Flower exhibition digital stylized reconstruction system using image sampling and artificial intelligence

Abstract

The invention provides a flower exhibition digital stylized reconstruction system utilizing image sampling and artificial intelligence, digital reconstruction of a flower exhibition site is realized by utilizing an image sampling and artificial intelligence technology, the system comprises six modules of image sampling, processing, style model, intelligent generation, interaction control and virtual guide, and the image sampling module captures a high-definition image; the image processing module optimizes and extracts features; the style model module accurately reconstructs style parameters through a combination algorithm of a diffusion model and a Lora model group; the intelligent generation module redraws the stylized image by using the diffusion model in combination with the parameters; the interaction control module is used for realizing interaction of cue words and parameters through a Touch designer in combination with a comfyUI component, sampling conversion feedback in real time, and improving interactivity in cooperation with touch and gesture feedback; the virtual navigation module assists the user in navigation, the system generates a smooth surface through a Poisson reconstruction algorithm, the model quality is ensured, and the overall design improves the user experience.

Description

Technical Field

[0001] The present invention relates to the fields of image processing and virtual reality, and more specifically, to a flower show digital stylization reconstruction system using image sampling and artificial intelligence. Background Art

[0002] Driven by modern science and technology, image processing and artificial intelligence (AI) technologies have made significant progress. The application of these technologies in the field of visual arts has gradually increased, especially in flower show design and digital stylization reconstruction. The flower show digital stylization reconstruction system combines advanced technologies such as image sampling, computer vision and machine learning to achieve digital and artistic reproduction of flower exhibitions.

[0003] The existing flower show reconstruction system cannot meet the specific stylization requirements, and the low degree of automation leads to low rendering efficiency. Summary of the invention

[0004] In order to overcome the problems that the existing flower show reconstruction system cannot meet the specific stylization requirements and the low degree of automation leads to low rendering efficiency, the present invention discloses a flower show digital stylization reconstruction system using image sampling and artificial intelligence, which can effectively solve the above problems.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] A flower show digital stylization reconstruction system using image sampling and artificial intelligence, image sampling module, image processing module, style model module, generation module, interactive control module and virtual tour module;

[0007] The image sampling module is used to obtain high-definition image data of the flower show site through an image acquisition device;

[0008] The image processing module preprocesses and optimizes the image data acquired by the image sampling module, segments and extracts features of the image data, and extracts information useful for generating stylized images;

[0009] The style model module uses the image data provided by the image processing module, combined with the self-training data Lora model, to generate the stylized parameters of the flower show site through the comfyUI automated workflow model combination algorithm;

[0010] The generation module reconstructs the image based on the image generated by the style model module to further construct a complete stylized redrawing;

[0011] The interactive control module is used to control the interactive operation between the user and the stylized description, adjust the parameters of the artificial intelligence generation system, and connect to the external sensor;

[0012] The virtual tour module provides navigation and guidance services for users, helping them to better understand the layout and features of the flower show site.

[0013] Preferably, the image sampling module includes an image segmentation submodule, a non-uniform sampling submodule and a module data serialization submodule;

[0014] The image segmentation submodule is used to segment the image data into a plurality of small blocks or regions;

[0015] The non-uniform sampling submodule performs image data sampling with emphasis according to the frequency and location of suspected defects in the image data, performing more sampling in areas with a higher frequency of suspected defects and reducing sampling in areas with a lower frequency of defects;

[0016] The module data serialization submodule aggregates and serializes the sampled image block data to form serialized data pairs.

[0017] Preferably, the style model module includes a style feature extraction submodule, a style library management submodule, a Lora submodule and a style parameter generation submodule;

[0018] The style feature extraction submodule is used to further extract features related to stylization from the image data provided by the image processing module;

[0019] The style library management submodule is used to manage and maintain a style library, which contains a variety of predefined style templates or style parameters;

[0020] The Lora submodule processes the extracted style features using a self-trained Lora model;

[0021] The style parameter generation submodule combines the features extracted by the image processing module and the templates in the style library, and generates stylized parameters for the current flower show scene image through the comfyUI automated workflow model combination algorithm.

