Demoisaicing is particularly tricky in parts of an image with a lot of texture, detail and colors. "Crowded mathematical calculations are required to perform the interpolation required to build an image," Adobe suggests in a white paper. "This takes time, even on the most powerful hardware. Therefore, software like Lightroom constantly balances balance between image fidelity and velocity deviation.
The problems that usually arise are the loss of small details, false colors over sharp edges, moire, edge release (zipping) and more. Using Adobe's Sensei, Enhance details have been trained a lot in the cloud so that it can use the hardware on your computer or Mac. "We trained a neural network to demosic problematic RAW images, since [use] machine learning built into the latest Mac OS and Windows 1
The result is up to 30 percent more resolution in Bayer and X-Trans RAW files, and it comes organically from the camera's sensor, rather than being artificially generated. There are some precautions: You need a relatively powerful computer, and the process is only useful for images that are designed to be printed in large size, or those with lots of details or artifacts.
On top of the enhancement feature, Adobe added HDR, Pano and HDR Pano fusion tools, along with a targeted customization tool and histogram clips. The HDR and Pano functions make it easier to combine multiple images to create more dynamic range images, either in standard or panoramic formats.
Lightroom CC now has a targeted adjustment tool that makes it easy to fine-tune specific parts of an image, such as the sky. The histogram clips show areas that are too dark (in blue) or too bright (red), which helps you manage under- and over-exposed parts of an image. Finally, Adobe added image sharing to its Lightroom iOS app and fixed errors in the Android and ChromeOS versions, "making room for new features coming soon". The release starts to roll out today to Creative Cloud users with access to Lightroom CC.