Fig 1 - uploaded by Alexandru Iosup
Content may be subject to copyright.
Types of game content that can be procedurally generated. The numbers in each box refer to the subsection dedicated to the content type in Sections 2 (definition) and 4 (application of procedural generation methods).
Source publication
Hundreds of millions of people play computer games every day. For them, game content—from 3D objects to abstract puzzles—plays a major entertainment role. Manual labor has so far ensured that the quality and quantity of game content matched the demands of the playing community, but is facing new scalability challenges due to the exponential growth...
Context in source publication
Context 1
... addition to classes of content present strictly inside games, we consider derived content, that is, content derived from the content or the state of a game with the goal of immersing the player further into the game world. We structure the six classes as a virtual pyramid (see Figure 1), in which classes of content closer to the top may be built with elements from the classes closer to the bottom. ...
Citations
... In the literature, several surveys elucidate advancements in PCG within gaming and offer valuable insights into specific game content generation. Hendrikx et al. [24] introduce a six-layered taxonomy of game content and map PCG methods to these game content layers. For newcomers to the field, this is an excellent starting point for understanding various PCG approaches within the gaming domain. ...
Procedural content generation (PCG) has significantly impacted game design by automating the creation of dynamic game environments, thereby saving time and effort while maintaining the freshness at each play which is required for games as a service. Recent advances in machine learning, particularly Generative Adversarial Networks (GANs), offer exciting possibilities for generating diverse and playable game levels that surpass traditional methods. Despite challenges such as training instability and ensuring playability, GANs present considerable potential for dynamic content generation. This paper explores the advantages of GAN-based approaches, addresses their limitations, and suggests improvement strategies, including combining algorithms or using solvers to mitigate poor generations. This paper explores the advantages and limitations of GAN-based approaches and suggests promising research directions to improve GAN-based procedural level generation, including combining algorithms or using solvers to mitigate poor generations. Future research directions are also identified, such as the need for user studies and improved GAN training techniques to fully harness the potential of GANs in game-level generation.
... Procedural generation techniques [10] offer even more potential, as exemplified by video games like No Man's Sky, where an entire universe was created with 18 quintillion planets, each can be visited by players to observe diverse ecosystems that are unique for each planet [11]. These advancements highlight the potential for creating high-fidelity VCs with dynamic features. ...
Virtual Cities (VCs) transcend simple digital replicas of real-world systems, emerging as complex socio-technical ecosystems where autonomous AI entities function as citizens. Agentic AI systems are on track to engage in cultural, economic, and political activities, effectively forming societal structure within VC. This paper proposes an integrated simulation framework that combines physical, structural, behavioral, cognitive, and data fidelity layers, allowing multi-scale simulation from microscopic interactions to macro-urban dynamics. A composite fidelity metric ( F 0) provides systematic approach to evaluate accuracy variations across applications in VCs. We also discuss autonomy of AI entities and classify them according to their capacity to modify goals—ranging from “tools” with fixed objectives to “entities” capable of redefining their very purpose. We also outline the requirements to define a coefficient to evaluate the degree of autonomy for AI beings. Our results demonstrate that such virtual environments can support the emergence of AI-driven societies, where governance mechanisms like Decentralized Autonomous Organizations (DAOs) and an Artificial Collective Consciousness (ACC) provide ethical and regulatory oversight. By blending horizon scanning with systems engineering method for defining novel AI governance models, this study reveals how VCs can catalyze breakthroughs in urban innovation while driving socially beneficial AI development - consequently opening a new frontier for exploring human–AI coexistence.
... Research on the use of procedural content generation (PCG) [10,11] shows new possibilities for personalising learning in educational applications. PCG enables the automatic creation of unique content, providing variety and novelty to the learning experience. ...
The article describes the application of virtual reality (VR) technologies in the training of hoisting vehicle operators, with a focus on improving the efficiency and safety of the educational process. Traditional methods of training are associated with high costs and significant risks, but virtual reality allows to create a safe and controlled environment for practising professional skills. The architecture of the developed VR simulator is discussed, including realistic crane reproduction, interaction with a virtual slinger and modelling of different weather conditions. An important aspect of the simulator is the procedural generation of content, which allows the creation of unique training scenarios, thus greatly diversifying the learning process. The use of virtual characters adds realism and interactivity to the learning process, allowing users to practice solving complex tasks while interacting with others in the workplace. Research shows that the use of VR in training tower crane operators not only improves the quality of training, but also helps to reduce errors in the initial stages of work. The article emphasises that dynamically variable levels play a key role in adapting the learning process to individual learner needs, making VR training a powerful tool for training professionals in the construction industry.
