De-noising is unquestionably high on the list of operations we like the least when it comes to post-production. For one thing, it’s extremely taxing on computers, which is why we disable the noise reduction node until we’re ready to render the entire project. For another, used incorrectly, it can introduce unwanted artifacts and destroy important detail. We’ve already expressed our horror in these pages at YouTubers who advocate enlarging the image 999%, indiscriminately blasting the clip with gobs of temporal and spatial noise reduction followed by heaping on shocking amounts of sharpening, making the footage resemble cheap camcorder video. Instead, we suggest using noise reduction sparingly and sharpening not at all. We certainly don’t want to eliminate noise altogether, as it helps prevent banding, among the most disagreeable artifacts known to mankind; and with certain cameras, noise can add depth and texture to an otherwise bland digital image. However, if de-noising is unavoidable, it’s desirable to evaluate the effect of noise reduction on a decent sized monitor or television at a normal viewing distance; but if you’re sharing your video to YouTube, we’d go even further and recommend uploading the video with visibility set to private first, dialing back noise reduction if necessary, then re-uploading. Sometimes, it’s not necessary to apply noise reduction to the entire clip, as the most offensive noise is usually found in the darker regions of the image. The following steps illustrate how to selectively apply noise reduction to the shadows.
Be sure to watch Cullen Kelly’s Grade School for more tips like this.