Game NPC drawing system based on artificial intelligence

Abstract

The invention discloses a game NPC drawing system based on artificial intelligence, and belongs to the computer game and artificial intelligence technology. A game NPC drawing system based on artificial intelligence forms a drawing model of an NPC painter in a game by utilizing an existing game server, and comprises a creation module, an NPC evaluation module and a memory flow module, a processing result and creation feedback are obtained through the memory flow module and the creation module so as to realize data storage and model calling; the NPC evaluation module is used for evaluating the portrait image by other NPCs set in the game to obtain corresponding evaluation data; the evaluation data is stored in the memory flow module; the self-boosting module calls the data in the memory flow module through the natural language processing module and extracts the data to generate new creation data; therefore, the NPC painter is more personalized, real-time and interactive.

Description

Technical Field

[0001] The present invention belongs to computer games and artificial intelligence technology, and more specifically, to a game NPC painting system based on artificial intelligence. Background Art

[0002] With the widespread application of artificial intelligence technology in the field of computer games, the behavior simulation and interaction capabilities of NPCs in virtual communities and video games have gradually attracted attention. However, current technology still has many shortcomings; The existing NPC image generation technology mainly relies on preset image libraries and style models, as well as simple trigger conditions and interaction logic. This approach has obvious limitations: First, the generated pictures lack personalization. The preset picture library and style model are often fixed and fail to fully consider the unique characteristics, personality, experience and specific context of each NPC. This results in the generated pictures being all the same and unable to meet the players' demand for a unique visual experience. Secondly, the real-time performance is seriously insufficient. Due to the reliance on fixed preset modes, it is difficult for NPCs to quickly and accurately generate pictures that are closely integrated with the current situation in the dynamic changes of the game. The situation in the game changes rapidly, such as the player's behavior, the progress of the game, changes in the environment, etc., and the existing technology cannot respond to these changes in time, resulting in a clear disconnect between the pictures and the game scenes, which greatly reduces the player's immersion. In addition, the existing technology lacks effective evaluation feedback and self-improvement mechanisms; this means that NPCs cannot improve their creations based on the evaluations of other NPC characters, and cannot continuously improve their own creation level and performance capabilities; this limitation restricts further optimization of the game experience, making the game content easily monotonous and lacking in innovation; More importantly, for example, the Stanford Town project only used one AI model in practice, which further limited the diversity and depth of NPC image generation and behavior simulation; a single model often cannot fully capture and express the complexity and diversity of NPCs, resulting in a lack of hierarchy and delicacy in the generated images and behaviors; In summary, the existing NPC image generation and behavior simulation technologies have deficiencies in personalization, real-time, realism, deep interactivity, as well as evaluation feedback and self-improvement mechanisms, and cannot meet the players' growing demand for high-quality, immersive gaming experiences. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a game NPC painting system based on artificial intelligence, which can achieve a more personalized, real-time and interactive gaming experience.

[0004] The present invention provides an artificial intelligence-based game NPC painting system, which utilizes an existing game server to form a painting model of an NPC painter set in the game, including a creation module, an NPC evaluation module, and a memory flow module; Creation module: used for game players to generate painting images through NPC painters; the creation module includes image generation module, natural language processing module, text embedding module, and player interaction module; players input instructions, which are transmitted to the game server, and the game server calls the natural language processing module, text embedding module, and player interaction module respectively to form the creation data of the game NPC painter, and the image generation module generates the painting image according to the creation data; NPC evaluation module: The NPC evaluation module is used for other NPCs set in the game to evaluate the portrait images and output evaluation data; Memory stream module: used for data storage, including game status, player behavior, NPC painters' works, NPCs' overall experience and evaluation data; evaluation data includes evaluation data of paintings by other NPCs in the game; Self-improvement module: The natural language processing module calls the relevant memory data in the memory flow module, refines it, and generates new creative data.

