amplitude of the light wave at position r in three-dimensional space;

[0097] E0 is the initial amplitude of light, which indicates the energy intensity of the light source;

[0098] e i(k·r-ωt) is the phase factor of the light wave, which describes the propagation of the light wave in space and time;

[0099] k is the wave vector of light, which describes the propagation direction of light in different media;

[0100] ω is the angular frequency of light, reflecting the time evolution characteristics of light waves;

[0101] t is the time variable.

[0102] The wave optics model is used to calculate the light intensity distribution, so that the light simulation conforms to the laws of physics and can truly reproduce the interference, diffraction and phase change characteristics of light.

[0103] Traditional geometric optics methods are mainly based on ray tracing and ignore the wave characteristics of light. However, the method of the present invention can accurately describe the propagation characteristics of light in different media and improve the accuracy of illumination calculations.

[0104] By calculating the propagation direction of light in different media using the light's wave vector k, the refraction and reflection behavior of light on transparent, translucent, and highly reflective materials can be dynamically adjusted, avoiding the errors that occur in refraction calculations using traditional methods.

[0105] The time variation of illumination is calculated in combination with the angular frequency ω, which makes the illumination variation of dynamic light sources smooth and reduces the problem of illumination jump during the rendering process.

[0106] The light propagation formula of wave optics is used to calculate the light intensity, making the light and shadow effects of shadows, diffuse reflections, and specular reflections natural, reducing the problems of harsh shadows and excessive uneven light and shadows in traditional methods.

[0107] By calculating the light wave amplitude E(r, t) and its changes, the light intensity distribution in the highlight and shadow areas is simulated, making the light and shadow contrast of the three-dimensional animation picture realistic.

[0108] The calculation of the illumination distribution in step 2 uses the Monte Carlo integration method to calculate the photon state. The calculation formula is as follows:

[0109]

[0110] Where I(r) is the light intensity at position r in three-dimensional space;

[0111] N is the number of Monte Carlo sampling times, which indicates the number of samples used when calculating lighting;

[0112] f(ri ) is the photon at position r i Contribution factor at

[0113] P(r i ) is the photon at position r i The probability distribution at .

[0114] The Monte Carlo integration method is used to calculate the illumination distribution, which makes the estimation of the photon state accurate and can effectively approximate the actual illumination distribution.

[0115] Through sampling technology, the calculation error caused by the fixed sampling method can be reduced, the lighting calculation can be stabilized, and it is suitable for global illumination calculation of complex three-dimensional scenes.

[0116] Traditional lighting calculation methods have high computational costs when dealing with complex lighting scenes, while the Monte Carlo method can achieve high-precision lighting estimation with limited computing resources.

[0117] By adjusting the number of sampling times N, the calculation accuracy and efficiency can be dynamically balanced, which is suitable for different rendering requirements and improves the calculation flexibility of 3D animation production.

[0118] By introducing the probability distribution P(r i ) Model the photon state so that the lighting calculation can conform to the laws of physics and reduce the generation of non-physical lighting artifacts.

[0119] Combined with the photon contribution factor f(r i ) Calculates the impact of photons on lighting at different positions, which can optimize the transition effect of lighting on complex geometric structures, reduce the sharp transition problem of light and shadow edges, and make the lighting distribution natural.

[0120] In step 3, the neural network optimizes the illumination data using a convolutional neural network to extract illumination features and calculate the optimization error of the illumination data. The loss function is as follows:

[0121]

[0122] Among them, L cnn Optimizing the loss function for lighting calculations for neural networks;

[0123] M is the number of training samples;

[0124] I cnn (s j ) is the light intensity calculated by the neural network;

[0125] I gt (s j ) is the actual light intensity;

[0126] s j is the spatial location of the training sample.

[0127] A convolutional neural network is used to extract illumination features, enabling the neural network to learn the spatial correlation in illumination data and accurately predict illumination distribution.

[0128] By optimizing the error calculation, the output of the neural network is made close to the actual light intensity, reducing the light distortion problem caused by insufficient sampling and calculation deviation in traditional light calculation methods.

[0129] Traditional lighting calculation methods are prone to generate high-frequency noise under low sampling conditions, while neural network optimization of lighting data can effectively remove this noise and improve rendering quality.

[0130] By calculating the neural network to optimize the loss function L cnn , can dynamically adjust the neural network parameters to make the lighting calculation robust and improve the consistency of lighting estimation in different scenes.

