Topaz DeNoise AI has been the hot topic lately and you might have seen a handful of ads across social media praising it. The ads claim that it performs miracles with noisy images but is that really true?
Does it deliver the results that the ads claim it does? Is it really the best noise reduction software?
That’s what I want to find out.
After spending a week guiding a photography workshop in Greenland I’ve returned home with hundreds of high-ISO images that are in desperate need of some noise reduction. This is exactly what DeNoise AI is made for so I went ahead and purchased it to see if it really can perform miracles.
Keep on reading this Topaz DeNoise AI review to find out whether noisy images can be salvaged or not (and if it works for night photography too)
Already know what DeNoise AI is? Click here to hop to the review.
What is Topaz DeNoise AI?
Some of you reading this might be wondering what on earth Topaz DeNoise AI is. So, before we open the first image, let’s take a quick look at what this software is and what its purpose is.
The short answer is that Topaz DeNoise AI is a noise reduction software.
The long answer is that DeNoise AI is an intelligent noise reduction software that uses deep learning technology to accurately distinguish between real image detail and noise.
According to Topaz Labs, this allows you to reduce noise while recovering crucial image detail. That sounds pretty good to me as a common issue with noise reduction has been images losing details and appearing soft.
It can be used as standalone software or as a plugin to Adobe Lightroom or Adobe Photoshop. The best results come from opening RAW files directly inside the standalone tool but I haven’t noticed a difference when, for example, opening the RAW file from Lightroom.
What I Am Looking for in a Noise Reduction Tool
I kind of touched on this topic above as the main thing that I look for in a noise reduction software is the ability to remove noise while protecting details. Far too often we see images appear soft after having denoising applied.
Details are important in the image. Landscape photographers don’t want the mountains to look as smooth as silky water. Portrait photographers want the skin to appear sharp.
Besides this, I do appreciate software that’s “dumbed down”. I’ve got nothing against learning a new tool but at the end of the day, the easier you can get good results, the better.
(Which is exactly why I use Raya Pro to create Luminosity Masks rather than doing it manually for each photo)
At the end of the day, a noise reduction software is exactly that; a tool that’s supposed to speed up your workflow and deliver the same, or better, results as if you were doing it manually.
Topaz DeNoise AI Review: Does It Live Up to the Hype?
Ok. Let’s get back to the question at hand: how good is Topaz DeNoise AI?
The best way to answer this question is to actually run images through the program. It’s easy to make things sound great in writing but I believe that the results will speak for themselves. Either they’re good or bad.
To keep this as real and honest as possible I’m only going to use RAW files in this review. None of these images have received any adjustments beforehand and are opened directly into Topaz DeNoise AI (i.e., not via Adobe Lightroom or Adobe Photoshop)
The images I’ve selected are all different as I want to see how well it performs on both “easy” and “complex” photos. How will it tackle images with lots of details? What about minimalistic images? Night photography? Let’s find out.
Topaz DeNoise AI Performance on Medium ISO Photography
I spend a fair amount of time photographing from a boat when traveling in Greenland. This often means photographing handheld and bumping the ISO up to levels I’m not used to from stationary landscape photography. Especially when closing up on sunset and blue hour.
My latest visit was no exception and I captured a bunch of photos at ISO1000 and up. This resulted in a fair bit of noise but that’s exactly what I want to see if DeNoise AI can help with.
This should be a fairly straightforward image to denoise as there isn’t a lot of detail in the image besides the iceberg itself. Let’s see what happens:
Well… That turned out a lot better than I expected. What just happened to the noise?! This could’ve been shot at ISO64. That sky looks noise-free!
At a closer look, though, I find the ice to be slightly softer than what I’d like. The noise is gone but it certainly lost some of the finer details. That’s what the Recover Original Details slider is for so let’s give that a go:
That looks a lot better. We might have brought a tiny bit of noise back into the sky but the ice looks better than it did before.
I’m actually quite surprised how quickly and accurately it “fixed” the image. I’ve tried many noise reduction tools in the past but this is looking really promising so far.
Topaz DeNoise AI Performance on High ISO Photography
It might not be that often that we use the highest ISO values but from time to time there is no other option (obviously, certain genres of photography are more familiar with extreme ISO use than others)
For this test, I selected an image that was shot at ISO25600 with a Fuji X-T2. It’s a good camera but not one that’s exceptionally good at dealing with such ISO values, so it’s fair to say that noise is an issue in this shot.
My biggest question is how Topaz DeNoise AI will handle the details in the rocks and cliffs.
The first thing I noticed when opening the image in DeNoise was that the AI Model was automatically set to Severe Noise, not RAW. I’m guessing that this is the best AI Model when dealing with such high ISO values, regardless of the file format.
