Always wanted to relive some of your vacation photos, but only have photos? Well, Google researchers may have the tool for you. They’ve managed to reconstruct detailed 3D scenes of famous landmarks, like the Trevi Fountain in Rome, using regular photographs and machine learning. These aren’t basic models, but rather 3D renderings that allows you to move the camera around like you’re actually there.
NeRF-W, which builds upon NeRF (Google Brain’s Neural Radiance Fields), is the tool they used and it can also reconstruct 3D scenes from a collection of photographs taken in a controlled setting. In other words, it may have trouble with different lighting, image exposure, post processing, random objects, etc.
- 40x optical zoom with optical image stabilizer and zoom framing assist
- 4K video and 4K time-lapse movie
- 20.3 Megapixel CMOS sensor
- High-speed continuous shooting
The Neural Radiance Fields (NeRF) approach implicitly models the radiance field and density of a scene within the weights of a neural network. Direct volume rendering is then used to synthesize new views, demonstrating a heretofore unprecedented level of fidelity on a range of challenging scenes,” said the researchers.