Google’s AI Noise Reduction Brings Images Out from the Dark into Light
For still photographers and videographers, Google’s latest AI image application brings great advantages for shooting in dark areas.
Matt Growcoot has a new piece in petapixel.com that says Google researchers have unveiled a new AI program that has incredible denoising ability allowing users to “see in the dark.”
RawNeRF can combine images taken from many different camera viewpoints to jointly denoise and reconstruct the scene. And since a picture is worth a thousand words take a look at the picture below to see the RawNeRF in action.
Ben Mildenhall, a research scientist, gives an example of a table lit only by candlelight that he took on his smartphone. Attempts to brighten the RAW file gave an awful amount of noise.
But, after Mildenhall ran it through RawNeRF the image came back without almost any noise and with far more detail in the scene.
“And of course, RawNeRF is not just a denoiser, which means we can vary the camera position to view the scene from different angles,” Mildenhall explains.“RawNeRF goes beyond this, we actually reconstruct the scene and render new views in a linear hdr color space. This means we can also do things, like vary the exposure, by rescaling renderings in linear space before they’re post-processed.”
To most of us, it appears they can capture reality and then bend it to suit the desired results of the photo. It’s just more ‘magic’ being delivered by AI to this generation of photographers.
Photographers Laud Noise Reduction Tools
From fstoppers.com which is a photographer’s website we found another man who is sold on the new noise reduction programs coming out. Alex Armitage posted a video and his thoughts on this tool.
Years ago when I tried software like this it would always give very mixed results and end up ruining parts of the image like the transitions from branches to background. Ever since adding learning-based computation, the AI part of all recent software suites, the results have become mind-blowingly powerful. Specifically when it comes to wildlife. When the software can detect what type of object, in this case, a bird, within the image. It does an even better job at enhancing the areas of the image after removing the noise.
The results of the nightscape are good, but it is limited by trying to guess and evaluate every object and detail within the image. Overall I’m very pleased with the results and it means I can print such an image much larger than I could before but where ON1’s NoNoise AI shines is high ISO wildlife photography. I’ve had varying success depending on the animal within the photo as well and it is dependent on the clarity of the original image.
Stand by. Things are going to become crystal clear for movies, photographs, and video gaming with the use of this AI magic.
read more at petapixel.com
or at fstoppers.com