Hyojun Go

Over the past years, generative models have rapidly advanced from producing high-quality 2D images to generating coherent videos and multi-modal content. The next frontier is to enable these models to directly generate and edit 3D worlds, capturing both geometric structure and realistic appearance. The research aims to push generative models toward this goal by developing methods that can create, reconstruct, and edit 3D scenes in a unified framework. Ultimately, the research goal seeks to build models that can generate immersive and realistic 3D environments, opening new possibilities for applications in AR/VR, robotics, and digital content creation.