Vehicle paint simulation method and system supporting dynamic forward and reverse bidirectional rendering

Abstract

The invention relates to the technical field of automobile coating, in particular to an automobile paint simulation method and system supporting dynamic forward and reverse two-way rendering, and the method comprises the steps: forward rendering: inputting automobile paint formula detail information, and analyzing paint layer rendering parameters and color master rendering parameters to output a rendering result in an automobile paint renderer; reverse rendering: inputting the original measured data, and performing reverse parameter analysis and parameter optimization to obtain vehicle paint physical parameters; forward and reverse fusion: taking the vehicle paint physical parameters as an analysis result of vehicle paint formula detail information newly input in forward rendering, and taking a rendering result as a prior reference in reverse rendering to constrain the range of the vehicle paint physical parameters; performing positive and negative consistency verification based on the rendering result and the original measured data to calculate a rendering deviation; and performing back propagation based on the rendering deviation to iteratively optimize the forward rendering and the reverse rendering until the rendering deviation is less than a threshold value. According to the invention, the precision and effect of vehicle paint rendering can be improved.

Description

Technical Field

[0001] This application relates to the technical field of automotive painting, and in particular to a method and system for simulating automotive paint that supports dynamic forward and reverse bidirectional rendering. Background Technology

[0002] With the surge in demand for personalized exterior designs in the automotive consumer market and the accelerated digital transformation of the automotive manufacturing industry, accurate simulation and efficient iteration of vehicle paint appearance have become core industry requirements. However, current vehicle paint rendering technologies still have significant limitations and are difficult to match the actual application scenarios of the automotive industry. The main problems are as follows:

[0003] The static limitations of existing renderers: Mainstream car paint renderers are centered on "static effect reproduction", relying on preset material templates or existing car paint effect libraries. They only support adjusting abstract material ball parameters (such as "highlight intensity" and "metallicity" and other graphics parameters), and cannot establish a direct relationship with the actual car paint formula details in the automotive industry (such as clear coat thickness, aluminum powder particle size distribution, color paint composition ratio and other specific parameters).

[0004] Disconnect from industrial processes: In the automotive industry, the core of paint research and color change design lies in the dynamic adjustment of formula parameters (such as increasing the proportion of aluminum powder by 5% or adjusting the thickness of clear coat), rather than modifying material parameters at the graphics level. Existing renderers lack mapping modeling between formula and optical properties, so they cannot automatically update the rendering effect when the formula changes. This requires manual readjustment, resulting in a break in the "formula adjustment - effect preview" closed loop, which seriously affects research and development efficiency.

[0005] Lack of dynamic responsiveness: Existing renderers cannot support interactive editing of recipe details and dynamic linkage of rendering effects. Requirements such as "adjusting color masterbatch ratio → previewing effect" in new vehicle color development and "customizing particle distribution → viewing iridescent effect" in personalized color changes cannot be met. The industry still needs to rely on physical sample verification, resulting in cost waste and contradicting the trend of digital R&D. Summary of the Invention

[0006] To improve the accuracy and effect of vehicle paint rendering, this application provides a vehicle paint simulation method and system that supports dynamic forward and inverse bidirectional rendering.

[0007] Firstly, this application provides a method for simulating vehicle paint that supports dynamic forward and inverse bidirectional rendering, employing the following technical solution:

[0008] A method for simulating vehicle paint that supports dynamic forward and inverse bidirectional rendering includes the following steps:

[0009] Forward rendering: Input the details of the paint formula and parse out the paint layer rendering parameters and color masterbatch rendering parameters to output the rendering results in the paint renderer;

[0010] Reverse rendering: Input the original measured data and perform reverse parameter parsing and optimization to obtain the physical parameters of the car paint;

[0011] Forward and reverse fusion: The physical parameters of the vehicle paint are used as the parsing result of the newly input vehicle paint formula details in the forward rendering, and the rendering result is used as the prior reference in the reverse rendering to constrain the range of the physical parameters of the vehicle paint;

[0012] Based on the rendering results and the original measured data, a forward and reverse consistency check is performed to calculate the rendering deviation.

[0013] Backpropagation is performed based on the rendering deviation to iteratively optimize the forward rendering and the reverse rendering until the rendering deviation is less than a threshold.

[0014] In some embodiments, resolving paint layer rendering parameters includes the following steps:

[0015] The detailed information of the vehicle paint formula is broken down into a physical layer structure that includes a primer layer, a color coat layer, a particle layer, and a clear coat layer;

[0016] Light transmission rules are modeled for each layer to assign independent rendering parameters for each layer, and the total reflectivity is calculated based on each of the rendering parameters for each layer, including reflectivity, refractive index, thickness, absorption coefficient, particle density, particle size distribution, and orientation angle.

[0017] The paint layer rendering parameters are generated based on the rendering parameters of each layer and the total reflectivity.

[0018] In some embodiments, parsing the master rendering parameters includes the following steps:

[0019] Based on the rendering parameters of each layer corresponding to the particle layer, the physical feature parameters of each type of particle color masterbatch in the particle layer are extracted;

[0020] The optical characteristics of the particulate colorant are calculated based on the physical characteristic parameters and a preset Mie scattering model.

[0021] The mapping relationship between the physical feature parameters and BSDF parameters is established based on a preset parametric regression algorithm;

[0022] The color master rendering parameters are analyzed based on the optical characteristics and the mapping relationship.

[0023] In some embodiments, the raw measured data includes raw multi-angle spectra and raw HDR images. Inputting the raw measured data and performing reverse parameter analysis and optimization to obtain the physical parameters of the vehicle paint includes the following steps:

[0024] The original multi-angle spectrum and the original HDR image are preprocessed.

