One of the by-products of digital photography is image noise. “Noise” is a consistent pattern of interference that occurs when an image is imprinted into a digicam’s CCD or CMOS sensor. It would be akin to the (sometimes) intentional noise that occurs when using high ISO based film, though tends to be persistent in digital photography even at a low ISO.
When printing images at 4×6 or 5×7 – the noise inherent in a digital photo is hardly perceptible. This changes dramatically when enlarging digital images to larger sizes – such as 8×10 and beyond. The noise becomes extremely noticeable in areas where there is a consistent color – such as a blue sky or the petal of a flower. To add to this problem, when one sharpens an image in the final phase of post-processing, the noise is sharpened right along with the rest of the image.
To combat this, many denoise tools have been introduced on the market – with one of the leaders being Noise Ninja by PictureCode. This tool does an excellent job at intelligently removing the noise in an image and allows you to import custom profiles for your specific digital camera (or scanner). The results from the default profile and using automatic noise removal is excellent and allows one to comfortably print enlargements in the 12×18 and 16×20 range with very little perceptible noise. Using a noise filter will also increase the quality of smaller prints as well.
There are two drawbacks to using a noise filter on an image. Many times I have seen the filter remove shadow details within an image and can sometimes be too extreme in that regard. Also, on several images with very fine detail (such as an image with tall grass in the foreground) I have seen a loss of edge detail and a blending of the images that, in my opinion, detracts from the crispness of the image. Despite these drawbacks, it is still an essential function within the digital workflow.
The big question – where does one apply “denoise” within their digital workflow. The accepted methodology is to denoise near the very beginning of the workflow. Not one to accept what’s generally accepted, I conducted my own tests and arrived at the same conclusion. The noise filter is better at removing noise before the image has been heavily modified and also prevents further post-processing from applying the effects to an all ready noisy image. I did find that in very few images that it was desired to do another low-impact pass of noise filtering after I do my final sharpen to clear up any noise artifacts that weren’t caught in the process.