[0005] As a further improvement of the present invention, the player interaction module is used to receive player input instructions and provided comments, and transmit them to the natural language processing module; Image generation module: including the Wensheng graph model, which combines the game state information and the memory experience in the memory flow module to generate images; Text embedding module: includes text embedding model, through which evaluation data is processed and analyzed; Natural language processing module: includes a natural language processing model. The natural language processing model provides detailed suggestions on creative style, theme selection, and expression techniques based on the game status, player behavior, and memory experience. The information is accurately transmitted to the NPC painter.

[0006] As a further improvement of the present invention, the comprehensive experience of the NPC in the memory flow module includes each creative process of the NPC painter and player feedback.

[0007] As a further improvement of the present invention, other NPCs provided in the game are NPCs of different types and personality characteristics.

[0008] As a further improvement of the present invention, the game server is a Stanford Town server.

[0009] As a further improvement of the present invention, the text graph model is the CogView-3-Plus text graph model; the natural language processing model is the GPT-3.5-turbo model; and the text embedding model is the text-embedding-3-small model.

[0010] As a further improvement of the present invention, the self-improvement module includes a reflection mechanism, the NPC painter regularly generates reflection questions, deeply retrieves the data stored in the memory flow module, and generates reflection results through a natural language processing model; the reflection questions are generated based on the NPC painter's experience.

[0011] As a further improvement of the present invention, the specific process of image generation, evaluation feedback and self-improvement includes the following steps: S1: The player inputs painting instructions to the NPC painter; S2: The NPC painter calls the data in the memory stream module according to the player's instructions to obtain the creation data; and creates the painting through the creation module according to the creation data to generate an image; S3: After the image is generated, the evaluation and feedback process is automatically triggered. Other NPCs will evaluate and feedback the painting image. All evaluation data will be transmitted to the system in data format and processed and analyzed through the text embedding model, which will be converted into a form that can be understood and applied by NPC painters, providing reference and improvement direction for the painters' subsequent creations. S4: Through S3 evaluation feedback, NPC painters adjust their creation data and complete self-improvement.

[0012] As a further improvement of the present invention, the evaluation of other NPCs is based on preset multi-dimensional evaluation criteria, including the creativity of the painting, the skill level, the fit with the current game situation, the use of colors, and the rationality of the composition.

[0013] As a further improvement of the present invention, the evaluation data is input into a text embedding model; the text embedding model processes the information, converts the text content into vector form and inputs it into a natural language processing model; the natural language processing model provides suggestions based on the current game situation, player needs and overall creative trends; the integrated information is passed to the NPC painter; the NPC painter conducts a comprehensive and comprehensive analysis of the information and identifies the parts that are valuable to his or her own creation.

[0014] Compared with the prior art, the present invention has the following beneficial effects: First, by integrating advanced AI models, the NPC painters can generate personalized and real-time pictures; the text-based model can generate realistic and diverse images, and the natural language processing model and text embedding model can accurately understand the player's needs and game context; the three work together, combined with the memory flow module, so that the generated pictures not only fit the NPC's personal preferences, but also respond to various changes in the game in real time, greatly enriching the visual experience of the game, allowing players to experience unique and context-appropriate pictures, thereby enhancing the fun and appeal of the game; The CogView-3-Plus document graph model is used as the document graph model, the GPT-3.5-turbo model is used as the natural language processing model, and the text embedding model is the text-embedding-3-small model. High-quality models are preferred. Secondly, the evaluation and interaction mechanism between NPCs significantly enhances the interactive depth and realism of the game. Different NPCs are evaluated based on preset multi-dimensional evaluation criteria, and the evaluation data is processed to provide valuable reference for painters. This allows players to no longer simply interact with a single NPC in the game, but to participate in a virtual community full of social interaction and feedback. It allows players to participate in the communication and interaction between NPCs, as if they were in a real art community, which greatly enhances the realism and richness of the game. Third, the introduction of the self-improvement mechanism enables NPC painters to continue to grow and improve; through the integration of reflection mechanisms, evaluation feedback and creative suggestions, NPC painters can continuously adjust their creative strategies and parameters to improve their creative level; providing players with more high-quality and constantly evolving game content, allowing players to witness the growth process of NPC painters, increasing the long-term playability of the game and the loyalty of players; Fourthly, by optimizing the image generation and interactive mechanism, the overall quality and competitiveness of the game are improved; high-quality image generation and in-depth interaction can attract more players, increase the market share and economic benefits of the game; and provide a more personalized, real-time and interactive gaming experience, thus standing out in the fiercely competitive gaming market.