[0022] Preferably, the Lora submodule is used to import the exported model into stable diffusion and perform preliminary integration. The detailed steps include:

[0023] Create a dataset project, collect relevant style images, annotate parameters and prompt words, create a new training model project, and train a stylized migration model;

[0024] Import the model. In ComfyUI, create a workflow. First, build the image import group, the pixel scaling group, and the automatic cropping group. Combine the diffusion model, select the self-trained Lora model file, click Generate, and adjust the parameters according to the effect.

[0025] Adjust the model's generate settings, including weight factors, calculation steps, and fixed prompt words, to ensure that the model can be read normally in stable diffusion;

[0026] Add interaction and VR support, install interaction support plugins, configure software collaboration, test the display and interaction functions of the model in comfyUI, and adjust the model parameters and model group settings as needed.

[0027] Preferably, the generation module includes a style application submodule, an image reconstruction submodule and a rendering optimization submodule;

[0028] The style application submodule applies the stylized parameters generated by the style model module to the image data provided by the image processing module, and adjusts the color, texture and contrast properties of the image to make it conform to the target style characteristics;

[0029] The image reconstruction submodule further optimizes and reconstructs the image as a whole based on the style application submodule;

[0030] The rendering optimization submodule is used to improve the quality and rendering efficiency of the generated image and optimize the image rendering process.

[0031] Preferably, the rendering optimization submodule uses the MSAA algorithm, and the detailed steps are as follows:

[0032] Enable MSAA and turn on the MSAA function in the rendering pipeline;

[0033] Create an MSAA framebuffer, create an MSAA-enabled framebuffer and assign the necessary color and depth and stencil attachments to it;

[0034] Configure the sampling mode, set the desired sampling mode, and determine the number of sampling points for each pixel:

[0035] Adjust the viewport and scissor rectangle to ensure that they take into account the MSAA sampling mode so that the rendered output is clipped correctly.

[0036] Perform depth and template tests on each sub-sampling point to generate coverage and occlusion information;

[0037] Perform shading operations at the pixel level, writing color values ​​to sub-sampled points that pass the depth and template tests;

[0038] Perform the MSAA Resolve operation to resolve the color value of the sub-sampling point into the final pixel color after rendering is completed.

[0039] Preferably, the interactive control module includes a prompt word module, a generation parameter module, and a visual interaction module;

[0040] The prompt word module is used to adjust the variable prompt words in the prompt words, self-label the prompt words in the model training, map them into optional parameters, and concentrate the stylization within a fixed range;

[0041] The generation parameter module links the generation parameter data set of the generation system to the physical world sensor, and uses touch or gesture to change the parameters to affect the generated image result;

[0042] The visual interaction module cooperates with the tactile and gesture feedback devices to provide physical touch, superimpose interactive visual effects on the basis of generated image content, and enhance the interactive feedback sense.

[0043] Compared with the prior art, the technical solution of the present invention has the following beneficial effects: the present invention further extracts features related to stylization from the image data provided by the image processing module through the style feature extraction submodule, thereby ensuring the accuracy and pertinence of subsequent stylization processing; the style library management submodule manages and maintains a style library containing a variety of predefined style templates or style parameters, which provides a rich style selection basis for the system; the Lora submodule processes the extracted style features using the self-trained Lora model; through deep learning technology, the Lora model can learn and simulate various complex artistic styles, thereby meeting the user's needs for specific stylization; the style parameter generation submodule combines the features extracted by the image processing module and the templates in the style library, and generates stylization parameters for the current flower show site image through the comfyUI automated workflow model combination algorithm; this step ensures the accuracy and personalization of the stylization parameters, and meets the user's pursuit of specific stylization effects; the generation module is refined, and the style application submodule applies the stylization parameters generated by the style model module to the image data, and adjusts the color, texture and contrast attributes of the image to make it conform to the target style features. This step ensures the accurate realization of the stylized effect. Based on the style application, the image reconstruction submodule optimizes and reconstructs the image as a whole, further improving the image quality and stylized effect, and solving the problem of low rendering efficiency caused by low automation. The non-uniform sampling submodule focuses on sampling according to the frequency and location of suspected defects in the image data, reducing unnecessary sampling work and improving sampling efficiency. The module data serialization submodule aggregates and serializes the sampled image block data to form serialized data pairs, providing efficient data input for subsequent processing. The entire system adopts a highly automated processing flow from image sampling to stylized parameter generation, and then to image reconstruction and rendering optimization. omfyUI automated workflow model combination algorithm, the system can automatically complete complex image processing and stylization tasks, greatly improving processing efficiency, using MSAA (multi-sampling anti-aliasing) algorithm to optimize the image rendering process, by increasing the number of sampling points and merging the color values ​​of these sampling points to reduce aliasing, thereby improving the quality of the generated image and rendering efficiency, the MSAA algorithm is turned on in the rendering pipeline, and a framebuffer that supports MSAA is created, by configuring the sampling mode, adjusting the viewport and scissor rectangle, performing depth and template tests and shading operations, and finally resolving the color value of the sub-sampling point into the final pixel color, achieving efficient and high-quality rendering effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived based on the provided drawings without paying any creative work.

