VGGT4D：无需训练，挖掘3D基础模型潜力，实现4D动态场景重建

VGGT4D 提出了一种无需训练的新范式，成功将 3D 基础模型的能力扩展至 4D 动态场景。该工作证明了通过合理挖掘模型内部的 Gram 相似度统计特性，可以有效解耦动态与静态信息。这不仅为低成本的 4D 重建提供了新思路，也展示了基础模型在零样本迁移任务中的潜力。
The Verge

‘The Fourth Dimension’ for iOS: learn to see in 4D (hands-on)

Posts from this topic will be added to your daily email digest and your homepage feed. Hands-on impressions of The Fourth Dimension, an app that attempts to explain 4D geometry. Hands-on impressions ...
Digital Production

Cinema 4D 2026.1: Fluids, Distributions, and Practical Fixes

Cinema 4D 2026.1 adds GPU-accelerated fluid simulation, new MoGraph distributions, and workflow improvements across Take ...