In our new paper, “A Neural Field-Based Approach for View Computation & Data Exploration in 3D Urban Environments,” recently accepted to IEEE Transactions on Visualization and Computer Graphics (TVCG) and led by UIC PhD student Stefan Cobeli, we introduce a neural field-based, view-driven method that encodes environments into an implicit representation.
This supports both direct queries (e.g., visibility and solar exposure analyses) and inverse queries (e.g., automatically suggesting promising viewpoints). The paper builds on our ongoing work in 3D urban analytics (https://urbantk.org/survey-3d/), and supports a range of urban analysis tasks, including evaluating outdoor spaces and exploring views for new developments.
Paper, code, and model are available here: https://urbantk.org/neural-3d/
