The Image Noise Reduction Challenge: We Compare the Best Noise Reduction Software for Photographers
My biggest post-processing challenge in photography is image noise. Unlike studio photographers who can precisely control every aspect of lighting, I’ve been photographing most of my campaigns handheld in natural light for more than 15 years.
I’m a travel photographer shooting regional lifestyle photography for different agencies all around the world. Because of how I shoot and travel, including the desire to be less conspicuous for my own safety, working with strobes generally isn’t appropriate for me. My primary light source is the sun in the golden hour, and my style is very breezy, natural, and spontaneous, which means I sometimes shoot at higher ISOs.
Why Shoot With High ISO?
Simply stated, higher ISOs dial up the sensitivity of your image sensor. When the right combination of aperture and shutter speed isn’t enough to achieve a proper exposure, bumping up your ISO to handle low-light conditions is a given.
The purists will argue: Why not add some artificial light to the mix, instead? Higher ISOs degrade the quality of your photo.
But controlled lighting is not always an option. Flash photography is often banned or unfeasible in certain environments. Can you imagine popping off a flash during a sacred wedding ceremony or climbing a tree in the wilderness to set up a nice rim light for a perched eagle?
Many wildlife photographers, wedding photographers, photojournalists, and lifestyle photographers can relate: when your job is to capture — rather than create — authentic moments, there will often be some trade-offs to achieve the shot.
The Problem With Higher ISOs: Image Noise
Shooting in low light with higher ISOs will introduce noise in the resulting image. Intentionally applied grain, like what you get from nostalgic old-school film, can be a beautiful aesthetic choice.
But unwanted noise can be quite extreme, and the detail in your photos will suffer. Some high-end cameras certainly handle noise better than prosumer models, but every high-ISO photo could be improved upon with the help of noise reduction tools in post-production.
Discovering the Best Noise Reduction Software
A few years ago I started a side project photographing musicians in Madrid. The shots were very casual — at night, in the city center, discovering locations on the fly, making use of ancient architecture, and using whatever ambient light was bouncing around the streets.
Shooting artistic portraits helped stretch my creativity beyond the commercial photography I shoot in the day, and since the photos were designated for low-res Instagram posts, I was able to focus more on the mood and feeling rather than precision output.
Then, the career of a jazz singer I regularly photograph took off, and suddenly, the work I had produced was being printed onto full-size posters and shown on Spanish TV. Awesome! But that meant reprocessing some of those low-light high-ISO portraits to a cleaner commercial standard suitable for high-resolution display.
As a lifelong Adobe fan, I’ve had limited experience with other photo apps and plugins. Between Photoshop and Lightroom, what more could you need?
Unfortunately, I learned Lightroom doesn’t quite cut it when you’re working with heavy noise. My goal was to achieve a silky-smooth background (which required pushing the Luminance slider way up) without losing texture in the details. Here was the result in Lightroom working with my ISO 6,400 portrait, which needed to be prepped for a large-format print:
I needed something more powerful than Lightroom. A photographer friend recommended DeNoise AI by Topaz Labs to solve my noise issue, and I gave it a try. I only needed a light touch of the noise reduction and sharpening sliders before the noise was fully corrected. As you can see, the DeNoise AI results were quite good:
The noise reduction output with DeNoise AI was exactly what I was looking for: a smooth, noise-free background without obliterating skin and hair texture. Here’s a comparison alongside the Lightroom output and the original noisy image:
Further testing with DeNoise AI showed that it continues to outperform Lightroom’s noise reduction even with less extreme noise issues. This tight crop of an ISO 1,250 photo looks sharper and more defined after running it through DeNoise — note the skin texture, the sharp hairs of the mustache, and defined lip.
I was so impressed with these results that I looked no further. DeNoise has been my go-to noise reduction application since discovering it last year.
I reached out to the CEO of Topaz Labs last summer to chat more about the technology and ended up partnering with them to supply ongoing images and feedback for all of their AI-powered photo editing applications. When they recently asked me to check out their latest beta of DeNoise, I decided it was high time to take a deep look at other popular noise reduction solutions. Do other applications offer better noise reduction, more intuitive editing options, or a better workflow? Let’s find out. On to the challenge!
The Image Noise Reduction Challenge
For this challenge, I’m going to work with a terrible photo — a handheld indoor snapshot taken at ISO 6,400 from a Canon 6D. I chose to use my family’s rescued parrot for this example so we could test noise reduction possibilities on the most challenging elements: eyes and feathers with a heavy blanket of thick color noise.
