Fei-Fei Li’s World Labs speeds up the world model race with Marble, its first commercial product

World laboratoriesthe startup founded by the AI ​​pioneer Fei-Fei Li, is launching its first commercial model product globally. Marble It is now available through freemium and paid tiers that allow users to convert text messages, photos, videos, 3D designs or panoramas into editable and downloadable 3D environments.

The launch of the generative world model, first released in limited beta previewed two months ago, arrives just over a year after World Labs came out of stealth with $230 million in funding and puts the startup ahead of its competitors building global models. World models are artificial intelligence systems that generate an internal representation of an environment and can be used to predict future outcomes and plan actions.

Startups like rule out and Odyssey have released free demos and The genius of Google It is still in a limited research preview. Marble differs from these, and even from World Labs’ own real-time model, RTFM – because it creates persistent, downloadable 3D environments instead of generating worlds on the fly as you explore. This, the company says, results in fewer transformations or inconsistencies and allows users to export worlds as symbols, meshes or Gaussian videos.

Marble is also the first model of its kind to offer native AI editing tools and a hybrid 3D editor that allows users to lock in spatial structures before AI fills in visual details.

“This is a new category of model that is generating 3D worlds, and it’s something that will improve over time. It’s something we’ve already improved quite a bit,” Justin Johnson, co-founder of World Labs, told TechCrunch.

Last December, World Labs showed how its first models It could generate interactive 3D scenes based on a single image. While impressive, the somewhat cartoonish scenes were not fully explorable as movements were limited to a small area and there were occasional rendering errors.

In my beta preview testing, I found that Marble generated stunning worlds just from image prompts, from game-like environments to photorealistic versions of my living room. Scenes were morphed around the edges, although that was apparently improved in today’s release. That said, a world I had generated in beta using a single prompt looked better and matched my intent more than the same prompt does now.

I haven’t tried the editing features yet, although Johnson says they make Marble practical for short-term gaming, VFX, and virtual reality (VR) projects.

“One of our main themes for Marble going forward is creative control,” Johnson said. “There should always be a quick path to generating something, but you need to be able to dig even deeper and have a lot of control over the things you’re generating. You don’t want the machine to just take the wheel and take away all that creativity.”

Marble’s take on creative control begins with flexibility of input. The beta version only accepted individual images, forcing the model to invent invisible details for a 360-degree view. With the full release, users can now upload multiple images or short clips to show a space from different angles and have the model generate quite realistic digital twins.

Then we have Chisel, an experimental 3D editor that allows users to block in rough spatial layouts (think walls, paintings, or plans) and then add text cues to guide the visual style. Marble generates the world, decoupling structure from style, similar to how HTML provides the structure of a website and CSS adds color. Unlike text-based editing, Chisel allows you to manipulate objects directly.

“I can just go in and take the 3D block that represents the couch and move it somewhere else,” Johnson said.

Another new feature that gives you more editing control is the ability to expand a world.

“Once you generate a world, you can expand it up to one time,” Johnson said. “When you move to a part of the world that’s starting to fragment, you can basically tell the model to expand there or generate more world in the vicinity of where you currently are, and then you can add more detail in that region.”

Users who want to create extremely large spaces can combine multiple worlds with “composer mode.” Johnson demonstrated it to me with two worlds he had already built: a room made of cheese with grape chairs and another of a futuristic meeting room in space.

The path to spatial intelligence

Marble is available through four subscription levels: Free (four generations of text, image or panorama), Standard ($20/month, 12 generations plus multi-image/video input and advanced editing), Pro ($35/month, 25 generations with scene expansion and commercial rights), and Max ($95/month, all features and 75 generations).

Johnson believes Marble’s initial use cases will be gaming, film visual effects, and virtual reality.

Game developers have mixed feelings about the technology. A recent Game Developers Conference Survey found that a third of respondents believed that generative AI has a negative impact on the games industry – 12% more than the previous year’s survey indicated. Among the main concerns expressed were intellectual property theft, energy consumption, and decreased quality of AI-generated content. And last year, a cabling The research found that gaming studios like Activision Blizzard are using AI to take shortcuts and combat burnout.

In games, Johnson sees developers using Marble to generate background environments and ambient spaces and then importing those assets into game engines like Unity or Unreal Engine to add interactive elements, logic, and code.

“It’s not designed to replace your entire existing gaming portfolio, but simply to give you assets that you can put into that portfolio,” he said.

For visual effects work, Marble avoids the inconsistency and poor camera control that plagues AI video generators, according to Johnson. Its 3D assets allow artists to assemble scenes and control camera movements with perfect precision, he said.

While Johnson said World Labs is not focusing on virtual reality (VR) applications at this time, he noted that the industry is “hungry for content” and enthusiastic about the launch. Marble is already compatible with the Vision Pro and Quest 3 VR headsets, and each generated world can be viewed in virtual reality today.

Marble may also have potential use cases for robotics. Johnson noted that unlike image and video generation, robotics does not have the benefit of a large repository of training data. But with generators like Marble, it’s easier to simulate training environments.

According to a recent manifesto According to Fei-Fei Li, CEO and co-founder of World Labs, Marble represents the first step toward creating “a truly spatially intelligent world model.”

Li believes that “the next generation of world models will allow machines to achieve spatial intelligence at a whole new level.” If large language models can teach machines to read and write, Li hopes that systems like Marble can teach them to see and build. She says the ability to understand how things exist and interact in three-dimensional spaces may eventually help machines make advances beyond gaming and robotics, and even into science and medicine.

“Our dreams of truly intelligent machines will not be complete without spatial intelligence,” Li wrote.

Do you have sensitive advice or confidential documents? We’re reporting on the inner workings of the AI ​​industry, from the companies shaping its future to the people affected by their decisions. Contact Rebecca Bellan at rebecca.bellan@techcrunch.com or Russell Brandom at russell.brandom@techcrunch.com. For secure communication, you can contact them via Signal at @rebeccabellan.491 and russellbrandom.49.