Neural Radiance Fields, or NeRFs, are a relatively new type of 3D model that can be generated from a set of 2D images for a scene. Introduced in 2020, NeRFs have shown remarkable capabilities in generating photorealistic images and 3D models from sparse input data. The key idea behind NeRFs is to represent a 3D scene as a continuous function that maps 3D coordinates to color and density values.
Mapillary's journey with NeRFs
We integrated NeRFs into our platform in early 2024—check out our introductory blog post. This integration enables users to generate high-quality 3D models through only photos they've captured. By leveraging NeRFs, Mapillary can now provide more accurate and detailed 3D reconstructions, which can be used for various applications such as urban planning, architecture, and virtual reality.
Currently, NeRFs are only visible on our web app. The NeRF point layer is on by default—simply click on an image thumbnail bubble to view a NeRF! If you'd like to toggle off the layer, you can do so via the 'Map data' button.