I Didn't Believe the Hype Until I Tried AI Noise Reduction Myself

As a landscape photographer, I don't usually deal with too much noise in my images unless I'm shooting nightscapes. Once I started shooting more wildlife during my time in Alaska, I decided to give it a try and I was absolutely blown away with the results. As someone quite new to wildlife photography which involves shooting in the exact opposite style as landscape photography, there was a lot to learn. One of those lessons was that I have to bump my ISO to gut-wrenching levels, at least by my standards which typically include never going above ISO 400. After capturing a few shots around ISO 5,000 and then pushing all the way to ISO 12,800, I felt it was time I tried one of the plethora of software options I'm continually getting ads for to see if they could do anything close to what they advertised. 

Throughout this article and video I'll be using ON1 NoNoise AI but fully suggest you try all the available options out there to find what gives you the best results. Many of them have free trials and you can test them at your leisure. 

Nightscape Noise Reduction

Previously I tried reducing the noise in a few nightscape images I had taken over the years and had mixed results. Sometimes it would give me decent results and other times it didn't do much at all. I ran a recent nightscape image I took in Oregon through ON1's NoNoise AI with decent results. No matter what software you're using they typically suggest that you process your images through the program as raw files, rather than after you're done editing them. I had already finished this image which involved a 20-star stacked Milkyway blended with a 3-minute exposure for the foreground, thus I wasn't delivering the best possible file to the program yet it still did a great job.

Above is a close-up of the before and after the noise reduction. The obvious difference is the amount of noise removed from the dark areas but what is important to pay attention to is the detail put back into those areas. Also, notice how great the water looks without reducing the ethereal feel of that three-minute exposure. If this smoothing is too much for you, you can always adjust the amounts applied or get specific and mask out certain areas.

Here is the final full image before and after, keep in mind that the starry area was stacked in both images meaning there isn't much noise to remove. What was nice is that the processing didn't remove detail within the milky way and star areas. Overall I was very happy with the results but I wasn't blown away... Until I started processing a few wildlife shots I took.

Mind Blown

Unlike the previous image, my workflow for these images involved running ON1's NoNoise AI first and then applying my edits which should give the best results than doing it in reverse. 

The above image was taken at ISO 12,800 and while the uncropped version would have been passable for online use, as soon as I cropped in on the eagle, the noise from the Canon R5 became much more apparent. This is when I tried processing it through the software and was quite literally blown away. Not only did it reduce the noise very well, but it also added and enhanced the details in subtle yet realistic ways. Pay attention to the definition of the eagle's feathers throughout the image, both brown and white. To my eye it not only reduced the grainy noise texture, it also provided a sheen-like texture back into the feathers that wasn't there before. Just as before also notice the water, how it keeps the definition of the ripples without removing detail yet completely removing noise. 

I took this image a few weeks prior and posted it to Instagram without adding any noise reduction outside of lightroom's integrated noise reduction. Even at small resolutions, I noticed the noise specifically in the blurry tree areas, and essentially wrote this image off as unusable because of it. Although it was taken at ISO 5000, I had to recover quite a bit of shadow detail and overall those out-of-focus areas were distractingly noisy. 

At almost 100% crop it becomes quite obvious how incredible the results are and to me, this feels like magic. The feathers, the beak, the out-of-focus areas — every single thing within the image looks better after processing it through the software. Look specifically at the areas where there are out-of-focus leaves in front of the eagle and how the transitional area from those blurry leaves to the eagle feathers has a lot of green in it, as you'd expect from blurry leaves. Then look at the noise-reduced version and how it kept the essence of those blurred green parts yet still reduced the noise correctly. Typically noise comes through as green or red grainy texture and the fact that it can differentiate between the noise and what is actual details within the image is amazing. 

If your results aren't as pleasing I'd suggest adjusting the amounts of processing you're applying to your image. The truth is, I don't think I changed a single setting on most of these images. You can find more examples in the video within this article.

Conclusion

Years ago when I tried software like this it would always give very mixed results and end up ruining parts of the image like the transitions from branches to background. Ever since adding learning-based computation, the AI part of all recent software suites, the results have become mind-blowingly powerful. Specifically when it comes to wildlife. When the software can detect what type of object, in this case, a bird, within the image. It does an even better job at enhancing the areas of the image after removing the noise.

The results of the nightscape are good, but it is limited by trying to guess and evaluate every object and detail within the image. Overall I'm very pleased with the results and it means I can print such an image much larger than I could before but where ON1's NoNoise AI really shines is high ISO wildlife photography. I've had varying success depending on the animal within the photo as well and it really is dependent on the clarity of the original image.

I'd love to know your experience with AI-based noise reduction software. Share some of your before and afters in the comments below as I'd love to see just how far you can push this technology. I'll have a future article coming out that takes a look at AI sharpening and how important it can be for landscapes so be on the lookout for that.