“Google’s Genie 2 AI: Transforming Single Images into Immersive 3D Worlds”
It may also evaluate varied viewpoints and interactive elements such as doors and explosive barrels.
Google’s AI tool, Genie 2, is described as a “large-scale foundational world model” capable of creating “an endless range of action-controllable, playable 3D environments” based on a single provided image.
Genie 2 can evaluate various viewpoints, akin to first-person view, isometric perspectives, or third-person visuals, and can generate “complex 3D visual scenes,” featuring interactive elements such as doors and explosive barrels.
Physical effects, including smoke, gravity, lighting, and reflections can also be “rapidly” prototyped and manipulated by either human users or “AI agents” utilizing keyboard and mouse controls. According to a report describing the advanced technology, this enables artists and designers to prototype swiftly, “which can enhance the creative process for environment design, thereby expediting research.”
“Thanks to Genie 2’s out-of-distribution generalization abilities, concept art and sketches can be transformed into fully interactive environments,” the report clarifies. “This empowers artists and designers to prototype promptly, which can stimulate the creative process for environment design, further accelerating research.”
“While this evaluation is still in its initial phase with significant potential for improvement for both the agent and the environment generation.”