A material modeling system and method based on PBR materials

By using a PBR-based material modeling system, combined with bionics and topology optimization algorithms, a closed-loop design from target color to microscopic geometry was achieved, solving the problems of low design efficiency and insufficient accuracy in structural color modeling, and providing a high-precision structural color rendering solution.

CN122199765BActive Publication Date: 2026-11-10SHANGHAI SIKAITUO ROBOT TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610272351.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-03-06
Publication Date
2026-11-10
Estimated Expiration
2046-03-06

AI Technical Summary

Technical Problem

Existing technologies struggle to efficiently and accurately simulate the interference and diffraction effects of light at the microscale when modeling structural colors, resulting in deviations between rendering effects and real physical phenomena. Furthermore, the design threshold is high, making it impossible to establish an inverse mapping relationship between color representation and microscopic geometric structure.

Method used

A material modeling system based on PBR material is adopted, combined with a bionic prior knowledge base and topology optimization algorithm. Through target spectrum definition unit, bionic microstructure generation unit, wave optics solving unit and material output unit, a closed-loop design process from target color representation to micro-geometric structure is realized. The wave optics solver is used to accurately simulate light propagation and generate a reflection spectrum response that conforms to the laws of wave optics. A manufacturing feasibility assessment module is introduced to ensure the feasibility of microstructure.

Benefits of technology

It significantly lowers the design threshold for complex structural color materials, accurately reproduces rainbow colors and holographic interference phenomena that change with viewing angle, enhances the physical realism and visual expressiveness of the rendering results, and ensures that the generated microstructures have practical processing possibilities. It is suitable for high-end industrial design, anti-counterfeiting labels, virtual reality and digital twins and other fields.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122199765B_ABST
    Figure CN122199765B_ABST
Patent Text Reader

Abstract

The present application relates to a kind of material modeling system and method based on PBR material, belong to computer graphics and three-dimensional modeling technical field.The system includes target spectrum definition unit, bionic microstructure generation unit, wave optics solving unit and material output unit: through the target spectrum reflectivity defined by user, in combination with biological micro-nano structure template in the prior knowledge base of bionics, using topological optimization algorithm, the micro-geometric structure that can realize target optical response is reversely generated;Then, the wave optics simulation of microstructure is carried out by finite-difference time-domain method, and the diffraction, interference and scattering effects thereof are accurately calculated;Finally, the verified microstructure and its perspective-dependent reflection data are packaged as standard PBR material format output.The present application integrates bionic design, physical simulation and manufacturing constraints, realizes high-fidelity, manufacturable structural color material automatic modeling, and improves the authenticity and engineering practicability of rendering.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of computer-aided design and 3D modeling technology, specifically relating to a material modeling system and method based on PBR materials. Background Technology

[0002] With the continuous evolution of computer graphics and realistic rendering technologies, physically based rendering has become a primary means of constructing highly realistic visual effects in digital media, industrial design, and virtual reality. This technology simulates the photophysical properties of the real world, such as energy conservation, surface roughness, and metallicity, enabling virtual objects to exhibit consistent and realistic material textures under different lighting conditions. Material modeling, as a key step in enhancing the realism of 3D scenes, focuses on accurately describing the interaction between light and object surfaces through mathematical models, thereby achieving a high degree of fidelity to complex physical phenomena.

[0003] Structural color material modeling is an important research direction in the field of advanced realistic rendering. It primarily studies optical phenomena such as thin-film interference, diffraction, and scattering caused by the micro- and nano-structures of object surfaces. Unlike traditional color-coding mechanisms that rely on chemical pigments absorbing light energy, structural color materials modulate light waves through the periodic or quasi-periodic micro-geometric morphology of their surfaces, producing unique visual effects such as iridescent and holographic colors that change with the viewing angle or direction of illumination. Achieving high-precision structural color modeling requires not only a system with complex micro-geometric representation capabilities but also precise simulation of wave-optical-level energy distribution at the rendering level.

[0004] Current technologies for structural color modeling primarily rely on semi-empirical anisotropic specular models or pre-defined texture maps for approximate simulation. This results in visual effects lacking a rigorous physical basis and failing to realistically reproduce dynamic spectral distributions with high angle dependence. Because the design of micro / nano structures involves complex optical and physical constraints, traditional manual parameter adjustment methods cannot intuitively translate specific target color requirements into corresponding microscopic geometric shapes, leading to extremely low modeling efficiency and excessively high design barriers for complex materials. Conventional material generation schemes are mostly based on forward generation logic, failing to establish an inverse mapping relationship between color representation and microscopic topology. Furthermore, the lack of an effective wave optics solution mechanism makes it difficult to capture the interference and diffraction effects of light at the microscale, causing the final rendering effect to deviate from real physical phenomena.

