Privacy-Preserving 3D Gaussian Splatting

In recent years, there has been notable progress in rendering techniques within the gaming industry, aiming to enhance the visual realism and immersion of gaming experiences. Among these advancements, 3D Gaussian Splatting (3DGS) has emerged as a prominent method for efficiently visualizing volumetric data in real-time. 3DGS is a powerful technique for reconstructing detailed geometric information from multiple images captured from different camera viewpoints. However, this approach raises serious privacy concerns due to the possible exploitation of the facial photos of people. Using user data for lifelike avatars raises concerns about unauthorized access and data misuse. This paper explores the application of 3DGS in gaming, examining its fundamental principles, advantages, potential impact on game graphics, and privacy protection. By employing a sequence of identity-obscuring images, the proposed approach successfully creates facial geometry, safeguarding facial privacy. By incorporating 3DGS, game developers have the opportunity to create dynamic lighting, immersive environments, and visually captivating effects, ultimately enhancing the overall gaming experience for players.