具有難度控制功能之電子遊戲程序化內容生成器

我們提出了一種名為「難度控制程序化生成」(DC-PCG) 的程序化內容生成演算法，用於生成電子遊戲關卡。我們的研究有兩個主要貢獻。首先，我們設計了一系列用於衡量電子遊戲關卡難度的標準。其次，DC-PCG 利用這些標準來控制生成關卡的難度。DC-PCG 使用基於搜索的演算法在生成空間中搜索能量最低的關卡。DC-PCG 包括兩個階段。第一階段，DC-PCG 生成關卡的拓撲結構。第二階段，DC-PCG 為關卡中的每個節點生成遭遇內容。對於每個階段，我們有不同的能量函數。我們透過調整能量函數中的參數來控制生成關卡的難度。我們進行了一項使用者研究，以了解生成內容的品質。在這項研究中，參與者玩了由DC-PCG 生成的關卡，並給這些遊戲關卡評分。通過分析使用者研究的數據，我們確定 DC-PCG 生成的關卡難度與使用者研究中的評分相符。這表明 DC-PCG 能夠生成具有可控難度的高品質電子遊戲關卡。

關鍵字

程序化內容生成；圖論；電子遊戲

並列摘要

We propose Difficulty Control Procedural Content Generation(DC-PCG), a procedural content generation algorithm that generates video game levels. Our research has two primary contributions. First, we design a series of metrics to measure the difficulty of the video game level. Second, DC-PCG uses these metrics to control the difficulty of the generated level. DC-PCG uses a search-based algorithm to search for the level with the lowest energy in the generation space. DC-PCG has two-phase. In the first phase, DC-PCG generates the topological structure of the level. In the second phase, DC-PCG generates the encounter for every vertex in the level. For each phase, we have different energy functions. We control the difficulty of the generated level by tweaking the parameters in the energy functions. We conduct a user study to understand the quality of the generated content. In the study, participants played the level generated by DC-PCG and scored the game levels. After analyzing the data from the user study, we determine that the difficulty of the level generated by DC-PCG aligns with the user’s score in the user study. It shows that DC-PCG can produce high-quality video game levels with controlled difficulty.