[0005] Therefore, a material modeling system and method based on PBR materials is desired. Summary of the Invention

[0006] The purpose of this invention is to provide a material modeling system based on PBR materials, which can solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A material modeling system based on PBR materials includes a target spectrum definition unit, a biomimetic microstructure generation unit, a wave optics solution unit, and a material output unit, wherein: The target spectral definition unit is configured to receive the target color performance requirements input by the user and convert them into target spectral reflectance data under corresponding viewing angle and lighting conditions. The biomimetic microstructure generation unit is configured to, based on the target spectral reflectance data and combined with the natural structural color biological micro-nano morphology features stored in the biomimetic prior knowledge base, reverse derive a microscopic geometric structure model capable of generating the target spectral reflectance through a topology optimization algorithm. The wave optics solving unit is configured to perform optical propagation physics simulation on the microscopic geometric structure model. It uses the finite-difference time-domain method to solve Maxwell's equations to accurately calculate the diffraction, interference and scattering effects of incident light on the microstructure surface and generate a reflection spectrum response that conforms to the laws of wave optics. The material output unit is configured to encapsulate the verified micro-geometric model and its corresponding optical response data into a standard PBR material format and output it to the rendering engine for real-time or offline rendering.

[0008] Preferably, the biomimetic prior knowledge base stores a variety of micro-nano structure templates of naturally occurring structurally colored organisms, including the multilayer thin film structure of butterfly scales, the two-dimensional photonic crystal array of peacock feathers, and the spiral Bragg reflector of beetle shells. When performing topology optimization, the biomimetic microstructure generation unit uses the micro-nano structure templates as the initial topological shape or constraint conditions to improve optimization efficiency and physical rationality.

[0009] Furthermore, the topology optimization algorithm adopts an iterative strategy that combines gradient descent and the adjoint field method. It uses the difference between the target spectral reflectance and the current simulated reflectance of the microstructure as the loss function. By continuously adjusting the material distribution density inside the microstructure, the simulated reflectance gradually approaches the target spectral reflectance while meeting the manufacturing feasibility constraints.

[0010] Furthermore, when performing light propagation simulation, the wave optics solving unit models the incident light as an electromagnetic wave containing a complete polarization state and time-frequency characteristics, and discretizes the time evolution process of the electric and magnetic field components on a three-dimensional spatial grid, thereby accurately capturing the phase change and energy redistribution of the light wave at the microstructure scale.

[0011] Preferably, the target spectral definition unit supports joint definition under multiple viewing angles and multiple light source conditions. Users can specify the desired color effect under a specific combination of viewing angle and illumination direction. The system generates a target spectral reflectance set with angle dependence to drive the reverse design of subsequent microstructures.

[0012] Furthermore, the PBR material generated by the material output unit includes a voxelized representation of the micro-geometry or a parametric surface description, and is accompanied by a view-dependent reflectance lookup table calculated by the wave optics solving unit, so that the rendering engine can reproduce high-fidelity structural color visual effects without running wave optics simulation in real time.

[0013] Furthermore, the system also includes a manufacturing feasibility assessment module, which is configured to detect the minimum feature size, overhang angle and material continuity of the generated microstructure in real time during the topology optimization process, and correct the optimization path according to the preset manufacturing process constraints to ensure that the final output microstructure can be realized under existing micro-nano processing technologies such as nanoimprinting, two-photon polymerization or electron beam lithography.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The material modeling system based on PBR material provided by this invention breaks through the limitations of traditional pigment absorption and empirical specular models. For the first time, it deeply integrates bionic principles, inverse topology optimization and wave optical physics simulation, realizing a closed-loop design process from "defining color" to "automatically generating physically realizable microstructures". 2. By constructing a priori models of bio-inspired micro-nano structures, the system significantly reduces the design threshold for complex structural color materials; by using a rigorous wave optics solver to replace the geometric optics approximation, it can accurately reproduce real physical phenomena such as rainbow colors and holographic interference that change with the viewing angle, thereby improving the physical realism and visual expressiveness of the rendering results. 3. By introducing a manufacturing feasibility constraint mechanism, it is ensured that the generated microstructures not only meet the target requirements optically, but also have the possibility of actual processing, providing a new generation of material modeling solutions that combine scientific rigor and engineering practicality for high-end industrial design, anti-counterfeiting labels, virtual reality and digital twins. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall technical solution architecture of the present invention; Figure 2 This is a schematic diagram of the core principle framework of reverse design of micro-geometric structures based on biomimetic prior knowledge and topology optimization in this invention. Figure 3 This is a flowchart illustrating the logical process of wave optics solution and spectral response calculation based on the finite-difference time-domain method in this invention. Figure 4 This is a schematic diagram of the multi-level interaction relationship and data flow between the material output unit and the rendering engine in this invention; Figure 5 This is a flowchart illustrating the logic framework for microstructure optimization path correction in conjunction with manufacturing process constraints in this invention. (Detailed implementation details follow.) Example 1: Please refer to the appendix Figure 1 To be continued Figure 5 To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0016] A material modeling system based on PBR material includes a target spectrum definition unit, a biomimetic microstructure generation unit, a wave optics solution unit, a manufacturing feasibility assessment module, and a material output unit. The target spectral definition unit receives the user's input target color performance requirements and converts them into target spectral reflectance data under corresponding viewing angles and lighting conditions. The target spectral definition unit is configured to provide an interactive multi-dimensional parameter input interface, allowing users to define the optical properties they wish to exhibit on a physical surface using digital coordinates or spectral curves. The target color performance requirements are not limited to a single color value but include a dynamic set of color descriptions that vary with the viewing angle and the angle of incident light. The target spectral definition unit includes a spectral discretization subunit, a multi-view parameter configuration subunit, and an ambient lighting simulation subunit.

