Search papers, labs, and topics across Lattice.
3
0
6
Flow matching in latent spaces can unlock new capabilities for 3D scene generation, outperforming traditional methods by leveraging learned geometric priors.
T2LDM++ generates realistic LiDAR scenes with rich geometric details, overcoming the limitations of existing models that struggle with insufficient training data and controllability.
Ditch the heavy 3D encoders: Fase3D uses Fourier transforms to build a surprisingly efficient and scalable 3D large multimodal model directly from point clouds.
