Full disclosure: I’m not affiliated with any brand or software developer. I’m just a regular landscape photographer who happens to shoot Nikon. In fact, I would love to learn which current software reduces noise the best myself.
There are many tricks and tools available to enhance the quality of our images, about as many as there are genres in photography. I’m choosing to compare those tools on a nightscape image, because its typical usage scenario is almost always the same. In which other genre do you max out the ISO as well as the shutter speed for almost every shot? Anyway, raw converters and Photoshop plugins will help to reduce noise, but what we really want is to increase the exposure of the stars in these nightscapes without increasing the noise. We want to increase the signal-to-noise ratio. I’ll touch on that in another article. But we can’t expose for the stars any longer, since the rotation of the Earth will stretch them from pinpricks to streaks across the sky. To avoid star trails, there’s a rule of thumb going around the internet, but I’ll warn in advance that there’s more to it.
To make a long and complicated story short: An increased exposure doesn’t necessarily mean more and brighter stars. If you can follow along, this next part may appeal to you. Don’t worry if it’s too complicated. Just skip to ISO and we’ll catch up there.
The rule of thumb is the 600-rule, which states that if you divide 600 by your focal length on a full-frame camera, the maximum shutter speed in seconds pops out at the other end of the equation. So, if I shoot at 14mm, this “rule” will tell me that I can expose for up to 43 seconds until the stars appear to trail. While this may be the case when your composition includes either celestial pole, it is definitely not the case when you capture the galactic core of the Milky Way. Even a shutter speed of 20 seconds with a 14mm lens on a full-frame body is really pushing it in terms of preventing star trails. The 500-rule is a better rule of thumb, but for the sake of the argument, think more along the lines of a 300-rule if the Milky Way is the subject for your full-frame.
Think of a star that at the beginning of the exposure is in one place. As it progresses along the night sky, its image is projected upon the camera sensor. A photodiode, a piece of a pixel on a camera sensor, translates the brightness of that star into an electric charge. The electric charge recorded over the exposure time multiplied by the ISO-setting is a measure for how bright that star actually is.
Let’s say a pixel recorded about 30% of its capacity. Now, consider that at the end of the exposure, our planet has rotated and the sky seemingly shifted one pixel. And it too records a value of 30%. While both record a value of 30%, neither of those will ever reach 100%, because of Earth’s day and night cycle.
The camera used to capture an image to test the capabilities of noise reduction software was the Nikon D750, with a native ISO range of 100-12,800. Above that range, the signal will be amplified once it’s been recorded by the sensor. At any lower setting than 100, that signal is attenuated. The maximum setting on the D750 will overexpose the city lights on the horizon (light pollution), so we avoided maxing out the camera.
Let’s say we don’t bump the ISO all the way to 12,800, but settle for a stop less at 6,400. That keeps the light-polluted horizon well within the histogram and certainly works wonders for noise reduction purposes.
We will work with Adobe Lightroom as a hub to find, compare, and eventually process the image. What we are doing in this step, is called pre-processing: creating a single file which you can work with in your favorite image editor. This comparison is intended for reducing noise before you do anything else to it in order to retain faint stars and nebulae. Let’s work on the above raw photo, zooming in on the red marked area of the image and applying an “s”-shaped contrast curve to better see what would happen if we processed the image after noise reduction. We’ll compare the seven applications listed here:
Plugins and Filters
And we’ll compare each of those with whichever comes out best of:
- The average of a stack of three images
- The median of a stack of three images
For the tested plugins, I have exported 16-bit TIFF files using Adobe Lightroom, leaving every develop setting turned off. As for the tested raw converters, I have only turned on the noise and sharpening modules in those, effectively exporting the same result as Lightroom should.