[0017] The spectral discretization subunit is configured to divide the visible light band according to a preset wavelength interval, generating a series of discrete spectral sampling points. In one specific implementation, the spectral discretization subunit resamples the visible light range from 380 nm to 780 nm with an accuracy of 1 nm or 5 nm, thereby constructing a high-dimensional spectral vector. The multi-view parameter configuration subunit is used to define the relative geometric relationship between the observer and the surface to be modeled, including the range of variation of the zenith angle and azimuth angle, ensuring that the defined target color has angle dependence. The ambient lighting simulation subunit is used to set the energy spectrum distribution of the incident light, such as simulating a standard D65 light source, an LED light source with a specific color temperature, or sunlight with a continuous energy spectrum, so that the generated reflectivity data can reflect the color performance of the material in a real physical environment.

[0018] The biomimetic microstructure generation unit, connected to the target spectrum definition unit, is used to derive, through a topology optimization algorithm, a microscopic geometric structure model capable of generating the target spectral reflectance, based on the target spectral reflectance data and combined with the structural color biological micro / nano morphological features stored in the biomimetic prior knowledge base. The biomimetic microstructure generation unit is the core logic component of the system, responsible for constructing a geometric topology capable of modulating the phase of light waves at the sub-micrometer scale.

[0019] The biomimetic microstructure generation unit integrates a biomimetic prior knowledge base subunit. This prior knowledge base stores various micro- and nano-structure templates from naturally occurring structurally colored organisms. These templates are typical physical morphological descriptions extracted from butterfly scales, peacock feathers, beetle shells, and marine organism epidermis. Specifically, the templates include, but are not limited to, multilayer thin-film interference structures of butterfly scales, two-dimensional photonic crystal arrays of peacock feathers, spiral Bragg reflectors of beetle shells, and light-trapping structures with random scattering characteristics. When performing topology optimization, the biomimetic microstructure generation unit does not start from a completely disordered state, but rather uses the micro- and nano-structure templates as initial topological forms or spatial constraints. This approach reduces the optimization search space, improves the algorithm's convergence speed, and ensures the physical rationality of the generated geometric structures.

[0020] The biomimetic microstructure generation unit also includes a topology optimization engine subunit. This subunit is configured to employ a gradient-based iterative search strategy. The engine utilizes a mechanism combining gradient descent and the adjoint field method, defining the sum of squared residuals between the target spectral reflectance and the reflectance obtained from the current microstructure simulation as the loss function. In each iteration, the topology optimization engine subunit calculates the sensitivity information of the material density at each spatial voxel point within the microstructure to the loss function. This sensitivity information is obtained by solving the overlap integral of the forward wave equation and the adjoint wave equation, thereby accurately guiding the redistribution of material density. By continuously adjusting the material distribution density within the microstructure, while satisfying preset material volume fraction constraints and spatial periodicity constraints, the simulated reflectance is guided to gradually approach the target spectral reflectance.

[0021] The wave optics solving unit is connected to the biomimetic microstructure generation unit and the material output unit. It is used to perform light propagation physics simulation on the micro-geometric structure model, and uses the finite-difference time-domain method to solve Maxwell's equations to accurately calculate the diffraction, interference and scattering effects of incident light on the microstructure surface, and generate a reflection spectrum response that conforms to the laws of wave optics.

[0022] The wave optics solution unit includes an electromagnetic field discretization subunit, a time evolution iteration subunit, and a frequency domain transformation analysis subunit. The electromagnetic field discretization subunit is configured using Yee mesh generation technology to divide the computational domain containing the microscopic geometric model into numerous cubic elements, defining electric and magnetic field components on the edges and faces of each element. This staggered mesh arrangement ensures the second-order central difference accuracy of the wave equation in the discrete space. The time evolution iteration subunit, based on Maxwell's curl equation, alternately updates the electric and magnetic fields along the time axis. To ensure numerical stability during computation, the time evolution iteration subunit is configured such that the time step satisfies the Courant stability condition, meaning the time step size must be less than a preset ratio of the spatial mesh size divided by the speed of light.

[0023] During the simulation, the wave optics solver modeled the incident light as an electromagnetic wave pulse containing complete polarization states and time-frequency characteristics. By tracking the temporal evolution of the electric and magnetic field components in real time on a three-dimensional spatial grid, the system can capture the phase lag, multipath reflection, and interference superposition that occur when the light wave passes through micro- and nano-scale geometries. The frequency domain transformation analysis subunit performs a discrete Fourier transform on the recorded time-domain electromagnetic signal, converting the time-domain data into a complex field distribution at a specific wavelength, and then calculates the far-field optical power distribution and anisotropic reflection coefficient to form the final reflection spectral response data.

[0024] The manufacturing feasibility assessment module, integrated into the iterative loop of the biomimetic microstructure generation unit, is used to detect the minimum feature size, overhang angle, and material continuity of the generated microstructure in real time during topology optimization, and to correct the optimization path based on preset manufacturing process constraints. The manufacturing feasibility assessment module includes a process parameter matcher, which incorporates technical limitations for different micro / nano fabrication processes such as nanoimprint lithography, electron beam lithography, and two-photon polymerization printing. For example, when the system detects a narrow slit smaller than 20 nanometers or an excessively large overhang ratio within the generated microstructure, the module generates a penalty term added to the loss function of topology optimization, prompting the optimization direction to shift towards a form more easily manufactured in practice. This ensures that the final output microstructure model not only meets optical performance standards but also possesses the potential for large-scale production.

