A dynamic scene generation method, terminal and storage medium based on submodular maximization

Abstract

The present application relates to a method for generating a dynamic scene based on submodular maximization, a terminal and a storage medium. Including: constructing a set number of scene beats composed of at least one event according to the game scene, generating a scene beat set; calculating the contribution value of each scene beat in the scene beat set to the task target intensity vector, and sending the scene beat set and contribution value to Scenario generator based on maximization of secondary grinding; through the scenario generator based on maximization of secondary grinding, the task target strength vector value of the game scene defined by the player side and the objective function defined by the scene design side are obtained, based on the task target strength vector value and the target function, using the greedy algorithm to iteratively select the scene beat that maximizes the objective function from the scene beat set and add it to the scene simulator for simulation training, and after the number of iterations reaches the cardinality constraint, a game scene that matches the task target intensity vector value is generated. The embodiment of the present application can meet the personalized needs of different players.

Description

technical field [0001] The present application belongs to the technical field of virtual scene generation, and in particular, relates to a dynamic scene generation method, terminal and storage medium based on sub-modular maximization. Background technique [0002] In recent years, with the development of computer animation and proxy technology, it has become a new trend to use digital games as educational resources. These digital games are games with training purpose that allow people to practice in the form of playing games in virtual scenes Measurable skills related to a given task. It has become an emerging challenge in the development of this type of game to efficiently and quickly generate matching game scenarios that meet the needs of different players and systems. These scenarios are abstracted into time series with different types of events. For example, a peace-keeping scenario may include crowd gathering, Incidents such as the arrival of instigators, the spread of...

AI Technical Summary

Problems solved by technology

However, the establishment of the player's emotional value evaluation model has unexplainable statistical performance, resulting in some game features being "wasted" when the scene is biased towards the player's experience generation based on this method, and the evaluation model requires time and data to train. The method needs to obtain the player's experience preference characteristics based on statistics, so that when designing the game levels, it is more inclined to choose positive levels as the next level of the game. This kind of scene generation method is not accurate enough and cannot adapt to the special needs of players.

[0005] 2. Task-based virtual training design Interactive scenario generation system [Interactive scenario generation for mission-based virtual training]; this method uses genetic algorithm to solve the optimal or approximate solution, and involves multiple redundant features when evaluating the scene. There is no further extraction of scene features, and the generated scene lacks consideration of the real needs of players

[0006] 3. A data-driven scene generation framework based on game training; this technology adds ANN (Artificial Neutral Network, artificial neural network) neural network, AI (Artificial Intelligence, artificial intelligence) players and scenes generated by scene feature evaluation, and uses multiple Real players participate in the evaluation experiment, and the model is more complex, but the core algorithm of the technology is the genetic algorithm, which is less efficient when the search space is large

Images