Google ’s newAI tool can create well-lit 3D images from dark 2D photos.- AI tool manipulates focus, exposure and tone mapping.
- Below, check out the power of
RawNeRF in a 6-minute long video.
RawNeRF is part of the larger MultiNeRF project that uses NeRF (Neural Radiance Fields) to scan a collection of images and reconstruct a 3D scene render. Earlier this year, Google Research presented its work on RawNeRF at the Computer Vision and Pattern Recognition (CVRP) 2022 conference.
This technology seems very similar to the movie scene in Iron Man 3, where Tony Stark uses 3D images to recreate a street explosion.
RawNeRF uses raw images taken from different camera positions, exposure and focus. RawNeRF combines images taken from different camera viewpoints and combines them to denoise and reconstruct a scene using AI. To enhance the picture, AI manipulates the focus, exposure and tone mapping, says Google researcher
RawNeRF can also handle scenes with a large dynamic range, switching focus and manipulating exposure levels and shooting angles.
The secret behind RawNeRF is not just manipulation with exposure level or denoising, it's how the AI is trained. The RawNeRF tool is trained in such a way that it gathers data from raw images rather than standard JPEG format. Captured raw images have more detail than standard photos, and those details can be used to enhance the image in post-processing.
Will the RawNeRF AI tool will come to our smartphone camera in the near future? This is a tough question to answer, but using the RawNeRF AI tool in smartphone cameras will completely change the perspective of smartphone photography.
SEE ALSO:
This herb is the magic fertilizer needed to grow vegetables on Mars
DRDO has developed a facial recognition system that claims to see through masks and disguise