[0025] The material output unit encapsulates the verified microscopic geometric model and its corresponding optical response data into a standard PBR material format and outputs it to the rendering engine for real-time or offline rendering. The material output unit includes a data encapsulation subunit and a rendering adaptation subunit. The data encapsulation subunit transforms complex wave optics calculations into an efficient lookup table structure, which records the mapping relationship between different viewing angles, incident light angles, and reflectivity intensity. This subunit also stores the microscopic geometry in the form of voxelized density distribution maps or parametric surfaces. The rendering adaptation subunit integrates this data into a physically based rendering shader. By querying the pre-calculated spectral lookup table during real-time rendering, it enables the reproduction of high-fidelity structural color materials with rainbow colors and holographic dynamic effects without requiring real-time wave optics simulation.

[0026] Example 2: Building upon Example 1, Example 2 provides an implementation of a material modeling system based on a distributed architecture. In this example, the system is deployed in a cluster environment composed of multiple high-performance computing nodes to handle the enormous computational overhead required for large-scale, high-resolution microstructure modeling.

[0027] In Example 2, the wave optics solving unit employs a distributed strategy combining spatial decomposition and task parallelism. The electromagnetic field discretization subunit divides the microscopic geometric region to be simulated into multiple subdomains, each assigned to an independent computing core within the cluster for processing. The computing cores exchange boundary layer data in real time via a message passing interface to ensure the continuity of electromagnetic wave propagation at the subdomain boundaries. This distributed design shortens the simulation time per iteration, enabling the system to handle material modeling tasks with larger physical dimensions or more complex topological features.

[0028] In Example 2, the biomimetic microstructure generation unit introduces a heterogeneous acceleration mechanism. The topology optimization engine subunit is configured to utilize a graphics processing unit (GPU) cluster for sensitivity analysis and material density update calculations. Since topology optimization involves a large number of parallel tensor operations, the streaming architecture of the GPU can achieve a speedup several times that of traditional general-purpose processors. The biomimetic prior knowledge base subunit is constructed as a globally shared high-speed cache database, supporting concurrent access by multiple optimization tasks, thereby improving the design collaboration efficiency across material schemes.

[0029] Furthermore, the target spectral definition unit in Example 2 supports real-time data acquisition from physical samples via a spectral sensor. This unit, connected to an external spectral scanner, can directly acquire the angular-related reflectance distribution of real biological specimens or existing structural color materials, using it as input for reverse engineering. This "physically driven" approach enables the system to accurately clone complex optical phenomena existing in nature.

[0030] In Embodiment 2, the material output unit adds a multi-platform format conversion module. To adapt to the special shading models of different mainstream rendering engines (such as Unreal Engine, Unity, or cinematic offline renderers), the multi-platform format conversion module can automatically remap the reflectivity data generated by wave optics to the specific parameter slots of each engine, including but not limited to base color, metallicity, roughness, and transparent coating parameters specifically used to describe interference effects.

[0031] Example 2 also includes a performance monitoring and automated load balancing module. This module monitors the memory usage, processor load, and network latency of each computing node in the cluster in real time, and dynamically adjusts the allocation of computing resources according to the complexity of the current topology optimization task to ensure that the high-precision material generation task is completed within a preset time period.

[0032] Example 3: Based on Example 1 or 2, Example 3 describes a material modeling system with an intelligent acceleration prediction mechanism. This system introduces a deep learning proxy model to replace part of the high-load wave optical physics simulation.

[0033] In Example 3, the wave optics solving unit and the neural network proxy computing module operate in parallel. The neural network proxy computing module is configured as a regression model based on a convolutional neural network or a graph neural network. During the offline training phase of the system, the wave optics solving unit generates a massive number of "microstructure-spectral response" pairs as training samples. Through deep learning training, the neural network proxy computing module masters the complex mapping law between microscopic geometric topology and macroscopic optical performance.

[0034] During the online design phase, when the biomimetic microstructure generation unit submits a new candidate geometry, the neural network proxy computing module first performs rapid inference to estimate the optical reflectivity of the structure. Only when the estimated result is close to the target spectral requirement and the confidence level exceeds a preset threshold will a high-precision wave optics solving unit be activated for physical verification. This dual-loop prediction mechanism reduces unnecessary physical simulations, enabling the system to maintain high physical accuracy while possessing extremely high response speed.

[0035] In Example 3, the biomimetic microstructure generation unit further integrates an adaptive mesh refinement function. The topology optimization engine subunit uses a coarser mesh for rapid morphological evolution in the early stages of iteration. When the loss function decreases gradually and structural details begin to appear, the system automatically triggers a mesh refinement process, increasing the mesh density in areas with large electromagnetic field gradients, such as material interfaces, thereby optimizing the overall computational resource consumption while ensuring local accuracy.

