Try the early preview of 3D mesh segmentation. Assistant now separates generated assets into individual parts, giving you full control to add behaviors, customize materials & textures, or swap pieces. Soon, you'll even be able to segment imported models. devforum.roblox.com/t/early-…

👋Introducing CubePart, our new work on part-controllable 3D generation from Roblox (to appear in SIGGRAPH later this year). CubePart generates semantically decomposed 3D assets directly from: a text prompt a user-defined part schema. This enables generated assets to plug directly into downstream animations, physics, and game-engine workflows. CubePart relies on strong text-to-3D pre-training to generalize to open-vocabulary part schemas. This rich semantic backbone allows it to flexibly generate coherent parts for previously unseen schemas. 🌐Project page: cubepart.github.io/ 💻Code: github.com/Roblox/cube/tree/… 📜Paper: arxiv.org/abs/2605.28763 🗞️Blog post: about.roblox.com/newsroom/20…

We've released the code for LegoGPT. This autoregressive model generates physically stable and buildable designs from text prompts, by integrating physics laws and assembly constraints into LLM training and inference. This work is led by PhD students @AvaLovelace0, @kangle_deng, @RuixuanLiu_, and in collaboration with CMU faculty Changliu Liu and Deva Ramanan. LegoGPT is a small first step towards the ultimate goal of generative manufacturing of physical objects. Our implementation is limited to 20x20x20 dimensions, 21 object categories, and simple brick types, but we are working on scaling it up! Code: github.com/AvaLovelace1/Lego… Website: avalovelace1.github.io/LegoG… Demo: huggingface.co/spaces/cmu-gi…

We shared some early work towards a multi-modal and multi-task 3D foundation model at Roblox. First release is a discrete shape tokenizer compatible with autoregressive modeling for text-to-shape. More to come soon Github: github.com/Roblox/cube Arxiv: arxiv.org/abs/2503.15475

Roblox just released Cube on Hugging Face A Roblox View of 3D Intelligence

🚀 Introducing NormalFlow: the state-of-the-art tactile tracking algorithm! Just by touch 👇, it precisely tracks objects—even with minimal texture—and can even reconstruct objects! 🌟 project page: joehjhuang.github.io/normalf… paper (RA-L): ieeexplore.ieee.org/document… ⬇️(1/4)

3D content creation with touch! We exploit tactile sensing to enhance geometric details for text- and image-to-3D generation. Check out our #NeurIPS2024 work on Tactile DreamFusion: Exploiting Tactile Sensing for 3D Generation: ruihangao.github.io/TactileD… 1/3