[0025] The original measured data is input into a preset model based on machine learning algorithms to infer the material parameters and particle distribution characteristics of each layer corresponding to the original measured data.

[0026] The difference between the initial output rendering result and the original measured data is minimized by a preset differentiable rendering engine, and the material parameters of each layer and the particle distribution characteristics are adjusted based on the optimization results.

[0027] In some embodiments, the rendering result includes a rendered spectrum and a rendered HDR image. A forward and inverse consistency check is performed based on the rendering result and the original measured data to calculate the rendering deviation, including the following steps:

[0028] Obtain preset consistency verification rules, and perform comparison calculations between the rendered spectrum and the original multi-angle spectrum based on the type of the consistency verification rules, as well as comparison calculations between the rendered HDR image and the original HDR. The type of the consistency verification rules includes at least band requirements, color difference standards, and texture detail comparison.

[0029] In some embodiments, the comparison calculation between the rendered spectrum and the original multi-angle spectrum, and the comparison calculation between the rendered HDR image and the original HDR, are performed based on the type of the consistency check rule, including the following steps:

[0030] Based on the band requirements, the mean square error between the rendered spectrum and the original multi-angle spectrum is calculated to generate spectral error results;

[0031] Based on the color difference standard, a color difference index is calculated between the rendered HDR image and the original HDR image to generate a color difference result;

[0032] Based on the texture detail comparison, the structural similarity index between the rendered HDR image and the original HDR image is calculated to generate texture error results;

[0033] The rendering deviation is generated based on the spectral error result, the color difference result, and the texture error result.

[0034] In some embodiments, backpropagation is performed based on the rendering deviation to iteratively optimize the forward rendering and the reverse rendering until the rendering deviation is less than a threshold, including the following steps:

[0035] Generate the initial weight coefficients corresponding to each item in the consistency verification rule;

[0036] An optimization objective function is generated based on the initial weighting coefficients, the consistency verification rules, and the detailed information of the vehicle paint formula.

[0037] Determine whether iterative optimization is needed based on the magnitude of the rendering deviation.

[0038] If necessary, the rendering deviation is backpropagated to the optimization objective function to adjust and correct the initial weight coefficients in the reverse rendering and the paint formula details in the forward rendering, until the rendering deviation is less than the threshold, at which point the iteration stops and the final paint formula details are output.

[0039] In some embodiments, the calculation formula for the optimization objective function is specifically as follows:

[0040] ,

[0041] Where θ represents the paint formulation parameters. , , These are the initial weight coefficients corresponding to each item. The representation is the rendered spectral data based on the paint formulation parameters. Characterized as raw multi-angle spectral data, The result is represented by the color difference standard, while SSIM is represented by the structural similarity index.

[0042] In some embodiments, the process of inputting raw measured data and performing reverse parameter parsing and optimization to obtain the physical parameters of the vehicle paint also includes the following steps:

[0043] Create an anchor point library, which stores optical feature data and paint formula parameters corresponding to several typical car paints. The optical feature data includes spectral curves, color information, and structural similarity index.

[0044] Based on the original measured data, the original spectral curve, original color, and original structural similarity index are extracted and compared with the optical feature data to calculate the feature distance between each anchor point.

[0045] Determine if there is a feature distance smaller than a preset value;

[0046] If it exists, select the anchor point with the smallest feature distance and extract its corresponding paint formula parameters as the paint physical parameters;

[0047] If not, the physical parameters of the vehicle paint are obtained based on reverse parameter parsing and parameter optimization.

[0048] In some of these embodiments,

[0049] Secondly, this application provides a vehicle paint simulation system that supports dynamic forward and inverse bidirectional rendering, employing the following technical solution:

[0050] A vehicle paint simulation system that supports dynamic forward and reverse bidirectional rendering is used to implement the above method.

[0051] The technical solutions provided by the embodiments of this application have the following technical effects:

[0052] Through a bidirectional rendering architecture, a dynamic correlation is established between the paint formula and optical effects during the forward rendering process, while the reverse rendering process addresses the problem of "how to provide feedback on the formula when the actual paint effect is known," providing support for quality inspection and parameter optimization. Furthermore, it doesn't simply focus on an independent unidirectional rendering process, but rather achieves dynamic linkage between "formula parameters and visual effects" through a closed loop of "forward rendering to generate effects → reverse analysis to deduce parameters → forward and reverse fusion to correct deviations." The output of each step provides the foundation for the input of the next step, and through bidirectional interaction and iterative optimization, it ensures the unity of rendering accuracy and industrial practicality, solving the core problem of "disconnect between formula and effect" in traditional technologies. Attached Figure Description

[0053] Figure 1 This is a schematic diagram illustrating the steps of a vehicle paint simulation method that supports dynamic forward and reverse bidirectional rendering, as provided in this embodiment.

[0054] Figure 2 This is a schematic diagram of the logic of the vehicle paint simulation method that supports dynamic forward and reverse bidirectional rendering provided in the embodiments of this application. Detailed Implementation

[0055] To better understand the purpose, technical solutions, and advantages of this application, it has been described and illustrated below with reference to the accompanying drawings and embodiments. However, those skilled in the art should understand that this application can be implemented without these details. In some cases, to avoid obscuring various aspects of this application due to unnecessary description, well-known methods, processes, systems, components, and / or circuits already described at a higher level will not be elaborated upon. It will be apparent to those skilled in the art that various modifications can be made to the embodiments disclosed in this application, and the general principles defined in this application can be applied to other embodiments and application scenarios without departing from the principles and scope of this application. Therefore, this application is not limited to the illustrated embodiments, but conforms to the broadest scope consistent with the scope of protection claimed in this application.

[0056] It should be noted that the descriptions of these embodiments are for the purpose of aiding understanding the present invention, but do not constitute a limitation thereof. Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0057] In the description of this application, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0058] In the description of this application, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples.

