World Labs’ AI Creates 3D Worlds from Static Images
A photo can say a thousand words. We’ve all heard this common phrase, and for many of us, it rings true. Although written prose can convey a medley of stories, relying on the boundless limits of human imagination to build worlds, it only takes one photo to bring such fantasies to life. However, before the advent of AI technology, a thousand words is all a photo could say. Now, thanks to World Labs, a spatial intelligence AI company based in London, pictures can tell a thousand stories.
Video by: World Labs
World Labs has recently pioneered (literally) groundbreaking Large World Models (LWMs) that can generate entire 3D worlds from a single image. Other spatial intelligence systems exist, mirrored in AI systems like NVIDIA Instant NeRF, which can conjure 3D scenes from 2D pictures similar to World Labs. However, World Labs’ model differs in some exciting ways, allowing users to actually interact with the generated 3D world. Whether you want to change the lighting in a particular backdrop, change the color of an object, or edit the world around you, World Labs has you covered! This brilliant stride in AI technology can open doors for game developers, artists, filmmakers, and engineers to brainstorm world-building concepts and fledge out scenes for their projects.
The feat of World Labs is no surprise, considering the company was co-founded by the “Godmother of AI” Fei-Fei Li. Li, an AI Researcher & Professor at Stanford University and Co-Director of the Stanford Human-Centered AI Institute, spoke to the power of spatial intelligence in a TED talk in April.
“These are prototypes of the first budding signs of a future possibility. One in which the human race can take our entire world and translate it into digital forms and model the richness and nuances,” Li said, presenting the World Labs prototype to the audience at TED2024. “What nature did to us implicitly in our individual minds, spatial intelligence technology can hope to do for our collective consciousness.”
Fei-Fei Li, Co-Founder and CEO of World Labs and AI Researcher & Professor at Stanford University.
This technology falls into a similar category as its Large Language Model (LLM) counterparts, like ChatGPT and Perplexity. The advancement of LLMs and LWMs has been accompanied by rightful criticism, highlighting the errors and limitations of these young systems. ChatGPT has been found to provide incorrect answers to computer programming questions an astounding 52% of the time and frequently fabricates references for essays. Similarly, World Labs’ model can render scenes incorrectly and is still not fully explorable. Despite these issues, AI companies like World Labs insist that their current systems are merely “previews” of what is yet to come. With enough time, effort, and patience, the LLMs and LWMs of today will flourish into the multi-purpose, effective machines of tomorrow.
Investors have recognized the potential of World Labs, having raised more than $230M in funding and valuing the company at over $1 billion this past August. The latest financing round was led by well-known venture capitalists: New Enterprise Associates (NEA) CEO and CIO Scott Sandell, Andreessen Horowitz’s Martin Casado and Sarah Wang, and Radical Ventures co-founder Jordan Jacobs. Other notable investors include Intel Capital, AMD Ventures, and former CEO & Chairman of Google, Eric Schmidt. Thanks to the support of these high-profile stakeholders, World Labs hopes to have a polished final product available to the public by 2025.
It is no exaggeration to say that a picture can convey far more than a thousand words. Enhanced by AI technology like World Labs’ LWM, photos have the potential to unlock realms of the world yet unexplored by the human mind.