I’ve long been a user of Topaz DeNoise AI, as well as DxO PureRAW, to help eliminate noise from my images. But now, there’s another company trying to take care of your noise images, with Skylum releasing its NoiselessAI extension today for Luminar Neo.
I ran some tests between NoiselessAI, Topaz DeNoise AI and DxO PureRAW to see which would perform better. While all three performed well and helped to lessen or eliminate the noise in my images, there was one that came out on top.
How to use NoiselessAI
Before I do a comparison, I wanted to run a few tests on higher ISO images to see how Luminar Neo’s newest extension would perform.
The tool is pretty simple to use. After installing NoiselessAI, switch over to the Edit tab and locate NoiselessAI in the list of tools. Expand this tool and you’ll see three preset adjustments — Low, Medium and High — to choose from. Luminar Neo will analyze your image and recommend one for you, which is great.
Once you choose one of the three presets, you’ll have options to further tweak the results. Three sliders are present in the Adjustments tab — Luminosity Denoise, Details and Sharpness. You can take it a step further, too, by going over to the Masking tab. This lets you use the brush, gradients and MaskAI tools to select only parts of your image to be affected by NoiselessAI.
Once NoiselessAI is complete, it’ll zoom in on your image, letting you see the noise that’s been reduced. You can use the eye icon in the tool to see what your image looked like before and after the adjustment.
Speed and results with NoiselessAI
Overall, Luminar Neo did a nice job here, and the results were effective in reducing noise in my images. Details stayed sharp, and NoiselessAI seemed to do what it promises.
There was one big flaw, though, at least with the beta version I had. NoiselessAI — especially when you first run an image through it — is pretty slow. Using one of Luminar Neo’s sample images from a Nikon D800, setting NoiselessAI to the High setting took 15 seconds to render. And that was on an M1 Ultra Mac Studio. I can’t imagine how long it would take on an older machine.
Speed issues aside, NoiselessAI does a nice job, and it’s a great option for those who rely on Luminar Neo on a daily basis. It’s great to have these tools as a part of the overall ecosystem, and it eliminates Luminar users from having to purchase third-party applications when everything can be done right inside the host program.
Comparing NoiselessAI to Topaz and DxO
Sure, Luminar Neo’s NoiselessAI is a great option. But how does it stand up to some of the industry’s leaders?
I used this image below of our very own Vanelli, which was shot on a Micro Four Thirds camera with an ISO of 5000. Without some type of noise reduction, this image would not be usable.
For all three programs, I used the standard and recommended settings. With Luminar Neo, that meant the Middle setting. I’ve zoomed in so you can see the differences between the three programs. For DxO PureRAW, I turned off sharpening.
All three results are incredibly similar, but we can see a few minor differences. The clearest and sharpest is definitely DxO PureRAW. The blacks here are also richer; you can really see Vanelli’s logo popping on the black of his shirt.
Topaz DeNoise and Luminar Neo are even closer, and honestly, it’s tough for me to pick a winner between those two. Topaz did retain more detail while taking out additional noise (especially Vanelli’s shirt and around his eyes), but it also left some noise in weird places, like the top of Vanelli’s hair, and the silver pieces in the upper right of the image. I’ve experienced this with Topaz before, and it’s probably my number one qualm about the software.
Based on this photo, DxO PureRAW is still the king of denoising — despite its lack of controls — followed by a tie between Topaz DeNoise and Luminar Neo NoiselessAI. The fact that I really had to zoom in to tell a difference should be telling, though. For its first crack at AI noise reduction, Luminar Neo is surprisingly good. If it can fix the speed issues and some of the lost details, it’ll definitely be a winner.
For photographers wondering which they should pick, I’d say it really comes down to your workflow, and what makes the most sense for you. I’d be happy with any three of these images when turning them into clients.