[0059] like Figure 1 and Figure 2 As shown in the figure, this application discloses a method for simulating vehicle paint that supports dynamic forward and inverse bidirectional rendering, including the following steps:

[0060] Forward rendering:

[0061] S100: Input the details of the paint formula and parse out the paint layer rendering parameters and color masterbatch rendering parameters to output the rendering results in the paint renderer.

[0062] Forward rendering is a rendering process from formula to effect. The input is the detailed data of the paint formula, and the formula is used to decompose the optical rendering parameters of different paint layers and the color masterbatch rendering parameters of the corresponding particle color masterbatch characteristics. The final output is the rendering of the paint surface data and color masterbatch data by the paint renderer to obtain the rendered spectral image and HDR image.

[0063] Furthermore, this application supports dynamic rendering linkage, allowing users to upload detailed information about the vehicle paint formula and then edit the details in real time (such as adjusting the clear coat thickness and aluminum powder ratio). This real-time editing can be done manually by the user or intelligently based on feedback from backpropagation during subsequent linkage with reverse rendering. The vehicle renderer automatically updates the rendered spectrum and HDR image based on parameter changes, outputting real-time, up-to-date rendering results.

[0064] Reverse rendering:

[0065] S200: Input the original measured data and perform reverse parameter parsing and parameter optimization to obtain the physical parameters of the vehicle paint.

[0066] Reverse rendering is a rendering process from measured image data to parameters. The input is raw measured data composed of spectral images actually measured by the user and HDR images actually captured. Through a series of preprocessing, parameter parsing and further parameter optimization and convergence of the raw measured data, the physical parameters of the car paint are output. The physical parameters of the car paint directly correspond to the formula details information during forward rendering.

[0067] Forward and reverse fusion:

[0068] S300 uses the physical parameters of the vehicle paint as the parsing result of the newly input vehicle paint formula details in the forward rendering, and uses the rendering result as a priori reference in the reverse rendering to constrain the range of the physical parameters of the vehicle paint.

[0069] In this application, forward rendering and reverse rendering are not independent processes, but rather form a closed loop through data interaction, feedback verification, and other actions to ensure that "the effect of forward rendering is reliable" and "the parameters of reverse parsing are usable." Specifically,

[0070] The spectral / image features generated by forward rendering serve as a "prior reference" for reverse analysis, constraining the parameter range. For example, when there are anomalies in the measured data (such as spectral jumps caused by equipment failure), the abnormal values ​​can be corrected through forward results, narrowing the parameter search range. The physical parameters (corresponding to the formula details) from reverse analysis are fed back to the forward rendering model as "dynamic input" for forward rendering, correcting model biases and reducing the deviation between model assumptions and reality.

[0071] S400 performs forward and reverse consistency checks based on the rendering results and the original measured data to calculate the rendering deviation.

[0072] Furthermore, a real-time feedback mechanism is designed to quantify deviations and pinpoint problems.

[0073] The consistency of the current forward rendering result (spectral + HDR image) and the original measured data (spectral + HDR image) detected by actual shooting is checked. The main checks include spectral deviation check, color difference check, and texture detail difference check corresponding to particles. The rendering deviation is calculated based on the check results.

[0074] Based on different rendering deviation results, the model can quickly quantify the deviation between the forward rendering of recipe-effect and the reverse rendering of image-parameter, and determine the specific problem causing the deviation difference based on the numerical relationship of the rendering deviation results.

[0075] S500 performs backpropagation based on rendering deviation to iteratively optimize forward and reverse rendering until the rendering deviation is less than a threshold.

[0076] The rendering deviation is backpropagated to participate in the model actions of forward and reverse rendering. Based on the numerical relationship of the deviation, the corresponding parameter sizes in forward and reverse rendering are adjusted and corrected. After each adjustment and update, the forward and reverse rendering processes are repeated according to the new parameter information, and the rendering deviation is recalculated. In this way, the model parameters are iteratively optimized through continuous feedback and negative feedback until the final deviation size is less than the threshold. Then, the iteration stops and the final data is output.

[0077] The above method, employing a bidirectional rendering architecture, establishes a dynamic correlation between paint formula and optical effects during the forward rendering process, and addresses the problem of "how to provide feedback on the formula given a known real paint effect" during the reverse rendering process. This provides support for quality inspection and parameter optimization. Furthermore, it doesn't simply focus on an independent unidirectional rendering process, but rather achieves dynamic linkage between "formula parameters and visual effects" through a closed loop of "forward rendering to generate effects → reverse analysis to deduce parameters → forward and reverse fusion to correct deviations." The output of each step provides the foundation for the input of the next step, and through bidirectional interaction and iterative optimization, it ensures the unity of rendering accuracy and industrial practicality, solving the core problem of "disconnect between formula and effect" in traditional technologies.

[0078] In other embodiments, resolving paint layer rendering parameters includes the following steps:

[0079] S110 breaks down the details of the vehicle paint formula into a physical layer structure that includes a primer layer, a color coat layer, a particle layer, and a clear coat layer.

[0080] Traditional car paint rendering typically treats the paint as a single "material sphere." However, in this embodiment, in order to restore the physical structural features of the paint as much as possible, layered modeling based on the paint formula details is used to decompose the physical model of several paint layers.

[0081] Primer layer: Opaque, its main function is to adhere to the car body and provide basic coverage.

[0082] Color layer: translucent, containing pigment molecules (such as organic pigments), which determines the basic hue of the car paint.

[0083] Granular layer: Contains sheet-like / granular color masterbatches such as aluminum powder and mica flakes, which are the core source of iridescent effects (such as metallic luster and color variation at different angles).

[0084] Clear coat: Transparent, covering the outermost layer, providing gloss and protection.

