GDC 2025 Session Viewer
Enhancing 3D Building Asset Creation: Machine Learning Meets Procedural Generation
Qiang Dai (Principal Engineer, LIGHTSPEED STUDIOS)
Pass Type: All Access Pass, Core Pass - Get your pass now!
Track: Programming, Visual Arts
Format: Lecture
Vault Recording: TBD
Audience Level: All
This session explores the integration of large language models and machine learning with procedural generation to enhance the creation of 3D building assets in game development. Attendees will discover how this approach supports various architectural styles within a unified framework, allowing for game-ready assets with higher quality and diversity. The presentation will cover the unified procedural framework supporting various architectural styles, the system's foundational modules, and the optimization of traditional workflows. By providing detailed insights, this session aims to equip game designers and artists with advanced tools and techniques to significantly improve efficiency and creativity in their projects.
Takeaway
Attendees will learn how integrating large language models and machine learning with procedural generation can streamline the creation of 3D building assets, enhancing efficiency and quality in game development. Practical insights include using natural language commands for asset generation and optimizing traditional workflows.
Intended Audience
This session is ideal for technical artists, game designers, artists and developers interested in procedural generation, and machine learning. Attendees should have a basic understanding of 3D asset creation and game development pipelines to fully benefit from the insights and examples provided.
