Scenario-centered interaction-driven scene generation system

Abstract

The invention, which relates to the technical field of interactive script generation, discloses a script-centered interaction-driven scene generation system comprising the following modules: a script management module, a script arrangement module, a dependency propagation module, a thesaurus interaction cooperation module and an execution output module. Creating a basic data structure of the script, distributing script attributes, and managing script elements, constraints and logics; identifying scene nodes and events, constructing three-layer dependency and weighting, and generating a reverse causal chain through recursive dependency and depth-first search; by maintaining a dependency table, event-driven updating and dirty node marking for each node, efficient dependency propagation and low-delay updating are realized through a priority queue and differential generation; dividing three levels to cooperatively realize dynamic script generation and interaction, and realizing script control based on gating residual connection, a reverse causal chain and a time attention mechanism; and constructing an interactive execution layer, integrating a real-time feedback mechanism, automatically generating a script, and outputting and displaying an executable format.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of interactive script generation, and in particular to a script-centered interactive driving scene generation system. BACKGROUND

[0002] With the rapid development of computer technology, artificial intelligence, virtual reality and augmented reality technology, the traditional linear narrative method has been unable to meet the needs of users for personalized and diversified interactive experiences. Although existing interactive media can provide basic user selection and plot branching, there are still certain limitations in handling complex plot dynamics, user behavior feedback and multi-dimensional interactive experiences. The traditional preset scene and plot development method cannot adapt to rapidly changing user behavior and interactive methods.

[0003] The prior art has the following disadvantages:

[0004] (1) There is a tendency to drift in terms of cross-scene style consistency, emotional trend and character behavior consistency, making it difficult to always follow the structure and style constraints set by the script, and lacking controllable coverage of complex plots;

[0005] (2) In multi-round dialogues and action decisions, it is often difficult to maintain a clear causal chain, leading to plot breaks, misplaced motives or contradictions, affecting the viewing experience and playability;

[0006] (3) The timing alignment between text, speech, expressions, actions and scene environment is still weak, making it difficult to maintain consistency between different modalities, resulting in a decline in scene immersion or instability in the execution level;

[0007] (4) There is a lack of clear explainability path, making it difficult for creators to understand the basis for generated decisions and to accurately adjust specific scene elements.

[0008] The above information disclosed in the background section is only used to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0009] The purpose of the present application is to provide a script-centered interactive driving scene generation system to solve the problems in the background art.

[0010] In order to achieve the above-mentioned purpose, the present application provides the following technical solution: a script-centered interactive driving scene generation system, specifically comprising the following modules: a script management module, a script arrangement module, a dependency propagation module, a narrative control interaction coordination module and an execution output module.

[0011] Script management module: create the basic data structure of the script, and assign the script attributes, manage the script elements, constraints and logic;

[0012] Script arrangement module: identify scene nodes and events, build three-layer dependencies and weight, generate reverse causal chains through recursive dependency and depth-first search;

[0013] Dependency propagation module: maintain dependency table for each node, event-driven update and dirty node marking, efficient dependency propagation, fast redraw and low latency update through priority queue and difference generation;

[0014] Narrative control interaction collaboration module: divide three levels of collaboration to realize dynamic script generation and interaction, realize script control based on gated residual connection, reverse causal chain and time attention mechanism;

[0015] Execution output module: build an interactive execution layer, integrate real-time feedback mechanism, automatically generate and output executable format.

[0016] As a preferred scheme of the script-centered interactive driving scene generation system described in the application, wherein:

[0017] Create the basic data structure of the script, including nodes and edges between nodes, and assign script attributes to each node, including role, intent, emotion vector, duration prompt, and modifiability flag.

[0018] Define the metadata of the role script attribute in the role library, including role id, role attribute, initial state, available action / permission;

[0019] When the role script attribute is selected, trigger the symbol checker, including checking and judging role conflicts, prompting alternatives or prohibitions;

[0020] Associate and inject the current state of the role script attribute into the context of the script;

[0021] Build an extensible label library of intent script attributes, provide intent selection controls through the UI interface and allow free text idea description, select the selected intent as the highest priority instruction for attention bias and generation prompt of the narrative layer of the script;

[0022] Build a dimension list of emotional script attributes, provide an emotional time series encoder through a visual editor;

[0023] According to the generation stage of the script, embed the emotional curve of the script into the script control layer through the emotional time series encoder, and jointly encode based on the time axis and the current world state of the role;

[0024] The duration prompt script attribute is taken as a constraint condition in a script generation pipeline, and in the presence of multiple time length constraint basis nodes, a final market strategy is autonomously defined according to the priority and intensity of the nodes;

[0025] Whether the node is marked as a hard constraint is judged when the node is created;

[0026] When generating a script, the node is taken as a must-satisfied rule of a script narrative layer;

[0027] Symbolic verification is performed after the script is generated, and if the hard constraint is not satisfied, local script repair is triggered or the script generation result is directly rejected.

[0028] As a preferred scheme of the script-centered interactive driving scene generation system provided by the application, wherein:

[0029] Each scene node and event node in the script is identified, defined and classified;

[0030] A dependency graph among a script narrative layer, a script control layer and a script interaction layer is constructed, and the weight values of the intention, emotion vector and duration prompt among each node in the dependency graph are increased;

[0031] Based on recursive dependency, the current node is dependent on multiple preceding nodes, and through depth-first search, the causal relationship of the script is traced according to the recursive dependency when the script is generated, and a reverse causal chain is generated;

[0032] The preceding and following relationship of each emotion node is traced through the reverse causal chain, and the logical consistency of the plot is checked;

[0033] A hybrid control architecture of a symbolic layer and a neural layer is constructed, and a large language model is used to enrich the script based on the rich script narrative layer of the script;

[0034] A strategy network is defined as a controller, and the hard constraint selection is prioritized according to the requirements of each decision point.

[0035] As a preferred scheme of the script-centered interactive driving scene generation system provided by the application, wherein:

[0036] When the script is interacted, a dependency relationship table is initialized for each node, and other nodes dependent on each node are recorded;

[0037] When the node changes, the downstream nodes of the node in the dependency graph are checked;

[0038] All affected nodes are found by traversing the dependency graph;

[0039] When the script is updated, all affected nodes are iteratively updated based on the modified node as an initial point;

[0040] Dynamically update dependency graph by event-driven manner;

[0041] Register event listener on each node, trigger update immediately when node content changes;

[0042] Mark affected nodes, and only update these marked dirty nodes in next rendering;

[0043] Update processing node by dispatch queue, queue all affected nodes by priority, and generate reduced latency by difference.

[0044] As a preferred scheme of the script-centered interactive driving scene generation system of the present application, wherein:

[0045] Link the script narrative layer as high-level context, and query the script control layer and the script interaction layer as low-level to the script narrative layer, and return the low-level feedback to the high-level through the gated residual connection;

[0046] Based on the continuous action and video sequence of the script, add time attention of the window through the reverse causal chain;

[0047] Set independent control strategy for each layer, specifically including:

[0048] Script narrative layer: based on the emotion curve and node priority of the script, generate the event sequence sketch of the script through the planner, and provide the script control layer for script unfolding;

[0049] Script control layer: generate the event chain of the script based on the event sequence sketch;

[0050] Script interaction layer: connect the fine-grained generator API and the physical post-processor API, allow direct modification of the key frame of the continuous action and video sequence of the script, and feedback to the high-level;

[0051] Close the narrative vector of the script narrative layer and the visual encoding of the corresponding script interaction layer in the same batch, and zoom out the mismatched pair;

[0052] Do similarity regularization on the event vector of the script control layer and the interaction action vector of the corresponding script interaction layer to encourage semantic matching;

[0053] Generate the expected middle-level statistics according to the script narrative layer, and compare the currently generated low-level statistics;

[0054] Based on the script interaction layer and the reverse causal chain, verify the periodic consistency of the narrative vector of the reverse prediction;

[0055] Based on the script-based interactive depth hierarchical unfolding, specifically including:

[0056] Low-level interaction: Only the script narrative layer and script control layer are used to automatically generate interactive actions for the script interaction layer;

[0057] Medium-level interaction: Through editing nodes in the script control layer, interactive actions in the script interaction layer are generated based on the script narrative layer;

[0058] Highly interactive: By directly editing the interactive actions of the script interaction layer, a bottom-up consistency check is performed and local script updates are sent to the script control layer and the script narrative layer.

[0059] As a preferred embodiment of the script-centric interactive scene generation system described in this invention, wherein:

[0060] Build an interactive execution layer to dynamically adjust the plot and the order of events based on user input;

[0061] It integrates a real-time feedback mechanism to quickly update the plot status and character behavior, while maintaining the consistency of the script's logic;

[0062] The automated script generation algorithm automatically generates script content under given context or input conditions and adjusts it according to preset conditions.

[0063] The final output of the script is converted into an executable format, including text, actions, camera transitions, sound, and background music;

[0064] The generated script's interactive scenes are displayed on different platforms.

[0065] On the other hand, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements the steps of the script-centric interactive scene generation system as described above.

[0066] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements the steps of the script-centric interactive scene generation system as described above.

[0067] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0068] (1) By combining emotional vectors, intentional script attributes and emotional time sequence encoders, the emotions and behaviors of the characters can be adjusted in real time during the script generation process to ensure the consistency and logic of the plot, while adapting to the input and interaction needs of users.

[0069] (2) By introducing recursive dependency relationships and reverse causal chains, the pre and post relationships of each node can be traced more accurately, reducing logical errors and inconsistencies in the plot. The affected nodes can be dynamically identified and adjusted, making the entire script generation process more natural and smooth.

[0070] (3) By embedding emotion curves into the script control layer and combining with duration prompts, the duration of character behavior in the script can be accurately adjusted. Based on the priority and intensity of different nodes, the script content can be adjusted to better achieve dynamic performance of characters and smooth transition of emotional changes.

[0071] (4) By using hard constraint markers, symbol verification and script repair mechanisms, the generation process can be forced to meet the preset rules, avoiding the generation of content that does not meet the constraints, ensuring the compliance of the generated script, and enabling immediate repair.

[0072] (5) By designing an interactive execution layer, the plot can be dynamically adjusted according to user input, and the behavior of characters and the order of plot events can be updated in real time, allowing the script generation to be highly personalized according to user needs, enhancing user experience and interactivity.

[0073] (6) By combining the symbol layer with the neural layer through a hybrid control architecture, a policy network is used as the controller to achieve a more intelligent decision-making mechanism. The generation strategy can be adjusted according to the complexity and requirements of the script, improving the quality and flexibility of script generation. BRIEF DESCRIPTION OF DRAWINGS

[0074] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art based on these drawings.

[0075] Figure 1 The method flowchart of the script-centered interactive driving scene generation system of the present application.

[0076] Figure 2 The module schematic diagram of the script-centered interactive driving scene generation system of the present application. DETAILED DESCRIPTION

[0077] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the scope of example implementations to those skilled in the art.

[0078] Embodiment 1, refer to Figure 1 and Figure 2 As the first embodiment of the present application, the embodiment provides a script-centered interactive driving scene generation system, which specifically comprises the following modules: a script management module, a script arrangement module, a dependency propagation module, a narrative control interaction coordination module, and an execution output module.

[0079] The script management module: creates the basic data structure of the script and assigns script attributes, manages script elements, constraints, and logic.

[0080] The basic data structure of the script is created, including nodes and edges between nodes, and each node is assigned script attributes, including roles, intentions, emotion vectors, duration hints, and modifiability flags.

[0081] The role library defines the metadata of the role script attributes, specifically including role id, role attributes, initial state, available actions / permissions.

[0082] When selecting the role script attributes, a symbol checker is triggered, including checking and determining role conflicts, and prompting alternatives or prohibitions.

[0083] The current state of the role script attributes is associated and injected into the context of the script.

[0084] A scalable label library of intention script attributes is constructed, and an intention selection control is provided through a UI interface and allows free text idea description to be filled in autonomously, and the selected intention is used as the highest priority instruction for attention bias and generation prompt of the narrative layer of the script.

[0085] A dimension list of emotional script attributes is constructed, and an emotional time sequence encoder is provided through a visual editor.

[0086] According to the generation stage of the script, the emotional curve of the script is embedded into the script control layer through the emotional time sequence encoder, and is jointly encoded based on the time axis and the current world state of the role.

[0087] The duration hint script attribute is used as a constraint condition in the script generation pipeline, and in the presence of multiple duration constraints, the final market strategy is autonomously defined based on the priority and intensity of the nodes.

[0088] When creating a node, it is determined whether it is marked as a hard constraint.

[0089] When generating the script, the node is used as a rule that must be satisfied in the narrative layer of the script.

[0090] After generating the script, a symbol check is performed, and if the hard constraint is not met, local script repair is triggered or the script generation result is directly rejected.

[0091] Script arrangement module: identify scene nodes and events, build three-layer dependencies and weight, generate reverse causal chain through recursive dependency and depth-first search;

[0092] Identify, define and classify each scene node and event node in the script;

[0093] Build a dependency graph between the narrative layer, control layer and interaction layer of the script, and increase the weight value of the intention, emotion vector and duration prompt between each node in the dependency graph;

[0094] Based on the recursive dependency relationship, the current node depends on multiple preceding nodes. Through depth-first search, the causal relationship of the script is traced according to the recursive dependency relationship when generating the script, and a reverse causal chain is generated;

[0095] Trace the relationship before and after each emotional node through the reverse causal chain, and check the logical consistency of the plot;

[0096] Build a hybrid control architecture of symbolic layer and neural layer, and enrich the script based on the rich script narrative layer through a large language model;

[0097] Define the policy network as a controller, and prioritize the selection of hard constraints according to the requirements of each decision point.

[0098] Dependency propagation module: maintain a dependency table for each node, update and mark dirty nodes through event-driven, and achieve efficient dependency propagation, fast redraw and low-latency update through priority queue and differential generation;

[0099] When interacting with the script, initialize a dependency table for each node to record the other nodes that each node depends on;

[0100] When a node changes, check the downstream nodes of the node in the dependency graph;

[0101] Find all affected nodes by traversing the dependency graph;

[0102] When triggering script update, based on the modified node as the initial point, iteratively update all affected nodes;

[0103] Update the dependency graph dynamically through event-driven;

[0104] Register event listeners on each node, and trigger updates immediately when the node content changes;

[0105] Mark the affected nodes, and only update these marked dirty nodes during the next rendering;

[0106] Update the processing nodes through the scheduling queue, queue all affected nodes according to priority, and reduce latency through differential generation.

[0107] Storytelling interaction coordination module: three levels of coordination are divided to realize dynamic script generation and interaction, script control is realized based on gated residual connection, reverse causal chain and time attention mechanism;

[0108] The script narrative layer is linked as the high-level context, and the script control layer and the script interaction layer are queried as the bottom layer to the script narrative layer, and the feedback from the bottom layer is returned to the high layer through the gated residual connection;

[0109] Based on the continuous action and video sequence of the script, the time attention of the window is added through the reverse causal chain;

[0110] Set up an independent control strategy for each layer, which specifically includes:

[0111] Script narrative layer: based on the emotion curve and node priority of the script, the event sequence sketch of the script is generated through the planner, and the script control layer is provided for script unfolding;

[0112] Script control layer: generate event chain of the script based on event sequence sketch;

[0113] Script interaction layer: connect fine-grained generator API and physical post-processor API, allow direct modification of key frames for continuous action and video sequence of the script, and feedback to the high layer;

[0114] The narrative vector of the script narrative layer and the visual encoding of the corresponding script interaction layer are zoomed in and out in the same batch, and the mismatch is zoomed out;

[0115] The event vector of the script control layer and the interaction action vector of the corresponding script interaction layer are similarity regularized to encourage semantic matching;

[0116] According to the script narrative layer, the expected middle layer statistics are generated, and the currently generated bottom layer statistics are compared;

[0117] Based on the script interaction layer and the reverse causal chain, the narrative vector of the reverse prediction is verified for periodic consistency;

[0118] Based on the depth layered unfolding of the script interaction, specifically including:

[0119] Low interaction: only use the script narrative layer and the script control layer to automatically generate the interaction action of the script interaction layer;

[0120] Medium interaction: edit nodes through the script control layer, generate interaction actions of the script interaction layer based on the script narrative layer;

[0121] High interaction: directly edit the interaction action of the script interaction layer, perform bottom-up consistency check and send local script update to the script control layer and the script narrative layer.

[0122] Execution output module: build interactive execution layer, integrate real-time feedback mechanism, automate script generation and output executable format;

[0123] Build interactive execution layer, dynamically adjust the direction of the plot and the order of event occurrence according to user input;

[0124] Integrate real-time feedback mechanism, quickly update plot status and character behavior, and maintain the coherence of the script logic;

[0125] Automatically generate script content under given context or input conditions through automated script generation algorithm, and adjust according to preset conditions;

[0126] Convert the final output of the script into executable format, including text, actions, shot switching, sound and background music;

[0127] Display the generated interactive scenes of the script through different platforms.