So, let’s take a look at the results:
Pretty impressive. We can see that the noise is successfully removed from the image and both the sky and water look a lot better than before. I wouldn’t say it’s perfect but it’s also not realistic to expect an ISO25600 image to look like an ISO64 image.
But what about the details? I don’t want the rocks and cliffs to appear soft or blurry.
Again, I am pretty impressed. It’s undoubtedly a bit on the softer side and it requires some further processing to look sharp but this is absolutely acceptable. After increasing the Recover Original Detail slider to +65 the image looks like this:
The difference isn’t huge but the rocks appear slightly sharper than before. Not bad for an extremely grainy photo shot at ISO25600 with an APS-C camera.
How Good is Topaz DeNoise AI for Night Photography?
We’ve seen that DeNoise does a good job for standard images but what about night photography?
This is a more complex type of image to denoise as it contains a sky full of stars that can be hard to distinguish from the noise. Traditionally, denoising software hasn’t worked great for this type of photography.
For this test, I used a Milky Way image shot at ISO4000 with a Nikon D800. Again, I’m opening the unedited RAW file directly inside Topaz DeNoise AI.
Now, I’ll admit that I don’t mind some noise in the night sky. That is natural and, in many cases, it can emphasize the stars. I’d rather have some noise than a completely smooth sky where the stars look faded.
Let’s see how good Topaz DeNoise AI is with night photography:
This looks quite good in my eyes. The Milky Way core is perhaps slightly softer than what I would like and when zooming into 200% I notice that there are a handful of stars that have lost brightness/appear blurrier but this is being really nit-picky.
I wanted to try DeNoise AI on one more night photograph as well. This one doesn’t have quite as many stars but it’s got some Northern Lights and a fair bit of details in both the mountain and foreground. It’s mainly the mountain and foreground I’m curious about what will happen to.
The result shocked me. To the point where I left the computer for a minute to share my surprise with my wife.
The sky is noise-free and no stars have been affected but more impressively the mountains and foreground appear sharper than they were before. We haven’t lost any details; we have improved them.
It’s fair to say that Topaz DeNoise AI does a good job with night photography as well.
Recovering Lost Details
While we haven’t had a big issue with huge loss of details in the images we’ve worked on so far, it does happen from time to time. In fact, some of the images above will benefit from a little sharpening too.
There are a couple of different ways to do this but, personally, I prefer to use layers and masks in Photoshop to better control the results.
It’s possible to create a mask in Topaz DeNoise AI to selectively denoise your images but I find Photoshop to be slightly more accurate when it comes to creating precise selections.
This is how I approach recovering lost detail or sharpness:
- Open the original and denoised files as layers in Adobe Photoshop
- Make sure that the denoised image is on top
- Choose a soft brush with a low opacity (+/- 20%)
- Optional: Create a selection of the area you want to recover detail in
- Gently brush over these areas to bring back some of the noise
The steps above are an easy way to bring back some of the noise and detail found in the original file. You want to be very careful not to bring back too much of the noise as this will look unnatural.
Another option, and perhaps a better alternative in most situations, is to create a High Pass sharpening layer (or use Topaz Sharpen AI) and apply it to the soft areas. This will bring back sharpness without introducing the noise you’ve already removed.
Noise Reduction in Adobe Lightroom vs. Topaz DeNoise AI
When I was drawing out this article, I had planned to do a close look at the performance of Adobe Lightroom versus Topaz DeNoise AI. After testing the software, however, I think that the results speak for themselves.
The results you get from Topaz DeNoise AI is far superior to what you can achieve in Adobe Lightroom. There really isn’t much to compare.
Regular readers of CaptureLandscapes might know that I’m a fan of the Nik Collection so I decided to test the same high ISO image in Nik Dfine 2 as well.
Again, the results speak for themselves. I’ve always considered Dfine to be a decent noise reduction tool but, I’m sorry to say, it can’t compete with Topaz DeNoise AI. At least not for high-ISO images.
Nik Dfine 2 did an ok job removing noise from the water and sky but the rocks have lost a lot of important detail. The image now appears soft and would require a lot of further work.
ARE YOU ENJOYING THIS ARTICLE? THEN HELP US OUT BY SHARING IT!
Conclusion: How Good is Topaz DeNoise AI?
I’m not going to lie. I’ve been actively trying to find images where Topaz DeNoise didn’t do a good job. I wanted to look at it with a critical eye to show that all those ads I’ve seen were overly hyped.
What I found was quite different, though. The results have gone beyond any expectations that I had. I think it’s fair to say that Topaz DeNoise AI is the best noise reduction tool I have tried. By far.
It will be a crucial part of my image editing workflow from now on. I can highly recommend it to you as well.
Do you want to try Topaz DeNoise AI yourself too? Click here to get your copy for $79.99 or download the free trial version.