Tripo AI raises $50M and launches new 3D generation models for design, XR, simulation, and production workflows | XR News from Auganix.org ...

Meet Tripo AI, a leader in developing foundational AI models that power 3D model creation across industries such as gaming, ...

IT之家 5 月 14 日消息，阶跃星辰正式发布并开源 3D 大模型 ——Step1X-3D，这是继图像、视频、语音、音乐等模态后，阶跃星辰在多模态方向的最新成果。Step1X-3D 模型总参数量达 4.8B（几何模块 1.3B，纹理模块 3.5B），凭借坚实的数据基础与先进的 3D 原生架构，可 ...

2025年5月，香港中文大学的郑铎、黄世佳、李岩洋和王立威发表了一项创新研究，题为《从视频中学习3D世界：用3D视觉几何先验增强多模态大语言模型》（Learning from Videos for 3D World: Enhancing MLLMs with 3D Vision Geometry Priors）。这篇研究论文已在arXiv预印本平台上公开 ...

A new SAR interpretation method helps reveal the specific sources on three-dimensional targets that correspond to strong ...

The model does not have access to 3D geometry, depth buffers, or PBR material properties, all of which are inferred or ...

NVIDIA confirms DLSS 5 uses 2D frames and motion vectors, not full 3D data, raising questions about accuracy and control.