[0128] Through the combination of emotion vector, intent script attribute, and emotion time series encoder, the emotions and behaviors of the characters can be adjusted in real time during the generation of the script, ensuring the consistency and logic of the plot, while adapting to the input and interaction needs of the user.

[0129] Through the introduction of recursive dependency relationships and reverse causal chains, the relationship between each node can be traced more accurately, reducing logical errors and inconsistencies in the plot, and dynamically identifying and adjusting affected nodes, making the entire script generation process more natural and smooth.

[0130] By embedding emotion curves into the script control layer and combining with duration prompts, the duration of character behavior in the script can be accurately adjusted, and the content of the script can be adjusted based on the priority and intensity of different nodes, better realizing the dynamic performance of the characters and the smooth transition of emotional changes.

[0131] Through hard constraint markers, symbol verification, and script repair mechanisms, the generation process can be forced to meet the preset rules, avoiding the generation of content that does not meet the constraints, ensuring the compliance of the generated script, and enabling immediate repair.

[0132] By designing an interactive execution layer, the direction of the plot can be dynamically adjusted according to user input, and the behavior of the characters and the order of plot events can be updated in real time, enabling highly personalized adjustments of the script generation according to user needs, enhancing user experience and interactivity.

[0133] Through the hybrid control architecture, the symbol layer and the neural layer are combined, and a policy network is used as the controller to realize a more intelligent decision-making mechanism, which can adjust the generation strategy according to the complexity and needs of the script, improving the quality and flexibility of script generation.

[0134] Certain exemplary embodiments of the present application have been described above by way of illustration, and it is to be understood that various modifications will be apparent to those of ordinary skill in the art, without departing from the spirit and scope of the present application. Accordingly, the above description is to be construed as illustrative only and not as limiting of the scope of the present application.

Claims

1. An interactive, drive scene generation system centered around a script, characterized in that, Specifically includes the following modules: script management module, script arrangement module, dependency propagation module, narrative control interaction collaboration module and execution output module; The script management module creates the basic data structure of the script and assigns script attributes, manages script elements, constraints and logic; The script arrangement module identifies scene nodes and events, constructs three-layer dependencies and weights, and generates reverse causal chains through recursive dependencies and depth-first search; The dependency propagation module maintains a dependency table for each node, updates and marks dirty nodes through event-driven, and implements efficient dependency propagation, fast redraw and low-latency update through priority queue and difference generation; The narrative control interaction collaboration module divides three levels of collaboration to realize dynamic script generation and interaction, and realizes script control based on gated residual connection, reverse causal chain and time attention mechanism; The execution output module constructs an interactive execution layer, integrates a real-time feedback mechanism, automatically generates and outputs a display executable format.

2. The script-centered interactive driving scene generation system according to claim 1, wherein: The script management module creates nodes and edges between nodes, assigns script attributes, including role, intent, emotion vector, duration prompt, and modifiability flag; The role script attribute: define the metadata of the role script attribute in the role library, when the role script attribute is selected, trigger the symbol checker, including checking and judging role conflicts, prompting alternatives or prohibiting; Associate and inject the current state of the role script attribute into the context of the script; The intent script attribute: build an extensible label library of intent script attributes, provide intent selection controls through the UI interface and allow free text idea descriptions to be filled in, and select the intent as the highest priority instruction for attention bias and generation prompt of the narrative layer of the script; The emotion vector attribute: build a dimension list of emotional script attributes, and provide an emotional time sequence encoder through a visual editor; According to the stage of script generation, embed the emotional curve of the script into the script control layer through the emotional time sequence encoder, and jointly encode based on the time axis and the current world state of the role; The duration prompt attribute: use the duration prompt script attribute as a constraint condition in the script generation pipeline, and autonomously define the final market strategy according to the priority and intensity of the node when there are multiple time length constraints; The modifiable flag attribute: determine whether to mark as a hard constraint when creating a node; When generating a script, use the node as a rule that must be met in the narrative layer of the script; After generating the script, perform symbol checking, and if the hard constraint is not met, immediately trigger local script repair or directly reject the script generation result.