[0015] Finally, the present invention provides new ideas and methods for the combination of computer games and artificial intelligence technology; it has positive significance for promoting technological development and innovation in related fields, and provides useful reference and examples for future game development and artificial intelligence applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the system architecture of the present invention; Figure 2 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0017] The present invention is further described in detail below in conjunction with embodiments. The following embodiments are for explanation of the present invention but the present invention is not limited to the following embodiments.

[0018] Specific embodiment 1: Please refer to Figure 1-2 An artificial intelligence-based game NPC painting system uses the existing game server to form a painting model of the NPC painter set in the game. The NPC painter is a virtual character that can create independently, including a creation module, an NPC evaluation module and a memory flow module; Creation module: used for game players to generate painting images through NPC painters; the creation module includes image generation module, natural language processing module, text embedding module, and player interaction module; players input instructions, which are transmitted to the game server, and the game server calls the natural language processing module, text embedding module, and player interaction module respectively to form the creation data of the game NPC painter, and the image generation module generates the painting image according to the creation data; NPC evaluation module: including NPC evaluation module; NPC evaluation module is used for evaluating portrait images of other NPCs set in the game and outputting evaluation data; Memory stream module: used for data storage, including game status, player behavior, NPC painters' works, NPCs' overall experience and evaluation data; evaluation data includes evaluation data of paintings by other NPCs in the game; Self-improvement module: The natural language processing module calls the data in the memory flow module, refines it, and generates new creative data.

[0019] Through the memory flow module and the creation module, the processing results and creation feedback are obtained to realize data storage and model calling.

[0020] Player interaction module: used to receive player input instructions and comments, and transmit them to the natural language processing module; Image generation module: including the Wensheng graph model, which combines the game state information and the memory experience in the memory flow module to generate images; Text embedding module: including text embedding model, through which evaluation data is finely processed and deeply analyzed; Natural language processing module: includes natural language processing model. The natural language processing model provides detailed suggestions covering creative style, theme selection, expression method and other aspects according to the game status, player behavior and memory experience. The information will be accurately transmitted to the NPC painter.

[0021] The comprehensive experience of the NPC in the memory stream module includes information such as each creative process of the NPC painter, player feedback, etc.

[0022] Preferably, the text graph model is the CogView-3-Plus text graph model; the natural language processing model is the GPT-3.5-turbo model; and the text embedding model is the text-embedding-3-small model.

[0023] Optionally, the text graph model can also be StableDiffusion, Midjourney, etc.; the language processing model can also be Skylark, Wenxinyiyan, etc.; the text embedding model can also be Sentence-BERT, etc.

[0024] Optionally, the other NPCs may be NPCs of different types and personality traits. For example, an art appreciation NPC may focus more on the subtlety of technique and composition, while a lively NPC may focus more on creativity and vividness of color.

[0025] Preferably, the game server is the Stanford Town server; through the Stanford Town server, the CogView-3-Plus text graph model, the GPT-3.5-turbo natural language processing model and the text-embedding-3-small text embedding model are integrated to support players to input commands, provide comments, and interact with NPC painters; the memory stream module is responsible for storing game status, player behavior and creative works, and through the memory module and the creation module, the processing results and creative feedback are obtained to realize data storage and the calling of each model.

[0026] The self-improvement module includes a reflection mechanism. The NPC painter regularly generates reflection questions, deeply retrieves relevant memory data stored in the memory stream module, and generates profound reflection results through the natural language processing model. These reflection questions are generated based on the NPC painter's recent experiences and aim to extract high-level thinking; enabling the NPC painter to deeply refine the memories in the memory stream and generate more insightful thinking, thereby making more wise and reasonable creative decisions.