I’ll share my results below, but you’re welcome to see for yourself. For testing purposes, you can download and work with my cropped TIFF file, a cropped JPG, or the original CR2 raw file. All of the applications reviewed below offer free trials, so you can try them all out and see which works best for you.
Test Image: ISO 6,400
I’ll attempt to get the best results from each application under the following guidelines:
- Increase noise reduction strength until noise is substantially removed from the background
- Sharpen, if available, without introducing artifacts
Website: Adobe Lightroom
Price: $9.99/mo USD
Pros: Robust industry-standard raw photo editor
Cons: Noise reduction can soften details at higher ISO, software is only available by monthly subscription
Lightroom is my default raw image editor, and it’s served me well for many years. While generally, the noise handling does just fine, I’ve found that increasing noise reduction to accommodate heavier noise muddles up the image.
Working with the raw CR2 file with some minor exposure adjustments, I had to pump up the Luminance slider quite high to smooth out the noise, and the result was a more buttery texture in the foreground. Sharpening was limited before artifacts became too extreme. Here’s the final output:
Topaz DeNoise AI
Website: Topaz Labs DeNoise AI
Price: $79.99 USD
Pros: Cutting-edge AI-powered software with exceptional noise reduction, easy-to-use interface
Cons: Resource-intensive application can run slow on less-equipped computers
DeNoise AI can be launched as a plugin or a standalone application. I loaded up a TIFF from Lightroom and made my adjustments.
I only needed about 30% power on the noise reduction slider before the background was clean. I was able to ramp up the Sharpening slider to 100 without any artifacts to help bring in detail from the feathers. Using the Recover Original Detail slider, I brought back some of the texture. As always, I was impressed with DeNoise AI. The results are sharp and clean. Take a look:
DxO PhotoLab 3
Website: DxO PhotoLab 3
Price: $129 USD
Pros: Feature-rich raw photo editor
Cons: Prime noise reduction tool is difficult to visualize, making precision output a challenge to achieve
DxO can also handle raw files, so I launched the original CR2 file and got to work.
Working with DxO is a little tricky. I adjusted the curves since the original raw file was quite dark and used their PRIME mode for the best noise reduction algorithm they offer. In the screenshot above, these were my final adjustments just before export. As you can see, the full-screen preview doesn’t reflect my final output below — the noise reduction effects are displayed in that tiny thumbnail to the right. It was difficult to fully gauge my adjustments. I used a touch of Lens Sharpness and Unsharp Mask, and the final output came out better than Lightroom, but on closer inspection, you’ll see some strange artifacts around the perimeter of the bird and an overall blockiness:
Website: Neat Image
Price: From $39.90 USD
Pros: Noise reduction tool with detailed adjustments for refining output
Cons: Low-quality output for higher ISO photos, nNo support for raw files
Neat Image is a standalone app, and while it doesn’t work with raw files, I was able to launch a TIFF.
Once you set up your profile and switch into Advanced mode, you’ll find Neat Image has a wide variety of noise reduction tools. Dialing down any of the Frequencies sliders reintroduced noise, so I kept them ramped up to 100%. I tweaked the Sharpening levels to try to improve the thickness of the detail, but this is the best output I could get:
Website: Dfine 2
Price: $149 USD (Included in Nik Collection 3 plugin suite)
Pros: Minimal plugin for quick and easy noise reduction
Cons: Poor-quality results with high-ISO photos, limited options to refine output
Dfine 2 can be launched from Lightroom as a plugin, so I imported a TIFF.
The options here were surprisingly limited (no sharpening tools), and the output quite poor. The entire image looks like it’s underwater. I had hoped that the output would be better than the preview. Nope. Dfine 2 couldn’t maintain any detail with its noise reduction tools:
Website: Phase One Capture One
Price: $349 USD or $24/mo
Pros: Powerful industry-leading raw photo editor
Cons: Noise reduction capability is not as powerful as other features
Capture One by Phase One is what professional photographers use as the most common alternative to Lightroom. I work with a big team of commercial photographers in South Africa, and they all swear by Capture One for its raw handling and superior color output. I loaded up a raw file and gave it a shot.
Capture One’s noise reduction tools are similar to Lightroom’s, but the results fell flat. The details across the image were muddy and the noise wasn’t entirely removed. From what know of Capture One’s user demographic, I suspect Phase One hasn’t felt the need to advance their development into noise reduction. Why would they need to when their users are typically meticulous studio photographers shooting at native ISO? Here’s the output:
We had a wide variety of results with the noise reduction challenge. I’d love for our audience to decide the winner! You’ll see our targeted problem areas in our original photo to compare the output across all these applications:
What do you think? Do you have a preferred noise reduction application? Do you use a different noise reduction technique we should explore? Let us know in the comments.