[0045] Figure 1 A structural diagram of a digital stylization reconstruction system for flower shows using image sampling and artificial intelligence;

[0046] Among them, 1. Image sampling module; 2. Image processing module; 3. Style model module; 4. Generation module; 5. Interaction control module; 6. Virtual tour module; 11. Image segmentation submodule; 12. Non-uniform sampling submodule; 13. Module data serialization submodule; 31. Style feature extraction submodule; 32. Style library management submodule; 33. Lora submodule; 34. Style parameter generation submodule; 41. Style application submodule; 42. Image reconstruction submodule; 43. Rendering optimization submodule; 51. Prompt word module; 52. Generation parameter module; 53. Visual interaction module. DETAILED DESCRIPTION

[0047] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;

[0048] In order to better illustrate the present embodiment, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product;

[0049] It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0050] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0051] Example

[0052] A digital stylized reconstruction system for flower shows using image sampling and artificial intelligence, such as Figure 1 As shown, image sampling module 1, image processing module 2, style model module 3, generation module 4, interactive control module 5 and virtual tour module 6;

[0053] The image sampling module 1 is used to obtain high-definition image data of the flower show site through an image acquisition device;

[0054] The image processing module 2 preprocesses and optimizes the image data acquired by the image sampling module 1, segments and extracts features of the image data, and extracts information useful for generating stylized images;

[0055] The style model module 3 uses the image data provided by the image processing module 2, combined with the self-training data Lora model, to generate the stylized parameters of the flower show site through the comfyUI automated workflow model combination algorithm;

[0056] The generation module 4 reconstructs the image based on the image generated by the style model module 3 to further construct a complete stylized redrawing;

[0057] The interactive control module 5 is used to control the interactive operation between the user and the stylized description, adjust the parameters of the artificial intelligence generation system, and connect to external sensors;

[0058] The virtual tour guide module 6 provides navigation and guidance services for users, helping them to better understand the layout and characteristics of the flower show site.

[0059] The image sampling module 1 includes an image segmentation submodule 11, a non-uniform sampling submodule 12 and a module data serialization submodule 13;

[0060] The image segmentation submodule 11 is used to segment the image data into multiple small blocks or regions;

[0061] The non-uniform sampling submodule 12 performs image data sampling with emphasis according to the frequency and location of suspected defects in the image data, performing more sampling in areas with a higher frequency of suspected defects and reducing sampling in areas with a lower frequency of defects;

[0062] The module data serialization submodule 13 aggregates and serializes the sampled image block data to form serialized data pairs.

[0063] The style model module 3 includes a style feature extraction submodule 31, a style library management submodule 32, a Lora submodule 33 and a style parameter generation submodule 34;

[0064] The style feature extraction submodule 31 is used to further extract features related to stylization from the image data provided by the image processing module 2; these features may include color distribution, texture pattern, shape elements, etc., which are crucial for the subsequent generation of a reconstructed image with a specific style.

[0065] The style library management submodule 32 is used to manage and maintain a style library, which contains a variety of predefined style templates or style parameters; these styles can be classic art styles (such as oil painting, watercolor painting), modern design styles, or specific styles learned through deep learning models. The style library management submodule 32 allows the system to select or combine different style templates as needed.