... Floorplan generation. In the field of building design, generating high-quality house layouts has been an important research direction [5,10,16,20,27,33], and many innovative approaches have emerged in recent years. Nauata et al. [18] proposed House-GAN, a method based on Generative Adversarial Networks (GANs) to generate house layouts. ...
... This reverse process is typically implemented as an iterative process, where each time step depends on the output of the previous step and incorporates conditional information y to guide the generation. In the reverse denoising process at time step t, the model f θ predicts the denoised imagex 0 : x0 = f θ (xt, t, e(y)) (5) f θ is the denoising network Transformer [29] parameterized by θ. x t is the noisy image at current time step t. The reverse process updates the state of x t each step to remove noise and gradually guide the generated image toward the structure of the condition y. ...
This paper proposes a two-phase text-to-floorplan generation method, which guides a Large Language Model (LLM) to generate an initial layout (Layout-LLM) and refines them into the final floorplans through conditional diffusion model. We incorporate a Chain-of-Thought approach to prompt the LLM based on user text specifications, enabling a more user-friendly and intuitive house layout design. This method allows users to describe their needs in natural language, enhancing accessibility and providing clearer geometric constraints. The final floorplans generated by Layout-LLM through conditional diffusion refinement are more accurate and better meet user requirements. Experimental results demonstrate that our approach achieves state-of-the-art performance across all metrics, validating its effectiveness in practical home design applications. We plan to release our code for public use.
... PCG techniques have been widely adopted in the gaming industry to enhance game replayability, increase content diversity, and reduce development costs. Hendrikx et al. (2013) provide a comprehensive survey of PCG methods used in games, highlighting their applicability across different game elements. ...
This paper proposes a novel pipeline for generating game levels that elicit predefined emotional experiences from players. Our approach uses evolutionary algorithms alongside data-driven persona agents, predictive emotional models, a PCG parametric level generator, and a newly defined language for the clear and computable definition of player emotional experiences: ExpREx (Experience Regular Expressions). Using these components, we evolve game levels to match the player experience goals specified using the ExpREx language, aiming to create levels that evoke specific emotional experiences for different subsets of players. The efficacy of our method was validated through a user study involving 101 participants, whose continuous annotations of emotional experience were collected and analyzed to assess the congruence between the actual emotional responses elicited and those targeted by our pipeline. We found that 93.73% of the ExpREx goals targeted were also reported by the user study subjects.
... There are a handful of surveys on PCG for games with different focuses and aims that have been published before our work. Some of them discuss the technical details of PCG algorithms (Zhang, Zhang, and Huang 2022), while others focus on content created by PCG (Hendrikx et al. 2013;De Carli et al. 2011). Some papers focus on specific types of PCG; for example, machine learning in PCG, while other works (Summerville et al. 2018) mainly focus on DL algorithms in PCG. ...
... Additionally, content can be categorized as necessary or optional (Togelius et al. 2011). For this survey, we use categories similar to the ones presented by Hendrikx et al. (2013) with some modifications. We divide content created for games into five different categories, each consisting of multiple items. ...
... PRNG was first used as a data compression method because the generated sequence, while appearing random, can be reproduced if the same seed and algorithm are used. Combined with other methods, PRNGbased techniques can be used to generate buildings, textures, and items (Hendrikx et al. 2013). One of the most famous forms of PRNGs is noise functions. ...
Procedural Content Generation (PCG) is defined as the automatic creation of game content using algorithms. PCG has a long history in both the game industry and the academic world. It can increase player engagement and ease the work of game designers. While recent advances in deep learning approaches in PCG have enabled researchers and practitioners to create more sophisticated content, it is the arrival of Large Language Models (LLMs) that truly disrupted the trajectory of PCG advancement. This survey explores the differences between various algorithms used for PCG, including search-based methods, machine learning-based methods, other frequently used methods (e.g., noise functions), and the newcomer, LLMs. We also provide a detailed discussion on combined methods. Furthermore, we compare these methods based on the type of content they generate and the publication dates of their respective papers. Finally, we identify gaps in the existing academic work and suggest possible directions for future research.
