Low light photos, and those taken with poor quality cameras, have a lot of noise. It is something inevitable, with physical limits imposed by the world of optics, but Artificial Intelligence can process the result to greatly improve the image.
That’s what they do from Google Research with the MultiNerf project, an open source project that works wonders to correct photo noise. Watch this video to see several examples of the results:
The objective is to obtain a raw image and analyze the existing components in it. The AI ​​is in charge of recognizing each object and redoing it in the best possible way, interpolating information to make everything clearer. It is something similar to what the programs that I presented in the article 5 sites to increase the quality of your photos do, but with Google’s AI behind it.
There is no software that incorporates this technology at the moment, since until now everything is research and laboratory, but surely the photos of the future taken from the mobile will have a before and after when activating the AI ​​icon that we usually have on smartphones .
In the video we see results that seem to come out of cartoon movies, realism is lost, textures are lost, but it’s all a matter of time. Logically, an AI can never be compared to a good quality photograph, since it can never represent reality 100%, it will only imagine it and process it so that the result is visually pleasing, but it can help a lot in vacation photos where neither not even our eye will see as much as the camera of our mobile.
