Despite recent advances in 3D modeling, reconstruction, and generation, many methods remain limited to static scenes or dense viewpoints, making them less effective in real-world, dynamic, and often sparse or noisy settings. This workshop aims to gather researchers and practitioners focused on modeling, reconstructing, and generating dynamic 3D objects or scenes under challenging, in-the-wild conditions. Leveraging progress in 3D learning, the abundance of 2D/3D data, and powerful generative models, now is an opportune time to make 3D vision more robust and accessible. The workshop encourages contributions from standard 3D topics and broader 4D directions involving dynamics and video generation.