New technology produces incredibly sharp images even in low light

A group of independent developers, backed by search technologies from Google Research, are developing technology capable of capturing photos and recording videos in great quality at night. The MultiNerf project uses advanced algorithms linked to artificial intelligence to eliminate noise generated by image sensors.

The result of this technology is a clean picture, true to the environment and almost as good as a record taken in natural daylight. Great insight lies in the connection between photography and computer graphics, producing a unique result. paying off:

& Ampere ; Ampere ; & nbsp;

One of the creators of AI, Ben Mildenhall, made the video above in which he records some of the techniques in action. It’s amazing to see how natural the result looks, without the traditional pixels or image distortions common in night photography.

The material that served as the basis was recorded by cameras configured at 30 frames per second. Of course, it’s still a long way from the 4K and 120fps capabilities that many modern cameras can achieve, but it’s a very important step forward for the world of digital night photography.

Nerf is an acronym adopted by researchers for the expression Neural Radiance Fields. Basically what the technology does is adjust exposure, gamma map, tone and even the right focus to deliver an image with the least amount of optical noise possible. The coolest part is that the camera can move at different angles, including blurring and shifting elements, without affecting the displayed image.

Computational photography on a mobile phone

The so-called computational imaging is already present in many modern smartphones and receives different names. in case if SamsungNight Photography allows you to capture high-quality, crisp photos and videos at night or in low light. Grandma is in line Galaxy S22 It is one of the most important features of the camera.

Already An apple It has night mode in its iPhone cameras since version 11. The result is not as optimized as the South Korean competitor, but it is the result of hardware optimization related to artificial intelligence and algorithms.

For now, MultiNerf is still part of the research, Available in open source on GithubIt is not available as a commercial product. However, it is interesting to see how photography still has room to evolve, especially with the increased graphics and processing power of modern mobile phones.

Trending on Canaltech: