Researchers create more precise 3D reconstructions using only two camera perspectives

https://techxplore.com/news/2024-06-precise-3d-reconstructions-camera-perspectives.html

"combines neural network of surface with illumination process model considering light absorption... image brightness determines angle/ distance of surface relative to light source... natural surroundings to reconstruct relatively textureless objects... neural networks that predict 3D point clouds for individual video images that are then merged into a large-scale model of the roads traveled... autonomous driving, preservation of historical artifacts (digitize museum exhibits)"

See also: In a first, 3D models created with just two images, can boost self-driving
https://interestingengineering.com/innovation/3d-reconstructions-two-images-tum