[0085] By using layered modeling, the transmission process of light in each layer of the paint can be broken down into a complete path of "air-clear coat interface reflection → clear coat layer transmission → particle layer scattering → color coat layer absorption → primer layer reflection → reverse penetration of each layer", which facilitates subsequent simulation of real light effects to generate the optical features corresponding to each layer of paint.

[0086] S120 models the light transmission rules for each layer to assign independent rendering parameters for each layer, and calculates the total reflectivity based on the rendering parameters for each layer.

[0087] Based on the different regular properties of light in each layer, such as reflection, refraction, and scattering, optical parameters required for rendering are added to each layer of the model and defined as rendering parameters for each layer. Subsequently, when the car paint renderer renders based on these parameters, each layer's rendering parameters provide the color and lighting foundation for the rendered image. Specifically...

[0088] The optical parameters corresponding to the primer layer mainly include diffuse reflectance (constant reflectance, no angle dependence), refractive index, and thickness;

[0089] The optical parameters corresponding to the paint layer include absorption coefficient (positively correlated with pigment concentration), refractive index, transmittance, and thickness;

[0090] The optical parameters corresponding to the particle layer include particle density, particle size distribution, orientation angle, refractive index, transmittance, absorption coefficient, and thickness.

[0091] The optical parameters corresponding to the varnish layer include refractive index, thickness, transmittance, and absorption coefficient.

[0092] The formula for generating the total reflectivity based on the rendering parameters of each layer is as follows:

[0093] .

[0094] in,

[0095] .

[0096] Characterized as total reflectance; The equivalent reflectivity is characterized by the combined effect of the clear coat and the underlying layer. Its physical meaning is that when light enters the clear coat from the air, it combines the total reflection effect of "direct reflection from the upper surface of the clear coat" and "light that passes through the clear coat is reflected by the underlying layer structure and then passes through the clear coat again". Characterized by the combined effect of the granular layer and the underlying layer, its physical meaning is that when light passing through the varnish layer reaches the granular layer, it combines the effects of "direct reflection from the upper surface of the granular layer" and "light passing through the granular layer being reflected by the underlying layer and then re-exiting the granular layer." It is calculated... The basic parameters.

[0097] Characterized by the interfacial reflectance of the air-varnish layer, Characterized by the interfacial reflectance of the varnish layer and the particle layer. Characterized by the interfacial reflectance of the particle layer and the paint layer. Characterized by the transmittance of the varnish layer. Characterized as the transmittance of the granular layer, Characterized by the transmittance of the paint layer. Characterized by the reflectivity of the primer layer, Characterized by the air refractive index (value 1.0). Characterized by the refractive index of the varnish layer, Characterized by the refractive index of the granular layer, Characterized by the refractive index of the paint layer, Characterized by the refractive index of the primer layer, Characterized by the absorption coefficient of the i-th layer, It is represented by the thickness of the i-th layer.

[0098] Meanwhile, the specific calculation process for some of the above parameters is as follows:

[0099] ;

[0100] ;

[0101] ;

[0102] ;

[0103] ;

[0104] ;

[0105] .

[0106] S130 generates paint layer rendering parameters based on the rendering parameters of each layer and the total reflectivity.

[0107] In other embodiments, parsing the master rendering parameters includes the following steps:

[0108] S140: Extract the physical feature parameters of each type of particle color masterbatch in the particle layer based on the rendering parameters of each layer corresponding to the particle layer.

[0109] The optical effects of particulate color masterbatches such as aluminum powder and mica flakes (such as "angle-dependent color") depend on their physical properties, but in industry, parameters such as "particle size 5-20μm" and "aluminum powder content 10%" are commonly used to describe them. These parameters need to be converted into BSDF (two-way scattering distribution function) parameters that can be calculated by computer graphics through algorithms.

[0110] Therefore, the first step is to target the particle layer containing color masterbatch features. Based on the rendering parameters of each layer and the particle characteristics of different particles (such as aluminum powder, mica flakes, etc.), physical feature parameters such as density (the number of particles per unit volume, corresponding to the addition ratio of different particles in the formula), particle size distribution (the distribution range and probability density of different particles), and orientation (the tilt angle between the particles and the paint surface, affecting the direction of the highlight) are extracted.

[0111] S150 calculates the optical characteristics of the particulate colorant based on the physical characteristic parameters and the preset Mie scattering model.

[0112] Based on Mie scattering theory (describing the scattering of light by spherical particles) and geometric optics (describing the specular reflection of sheet-like particles), the scattering and reflection characteristics of light by particles are calculated, including:

[0113] Scattering intensity (the ability of a particle to scatter light of different wavelengths).

[0114] Angle dependence (the variation of scattered light with incident angle and observation angle directly affects the "angle-dependent color" effect);

[0115] Polarization characteristics (the polarization state of light scattered by particles, which affects the "iridescent" effect of highlights).

[0116] Specifically, the core of Mie scattering theory is to analyze particle size (d), particle refractive index (d), and other factors. ), refractive index of the surrounding medium ( Such as the refractive index of the paint layer ) and incident light wavelength ( The scattering efficiency factor is calculated using the following formula:

[0117] .

[0118] in, These are dimensional parameters; , The Mie scattering coefficient is determined by the refractive index difference between the particles and the medium, as well as their size parameters. It reflects the particle's ability to scatter light and is related to the "scattering intensity parameter" used in subsequent BSDF parameter mapping.

[0119] S160 establishes a mapping relationship between physical feature parameters and BSDF parameters based on a preset parametric regression algorithm.

[0120] A mapping between the physical properties of particles and BSDF parameters is established using a regression algorithm. Specifically,

[0121] The particle size distribution of aluminum powder is regressed to correspond to the "high-gloss bandwidth" of BSDF (the smaller the particle size, the more concentrated the highlights and the stronger the metallic feel).

