@pritipatelor: NVIDIA just released a really interesting new tool: with any random image, you can generate a 3D world that's freely ex…
Summary
NVIDIA开源了名为Lyra 2.0的模型，可将任意单张图片直接转换为可自由探索的交互式3D世界，甚至可用于向生成的场景中放入机器人进行仿真训练。模型已发布到Hugging Face，UI代码也已在GitHub开源。
View Cached Full Text
Cached at: 10/04/26, 07:18 PM
NVIDIA just released a really interesting new tool: with any random image, you can generate a 3D world that’s freely explorable
And the project is 100% open source 🔥
The model is already up on Hugging Face The UI code has been open sourced to GitHub
This model called Lyra 2.0 can turn an ordinary image directly into an interactive 3D scene
You can not only walk around freely inside, turn to explore, but even put robots into the generated world for simulation training
In simple terms: one image → a 3D world you can step into
GitHub：
Similar Articles
@itsPaulAi: Woow Nvidia has just released a 2.6B open-source world model You can turn a single image, text prompt and trajectory in…
Nvidia released a 2.6B open-source world model that can generate controllable worlds from a single image, text prompt, and trajectory, running on a single GPU.
Meet Helix World: A Real-Time Audio-Visual World Model (13 minute read)
Noiz AI 与香港科技大学联合发布 Helix World，一个基于单张图像的实时交互式视听世界模型，可在单块 NVIDIA H800 GPU 上以 24fps 同时生成可探索的视频场景与空间立体声，代码、checkpoint 和在线 demo 均已开源。
nvidia/Lyra-2.0
Lyra 2.0 is NVIDIA's framework for generating persistent, explorable 3D worlds from a single image, combining long-range video synthesis with explicit 3D reconstruction while addressing spatial forgetting and temporal drifting through novel training techniques.
@Huahuazo: While browsing GitHub, I found an eye-catching project — upload an ordinary photo, and it automatically recognizes the object's structure, proportions, and materials, directly generating a complete 3D model. Let me mention a few designs I found impressively unconventional: It comes with a "semi-finished" attribute out of the box — the generated model isn't a rigid photo texture, but includes clear hierarchy and collision body components...
Introducing the open-source project img2threejs on GitHub. Upload an ordinary photo and it automatically generates a code-based Three.js 3D model with hierarchy and collision bodies, suitable for direct use by developers and game creators.
@seclink: 5. Open-Source Acceleration of Robot World Models - NVIDIA Cosmos 3 + Isaac GR00T: Physical AI Foundation Models - AGIBOT Genie Sim 3.0: The First Fully Open-Source Robot Simulation Platform (Complete Open Source of Code, Data, and Assets) - VLA (Vision-…
Robot world models and simulation platforms are experiencing open-source acceleration: NVIDIA launched Cosmos 3 and Isaac GR00T physical AI foundation models, AGIBOT released Genie Sim 3.0, a fully open-source simulation platform, VLA models become mainstream for manipulation policies, collectively lowering the entry barrier for the robotics field.