Noise Reduction Software – Is it worth the money?

Before & After Noise Reduction

A friend recently sent me a RAW image taken on a very old Lumix digital camera with a 1/1.7″ sensor. The image was a really tricky one, very under exposed and while it was only shot at ISO 400, combined with the small sensor this created quite a messy, noisy image. What this did though was to formulate the idea in my head of writing a short article on how to improve a difficult, noisy image using some of the noise reduction tools I had available to me to clean up this image. These are basically On1 2022 NoNoise, Nik Dfine (the old stand alone verion not the newer DxO version) and Affinity Denoise. However I eventually decided just to keep this article short and sweet by focusing on one of my images and just looking at the new kid on the block which is On1 2022 NoNoise Ai.

Why bother about noise? Well, the vast majority of photographers will tell you that noise is ugly and that therefore you should do everything possible to remove it. For this reason trying to control noise in-camera, and if this fails, later in post, becomes hugely important. This is in sharp contrast (excuse the pun) to grain which can add enormous character to photos giving them an ageless filmic look which can be very attractive, even in a digital image. The two are therefore very different but beware, overdone digital grain can be equally distracting so it needs to be used with care.

Getting back to the main theme of this article, as a start point I needed a suitable image to work with. Racking my brain I recalled an image from a night shoot a couple of years back which was taken on a Lumix GX-80 and exposed at ISO 25,600. It was of revellers coming out of the theatre here in Torquay after The Rocky Horror Show. Regrettably, the RAW in question was inadvertantly deleted a year or so ago but I was able to locate a jpeg, which at just 2400px wide offered even more of a challange than the RAW would have done as NoNoise is really designed to process large RAW files not small jpegs. Anyway, as you can image the jpeg was full of really horrible noise so presented a stiff test to On1 NoNoise Ai.

Now for those that don’t know On1 NoNoise, this application is a core feature found in the latest release of On1 PhotoRAW 2022. It’s embedded in the development functionality although since it is usually the first process to be applied, it’s somewhat strange that On1 have chosen to embed it towards the end of the development tools. That being said, it’s perhaps also worth noting that there is now a stand-alone / plug-in version available which works with Photoshop, Capture One, Affinity Photo and others although unlike previous On1 releases, you now have to buy the plug-ins seperately, either individually or as a plug-in bundle. That’s good news for say Lightroom / Photoshop users who only want the plug-ins but not so good if you are an exisitng On1 customer or looking to buy On1 PhotoRAW 2022 as a new customer. Personally I have some strong thoughts on the way On1 have treated their existing client base but rather than bloat this article, feel free to visit http://thecreativecamera.co.uk/groups/bonjour/ to learn what they are!

Now, with regards to the results I obtained, in all fairness to On1 2022 NoNoise Ai did a pretty good job – see the split image above which is a before and after of the image I was working on. Here the noise in the original was completely removed using just the base (default) settings in the NoNoise panel. As I didn’t want to phaff around with this image, I just wanted to get a good comparison, I only really adjusted the white balance after the application of NoNoise simply to make the image look a little more realistic. I did this simply by picking on something white in the image, in this case the mans shirt,and applying a colour shift to it. You could also use Curves if you prefer or the colour settings available in the Effects section. This adjustment reduced the greenish tinge between the original and the processed image. Whether it’s worth working on the image any more than this is debatable as it’s such a low resolution but it does make a great image for comparison purposes.

In Summary

So, what are my thoughts on the results? Well, this is an image that isn’t going to stand a lot of close inspection. Had I started with the RAW, I think the results would have been much much better. If you do a deep dive into the image you’ll see that yes, the noise is really well controlled but at the expense of any real details in the faces. They look very plastic close up although that is as much a result of what I started with as what NoNoise was able to achieve so please don’t go away thinking that this is not a good result. As I mentioned, I made no effort to fine tune the image using either the sharpening or details sliders that are available. Stand back and look at the image from a distance though and it looks pretty good, especially given its history.

UPDATE

Although I initially decided not to do comparisons with other software I did throw the original jpeg into my FREE version of Nik Dfine2 to see what it could achieve. Now bear in mind that this version of Nik has received no updates since it was acquired by DxO in late 2017 so you are looking at the results here from what is now classed as old, defunct technology. Make you’re own mind up as to how far digital noise reduction has come in the past few years or so.

Processed using the unsupported, non DxO version of NIK Dfine2

Something else I noticed about On1 NoNoise, this was on a completely different photo, was that where the software was confused by shadows, in this case rain on the window, it caused some difficulty with the colours and tones behind it. This is best explained by showing you a before and after – see below. I’m not sure if this is a one off or it can be modified in some way but it does show some limitations with the technology.

Original on the left / with NoNoise applied on the right. Notice the degradation in skin tones in the NoNoise image.