[0036] The manufacturing feasibility assessment module in Example 3 adds support for multi-material composite processes. This module is configured to handle the spatial arrangement of multiple media materials with different refractive indices within the same microstructure and to evaluate the interfacial adhesion and thermal expansion coefficient matching between different materials. By introducing multiphysics constraints into the material modeling process, this system can generate more robust and mass-producible complex photonic crystal material schemes.

[0037] The material output unit supports real-time generation logic for holographic textures. Based on the generated microstructure model, this unit can automatically calculate the corresponding spatial light modulator or nanoimprint mask graphic file. This means that the system outputs not only rendering materials for the virtual world, but also production guidance documents that can be directly used in micro / nano fabrication equipment, truly achieving a closed-loop transformation from digital design to physical manufacturing.

[0038] Example 4: Based on any one of Examples 1 to 3, Example 4 describes a lightweight material rendering system optimized for mobile terminals or augmented reality devices. In this example, material generation is completed on a high-performance server in the cloud, while its rendering and reproduction are specially reconstructed for edge devices with limited computing power.

[0039] In Example 4, the material output unit employs a compression encoding strategy based on spherical harmonic functions or radial basis functions. The high-dimensional view-related spectral data generated by the wave optics solving unit is compressed into a compact set of basis function coefficients. This encoding method significantly reduces data storage capacity while allowing mobile renderers to quickly reconstruct complex rainbow-colored visual effects with minimal floating-point operations during the pixel shading stage.

[0040] The rendering adaptation subunit is implemented as a lightweight shader on mobile devices. This shader is configured with adaptive level-of-detail logic, dynamically switching between different precision spectral lookup tables based on the distance between the virtual camera and the object's surface. For close-up viewing, the system uses high-resolution textures to represent the fine interference fringes of micro / nano structures; while for distant viewing, it smoothly transitions to an approximate calculation based on average reflectivity, thus maintaining a stable rendering frame rate on mobile devices.

[0041] The system also includes a cloud-based interactive interface. Users can select a target hue or upload a photo of a natural organism through a simple interface on their mobile device. Upon receiving the request, the cloud system immediately executes a biomimetic topology optimization and wave optics solution process, and pushes the generated compressed material package to the terminal device in real time. This architecture combines the powerful physical computing capabilities of the cloud with a smooth interactive experience at the edge.

[0042] Furthermore, the target spectral definition unit in Example 4 incorporates a weighting mechanism based on visual perception. When translating user requirements into target spectral reflectance, the system references the sensitivity curves of the human eye to different wavelengths of light, assigning higher weights and finer sampling frequencies to wavelengths sensitive to human vision. This ensures that, within limited computational bandwidth, the final generated material visually better aligns with human aesthetic expectations, enhancing the user's immersion in the virtual environment.

[0043] Example 5: Based on the above examples, this Example 5 describes in detail the specific logical implementation of the deep integration of the bionic microstructure generation unit with the bionic prior knowledge base and the topology optimization algorithm.

[0044] The biomimetic prior knowledge base subunit not only stores geometric shapes but also indexes of optical physical properties related to specific biological microstructures. When the target spectrum definition unit outputs a narrowband reflection requirement with strong angular selectivity, the biomimetic microstructure generation unit automatically searches the knowledge base and identifies helical or layered structures with Bragg reflection characteristics as the optimization basis. This feature-matching-based heuristic initiation mechanism avoids blind searching in the solution space for topology optimization.

[0045] When performing iterative updates, the topology optimization engine subunit introduces a physical term called "spatial consistency constraint." This constraint term is defined as the spatial integral of the material density field gradient, and its function is to suppress isolated floating points and jagged edges generated during the optimization process. By adjusting the weight coefficient of this constraint term, the biomimetic microstructure generation unit can control the smoothness of the generated geometry, thereby adapting to different resolution processing techniques.

[0046] In wave propagation simulation, the wave optics solver unit in Example 5 introduces a perfectly matched layer (PML) absorbing boundary condition. The electromagnetic field discretization subunit has multiple absorbing layers with imaginary refractive indices at the edges of the computational region, effectively absorbing outward-propagating electromagnetic waves and preventing spurious reflection interference at the simulation boundaries, thereby improving the accuracy of far-field diffraction mode calculations.

[0047] In Embodiment 5, the material output unit also provides a material aging simulation module. Based on the physical properties of micro- and nano-structures, this module can simulate the effects of environmental factors (such as refractive index drift caused by ultraviolet radiation and structural collapse caused by physical wear) on optical performance. By adjusting the parametric description of the microstructure in real time in the rendering engine, the system can reproduce the physical evolution of structural color materials during long-term use, which has extremely important application value for digital twins and high-fidelity product displays.

[0048] Example 6: Building upon the above examples, Example 6 focuses on how the system handles highly complex non-periodic random structured color modeling. Unlike regularly occurring photonic crystals, many soft metallic colors in nature originate from quasi-randomly arranged nanoparticles.

[0049] In Example 6, the biomimetic microstructure generation unit integrates a probability distribution generator. This generator is configured to generate a series of nanoparticle position and size parameters that conform to a specific statistical distribution. The optimization objective of the topology optimization engine subunit is no longer a single geometric coordinate, but rather the parameters of the statistical distribution function, such as the mean and standard deviation of the particle size distribution, and the coherence length of the spatial correlation function.