[0122] The thickness distribution of the mica sheet is regressed to correspond to the "interference color shift" of the BSDF (as the thickness increases, the wavelength of the reflected light shifts to longer wavelengths, such as from blue-violet to golden yellow).

[0123] The particle density distribution is regressed to the "scattering intensity parameter" corresponding to the BSDF (the higher the density, the stronger the reflected light and the brighter the color).

[0124] These mapping relationships directly affect the scattering and reflection of light by the particle layer, thus determining the special optical effects that the car paint ultimately presents.

[0125] Taking the regression of aluminum powder particle size distribution to the "high bandwidth" of BSDF as an example, the mapping formula is:

[0126]

[0127] in, These are the specular angle range parameters in the BSDF model. Characterized by the particle size distribution of aluminum powder, This is the particle size value. Characterized as the probability of a corresponding particle size. The average particle size is Characterized by the standard deviation of particle size distribution, , , These are the regression coefficients (obtained through training with experimental data). It is characterized as regression error (minimized through training with a large number of samples).

[0128] For novel color masterbatches (such as nanoparticles), color masterbatch parameterization is performed by calibrating the mapping model using spectral data from a small number of physical samples to ensure the accuracy of parameter conversion. This process also takes place at the particle layer, because when novel color masterbatches are added to automotive paint, they are primarily present in the particle layer. By adjusting the relevant parameters of the particle layer, the optical effects of the novel color masterbatch can be simulated and presented.

[0129] S170 analyzes color master rendering parameters based on optical features and mapping relationships.

[0130] In this application, both forward and reverse rendering share the same multi-layer BSDF model and particle Mie scattering model to ensure physical consistency.

[0131] In other embodiments, the raw measured data includes raw multi-angle spectra and raw HDR images. Inputting the raw measured data and performing reverse parameter analysis and optimization to obtain the physical parameters of the vehicle paint includes the following steps:

[0132] S210 preprocesses its original multi-angle spectrum and original HDR image.

[0133] First, the preprocessing of the original multi-angle spectra and original HDR images includes multi-angle alignment, noise reduction, and data normalization.

[0134] Multi-angle alignment specifically involves mapping spectral data from multiple angles (such as the reflectance spectra under incident light at 0 degrees, 45 degrees, and 60 degrees) to the same angular coordinate system using the normal to the paint surface as a reference, thereby avoiding analytical errors caused by measurement angle deviations.

[0135] Noise reduction is characterized by filtering the spectral data using a Gaussian filter, while HDR images are filtered using a bilateral filter (preserving edges while removing speckle noise).

[0136] Data normalization is characterized by mapping spectral intensity (0, -1) and HDR image pixel values ​​(0-255) uniformly to the [0, 1] interval, providing a consistent scale for subsequent fusion.

[0137] The above preprocessing can significantly improve the parsing accuracy and provide high-quality input for the subsequent "reverse parameter parsing".

[0138] S220 uses machine learning algorithms to input raw measured data into a preset model to infer the material parameters and particle distribution characteristics of each layer corresponding to the raw measured data.

[0139] After preprocessing, the specific physical characteristic parameters of each layer of the paint are deduced from the measured data. Specifically, the optical characteristic parameters (such as total reflectivity) analyzed during the forward rendering process are used as physical constraints to narrow down the search range of parameters.

[0140] The neural network model is trained with measured spectra and image features as inputs and parameters of each layer as outputs. By combining the physical model with the neural network results, the image information in the measured data is transformed into specific physical feature parameters that include material parameters and particle distribution characteristics of each layer.

[0141] Machine learning algorithms play a crucial role in reverse rendering by establishing a mapping relationship between the features of the measurement data and the material parameters and particle distribution characteristics of each layer through a neural network model.

[0142] For example, a well-trained neural network can infer the pigment composition ratio in a paint layer based on changes in the reflection intensity of specific wavelengths in spectral data; and it can infer parameters such as the particle size distribution and density of aluminum powder or mica flakes in the granular layer based on the distribution and intensity of highlights in an HDR image. This mapping relationship is obtained based on a large amount of experimental data and machine learning training, enabling the transformation of complex measurement data into specific formulation parameters. During training, a large amount of automotive paint measurement data with known formulation details is input, allowing the neural network to learn the intrinsic relationship between measurement data features and formulation parameters. This allows it to accurately deduce the corresponding formulation parameters from the measurement data during actual reverse rendering.

[0143] The S230 minimizes the difference between the initial output rendering result and the original measured data through a preset differentiable rendering engine, and adjusts the material parameters and particle distribution characteristics of each layer based on the optimization results.

[0144] By minimizing the difference between the rendered results and the measured data through a differentiable rendering engine, parameters are refined to ensure that the reverse-engineered recipe details can be directly used for forward rendering.

[0145] Specifically, the principle of differentiable rendering is to transform the rendering process into a differentiable function, calculate the derivative (gradient) of the "difference between the rendering result and the measured data" with respect to the parameters, minimize the above differentiable function, and adjust the physical parameters according to the gradient direction until the difference between the output rendering result and the original measured data is less than a threshold.

[0146] The explanation of differentiable functions will be provided in detail later.

[0147] In other embodiments, the rendering results include a rendered spectrum and a rendered HDR image. A forward and inverse consistency check is performed based on the rendering results and the original measured data to calculate the rendering deviation, including the following steps:

[0148] S410: Obtain the preset consistency verification rules, and perform comparison calculations on the rendered spectrum and the original multi-angle spectrum based on the type of consistency verification rules, as well as comparison calculations on the rendered HDR image and the original HDR.

[0149] By comparing the rendering results driven by the reverse parameters with the original measured data through forward and reverse consistency verification, the deviation between the two is calculated. The consistency verification rules include at least band requirements, color difference standards, and texture detail comparison.