[0027] In the game, when an NPC painter completes a work, the system will automatically trigger the NPC evaluation module. After the NPC evaluation module is triggered, other NPCs will evaluate the work. The evaluation of other NPCs is based on preset multi-dimensional evaluation criteria, including the creativity of the painting, skill level, fit with the current game situation, color application, rationality of composition and many other aspects. All evaluation data will be transmitted to the system in a specific data format, and will be processed and analyzed in depth through the text embedding model, and converted into a form that can be easily understood and effectively applied by NPC painters, providing valuable references and clear improvement directions for the painters' subsequent creations. This evaluation and interaction mechanism between NPCs breaks the rigid limitations of traditional NPC behavior patterns, allowing NPCs to communicate and give feedback like a real artist community, greatly enhancing the realism and interactive depth of the game.

[0028] When other NPCs evaluate the painter's work, the evaluation data will first be fully collected. This information not only includes the evaluation content in text form, but may also contain rich information such as emotional tendencies and semantic emphasis; the evaluation data is input into the text embedding model; the text embedding model processes this information and converts the text content into vector form and inputs it into the natural language processing model; the natural language processing model provides detailed suggestions covering creative style, theme selection, expression techniques, etc. based on the current complex and changing game situation, player needs and overall creative trends; the integrated information is passed to the NPC painter; the NPC painter conducts a comprehensive and comprehensive analysis of this information and identifies the parts that are of great value to his own creation.

[0029] like Figure 2 The flowchart shown in the figure describes the specific process of image generation, evaluation feedback and self-improvement of the system, including the following steps: S1: The player inputs painting instructions to the NPC painter; S2: The NPC painter calls the data in the memory stream module according to the player's instructions to obtain the creation data; and creates the painting through the creation module according to the creation data to generate an image; S3: After the image is generated, the evaluation and feedback process is automatically triggered. Other NPCs will evaluate and feedback the painting image. All evaluation data will be transmitted to the system in a specific data format and processed and analyzed in depth through the text embedding model. It will be converted into a form that can be easily understood and effectively applied by NPC painters, providing valuable references and clear improvement directions for the painters' subsequent creations. S4: Through S3 evaluation feedback, NPC painters adjust their creation data and complete self-improvement.

[0030] Based on the painting model, the NPC painter becomes a virtual character that can not only create independently but also improve itself based on evaluation feedback.

[0031] NPC painters have the ability to dynamically generate paintings that match their identity and artistic style based on game logic, player interaction, historical memory, and other NPC evaluations.

[0032] Specific examples: Players input commands, which are transmitted to the game server. The game server calls the image generation module, natural language processing module, and text embedding module respectively. The image generation model, natural language processing model, and text embedding model work together, and the NPC painter generates personalized pictures based on the real-time game status, player interaction, and memory flow, combined with the NPC's comprehensive experience recorded by the memory flow module. This personalized generation is no longer limited to fixed modes and preset styles, but can truly fit each game situation and player needs, bringing players unprecedented visual enjoyment.

[0033] After the picture is generated, the system automatically triggers the process of other NPCs evaluating the work; for example: if the evaluation mentions the color matching is not harmonious many times, and the natural language processing model suggests trying a new color combination, the NPC painter will take this key information into consideration and flexibly adjust his or her color application strategy; in subsequent creations, the NPC painter will carefully adjust his or her own creative parameters such as line thickness, color saturation, composition method, etc. according to these precise feedback and valuable suggestions; at the same time, the NPC painter will also closely combine previous creative experience and memory with new feedback, constantly optimize his or her own creative process and unique style, and achieve continuous iterative improvement of creative ability.

[0034] The above is only a preferred specific implementation method of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes within the technical scope disclosed in the present application according to the technical solution and improved ideas of the present application, which should be covered by the protection scope of the present application.

