News hardware Smartphones: with this, Google photos will crush those of the iPhone 14
When we talk about mobile photography, we often think of Apple or Samsung, but since the release of the Pixel 6 last year, the cards have been reshuffled. And obviously, the Mountain View firm does not intend to stop there.
The Google Pixel 6 made everyone agree when it was released
Since Google’s Pixel 6 released in October 2021, the smartphone world was forced to recognize that Google had made drastic progress in terms of photography. A trend that was further confirmed with the Google Pixel 6a which was launched in July 2022.
But even if the lenses go up in range over the years and releases, the bulk of the work is done by the software part of smartphones. Thanks to increasingly advanced artificial intelligence, the latter are able to take photos or videos of quite remarkable quality. The functionalities are already multiple and yet Google does not want to stop there.
Through hard work and research, Mountain View researchers continue to experiment with new methods to further boost the quality of videos and photos. A job that probably ended up paying off, since if we are to believe our colleagues from 9to5Google, the member of GAFAM would have found a way to sublimate the night photos a little more.
RAWNeRF: Google’s new cheat code
What could look like a combination of hazardous keys, is in fact the name of a technology drastically worked by Google. Inspired by an AI that processes 2D images to generate 3D scenes called NeRF, RAWNeRF is able to process raw images. According to the words of 9to5Google, this technology is able to generate 3D scenes and also improve photos taken in the dark. How ? Simply by removing the noise (= phenomenon caused by insufficient lighting when shooting).
When optimized on many noisy raw inputs (25-200), NeRF produces such accurate scene representation that its new renderings outperform those of single-frame or multiple-frame dedicated raw denoisers running on the same frames. As a result, our method, which we call RawNeRF, can reconstruct scenes from extremely noisy images captured in near darkness.
Photo credit: Ben Mildenhall, Peter Hedman, Ricardo Martin-Brualla, Pratul Srinivasan and Jonathan T. Barron
So understand here that the RAWNeRF technology is supposed to be more powerful than some software specialized in noise suppression. That said, don’t expect to find this little gem in the Google Pixel right away, since processing raw images requires a lot of resources. Nevertheless, it is not impossible to see in the near future, a capable chip as well as a whole software part capable of supporting RAWNeRF. If that day comes, Samsung and Apple will therefore have to worry, because to compete with such technology it will certainly be necessary to mourn several nights of sleep.
Source : 9to5Google and Cornell University