[0050] The wave optics solution unit uses the Monte Carlo method to perform parallel simulations on multiple sets of random structure instances conforming to statistical distributions, and takes their average power spectrum as the final optical response. To accelerate this process, the system utilizes the principle of linear superposition of electromagnetic scattering, simplifying the computational complexity of multiple scattering under specific conditions.

[0051] The manufacturing feasibility assessment module provides compatibility testing for particle self-assembly processes for such random structures. This module can evaluate whether the currently optimized statistical parameters can be achieved through bottom-up processing methods such as sol-gel, electrochemical deposition, or particle coating. This allows the material solutions generated by this system to be extended beyond precision photolithography to the more cost-effective field of chemical coatings.

[0052] In Example 6, the material output unit introduces a random phase noise map to simulate the diffuse scattering effect generated by random structures in the rendering engine. This results in a macroscopically soft, silky-textured surface rather than a strong specular interference rainbow color, enriching the system's ability to model and cover different textured structural colors.

[0053] Example 7: Based on any one of Examples 1 to 6, Example 7 details the internal logic of the system when processing multi-layer composite structure color modeling.

[0054] In some advanced applications, the material may be composed of multiple layers of micro / nanostructures with different optical functions, such as a bottom layer being a highly reflective mirror, a middle layer being a color interference layer, and a top layer being an anti-reflection or protective layer. In Example 7, the biomimetic microstructure generation unit is configured to support a multi-layer independent optimization mechanism. The topology optimization engine subunit can assign different material properties and optimization objectives to each layer, while simultaneously considering the near-field coupling effect between layers.

[0055] The wave optics solver employs a hierarchical divide-and-conquer strategy when handling this multi-layered structure. For the fine intra-layer structure, the full-wave FDTD algorithm is used to capture complex local diffraction; while for long-distance light propagation between layers, a method based on the scattering matrix or transmission matrix is ​​used for fast analytical calculation. This hybrid simulation mode improves computational efficiency by an order of magnitude while maintaining physical accuracy.

[0056] The manufacturing feasibility assessment module in Example 7 adds an interlayer alignment tolerance assessment function. This module simulates translational, rotational, or tilting errors that may occur during actual layer stacking and analyzes the sensitivity of these errors to the final visual effect. If a certain structural scheme is found to have excessively stringent requirements for processing accuracy, the system will automatically adjust the structure through topology optimization to find an alternative with higher tolerance.

[0057] The material output unit generates composite shader code for this multi-layered material, allowing independent control of the optical parameters of each layer within the rendering engine. For example, users can adjust the matte finish of the top protective layer in post-processing without disrupting the spectral colors generated by the underlying interference structure. This modular material output method provides artists and designers with creative freedom.

[0058] Example 8: Based on the above examples, Example 8 further describes how the system handles polarization-sensitive structural color modeling.

[0059] Many organisms with biological structures (such as certain scarab beetles) can produce circularly polarized reflected light with specific chirality. In Example 8, the target spectrum definition unit adds a polarization state definition parameter, allowing the user to specify the Stokes vector of the reflected light. Based on this, the biomimetic microstructure generation unit generates micro / nano structures with chiral helical features through topology optimization, such as simulating the helical arrangement of a cholesterol liquid crystal phase.

[0060] The wave optics solving unit adopts a full vector calculation mode during simulation, tracks the rotation trajectory of the electric field vector in three-dimensional space in real time, and accurately calculates the reflectivity difference (i.e., circular dichroism) of left and right circularly polarized light.

[0061] The material output unit encapsulates this polarization characteristic into a specialized polarization parameter atlas. In engines that support polarization rendering, this material can realistically reproduce the brightness and color flickering phenomena that occur when viewed through a polarizer. This feature has significant application value in the fields of anti-counterfeiting label design and high-security document production.

[0062] Example 9: Based on Examples 1 to 8, Example 9 describes the self-learning evolution mechanism of the system.

[0063] The system includes a closed-loop module for modeling result feedback. Whenever the system generates a successful material modeling scheme and obtains user approval, the scheme's microstructure topological features, target spectral data, and final rendering performance indicators are stored as a new knowledge entry in the biomimetic prior knowledge base.

[0064] As usage time increases, the biomimetic prior knowledge base continuously evolves from a purely biological template library into a hybrid enhancement library containing numerous artificially optimized solutions. When faced with new modeling tasks, the topology optimization engine subunit utilizes an analogical reasoning algorithm to find the most similar successful cases from the enhancement library as a starting point. This evolutionary mechanism endows the system with the technical characteristics of "getting smarter and faster with use," enabling it to continuously adapt to increasingly complex aesthetic and industrial design needs.

[0065] Example 10: This Example 10 summarizes and details the hardware architecture and software stack configuration of the system of the present invention in specific engineering deployment, so as to ensure the stable implementation of the functions of each unit.

[0066] The system logically runs on a high-performance computing platform containing a cluster of servers equipped with high-capacity graphics processors. The servers are interconnected via 10 Gigabit Ethernet or Infiniband technology to support large-scale distributed parallel computing of the wave optics solver.