[0150] S420 calculates the mean square error between the rendered spectrum and the original multi-angle spectrum based on band requirements to generate spectral error results.

[0151] The mean square error (MSE) of the rendered spectrum and the measured spectrum is calculated wavelength by wavelength in the 400-700nm band, reflecting the deviation of the basic color (such as insufficient red pigment will lead to a large error in the 600nm band).

[0152] S430 calculates the color difference index between the rendered HDR image and the original HDR image based on the color difference standard to generate color difference results.

[0153] A color difference index based on the characteristics of human visual perception (the human eye cannot distinguish when ΔE<1.5) is used to assess the overall consistency of color perception.

[0154] S440 calculates the structural similarity index between the rendered HDR image and the original HDR image based on texture detail comparison to generate texture error results.

[0155] When evaluating texture details (such as the uniformity of particle distribution), an SSIM value < 0.95 indicates a deviation in particle layer parameters (such as particle size distribution).

[0156] S450 generates rendering biases based on spectral error results, color difference results, and texture error results.

[0157] The source of the deviation can be analyzed based on the different values ​​and states of the rendering deviation. For example, if the spectral error is concentrated in the blue light band, it indicates that the blue pigment parameters of the paint layer are abnormal.

[0158] In other embodiments, backpropagation is performed based on rendering deviation to iteratively optimize forward and reverse rendering until the rendering deviation is less than a threshold, including the following steps:

[0159] S510 generates the initial weight coefficients corresponding to each item in the consistency verification rule.

[0160] S520 generates an optimization objective function based on initial weighting coefficients, consistency verification rules, and detailed information on vehicle paint formula.

[0161] S530 determines whether iterative optimization is needed based on the magnitude of rendering deviation.

[0162] S540, if necessary, backpropagates the rendering deviation to the optimization objective function to adjust and correct the initial weight coefficients in the reverse rendering and the paint formula details in the forward rendering, until the rendering deviation is less than the threshold, at which point the iteration stops and the final paint formula details are output.

[0163] Based on the back propagation of the deviation signal, the weight factors of the reverse analysis (such as spectral / texture weights) and the paint layer rendering parameters and color masterbatch rendering parameters related to the paint formula details during forward rendering are dynamically adjusted until the deviation is less than the threshold.

[0164] The specific formula for calculating the objective function is as follows:

[0165] ,

[0166] Here, θ represents the paint formulation parameters, which are variables to be optimized. These include the clear coat thickness, the particle size distribution of aluminum powder / mica flakes, the absorption coefficient of the paint layer, and the particle density. These parameters determine the optical and physical properties of the paint. By adjusting θ, the rendering effect can be made closer to reality.

[0167] , , These are the initial weight coefficients corresponding to each item. This is a weighting factor for spectral error, controlling the importance of the difference between the rendered spectrum and the measured spectrum in the overall optimization objective. A larger value indicates greater emphasis on spectral accuracy. As a weighting factor for color difference, it controls the difference between the rendered color and the measured color. The influence weight of ")" in the objective function highlights the importance attached to the consistency of color perception by the human eye; As a weighting factor for texture differences, it controls the weight of "difference between rendered texture and measured texture (1−SSIM)" and emphasizes the attention to the consistency of the paint surface texture (such as particle distribution and gloss texture).

[0168] Characterized by the rendered spectral data based on the paint formulation parameter θ, and the virtual spectral data calculated by the forward rendering model (combining theories of light transmission and Mie scattering), it describes the reflection / transmission of different wavelengths of light by the paint and is the "expected optical performance" in the virtual environment.

[0169] Characterized by raw multi-angle spectral data, which is real vehicle paint spectral data obtained through actual measurement (using equipment such as spectrometers), it is a "real feedback" of the optical properties of vehicle paint in the physical world, serving as a "target reference" for optimization.

[0170] The result is represented by the color difference standard calculated based on the CIELAB color space, while SSIM is represented by the structural similarity index.

[0171] Based on the formula explanation, the process of iteratively optimizing model parameters through backpropagation using bias data includes:

[0172] Weighting factor adjustment: If the proportion of spectral error is high, increase the weighting factor. (Increase the weight of spectral fitting); if the texture difference is large, increase the weight. ;

[0173] Light transmission parameter correction: If the interlayer reflection error is large (such as the deviation in calculating the reflectance T1 at the varnish-particle layer interface), then correct the refractive index in the reflectance formula. , );

[0174] Convergence condition: Repeat "inverse parameter → forward rendering → deviation calculation → parameter adjustment". Stop the iteration when all three types of deviations are less than the threshold (spectral error <1.2%, ΔE <1.5, SSIM >0.95) and output the final parameters.

[0175] It should be noted that both steps S230 and S540 are based on the above-mentioned objective function for parameter optimization and iteration, but the difference lies in:

[0176] The core of step S230 is to refine the "formulation parameters" obtained from reverse parameter analysis. Its purpose is "local optimization after reverse analysis," adjusting only the physical parameters derived from the reverse analysis (such as clear coat thickness, aluminum powder particle size distribution, etc.). This means adjusting and optimizing θ in the objective function, while the weighting factors... , , These are fixed values, the purpose of which is to make the results generated by forward rendering of these parameters as close as possible to the measured data;

[0177] The deviation threshold is the criterion for determining whether the reverse parameters are "usable". When the deviation is less than the threshold, it means that the reverse-engineered formula parameters are accurate enough and can be directly used for forward rendering to reproduce the measured effect, meeting the needs of "deriving formula from actual measurement" in industrial scenarios (such as quality inspection and formula replication).

