SCENIC is a text-conditioned scene interaction model that adapts to complex environments with varying terrains, supporting user-specified semantic control through natural language. The model navigates 3D scenes using user-defined trajectories as sub-goals and textual prompts. SCENIC employs hierarchical reasoning methods in scene understanding, achieving seamless transitions between different motion styles through frame alignment of movement and text. This technology is significant as it generates character navigation movements that comply with real-world physics and user instructions, playing a crucial role in virtual reality, augmented reality, and game development.