Claims

1. An artificial intelligence-based game NPC painting system, which uses an existing game server to form a painting model of an NPC painter set in the game, characterized in that: Includes creation module, NPC evaluation module and memory flow module; Creation module: used for game players to generate painting images through NPC painters; the creation module includes image generation module, natural language processing module, text embedding module, and player interaction module; players input instructions, which are transmitted to the game server, and the game server calls the natural language processing module, text embedding module, and player interaction module respectively to form the creation data of the game NPC painter, and the image generation module generates the painting image according to the creation data; NPC evaluation module: The NPC evaluation module is used for other NPCs set in the game to evaluate the portrait images and output evaluation data; Memory stream module: used for data storage, including game status, player behavior, NPC painters' works, NPCs' overall experience and evaluation data; evaluation data includes evaluation data of paintings by other NPCs in the game; Self-improvement module: The natural language processing module calls the relevant memory data in the memory flow module, refines it, and generates new creative data.

2. The game NPC drawing system based on artificial intelligence according to claim 1, characterized in that: Player interaction module: used to receive player input instructions and comments, and transmit them to the natural language processing module; Image generation module: including the Wensheng graph model, which combines the game state information and the memory experience in the memory flow module to generate images; Text embedding module: includes text embedding model, through which evaluation data is processed and analyzed; Natural language processing module: includes a natural language processing model. The natural language processing model provides detailed suggestions covering creative style, theme selection, and expression techniques based on game status, player behavior, and memory experience. The information is accurately transmitted to the NPC painter.

3. The game NPC drawing system based on artificial intelligence according to claim 1, characterized in that: The comprehensive experience of the NPC in the memory stream module includes each creative process of the NPC painter and player feedback.

4. The game NPC drawing system based on artificial intelligence according to claim 1, characterized in that: The other NPCs set in the game are NPCs of different types and personality characteristics.

5. The game NPC drawing system based on artificial intelligence according to claim 1, characterized in that: The game server is the Stanford Town server.

6. The game NPC drawing system based on artificial intelligence according to claim 2, characterized in that: The document graph model is the CogView-3-Plus document graph model; the natural language processing model is the GPT-3.5-turbo model; and the text embedding model is the text-embedding-3-small model.

7. The game NPC drawing system based on artificial intelligence according to claim 1, characterized in that: The self-improvement module includes a reflection mechanism. The NPC painter regularly generates reflection questions, deeply retrieves the data stored in the memory stream module, and generates reflection results through the natural language processing model; the reflection questions are generated based on the NPC painter's experience.

8. The game NPC drawing system based on artificial intelligence according to claim 1, characterized in that: The specific process of image generation, evaluation feedback and self-improvement includes the following steps: S1: The player inputs painting instructions to the NPC painter; S2: The NPC painter calls the data in the memory stream module according to the player's instructions to obtain the creation data; and creates the painting through the creation module according to the creation data to generate an image; S3: After the image is generated, the evaluation and feedback process is automatically triggered. Other NPCs will evaluate and feedback the painting image. All evaluation data will be transmitted to the system in data format and processed and analyzed through the text embedding model, which will be converted into a form that can be understood and applied by NPC painters, providing reference and improvement direction for the painters' subsequent creations. S4: Through S3 evaluation feedback, NPC painters adjust their creation data and complete self-improvement.

9. The game NPC drawing system based on artificial intelligence according to claim 1, characterized in that: The evaluation of other NPCs is based on preset multi-dimensional evaluation criteria, including the creativity of the painting, skill level, fit with the current game situation, use of color, and reasonableness of composition.

10. The game NPC drawing system based on artificial intelligence according to claim 8, characterized in that: The evaluation data is input into the text embedding model; the text embedding model processes this information, converts the text content into vector form and inputs it into the natural language processing model; the natural language processing model provides suggestions based on the current game situation, player needs and overall creative trends; the integrated information is passed to the NPC painter; the NPC painter conducts a comprehensive and comprehensive analysis of this information to identify the parts that are valuable to his or her own creation.

Images