[0131] By learning the characteristics of large amounts of lighting data through neural networks, it is possible to quickly predict lighting distribution in new 3D animation scenes without relying on the large amount of ray tracing calculations used in traditional methods.

[0132] The convolutional neural network structure has a weight sharing mechanism that can efficiently process lighting data, reduce redundant calculations, improve the real-time performance of lighting rendering, and enable efficient production of high-quality 3D animation.

[0133] In step 4, the light path optimization uses the Markov chain Monte Carlo method to adjust the light propagation path. The probability of the photon path is calculated as follows:

[0134] P(x k+1 |x k )=A(x k , x k+1 )P(x k+1 ),

[0135] Among them, P(x k+1 |x k ) is the photon from x k Propagate to x k+1 probability;

[0136] x k is the current photon position, indicating the propagation point of the photon in space;

[0137] x k+1 The next photon propagation bit for the candidate;

[0138] A(x k , x k+1 ) is the acceptance probability, which is calculated as follows:

[0139]

[0140] Among them, P(x k+1 ) is the candidate new path point x k+1 probability;

[0141] P(x k ) is the current path point x k probability.

[0142] The Markov chain Monte Carlo method is used to optimize light path sampling, which can find the optimal photon propagation path in a few calculation steps and make the lighting calculation converge quickly.

[0143] The photon state transition probability P(x k+1 |x k ) calculates the propagation of light in three-dimensional space, enabling adaptive optimization of lighting calculations rather than relying on random ray tracing, significantly reducing invalid light path calculations.

[0144] Traditional ray tracing methods usually use uniform sampling, which results in a large amount of computing resources being wasted on invalid paths. However, this method uses the acceptance probability A(x k , x k+1 ), can dynamically adjust the direction of light propagation, give priority to calculating important lighting paths, and improve calculation efficiency.

[0145] By adaptively adjusting the sampling probability of the photon path, unnecessary lighting calculations are reduced, so that calculation results can be obtained quickly for complex lighting scenes.

[0146] Since Markov Chain Monte Carlo uses a probability-based optimal path search, it can effectively reduce the sampling error in traditional ray tracing methods and avoid the problem of uneven lighting caused by insufficient sampling.

[0147] By calculating the acceptance probability It can ensure smooth light propagation during lighting calculation, reduce the appearance of artifacts, and make the lighting and shadow effects of three-dimensional animation scenes natural.

[0148] In step 5, the global illumination calculation uses the finite element method to solve the illumination distribution. The calculation formula is as follows:

[0149]

[0150] Among them, K lm is a finite element matrix, which represents the mutual influence of light in different spatial units;

[0151] I m is the light intensity of the mth unit;

[0152] F l Contribution to the external light source of unit l;

[0153] Q is the total number of elements divided by the finite element.

[0154] The finite element method is used to divide the three-dimensional space into multiple groups of finite elements. The illumination intensity of each element is calculated by solving a set of equations, making the illumination calculation accurate. It is particularly suitable for global illumination calculation of complex geometric structures.

[0155] Through the finite element matrix K lm Establish the mutual influence relationship between different units of lighting, so that the lighting calculation can take into account multiple reflections and indirect lighting, and improve the physical reality of global lighting.

[0156] Traditional path tracing methods require a large amount of scene sampling, resulting in high computational costs. This method uses finite element meshing and performs calculations on key units, reducing unnecessary calculations and improving computational efficiency.

[0157] Through the matrix solving method, the calculation time is greatly reduced, making the calculation of global illumination fast and stable in large 3D animation scenes.

[0158] Since the finite element method can perform lighting interpolation calculations in space, it can effectively reduce the problem of uneven lighting caused by discrete sampling, making the light and shadow transitions in three-dimensional animation scenes natural.

[0159] Using the inter-unit lighting influence matrix K lm Calculate the lighting interactions in different spatial areas, reduce errors in lighting boundaries, make lighting calculations smooth, and improve the overall lighting quality of 3D animation images.

[0160] In step 6, the animation scene rendering uses the boundary element method to optimize the local lighting calculation. The calculation formula is as follows:

[0161] K·I=F,

[0162] Among them, K is the boundary element matrix, which represents the propagation relationship of boundary illumination;

[0163] I is the boundary light intensity distribution vector;

[0164] F is the boundary light source contribution vector.

