Noise reduction is a critical component in the post-processing workflow for photographers, especially those frequently dealing with high-ISO images or challenging lighting conditions. Understanding where and when to apply noise reduction can significantly impact the quality of the final image. This helpful video will show you the optimal time to do to ensure you get the best possible shots.
Coming to you from Anthony Morganti, this educational video addresses a common confusion among photographers regarding the optimal stage in the workflow for noise reduction. Historically, noise reduction was recommended to be done as early as possible in the editing process. However, with the advent of AI noise reduction in Adobe Lightroom, Morganti explains that the timing has become more flexible. He demonstrates this with a Nikon raw file of a bald eagle shot at a high ISO. The image, exhibiting considerable noise, serves as an excellent example to showcase the effectiveness of AI noise reduction at different stages of editing.
Morganti further elaborates on traditional versus AI noise reduction techniques. He points out that while older methods and third-party apps like Denoise AI may require early noise reduction application, AI noise reduction in Lightroom is more versatile. To illustrate, he compares two versions of the same image – one edited and the other unedited – both processed for noise reduction at different stages. This comparison effectively demonstrates that AI noise reduction can be applied after other edits without compromising the image quality. It's nice to see workflows becoming increasingly flexible to suit the style of individual photographers. Check out the video above for the full rundown from Morganti.
I like doing selective noise reduction at higher ISO, but don't mind the noise at lower ISO, as well as some more noticeable luminance noise at moderate ISOs for low light, since the higher ISO will reduce the amount of high frequency detail. At moderately high ISOs, too much luminance noise reduction, even with AI, can make things look off, but the randomness of some of the luminance noise left in, provides perceptual detail that is subjectively better than strong noise reduction.
The AI noise reduction built into the latest Lightroom and Adobe Camera Raw, is quite good, though it is not perfect, especially if you want to get a near noise-free result. Its AI model seems more focused on preserving detail rather than filling in detail that could be destroyed by noise reduction, and it can sometimes lead to artifacts, even if used in a part of an image with adequate lighting. Overall, it seems to prefer larger amounts of noise, and even then, a close eye on looking for artifacts.
I attached a random example, though I wouldn't use that noise reduction on an image like this anyway, as it is often unnecessary, and AI based noise reduction tend to have trouble with certain surfaces and textures.
(It also hates older laptops, thus not as useful when I am not on my main PC)