... 1) GPT for Procedural Content Generation (PCG): In this use case, GPT generated game content during gameplay according to constraints defined by the designers during development, without any iteration or human input involved in the generation process [24]. ...
Due to GPT's impressive generative capabilities, its applications in games are expanding rapidly. To offer researchers a comprehensive understanding of the current applications and identify both emerging trends and unexplored areas, this paper introduces an updated scoping review of 131 articles, 76 of which were published in 2024, to explore GPT's potential for games. By coding and synthesizing the papers, we identify five prominent applications of GPT in current game research: procedural content generation, mixed-initiative game design, mixed-initiative gameplay, playing games, and game user research. Drawing on insights from these application areas and emerging research, we propose future studies should focus on expanding the technical boundaries of the GPT models and exploring the complex interaction dynamics between them and users. This review aims to illustrate the state of the art in innovative GPT applications in games, offering a foundation to enrich game development and enhance player experiences through cutting-edge AI innovations.
... Video games are based on spatial models, which in recent years have taken the form of very complex and accurate representations of the fictional or real world (Chądzyńska & Gotlib, 2015). Hendrikx et al. (2012) consider game space as the one, two, or three-dimensional area where residing artifacts have relative position and direction. According to Toups et al. (2019), virtual spaces that games constructs and players which engage game mechanics are referred to as game worlds. ...
... The authors of this study elaborate on this thesis. Video games are primarily a form of entertainment and a ubiquitous activity in everyday life (Hendrikx et al., 2012;Quiroga et al., 2009;Vorderer et al., 2006). Movement in virtual space is much more dynamic than movement in real space, so the use of the map itself should be short and for a specific purpose. ...
This article examines the significance of the world map in video games for the interpretation of spatial situations. An example is the popular role-playing game The Witcher 3: Wild Hunt (TW3). Nowadays, most video games are characterized by the presence of a spatial aspect. The game world map is the most important navigational element of the game that the gamer can use. To this end, the authors decided to test the importance of the game world map in the context of analyzing different examples of spatial situations that appear in TW3 by the respondents. Eye movement tracking was chosen as the research method. The analysis was conducted using statistical tests. Both gamers and non-gamers of TW3, gamers and non-gamers in general, participated in the survey. Each subject was shown 5 movies (1 introductory movie, 4 movies in the main part of the study) from the gameplay of the game, in which the game world map was opened. After each video, a question was asked about both the gameplay and the game world map. It was found that familiarity with TW3, frequency of playing video games influenced the correctness and time of answering the questions asked. In addition, by analyzing mean pupil diameter it was found that the game world map and gameplay segments do not cognitively burden the users. By analyzing the duration of fixations and the number of fixations, differences in visual strategy were observed between the groups of test subjects.
... The first classifies solvable Sokoban boards and the second is a regressor that predicts the number of pushes to solve them. One of the motivations for carrying out this study is Procedural Content Generation [3] (PCG). Developing intelligent prediction methods can help us in the task of generating quality Sokoban levels in a short period of time [4]. ...
Knowing if a Sokoban level has a solution requires a high investment of computational resources. This also happens when estimating difficulty; heuristics may not be precise, only work for simplified cases, be discontinuous and take much computing time. The calculation of high-fidelity heuristics is expensive and the performance of optimization systems is reduced by the time they take to deliver the result. In this article, two Convolutional Neural Networks (CNN) are implemented, a classifier of resolvable Sokoban boards and a regressor that predicts the number of pushes necessary to solve them. These obtained an accuracy of 79% and MAE of 28, respectively. Trained models are useful in systems that do not require optimality, but rather guidance within the search space that balances time and quality. These networks can be used as surrogate methods, replacing high-fidelity heuristics that take too long.
... Procedural Content Generation (PCG) is a leading technique owing to its versatile applications across various industries (e.g., gaming) [4,17,20,45]. It is particularly effective for generating modular, highly repetitive, or rule-based content [14]. ...