[0066] The Lora submodule 33 processes the extracted style features using the self-trained Lora model;

[0067] The Lora model is a fine-tuning technology based on pre-trained models (such as GANs, StyleGAN, etc.), which can capture and simulate specific style features. By adjusting the parameters of the Lora model, reconstruction effects of different styles can be achieved;

[0068] The style parameter generation submodule 34 combines the features extracted by the image processing module and the templates in the style library, and generates stylized parameters for the current flower show scene image through the comfyUI automated workflow model combination algorithm.

[0069] The Lora submodule 33 is used to import the exported model into stable diffusion and perform preliminary integration. The detailed steps include:

[0070] Create a dataset project, collect relevant style images, annotate parameters and prompt words, create a new training model project, and train a stylized migration model;

[0071] Import the model. In ComfyUI, create a workflow. First, build the image import group, the pixel scaling group, and the automatic cropping group. Combine the diffusion model, select the self-trained Lora model file, click Generate, and adjust the parameters according to the effect.

[0072] Adjust the model's generate settings, including weight factors, calculation steps, and fixed prompt words, to ensure that the model can be read normally in stable diffusion;

[0073] Add interaction and VR support, install interaction support plugins, configure software collaboration, test the display and interaction functions of the model in comfyUI, and adjust the model parameters and model group settings as needed.

[0074] The generation module 4 includes a style application submodule 41, an image reconstruction submodule, and a rendering optimization submodule 43;

[0075] The style application submodule 41 applies the stylized parameters generated by the style model module 3 to the image data provided by the image processing module 2, and adjusts the color, texture and contrast properties of the image to make it conform to the target style characteristics;

[0076] Based on the style application submodule 41, the image reconstruction submodule 42 further optimizes and reconstructs the image as a whole; it may include operations such as image scaling, cropping, and rotation to ensure that the stylized image achieves the best visual effect. In addition, this submodule may also be responsible for processing details such as edge effects and color transitions in the image to enhance the overall texture of the image.

[0077] The rendering optimization submodule 43 is used to improve the quality and rendering efficiency of the generated image and optimize the image rendering process.

[0078] 7. The rendering optimization submodule 43 uses the MSAA algorithm, and its detailed steps are as follows:

[0079] Enable MSAA and turn on the MSAA function in the rendering pipeline;

[0080] This is typically done in a graphics API such as OpenGL or DirectX by setting the appropriate state or configuration flags.

[0081] Create an MSAA framebuffer, create an MSAA-enabled framebuffer and assign the necessary color and depth and stencil attachments to it;

[0082] In some cases, the system-provided framebuffer may already support MSAA, without the need for additional video memory.

[0083] Configure the sampling mode, set the desired sampling mode, and determine the number of samples per pixel: for example, 4x MSAA means 4 sub-sampling points per pixel.

[0084] Adjust the viewport and scissor rectangle to ensure that they take into account the MSAA sampling mode so that the rendered output is clipped correctly.

[0085] Perform depth and template tests on each sub-sampling point to generate coverage and occlusion information;

[0086] Perform shading operations at the pixel level, writing color values ​​to sub-sampled points that pass the depth and template tests;

[0087] Perform an MSAAResolve operation to resolve the color value of the sub-sampling point into the final pixel color after rendering is completed.

[0088] The interactive control module 5 includes a prompt word module 51, a parameter generation module 52, and a visual interaction module 53;

[0089] The prompt word module 51 is used to adjust the variable prompt words in the prompt words, self-label the prompt words in the model training, map them into optional parameters, and concentrate the stylization within a fixed range;

[0090] The generation parameter module 52 links the generation parameter data set of the generation system to the physical world sensor, and uses touch or gesture to change the parameters to affect the generated image result;

[0091] The visual interaction module 53 cooperates with the tactile and gesture feedback devices to provide physical touch, superimpose interactive visual effects on the generated image content, and enhance the interactive feedback sense.

[0092] In a specific implementation manner, the image segmentation submodule 11 uses an advanced image segmentation algorithm (such as DeepLabV3+) to segment the high-definition image data of the flower show site into multiple areas with obvious boundaries, such as flowers, backgrounds, decorations, etc. The non-uniform sampling submodule 12 determines the high-incidence areas of suspected defects (such as blur, occlusion) based on historical data or preliminary image analysis results, and increases the sampling point density in these areas, while reducing the sampling points in clear, unobstructed areas to optimize the image data acquisition efficiency. The module data serialization submodule 13 serializes the segmented image blocks and their related information (such as position, size, sampling density) to form data pairs that are convenient for subsequent processing.