3. The script-centered interactive driving scene generation system according to claim 1, wherein: The script arrangement module identifies, defines and classifies each scene node and event node in the script; Construct a dependency graph between the narrative layer, the control layer and the interaction layer of the script, and increase the weight values of intent, emotion vector and duration prompt according to the edges between each node in the dependency graph; Based on the recursive dependency relationship, the current node is dependent on multiple preceding nodes. During the generation of the script, the causal relationship of the script is traced according to the recursive dependency relationship through a depth-first search, and a reverse causal chain is generated. The forward and backward relationships of each emotional node are traced through the reverse causal chain, and the logical consistency of the plot is checked. A hybrid control architecture of symbolic layer and neural layer is constructed, and the script is enriched based on the rich narrative layer of the script by a large language model.

4. The script-centered interactive driving scene generation system according to claim 1, wherein: The propagation dependency module maintains a dependency table for each node, event-driven updates, and dirty node markers to achieve efficient dependency propagation, fast redraw, and low-latency updates through priority queues and differential generation; During script interaction, a dependency relationship table is initialized for each node to record the other nodes that each node depends on; When a node changes, the downstream nodes of the node in the dependency graph are checked; All affected nodes are found by traversing the dependency graph; When the script is updated, all affected nodes are iteratively updated based on the modified node as the initial point; The dependency graph is dynamically updated through event-driven mode; An event listener is registered on each node, and when the node content changes, the update is triggered immediately; Affected nodes are marked, and only these marked dirty nodes are updated during the next rendering; All affected nodes are queued according to priority through a scheduling queue update processing node, and the latency is reduced through differential generation.

5. The script-centered, interactive driving scene generation system of claim 1, wherein: The narrative control interaction coordination module links the script narrative layer as a high-level context, and queries the script control layer and the script interaction layer as a bottom layer to the script narrative layer, and returns the bottom layer feedback to the high layer through the gate residual connection; Based on the continuous action and video sequence of the script, the time attention of the window is added through the reverse causal chain; Set up an independent control strategy for each layer; The narrative vector of the script narrative layer and the visual encoding of the corresponding script interaction layer are zoomed in in the same batch, and the mismatch is zoomed out; The similarity regularization is performed on the event vector of the script control layer and the interaction action vector of the corresponding script interaction layer to encourage semantic matching; According to the script narrative layer, the expected middle layer statistics are generated, and the current generated bottom layer statistics are compared; Based on the script interaction layer and the reverse causal chain, the reverse predicted narrative vector verifies the periodic consistency; Based on the interactive depth hierarchical expansion of the script.

6. The script-centered, interactive driving scene generation system of claim 1, wherein: The independent control strategy of each layer specifically includes: Script narrative layer: based on the emotional curve and node priority of the script, the event sequence sketch of the script is generated through the planner, and the script control layer is used for script expansion; Script control layer: generating the event chain of the script based on the event sequence sketch; Script interaction layer: connecting the fine-grained generator API and the physical postprocessor API, allowing direct modification of the key frames of the continuous action and video sequence of the script, and feeding back to the high layer; The interactive depth hierarchical expansion based on the script specifically includes: Low-level interaction: only using the script narrative layer and the script control layer to automatically generate the interaction action of the script interaction layer; Medium-level interaction: editing nodes based on the script narrative layer to generate the interaction action of the script interaction layer; Highly interactive: interact with the script by directly editing the interactive layer, perform bottom-up consistency checks and send partial script updates to the script control layer and narrative layer.

7. The script-centered interactive-driven scene generation system according to claim 1, wherein: The execution output module constructs an interactive execution layer, dynamically adjusts the plot and the sequence of event occurrence according to user input; Integrates real-time feedback mechanism, quickly updates the plot state and character behavior, and maintains the coherence of the script logic; Automatically generates script content under given context or input conditions through automated script generation algorithm, and adjusts according to preset conditions; Converts the final output of the script into executable format, including text, actions, shot switching, sound and background music; Display the generated script interactive scene through different platforms.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that: The processor executes the computer program to realize the modules of the script-centered interactive-driven scene generation system according to any one of claims 1-7.

9. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the modules of the script-centered interactive-driven scene generation system according to any one of claims 1-7.

Images