Registering and Stacking
What can we do with three images, shot in quick succession at exactly the same camera settings? Of course we can try to auto-align them in Photoshop, but that will not get us very far. The sky shifts with each exposure, while the foreground is static. A dedicated free program will aid us in the alignment. We will register the initial three exposures (untouched TIFF files from Lightroom) in DeepSkyStacker, a program dedicated to astrophotography. It both registers (detects stars in exposures and aligns them to compensate for the rotation of the Earth) and stacks the images you’ve shot. The are many more applications that will do the same, like PixInsight, Starry Landscape Stacker for Mac, and Registar for Windows, but DSS is completely free. I ask that you trust me when I say that using multiple exposures that are offset slightly from each other result in the best possible noise reduction method without sacrificing small details. And the more images you have, the better your result will be. The true workings of stacking software are very complicated and are not within the scope of this comparison. However, I have only used DeepSkyStacker to align 3 images, which I’ve stacked in Adobe Photoshop CC. It’s really important that you align every star perfectly before you do, whether you do this automatically in DeepSkyStacker or manually in Photoshop with layer masks.
Average Versus Median
Stacking the exposures in Photoshop through a median mode is achieved by importing each registered exposure into a separate layer, then selecting all three layers and going to to Layer, Smart Objects, Convert to Smart Object. Now, go to Layer, Smart Objects, Stack Mode, Median.
The median method above works pretty well when you have a large number of exposures. Another method to stack your images in Photoshop is by averaging them. Averaging the three exposures works by putting each registered one in a separate layer as well. Set the bottom layer to 100% opacity, and each subsequent layer to an opacity value that is 100% divided by its exposure number. So, the second exposure you add will be set to an opacity of 50%, as 50 equals 100 divided by 2. A third exposure will be set to 33%, because 100 divided by 3 equals a rounded 33%. A fourth should be set to 25% opacity and so on.
Eventually you end up with something like this:
Average vs. Median.
While the stars are perfectly aligned with the help of DeepSkyStacker before, the foreground is all jumbled up. It’s easy to explain, but harder to correct. First, I can hardly sit still for a couple of seconds (and yes, this is a glorified selfie). Second, DeepSkyStacker detects stars and aligns the images based on their position. The foreground then gets distorted due to the rotation of the Earth. To compensate, I typically shoot one extra exposure just for the foreground at a lower ISO setting and longer exposure length. The foreground isn’t affected by the rotation of the Earth, so you can easily expose for a couple of minutes in low winds. Do be aware that increasing the shutter speed will also increase the temperature of the sensor, and that will introduce thermal noise in your foreground.
As you can see, producing a clean, noiseless nightscape involves advanced planning, shooting, and processing, but that’s not why you’re here. Let’s get to the actual noise comparison. Averaging the three images does produce a marginally smoother image, as seen above. Here is stacking versus raw with added contrast:
Stacking vs. raw with added contrast.
So, the above comparison shows our best stacking method with three images. Mind you, three images is not a very good stack, but it sure does help with suppressing noise. We will move on to comparing each application with that average stack, so we can see how well they reduce noise without sacrificing detail.
These suites perform a host of features. Noise reduction is one of them. The idea is that raw files contain more data, and when you get rid of noise during this step, it is more accurate. Let’s see if there’s any truth to that.
DxO Optics Pro 10 Elite
The “Elite” version of Optics Pro comes at a cost of $70 more than the essential edition, but also comes with some great perks. PRIME noise reduction is one of them, and it is supposed to yield amazing results. I’ve used these settings, which gave me the best results for this image:
Average stack vs. PRIME noise reduction on raw.
It’s nice and smooth, but optical aberrations at the edges of bright stars become more apparent, which is easy to correct with the built-in lens correction feature at the press of a button. However, faint stars are getting drowned out by the higher contrast that its raw conversion algorithm produces. Optics Pro also leaves some color to be desired in the stars, but the structure in the Milky Way nebulae is very pleasing to the eye, especially when we’re not zoomed in 300%.
Lightroom 6 comes in at $149, or $9.99 a month for its CC counterpart as well as Photoshop, but does so much more than noise reduction. In Lightroom, I’ve applied these settings, which yielded the best results for this image:
Average stack vs. Lightroom noise reduction.