[0093] Image processing module 2: This module preprocesses and optimizes the collected image data, including denoising, contrast enhancement, color correction, etc. It also uses the image segmentation results to extract features and provide basic data for stylized generation.

[0094] Style model module 3, style feature extraction submodule 31 extracts key features related to stylization from the preprocessed image, such as color distribution, texture pattern, etc., style library management submodule 32 establishes and maintains a style library containing a variety of predefined style templates, such as oil painting style, watercolor style, etc., Lora submodule 33 first trains the Lora model according to the collected style image data set, and then imports the trained Lora model into the stable diffusion framework and integrates it through the ComfyUI automated workflow. In ComfyUI, set the workflow, including image import, scaling, cropping, style transfer and other steps, and finally generate the stylized parameters of the image for the current flower show scene, style parameter generation submodule 34 combines the templates in the style library and the stylized parameters generated by the Lora model, and generates personalized stylized parameters adapted to the current image content through the comfyUI automated workflow model combination algorithm.

[0095] Generation module 4, style application submodule 41 applies stylized parameters to the preprocessed image data, and makes the image present the target style characteristics by adjusting the attributes such as color, texture and contrast. Image reconstruction submodule 42 performs global optimization and reconstruction on the image based on the style application to ensure the overall harmony and visual beauty of the image. Rendering optimization submodule 43 uses MSAA (multi-sampling anti-aliasing) algorithm to optimize the rendering of the generated image. Turn on the MSAA function, create a framebuffer that supports MSAA, set the sampling mode, adjust the viewport and scissor rectangle, perform depth and template tests, shading operations, and finally perform MSAAResolve operations to generate high-quality rendered images.

[0096] The interactive control module 5 and the prompt word module 51 provide the user with adjustable prompt word parameters. The user can choose different stylized prompt words according to personal preferences, and the system automatically maps them to corresponding stylized parameters. The generation parameter module 52 connects the parameters of the generation system with the sensors of the physical world (such as touch screens and gesture recognition devices). The user can adjust the generation parameters in real time through touch or gestures to affect the image generation results. The visual interaction module 53 combines tactile and gesture feedback devices to provide users with an intuitive interactive experience. On the basis of generating images, interactive visual effects are superimposed, such as changing the image color when touched, adjusting the image style when the gesture slides, etc., to enhance the user's immersion and participation.

[0097] Virtual Guide Module 6 This module constructs a virtual flower show environment based on the generated stylized images. Users can enter the virtual flower show through mobile devices or VR devices, freely navigate and browse various exhibition areas, and understand the layout and characteristics of the flower show. The system also provides voice guide and graphic introduction functions to help users understand the flower show information more comprehensively.

[0098] The same or similar reference numerals correspond to the same or similar components;

[0099] The terms used in the drawings to describe positional relationships are only used for illustrative purposes and should not be construed as limiting this patent;

[0100] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not limitations on the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made on the basis of the above description. It is not necessary and impossible to list all the implementation methods here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A flower show digital stylization reconstruction system using image sampling and artificial intelligence, characterized in that: Image sampling module (1), image processing module (2), style model module (3), generation module (4), interactive control module (5) and virtual tour module (6); The image sampling module (1) is used to obtain high-definition image data of the flower exhibition site through an image acquisition device; The image processing module (2) preprocesses and optimizes the image data acquired by the image sampling module (1), segments and extracts features of the image data, and extracts information useful for generating stylized images; The style model module (3) uses the image data provided by the image processing module (2) and combines it with the self-training data Lora model to generate the stylized parameters of the flower show site through the comfyUI automated workflow model combination algorithm; The generation module (4) reconstructs the image based on the image generated by the style model module (3) to further construct a complete stylized redrawing; The interactive control module (5) is used to control and process the interactive operation between the user and the stylized description, adjust the parameters of the artificial intelligence generation system, and connect to external sensors; The virtual tour guide module (6) provides navigation and guidance services for users, helping them to better understand the layout and features of the flower show site.