[0165] In step 6, when rendering the animation scene, the adaptive light mapping method is used to adjust the light reflectivity of the material. The calculation formula is as follows:

[0166] R mat (λ)=R0+α(λ-λ0),

[0167] Among them, R mat (λ) is the material light reflectance at wavelength λ;

[0168] R0 is the reflectivity at the reference wavelength λ0;

[0169] α is the reflectivity variation coefficient of the material;

[0170] λ is the wavelength of the current light; λ0 is the reference wavelength.

[0171] Step 6 During the animation scene rendering process, the shadow distribution of the object is adjusted based on the light direction. The shadow generation function is calculated as follows:

[0172] S(p,θ)=max(0,cosθ),

[0173] Where S(p, θ) is the shadow intensity at position p in three-dimensional space;

[0174] p is the spatial position in the three-dimensional scene, indicating the coordinates of the point where the shadow needs to be calculated;

[0175] θ is the incident angle of light; cosθ is the angle between the object surface and the light source direction.

[0176] Step 6: In the animation scene rendering process, in order to enhance the light and shadow details, high dynamic range is used to adjust the light brightness. The calculation formula is as follows:

[0177]

[0178] Among them, L hdr (x) is the brightness of the light after high dynamic range adjustment;

[0179] L raw (x) is the original calculated light intensity.

[0180] Benefits of using boundary element method to optimize local lighting calculations:

[0181] The boundary element matrix K is used to establish the illumination propagation relationship on the boundary, making the local illumination calculation accurate. It is particularly suitable for scenes with complex material surfaces, fine geometric structures and uneven illumination.

[0182] Traditional global illumination calculation methods are difficult to provide high-precision illumination in local detail areas. However, this method calculates the local illumination intensity I through boundary elements, optimizes the local light and shadow transition, and makes the lighting effect natural.

[0183] The boundary element method only needs to perform calculations on the boundary, without calculating the lighting distribution in the three-dimensional scene. Therefore, it reduces the amount of calculation and improves rendering efficiency, making it suitable for high-precision animation production.

[0184] The benefits of using adaptive light mapping to adjust material light reflectivity:

[0185] By calculating the wavelength-dependent material reflectivity R mat(λ)=R0+α(λ-λ0), which can dynamically adjust the light adaptability of the material under different light sources.

[0186] Traditional methods use a fixed reflectance model and cannot adapt to multiple light sources and dynamic lighting environments. However, this method can adaptively adjust according to spectral changes to ensure the realism of material lighting in three-dimensional animation scenes.

[0187] Benefits of using dynamic shadow adjustment based on light direction:

[0188] The shadow calculation formula S(p, θ) = max(0, cosθ) ensures that the shadow is adjusted in real time according to the change of the light source direction, making the shadow natural and avoiding the problem of static shadow distortion.

[0189] Traditional shadow mapping methods easily lead to harsh shadow edges and jagged artifacts. However, this method is calculated based on the lighting angle, which can effectively optimize the shadow transition and make the light and shadow of the three-dimensional animation scene smooth.

[0190] Benefits of using high dynamic range to adjust light brightness:

[0191] The high dynamic range calculation formula can avoid overexposure in highlight areas while retaining details in dark areas, making the picture clear and vivid.

[0192] Traditional lighting calculation methods are prone to problems such as bright spot overflow and loss of details in dark areas. However, this method automatically balances light intensity through nonlinear high dynamic range mapping, optimizes lighting levels, and makes animation images more natural.