[0178] The core of step S540 is to perform global optimization of the "forward and inverse bidirectional rendering closed loop," which is positioned as "overall convergence after forward and inverse interaction." It involves not only the adjustment of inverse parameters but also the light transmission parameters (such as interlayer transmittance) and multimodal weighting factors in forward rendering. , , The dynamic correction aims to ensure physical and visual consistency between forward rendering and reverse parsing.

[0179] The deviation threshold is the criterion for determining the consistency of forward and reverse rendering. When the deviation is less than the threshold, it indicates that forward rendering (based on recipe parameters) and reverse analysis (based on measured data) form a stable closed loop—forward rendering can accurately map recipe changes, and reverse analysis can reliably deduce recipe parameters. The physical logic and effect output of the two are consistent, meeting the core requirements of "dynamic editing and real-time feedback".

[0180] In other embodiments, the raw measured data includes raw multi-angle spectra and raw HDR images. The process of inputting the raw measured data and performing reverse parameter analysis and optimization to obtain the physical parameters of the vehicle paint also includes the following steps:

[0181] S240, Create an anchor point library. The anchor point library stores optical feature data and paint formula parameters corresponding to several typical car paints. The optical feature data includes spectral curves, color information, and structural similarity index.

[0182] A certain number of classic car paints, typical car paints, and popular car paints (such as pure black paint, bright silver paint, matte white paint, pearl red paint, etc.) are pre-stored to form an anchor point library, which serves as the "basic template" for subsequent rapid matching.

[0183] The structure of the anchor point library is shown in the diagram:

[0184] .

[0185] in, Characterized as a full-band spectrum (400-700 nm, λ is the wavelength). It is characterized as a set of formulation parameters.

[0186] S250 extracts the original spectral curve, original color, and original structural similarity index based on the original measured data, and compares them with optical feature data to calculate the feature distance between each anchor point.

[0187] The differences between the measured data and the anchor points in the anchor point library are quantified, and the most similar anchor point is found.

[0188] Optical features are extracted from the new measured image, and the total distance is defined as a weighted sum of spectral distance, color distance, and texture distance. Specifically,

[0189] .

[0190] in, The similarity is represented by the distance between spectral curves as measured by cosine similarity; the closer the value is to 1, the more similar the curves are. Characterized by the color difference formula calculated using CIELAB, the smaller the value, the more similar the colors are. It is represented by the texture difference distance; the smaller the value, the more similar the textures.

[0191] S260, determine whether there is a feature distance less than a preset value.

[0192] S270, if it exists, select the anchor point with the smallest feature distance and extract its corresponding paint formula parameters as paint physical parameters.

[0193] If S280 does not exist, the physical parameters of the vehicle paint will be obtained based on reverse parameter parsing and parameter optimization.

[0194] After calculating the distances between the measured data and each anchor point, the data is iterated through to find the distances that make the anchor points equal to the measured data. The smallest anchor point, while also paying attention to the smallest The value needs to be less than a preset value. If it is not less than a preset value, it means that the existing anchor point data with the closest feature distance is also dissimilar to or has a low degree of similarity with the measured data.

[0195] When the minimum feature distance is less than the preset value, it means that the measured data is extremely similar to the optical features corresponding to the anchor point. In this case, in order to change gradient descent from blind search to targeted refinement, the formula parameters corresponding to the anchor point can be directly selected as the physical parameters of the car paint during reverse rendering and the gradient descent iterative optimization process can be directly performed. In this way, for some classic or common car paints, the reverse rendering results can be quickly searched based on the above method without going through the conventional repeated refinement iterations, thus reducing the number of iteration steps.

[0196] This application also discloses a vehicle paint simulation system that supports dynamic forward and reverse bidirectional rendering, used to implement the above-mentioned method.

[0197] Furthermore, the basic data acquisition equipment for this system includes the XRite MA5C multi-angle spectrophotometer, optical transmittance meter, and paint film analyzer.

[0198] The specific efficiency indicators are:

[0199] For forward rendering, the single-frame rendering time is less than or equal to 0.5 seconds, and for reverse rendering, the convergence time with full parameter optimization is less than or equal to 10 minutes.

[0200] The implementation principle is as follows:

[0201] Through a bidirectional rendering architecture, a dynamic correlation is established between the paint formula and optical effects during the forward rendering process, while the reverse rendering process addresses the problem of "how to provide feedback on the formula when the actual paint effect is known," providing support for quality inspection and parameter optimization. Furthermore, it doesn't simply focus on an independent unidirectional rendering process, but rather achieves dynamic linkage between "formula parameters and visual effects" through a closed loop of "forward rendering to generate effects → reverse analysis to deduce parameters → forward and reverse fusion to correct deviations." The output of each step provides the foundation for the input of the next step, and through bidirectional interaction and iterative optimization, it ensures the unity of rendering accuracy and industrial practicality, solving the core problem of "disconnect between formula and effect" in traditional technologies.

[0202] It should be understood that although the steps in the flowcharts in the accompanying drawings are shown sequentially as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise expressly stated herein, there is no strict order in which these steps are performed, and they may be performed in other orders.

[0203] The above are all preferred embodiments of the present application, and are not intended to limit the scope of protection of the present application. Therefore, any equivalent changes made based on the structure, shape, and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A method for simulating vehicle paint performance that supports dynamic forward and inverse bidirectional rendering, characterized in that, Includes the following steps: Forward rendering: Input the details of the paint formula and parse out the paint layer rendering parameters and color masterbatch rendering parameters to output the rendering results in the paint renderer; Reverse rendering: Input the original measured data and perform reverse parameter parsing and optimization to obtain the physical parameters of the car paint; Forward and reverse fusion: The physical parameters of the vehicle paint are used as the parsing result of the newly input vehicle paint formula details in the forward rendering, and the rendering result is used as the prior reference in the reverse rendering to constrain the range of the physical parameters of the vehicle paint; Based on the rendering results and the original measured data, a forward and reverse consistency check is performed to calculate the rendering deviation. Backpropagation is performed based on the rendering deviation to iteratively optimize the forward rendering and the reverse rendering until the rendering deviation is less than a threshold.