The careful color noise reduction comes at a cost, but the Masking setting is great. The latter works by not applying any sharpening on smooth areas, thus only sharpening fine detail. There’s a lot of color noise going on, but the preservation of colour in the stars is quite good, just like the contrast in the nebulae.
PhaseOne Capture One
A raw conversion suite as well as a library for all your work it’s a true contender against Adobe Lightroom, but at a whopping $299.
Average stack vs. Capture One noise reduction.
Capture One’s sharpening is quite similar to Lightroom’s, with the Threshold slider preventing smooth areas from becoming overly sharpened. The amount of noise reduced is very similar to the average stack, but there’s much more random color in large swaths, while there’s almost no color at all in the stars. I personally do like the pattern of this noise, since it resembles film grain and doesn’t look quite as digital, but that’s a matter of taste.
Plugins and Filters
A plugin is something that works in or around your image editing workflow. You can apply a specific filter in a standalone program and come back to the image editor with your photo hopefully looking better than before. I’ve racked up the most popular noise reducers that work together with Photoshop.
Photoshop CC 2014: Reduce Noise Filter
Photoshop has its own noise reduction filter. So, why go through the trouble and price of a dedicated noise reduction program? I’ve found that these settings with this filter give the best results for this image:
Average stack vs. Photoshop’s Reduce Noise Filter.
The amount of noise reduction and color almost look identical to the stack, but that’s where the similarities end. Some of the larger, faint stars are pulled from the sky. Large stars are reduced in size and the structure of the nebulae becomes much harder to see, and there’s a distinct swirly thing going on with a good amount of introduced local contrast.
Nik Dfine 2
Recently, Google announced that the Nik Collection is now free to download. There’s absolutely no reason not to use and discover the power of both Color Efex Pro and Silver Efex Pro, but what about the collection’s noise reduction plugin? I’ve left the measurement on auto and increased the reduction of Contrast Noise, while I toned down Color Noise for best results.
Average stack vs.Nik Dfine.
That’s interesting. Faint stars in the darker areas are erased, while stars in the brighter nebula are mostly left alone. The pattern looks slightly swirly, but more natural than Photoshop’s attempt. There’s also a slight increase in saturation on some of the stars, while others have been desaturated. Finally, we see less structure in the nebulae, and medium brightness stars are reduced in size. But not bad for a free plugin!
Noise Ninja 1.2
Noise Ninja is yours as a part of PictureCode’s Photo Ninja Suite for $129. As with most of these suites, you’ll get a host of applications along with Noise Ninja, one of the most talked about noise reduction algorithms in recent times. Does the Noise Ninja 4 Turbo engine live up to the hype as we apply these settings to our night sky?
Average stack vs. Noise Ninja.
The noise pattern looks identical to me. That’s a good thing. And the reduction does a good job at smoothing out the background. However, the stars in that background are smoothed until there are none left, while the overall image looks to me like it’s pockmarked rather than star-speckled. The nebulae do look more cloudy and contrasty than some of the contenders, but not by much.
Topaz DeNoise 6
The latest upgrade to DeNoise is yours for $79.99, while the whole suite of Topaz software, which includes 16 more plugins, sells for $499. But just look at the amount of settings. I tweaked these for the better part of two hours to get the best result I could squeeze out from DeNoise.
Average stack vs. DeNoise.
It’s buttery smooth, but with a lot of digital-looking artifacts. I like the fact that this algorithm works by analyzing things locally. Not every part of the image is the same, and some parts benefit from more reduction than others. With that being said, the result isn’t natural-looking anymore. The nebulosity is more foggy than cloudy, and stars that are close together clump together with lots of noise in between them. There are obvious squares in the pattern as well, and faint stars are erased here too.
Put Them Together
So, what’s the verdict here? Let’s compare all of them in one composite to reach a conclusion.
