NVIDIA Unveils Lyra 2.0, Creating Expansive 3D Scenes from Single Images

On April 16, 2026, NVIDIA's research team officially launched Lyra2.0, a 3D scene generation system. This technology constructs expansive, highly coherent virtual environments from a single photograph, effectively solving the problem of image distortion over long camera paths. With the growing demand for embodied AI training, Lyra2.0 represents a significant breakthrough in AI's ability to understand 3D space and simulate environments in real time.
Technically, Lyra2.0 can generate 3D environments extending up to 90 meters from a single image. To combat the spatial distortion and error accumulation caused by "forgetting" in traditional video models, researchers implemented two key innovations. The system stores 3D geometry data for each frame in real time, ensuring environmental consistency when the camera revisits previous locations. It also trains on defective output data, enabling the model to learn self-correction. Benchmark tests show Lyra2.0 outperforms six competing models, including GEN3C and Yume-1.5, in image quality and camera control, while its fast version boosts generation efficiency by 13 times.
Lyra2.0 now integrates seamlessly with physics engines like Nvidia Isaac Sim, allowing generated 3D scenes to be exported directly as mesh models. This closed-loop workflow enables robots to conduct efficient simulation training within AI-generated environments, drastically reducing the need for large-scale real-world 3D data collection. Although currently limited to static scenes, Lyra2.0's advancements in generation scale and stability provide a more robust and imaginative infrastructure for advancing physical perception in autonomous driving and general-purpose robots (AGI).
Related article
ByteDance’s Seed launches global campus drive, offering virtual shares to win top large model talent
In the competitive landscape of large language models, securing top-tier talent remains the most critical strategic asset.On April 1st, ByteDance announced the launch of its Seed global campus recruitment initiative, part of its large model talent de
Suno to Watermark Songs Amid Legal Battles
Suno, the platform enabling users to generate AI-created music, has unveiled new features to label platform-produced tracks, restrict downloads, and update community standards to curb unauthorized replicas. These updates arrive as Suno confronts mult
Musk Admits Grok Build Leaked User Code, Promises to Erase All Historical Data
Elon Musk directly addressed the privacy controversy surrounding Grok Build, beginning with a simple "True" to confirm the incident's validity. He pledged that all user data previously uploaded to SpaceXAI would be permanently erased, stating, "not a
Related Special Topic Recommendations
Comments (0)
0/500

On April 16, 2026, NVIDIA's research team officially launched Lyra2.0, a 3D scene generation system. This technology constructs expansive, highly coherent virtual environments from a single photograph, effectively solving the problem of image distortion over long camera paths. With the growing demand for embodied AI training, Lyra2.0 represents a significant breakthrough in AI's ability to understand 3D space and simulate environments in real time.
Technically, Lyra2.0 can generate 3D environments extending up to 90 meters from a single image. To combat the spatial distortion and error accumulation caused by "forgetting" in traditional video models, researchers implemented two key innovations. The system stores 3D geometry data for each frame in real time, ensuring environmental consistency when the camera revisits previous locations. It also trains on defective output data, enabling the model to learn self-correction. Benchmark tests show Lyra2.0 outperforms six competing models, including GEN3C and Yume-1.5, in image quality and camera control, while its fast version boosts generation efficiency by 13 times.
Lyra2.0 now integrates seamlessly with physics engines like Nvidia Isaac Sim, allowing generated 3D scenes to be exported directly as mesh models. This closed-loop workflow enables robots to conduct efficient simulation training within AI-generated environments, drastically reducing the need for large-scale real-world 3D data collection. Although currently limited to static scenes, Lyra2.0's advancements in generation scale and stability provide a more robust and imaginative infrastructure for advancing physical perception in autonomous driving and general-purpose robots (AGI).
ByteDance’s Seed launches global campus drive, offering virtual shares to win top large model talent
In the competitive landscape of large language models, securing top-tier talent remains the most critical strategic asset.On April 1st, ByteDance announced the launch of its Seed global campus recruitment initiative, part of its large model talent de
Suno to Watermark Songs Amid Legal Battles
Suno, the platform enabling users to generate AI-created music, has unveiled new features to label platform-produced tracks, restrict downloads, and update community standards to curb unauthorized replicas. These updates arrive as Suno confronts mult
Musk Admits Grok Build Leaked User Code, Promises to Erase All Historical Data
Elon Musk directly addressed the privacy controversy surrounding Grok Build, beginning with a simple "True" to confirm the incident's validity. He pledged that all user data previously uploaded to SpaceXAI would be permanently erased, stating, "not a





Home