[0193] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A high-efficiency 3D animation production method based on artificial intelligence, characterized in that: include: Step 1: Construct a light propagation calculation model. Based on the propagation characteristics of photons in three-dimensional space, a light propagation calculation model is established to describe the distribution relationship of light in different media. The light propagation calculation model includes a description of the motion state of photons and calculates the light distribution by combining scene geometry information and material properties. Step 2: Calculate the illumination distribution based on the light propagation calculation model. By solving the motion state of photons, determine the distribution of illumination in the 3D scene. Adjust the attenuation and scattering characteristics of illumination based on the material parameters of the scene to generate preliminary illumination data. Step 3: Perform neural network optimization on the preliminary illumination data. Based on the constructed light propagation calculation model and illumination distribution, a neural network is used to learn the characteristics of the illumination data. The calculated illumination distribution is adjusted to adapt the illumination data to the characteristics of different scene light sources. The neural network input is the illumination distribution data, and the output is the optimized illumination distribution information. Step 4: Perform light path optimization based on the optimized light distribution. Based on the light distribution information obtained by neural network optimization, a path optimization method is used to adjust the light propagation path in three-dimensional space. The directionality of the light is optimized in combination with the photon propagation state, so that the light propagation on the surface of different media conforms to the optical characteristics. Step 5: Calculate global illumination based on the light path optimization results. Based on the propagation information obtained from the light path optimization and the geometric structure of the 3D scene, use the illumination calculation method to solve the global illumination and generate global illumination data. The global illumination data serves as the input information for the final illumination rendering. In step 5, the global illumination calculation uses the finite element method to solve the illumination distribution, and the calculation formula is as follows: ， in, is a finite element matrix, which represents the mutual influence of light in different spatial units; For the Light intensity of the unit; For the The unit's external light source contribution; The total number of elements divided into finite elements; Step 6: Render the animation scene based on the global illumination data. Based on the calculated global illumination data and combined with the material properties and lighting direction in the 3D animation scene, perform a light mapping calculation on the objects in the scene to generate a 3D animation image that conforms to the lighting characteristics. The light mapping calculation combines the lighting distribution information in the global illumination data to perform light and shadow adjustments to ensure that the rendering result conforms to the lighting changes in the 3D scene. In step 6, the animation scene rendering uses the boundary element method to optimize the local illumination calculation, and the calculation formula is as follows: ， in, is the boundary element matrix, which represents the propagation relationship of boundary illumination; is the boundary light intensity distribution vector; Contribution vector for the boundary light.

2. The high-efficiency 3D animation production method based on artificial intelligence according to claim 1, characterized in that: The light propagation calculation model in step 1 uses the light propagation formula based on wave optics to calculate the light distribution. The calculation formula is as follows: ， in, Represents a three-dimensional spatial position The amplitude of the light wave at is the initial amplitude of the light, indicating the energy intensity of the light source; is the phase factor of the light wave, which describes the propagation of the light wave in space and time; is the wave vector of light, which describes the propagation direction of light in different media; is the angular frequency of light, reflecting the time evolution characteristics of light waves; is the time variable.

3. The high-efficiency 3D animation production method based on artificial intelligence according to claim 1, characterized in that: The calculation of the illumination distribution in step 2 uses the Monte Carlo integration method to calculate the photon state, and the calculation formula is as follows: ， in, is the light intensity at position r in three-dimensional space; is the Monte Carlo sampling number, which indicates the number of samples used when calculating lighting; For photons at position Contribution factor at For photons at position The probability distribution at .

4. The high-efficiency 3D animation production method based on artificial intelligence according to claim 1, characterized in that: In step 3, the neural network optimizes the illumination data using a convolutional neural network to extract illumination features and calculate the optimization error of the illumination data. The loss function is as follows: ， in, Optimizing the loss function for lighting calculations for neural networks; is the number of training samples; Light intensity calculated for the neural network; is the real light intensity; is the spatial location of the training sample.

5. The high-efficiency 3D animation production method based on artificial intelligence according to claim 1 is characterized in that: In step 4, the light path optimization uses the Markov chain Monte Carlo method to adjust the light propagation path. The probability of the photon path is calculated as follows: ， in, For photons Spread to probability; is the current photon position, indicating the propagation point of the photon in space; The next photon propagation bit for the candidate; is the probability of acceptance, calculated as follows: ， in, Candidate new path points probability; Current path point probability.

6. The high-efficiency 3D animation production method based on artificial intelligence according to claim 1, characterized in that: In step 6, when rendering the animation scene, the adaptive light mapping method is used to adjust the light reflectivity of the material. The calculation formula is as follows: ， in, is the wavelength The material light reflectivity at Reference wavelength Reflectivity at is the reflectivity variation coefficient of the material; is the wavelength of the current light; is the reference wavelength.

7. The high-efficiency 3D animation production method based on artificial intelligence according to claim 1, characterized in that: In step 6, during the animation scene rendering process, the shadow distribution of the object is adjusted based on the lighting direction. The shadow generation function is calculated as follows: ， in, is the three-dimensional space position The shadow intensity at is the spatial position in the three-dimensional scene, indicating the coordinates of the point where the shadow needs to be calculated; is the incident angle of light; To calculate the influence of the angle between the object surface and the light source direction.

8. The high-efficiency 3D animation production method based on artificial intelligence according to claim 1 is characterized in that: In the animation scene rendering process, step 6 uses high dynamic range to adjust the light brightness in order to enhance the light and shadow details. The calculation formula is as follows: ， in, Lighting brightness adjusted for high dynamic range; The original calculated light intensity.

Images