Keiki: Towards Realistic Danmaku Generation via Sequential GANs

TL;DR

This paper introduces Keiki, a new platform for generating realistic danmakus in bullet hell games using sequential GANs, demonstrating promising results in modeling complex patterns.

Contribution

The paper presents Keiki, a novel platform that employs three types of GANs to generate realistic and diverse danmaku patterns in bullet hell games.

Findings

Time-series GANs outperform in modeling sequential patterns.

Periodic spatial GANs produce diverse danmakus.

Generated danmakus deviate from human-designed patterns.

Abstract

Search-based procedural content generation methods have recently been introduced for the autonomous creation of bullet hell games. Search-based methods, however, can hardly model patterns of danmakus -- the bullet hell shooting entity -- explicitly and the resulting levels often look non-realistic. In this paper, we present a novel bullet hell game platform named Keiki, which allows the representation of danmakus as a parametric sequence which, in turn, can model the sequential behaviours of danmakus. We employ three types of generative adversarial networks (GANs) and test Keiki across three metrics designed to quantify the quality of the generated danmakus. The time-series GAN and periodic spatial GAN show different yet competitive performance in terms of the evaluation metrics adopted, their deviation from human-designed danmakus, and the diversity of generated danmakus. The preliminary experimental studies presented here showcase that potential of time-series GANs for sequential content generation in games.