2. The vehicle paint simulation method supporting dynamic forward and inverse bidirectional rendering according to claim 1, characterized in that, The steps to parse the paint layer rendering parameters include: The detailed information of the vehicle paint formula is broken down into a physical layer structure that includes a primer layer, a color coat layer, a particle layer, and a clear coat layer; Light transmission rules are modeled for each layer to assign independent rendering parameters for each layer, and the total reflectivity is calculated based on each of the rendering parameters for each layer, including reflectivity, refractive index, thickness, absorption coefficient, particle density, particle size distribution, and orientation angle. The paint layer rendering parameters are generated based on the rendering parameters of each layer and the total reflectivity.

3. The vehicle paint simulation method supporting dynamic forward and inverse bidirectional rendering according to claim 2, characterized in that, Analyzing excellent master rendering parameters includes the following steps: Based on the rendering parameters of each layer corresponding to the particle layer, the physical feature parameters of each type of particle color masterbatch in the particle layer are extracted; The optical characteristics of the particulate colorant are calculated based on the physical characteristic parameters and a preset Mie scattering model. The mapping relationship between the physical feature parameters and BSDF parameters is established based on a preset parametric regression algorithm; The color master rendering parameters are analyzed based on the optical characteristics and the mapping relationship.

4. The vehicle paint simulation method supporting dynamic forward and inverse bidirectional rendering according to claim 1, characterized in that, The original measured data includes original multi-angle spectra and original HDR images. Inputting the original measured data and performing reverse parameter analysis and optimization to obtain the physical parameters of the vehicle paint includes the following steps: The original multi-angle spectrum and the original HDR image are preprocessed. The original measured data is input into a preset model based on machine learning algorithms to infer the material parameters and particle distribution characteristics of each layer corresponding to the original measured data. The difference between the initial output rendering result and the original measured data is minimized by a preset differentiable rendering engine, and the material parameters of each layer and the particle distribution characteristics are adjusted based on the optimization results.

5. The vehicle paint simulation method supporting dynamic forward and inverse bidirectional rendering according to claim 1, characterized in that, The rendering results include a rendered spectrum and a rendered HDR image. Based on the rendering results and the original measured data, a forward and inverse consistency check is performed to calculate the rendering deviation, including the following steps: Obtain preset consistency verification rules, and perform comparison calculations between the rendered spectrum and the original multi-angle spectrum based on the type of the consistency verification rules, as well as comparison calculations between the rendered HDR image and the original HDR. The type of the consistency verification rules includes at least band requirements, color difference standards, and texture detail comparison.

6. The vehicle paint simulation method supporting dynamic forward and inverse bidirectional rendering according to claim 5, characterized in that, Based on the type of the consistency check rule, the rendered spectrum is compared with the original multi-angle spectrum, and the rendered HDR image is compared with the original HDR, including the following steps: Based on the band requirements, the mean square error between the rendered spectrum and the original multi-angle spectrum is calculated to generate spectral error results; Based on the color difference standard, a color difference index is calculated between the rendered HDR image and the original HDR image to generate a color difference result; Based on the texture detail comparison, the structural similarity index between the rendered HDR image and the original HDR image is calculated to generate texture error results; The rendering deviation is generated based on the spectral error result, the color difference result, and the texture error result.

7. The vehicle paint simulation method supporting dynamic forward and inverse bidirectional rendering according to claim 6, characterized in that, Backpropagation is performed based on the rendering deviation to iteratively optimize the forward rendering and the reverse rendering until the rendering deviation is less than a threshold, including the following steps: Generate the initial weight coefficients corresponding to each item in the consistency verification rule; An optimization objective function is generated based on the initial weighting coefficients, the consistency verification rules, and the detailed information of the vehicle paint formula. Determine whether iterative optimization is needed based on the magnitude of the rendering deviation. If necessary, the rendering deviation is backpropagated to the optimization objective function to adjust and correct the initial weight coefficients in the reverse rendering and the paint formula details in the forward rendering, until the rendering deviation is less than the threshold, at which point the iteration stops and the final paint formula details are output.

8. The vehicle paint simulation method supporting dynamic forward and inverse bidirectional rendering according to claim 7, characterized in that, The specific formula for calculating the optimization objective function is as follows: ， Where θ represents the paint formulation parameters. , , These are the initial weight coefficients corresponding to each item. The representation is the rendered spectral data based on the paint formulation parameters. Characterized as raw multi-angle spectral data, The result is represented by the color difference standard, while SSIM is represented by the structural similarity index.

9. The vehicle paint simulation method supporting dynamic forward and inverse bidirectional rendering according to claim 4, characterized in that, The process of inputting raw measured data and performing reverse parameter analysis and optimization to obtain the physical parameters of the vehicle paint also includes the following steps: Create an anchor point library, which stores optical feature data and paint formula parameters corresponding to several typical car paints. The optical feature data includes spectral curves, color information, and structural similarity index. Based on the original measured data, the original spectral curve, original color, and original structural similarity index are extracted and compared with the optical feature data to calculate the feature distance between each anchor point. Determine if there is a feature distance smaller than a preset value; If it exists, select the anchor point with the smallest feature distance and extract its corresponding paint formula parameters as the paint physical parameters; If not, the physical parameters of the vehicle paint are obtained based on reverse parameter parsing and parameter optimization.

10. A vehicle paint simulation system supporting dynamic forward and inverse bidirectional rendering, characterized in that, Used to implement the method as described in any one of claims 1-9.

Images