[0067] At the software level, the biomimetic microstructure generation unit adopts a hybrid programming model. Its underlying topology optimization algorithm is implemented using the efficient C++ language and calls a highly optimized mathematical computing library for linear algebra operations. The upper-level logic control and user interface are built using a high-level scripting language to provide flexible scalability. The wave optics solving unit utilizes the Unified Computing Device Architecture (CUDA) to achieve deep access to the graphics processor, distributing the computation of hundreds of millions of grid points across tens of thousands of processing cores for parallel execution.

[0068] The material output unit and the rendering engine interact with each other through a standard material exchange format (such as USD or MDL), ensuring that the materials generated by the system can be seamlessly transferred in mainstream 3D creation and digital twin platforms worldwide.

[0069] This invention provides a PBR-based material modeling system that breaks through the limitations of traditional pigment absorption and empirical specular models. For the first time, it deeply integrates bionic principles, inverse topology optimization, and wave optics physical simulation, achieving a closed-loop design process from "defining color" to "automatically generating physically realizable microstructures." The system lowers the design threshold for complex structural color materials by constructing bio-inspired micro / nano structure prior models; it accurately reproduces real physical phenomena such as rainbow colors and holographic interference that change with viewing angle by using a rigorous wave optics solver to replace geometric optics approximations, enhancing the physical realism and visual expressiveness of the rendering results; and it ensures that the generated microstructures not only meet optical requirements but also possess the potential for actual fabrication by introducing a manufacturing feasibility constraint mechanism. This provides a new generation of material modeling solutions that combine scientific rigor and engineering practicality for high-end industrial design, anti-counterfeiting, virtual reality, and digital twins.

[0070] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention. Various variations and improvements can be made by those skilled in the art without departing from the spirit of the present invention, and these variations and improvements should also be considered within the scope of protection of the present invention.

Claims

1. A material modeling system based on PBR materials, characterized in that, The system includes: The target spectral definition unit is configured to receive the target color performance requirements input by the user and convert them into target spectral reflectance data under the corresponding viewing angle and lighting conditions. A biomimetic microstructure generation unit is configured to, based on the target spectral reflectance data and combined with the structural color biological micro-nano morphology features stored in the biomimetic prior knowledge base, reverse derive a micro-geometric structure model capable of generating the target spectral reflectance through a topology optimization algorithm. The wave optics solution unit is configured to perform physical simulation of light propagation on the microscopic geometric structure model. It uses the finite-difference time-domain method to solve Maxwell's equations to accurately calculate the diffraction, interference and scattering effects of incident light on the microstructure surface and generate a reflection spectrum response that conforms to the laws of wave optics. The material output unit is configured to encapsulate the validated micro-geometry model and its corresponding optical response data into a standard PBR material format and output it to the rendering engine for real-time or offline rendering. The biomimetic microstructure generation unit is configured to support multi-layer composite structure modeling, the multi-layer composite structure including a bottom reflective mirror layer, an intermediate color interference layer and a top protective layer; the biomimetic microstructure generation unit also includes a topology optimization engine subunit. The topology optimization engine sub-unit configuration is used to assign different material properties and optimization objectives to each layer, while also considering the near-field coupling effect between layers. The wave optics solution unit is configured to use a hierarchical divide-and-conquer strategy when processing the multilayer composite structure. For the microstructure within the layer, the full-wave time-domain finite difference algorithm is used, while for long-distance light propagation between layers, the analytical calculation method based on the transmission matrix is ​​switched. The target spectrum definition unit is also configured to define the Stokes vector of the reflected light to describe the polarization state. The biomimetic microstructure generation unit generates micro-nano structures with chiral helical characteristics accordingly. The wave optics solution unit is configured to use a full vector calculation mode to track the rotation trajectory of the electric field vector in three-dimensional space and calculate the reflectivity difference of circularly polarized light.

2. The material modeling system based on PBR material according to claim 1, characterized in that, The target spectrum definition unit includes a spectrum discretization subunit, a multi-view parameter configuration subunit, and an ambient lighting simulation subunit. The spectral discretization subunit is configured to divide the visible light band according to a preset wavelength interval to generate a series of discrete spectral sampling points, wherein the visible light range is limited to between 380 nanometers and 780 nanometers, and is resampled with a preset nanometer precision to construct a high-dimensional spectral vector. The multi-view parameter configuration subunit is configured to define the relative geometric relationship between the observer and the surface to be modeled, including the range of variation of the zenith angle and azimuth angle, and to give the target color representation requirements an angle-dependent feature; The ambient light simulation subunit is configured to set the energy spectrum distribution of the incident light. By simulating a standard light source, a solid-state light source with a specific color temperature, or a natural light source with a continuous energy spectrum, the generated reflectivity data reflects the color performance of the material under a preset physical environment.

3. The material modeling system based on PBR material according to claim 2, characterized in that, The biomimetic microstructure generation unit integrates a biomimetic prior knowledge base subunit. The bionic prior knowledge base subunit stores a variety of micro-nano structure templates of natural structural color organisms, including a multilayer thin film interference structure template for butterfly scales, a two-dimensional photonic crystal array template for peacock feathers, a spiral Bragg reflector template for beetle shells, and a light-trapping structure template with random scattering characteristics. When performing topology optimization, the biomimetic microstructure generation unit is configured to retrieve the biomimetic prior knowledge base subunit, identify micro / nano structure templates that match the angular selectivity features in the target spectral reflectance data, and use the micro / nano structure templates as initial topological morphology or spatial constraints to narrow the optimization search space and ensure the physical rationality of the generated geometric structure.

