Having good lighting when taking photos or videos ensures quality images even with the worst camera. However, especially for photos taken with smartphones, digital noise, i.e. scattered, mostly bright colored spots in the photos taken, remains a big problem. The Google Research team may have found a solution to this problem by adding a new open source to the artificial intelligence project called MultiNerf.
Technology giant Google recently announced an artificial intelligence that can produce 30-second 3D videos from a single photo by the DeepMind artificial intelligence team. Continuing to work on video and photos, Google teams will take the infrastructure “viewing in the dark” to a whole new level with a new artificial intelligence program called RawNeRF.
The infrastructure was built on a special system
The RawNeRF program can read images using artificial intelligence to add a higher level of detail to photos taken in darker conditions and low light settings. According to the team behind the project, RawNeRF works better than any other noise reduction tool.
This infrastructure was built on specially trained machine learning and artificial intelligence systems. In this way, it ensures that the photos taken are cleaner and clearer. RawNeRF helps to clean even the photos taken with high ISO value in very low light and have “noise” problem in seconds. The most striking point of the system is that it does this by protecting the tissues.
Google researchers explained that RawNeRF can combine images taken from many different camera perspectives to de-noise and reconstruct the scene together. Reconstructed images are rendered in a linear HDR color space, allowing users to further manipulate angles, exposures, tone mapping and focus. Mildenhall states in his video that each of these changing together “creates an atmospheric effect that can draw attention to different parts of the scene.”
Still in the research and development phase, RawNeRF is seen as an exciting development in how AI can help creators better reflect the world around them.
3685