2. A flower show digital stylization reconstruction system using image sampling and artificial intelligence according to claim 1, characterized in that: The image sampling module (1) comprises an image segmentation submodule (11), a non-uniform sampling submodule (12) and a module data serialization submodule (13); The image segmentation submodule (11) is used to segment the image data into a plurality of small blocks or regions; The non-uniform sampling submodule (12) performs image data sampling with emphasis according to the frequency and location of suspected defects in the image data, performing more sampling in areas with a higher frequency of suspected defects and reducing sampling in areas with a lower frequency of defects; The module data serialization submodule (13) aggregates and serializes the sampled image block data to form serialized data pairs.

3. The flower show digital stylization reconstruction system using image sampling and artificial intelligence according to claim 1, characterized in that: The style model module (3) includes a style feature extraction submodule (31), a style library management submodule (32), a Lora submodule (33) and a style parameter generation submodule (34); The style feature extraction submodule (31) is used to further extract features related to stylization from the image data provided by the image processing module (2); The style library management submodule (32) is used to manage and maintain a style library, which contains a variety of predefined style templates or style parameters; The Lora submodule (33) processes the extracted style features using a self-trained Lora model; The style parameter generation submodule (34) combines the features extracted by the image processing module and the templates in the style library, and generates stylized parameters for the current flower show scene image through the comfyUI automated workflow model combination algorithm.

4. The flower show digital stylization reconstruction system using image sampling and artificial intelligence according to claim 3, characterized in that: The Lora submodule (33) is used to import the exported model into stable diffusion and perform preliminary integration. The detailed steps include: Create a dataset project, collect relevant style images, annotate parameters and prompt words, create a new training model project, and train a stylized migration model; Import the model. In ComfyUI, create a workflow. First, build the image import group, the pixel scaling group, and the automatic cropping group. Combine the diffusion model, select the self-trained Lora model file, click Generate, and adjust the parameters according to the effect. Adjust the model's generate settings, including weight factors, calculation steps, and fixed prompt words, to ensure that the model can be read normally in stable diffusion; Add interaction and VR support, install interaction support plugins, configure software collaboration, test the display and interaction functions of the model in comfyUI, and adjust the model parameters and model group settings as needed.

5. The flower show digital stylization reconstruction system using image sampling and artificial intelligence according to claim 1, characterized in that: The generation module (4) includes a style application submodule (41), an image reconstruction submodule (42) and a rendering optimization submodule (43); The style application submodule (41) applies the stylized parameters generated by the style model module (3) to the image data provided by the image processing module (2), and adjusts the color, texture and contrast properties of the image to make it conform to the target style characteristics; The image reconstruction submodule (42) further optimizes and reconstructs the image as a whole based on the style application submodule (41); The rendering optimization submodule (43) is used to improve the quality and rendering efficiency of the generated image and optimize the image rendering process.

6. The flower show digital stylization reconstruction system using image sampling and artificial intelligence according to claim 5, characterized in that: The rendering optimization submodule (43) uses the MSAA algorithm, and its detailed steps are as follows: Enable MSAA and turn on the MSAA function in the rendering pipeline; Create an MSAA framebuffer, create an MSAA-enabled framebuffer and assign the necessary color and depth and stencil attachments to it; Configure the sampling mode, set the desired sampling mode, and determine the number of sampling points for each pixel: Adjust the viewport and scissor rectangle to ensure that they take into account the MSAA sampling mode so that the rendered output is clipped correctly. Perform depth and template tests on each sub-sampling point to generate coverage and occlusion information; Perform shading operations at the pixel level, writing color values ​​to sub-sampled points that pass the depth and template tests; Perform the MSAA Resolve operation to resolve the color value of the sub-sampling point into the final pixel color after rendering is completed.

7. The flower show digital stylization reconstruction system using image sampling and artificial intelligence according to claim 1, characterized in that: The interactive control module (5) includes a prompt word module (51), a parameter generation module (52), and a visual interaction module (53); The prompt word module (51) is used to adjust the variable prompt words in the prompt words, self-label the prompt words in the model training, map them into optional parameters, and concentrate the stylization within a fixed range; The generation parameter module (52) links the generation parameter data set of the generation system to the physical world sensor, and uses touch or gesture to change the parameters to affect the generated image result; The visual interaction module (53) cooperates with the tactile and gesture feedback devices to provide physical touch, and superimposes interactive visual effects on the basis of generated image content to enhance the interactive feedback sense.

Images