Text2Room Where Words Become Worlds

Created by team SpaceX on August 11, 2023

Text2Room is a groundbreaking innovation that redefines the process of 3D content creation. This transformative method harnesses the power of pre-trained 2D text-to-image models to generate room-scale textured 3D meshes from simple text prompts. Through an iterative scene generation process, Text2Room renders 3D meshes from diverse camera angles and seamlessly fuses missing details using advanced algorithms, resulting in immersive and captivating environments. What sets Text2Room apart is its two-stage viewpoint selection strategy – the Generation Stage creates the main scene layout, while the Completion Stage intelligently fills gaps, ensuring a complete and coherent 3D representation. By democratizing 3D content creation, Text2Room eliminates complexities and accelerates the process, making it accessible to a wide range of industries including AR/VR content, gaming, and architectural visualization. Text2Room isn't just a product; it's a creative revolution that empowers users to turn their imagination into tangible, interactive experiences. With limitless applications and an ever-growing market demand for immersive 3D content, Text2Room stands at the forefront of innovation, shaping the future of content creation.

Category tags:

"Nice project idea, the model needs more finetuning in the future to really shine. Excited to see the improvements for this project!"


Skander Karoui

Data Science and AI enthusiast || ICT Engineer

"I appreciate the team's effort in delving into the innovative text2room model. The presentation would have been more insightful if the presentation included details about the specific problem the technology addresses, its target audience, and the underlying business model. The visual representation of the 3D complex scene in the video was impressive, although I encountered difficulty running the demo using the provided code. Perhaps clearer instructions in that regard could enhance the overall experience. "


Meghana Reddy Guduru

Computer Vision Engineer