4. The material modeling system based on PBR material according to claim 3, characterized in that, The topology optimization engine subunit is configured to execute a gradient-based iterative search strategy, employing a mechanism that combines gradient descent with the adjoint field method, and defining the sum of squared residuals between the target spectral reflectance and the reflectance obtained from the current microstructure simulation as the loss function. During the iteration process, the topology optimization engine sub-unit is configured to calculate the sensitivity information of the material density at each spatial voxel point inside the microstructure to the loss function. The sensitivity information is obtained by solving the overlapping integral of the forward wave equation and the adjoint wave equation, which is used to guide the redistribution of the material density. The topology optimization engine subunit is further configured to introduce a spatial consistency constraint term, which is defined as the spatial integral of the material density field gradient. This constraint term is used to suppress isolated floating points and jagged edges generated during the optimization process. Under the premise of satisfying the material volume fraction constraint and the spatial periodicity constraint, it guides the simulated reflectance to gradually approach the target spectral reflectance.

5. A material modeling system based on PBR material according to claim 4, characterized in that, The wave optics solution unit includes an electromagnetic field discretization subunit, a time evolution iteration subunit, and a frequency domain transformation analysis subunit. The electromagnetic field discretization sub-unit configuration is used to divide the computational region containing the microscopic geometric structure model into a large number of cubic units using staggered mesh generation technology, and to define electric field components and magnetic field components on the edges and faces of each unit respectively, so as to ensure the second-order central difference accuracy of the wave equation in the discrete space. The time evolution iteration sub-unit is configured to alternately advance the update of electric and magnetic fields on the time axis according to Maxwell's curl equation, and is configured to satisfy the Courant stability condition in terms of the size of the time step, that is, the size of the time step must be less than the preset ratio of the spatial grid size divided by the speed of light. The frequency domain transformation analysis subunit is configured to perform discrete Fourier transform on the recorded time-domain electromagnetic signal, transforming the time-domain data into a complex field distribution at a specific wavelength, and then calculating the optical power distribution and anisotropic reflection coefficient in the far field to form the final reflection spectral response data.

6. A material modeling system based on PBR material according to claim 5, characterized in that, The system also includes a manufacturing feasibility assessment module, which is integrated into the iterative loop of the biomimetic microstructure generation unit; The manufacturing feasibility assessment module includes a process parameter matcher, which has built-in technical limitation parameters for nanoimprint lithography, electron beam exposure, and two-photon polymerization printing processes. The manufacturing feasibility assessment module is configured to detect the minimum feature size, overhang angle, and material continuity of the generated microstructure in real time during the topology optimization process. When a narrow slit smaller than a preset threshold or a sag exceeding a preset proportion is detected inside the microstructure, the manufacturing feasibility assessment module is configured to generate a penalty term and add it to the loss function of topology optimization, prompting the optimization direction to shift to a direction that conforms to the actual manufacturing form, ensuring that the final output microstructure model has the possibility of large-scale production.

7. A material modeling system based on PBR material according to claim 6, characterized in that, The material output unit includes a data encapsulation subunit and a rendering adaptation subunit; The data encapsulation subunit is configured to convert the calculation results generated by the wave optics solving unit into a view-dependent reflectance lookup table. The view-dependent reflectance lookup table records the mapping relationship between different viewing angles, incident light angles and reflectance intensity, and stores the micro-geometry in the form of a voxelized density distribution map or a parametric surface. The rendering adaptation subunit is configured to integrate the view-dependent reflectivity lookup table into the physically based rendering shader, and reproduce structural color materials with rainbow colors or holographic dynamic effects by querying the view-dependent reflectivity lookup table during real-time rendering. The material output unit also includes a material aging simulation module, configured to simulate the effects of refractive index drift caused by ultraviolet irradiation or structural collapse caused by physical wear on optical performance based on the physical properties of micro-nano structures.

8. A material modeling system based on PBR material according to claim 7, characterized in that, The system is deployed based on a distributed architecture, and the wave optics solving unit is configured to adopt a distributed strategy that combines spatial decomposition and task parallelism. The electromagnetic field discretization sub-unit configuration is used to divide the micro-geometric structure region to be simulated into multiple subdomains. Each subdomain is assigned to an independent computing core in the cluster, and the computing cores exchange boundary layer data in real time through a message passing interface. The system also includes a neural network proxy computing module, which runs in parallel with the wave optics solving unit. The neural network proxy computing module is configured as a deep learning-based regression model, which is used to learn the mapping law between microscopic geometric topology and macroscopic optical performance during the offline training stage, and to quickly infer and predict the optical reflectivity of candidate geometric structures during the online design stage. When the estimated result is close to the target requirement and the confidence level exceeds the preset threshold, the system is configured to start the wave optics solving unit for physical verification.

9. A material modeling method based on PBR materials, characterized in that, Material modeling based on PBR materials as described in any one of claims 1-8.