The big surprise is that Photoshop’s built-in filter retains the most stars after the median stacking method and the average stacking method. The background produced by Lightroom is rather grey and cold, while Capture One’s background is the most colorful of the bunch. Lightroom and Noise Ninja provide the highest contrast between the nebulae and background in this area.
Stacking the foreground doesn’t work with the methods used in this comparison, so we’ll leave them here. Topaz DeNoize produces a foreground that looks out of focus, but there’s no noise left. Adobe’s Lightroom and Photoshop both produce the most detail, especially in the shadows. Capture One and Optics Pro add in more contrast than the other methods, but that can easily be balanced by lowering the contrast since they are raw converters. Noise Ninja and Dfine are the balanced winners in this case, with Dfine showing more detail and Noise Ninja more noise reduction.
This is where it gets hard to count the stars. So, let’s look at the obvious differences first. Both Capture One and Optics Pro are the higher contrast images. That may be to your liking, but again, it can be dialed down easily if it’s not. The first three images, the plugins, sure have the least amount of stars. And of those, Dfine is more grainy, and DeNoise is more blurry. The raw converters do better, but Photoshop’s filter leaves the most stars intact when compared to the stacks. Noise Ninja and Lightroom come really close here, so let’s compare those two in detail, without the added contrast.
Lightroom vs. Noise Ninja.
With Lightroom showing more color detail and Noise Ninja showing more reduction, I’d say this is matter of taste. The blue shift in the stars is also due to the fact that this comparison is a crop of an area close to the edge of the frame. Aberrations in the lens (coma and chromatic aberration), as well as an aperture setting that shows the most of those aberrations are the reasons for that. Noise Ninja handles those aberrations quite well without any other correction at all. Lightroom, however, respects those color shifts in order for you to work on them separately.
So, which will you choose? My recommendation is that you at least do some amount of noise reduction with a program you currently have. And by all means, give Nik Dfine a shot. It’s free, remember? The prize for most aggressive noise reduction method goes to Topaz Denoise, and the prize for most conservative to Photoshop‘s Reduce Noise filter.
The very best results, however, are achieved by shooting many more images, editing those frames simultaneously in a raw converter of your choice without applying noise reduction, and stacking them, perfectly aligned. Any residual noise you can then reduce with the aid of whatever plugin or filter you already use, but with restraint. It’s a lot of work, but it produces the highest quality image, one that’s representative of that great night sky.
My opinion? Stacking only two images increases the signal-to-noise ratio about 1.4 times, but it also increases the dynamic range, so you’ll have more leeway in post-processing. I’m sure eager to try Noise Ninja after I have stacked at least two images or if I haven’t had the time to shoot multiple images in the field. I’ve previously always used PRIME noise reduction in DxO Optics Pro on my single-shot nightscapes, but that’s about to change due to the results I’ve seen here. Thank you for your time. I know it has been a long read!
Like most of us, you likely have dozens, if not hundreds of pictures stored on your phone or your camera’s SD card. But are you making the most of them? Most of them probably look okay, but with a free photo editor and a little time you can transform them into something amazing that you won’t be able to wait to share online, or even frame on your wall.
Whether you’re looking for a total Photoshop alternative that gives you fine control over every minuscule detail of your photos, or a basic Instagram-style free photo editor that offers a range of one-click filters, you’ll find something interesting here.
If you’re already happy with the general look of your photos and just want to tweak them a little before sharing them, a simple free photo editor that allows you to resize, crop and apply filters will be ideal. Many of these have social media sharing built in, removing the hassle of uploading your snaps manually later. They also have the benefit of being easier to use.
If you’re planning to print your work, you’ll be better off considering a more nuanced free photo editor that will give you the fine control you need to make your photos look perfect, even at high resolutions.
A free photo editor that’s a worthy rival to premium software
GIMP (the GNU Image Manipulation Program) is the most powerful free photo editor around. It’s packed with the kind of image-enhancing tools you’d find in premium software, and more are being added every day.
GIMP’s interface will be instantly familiar if you have ever used Photoshop – especially if you select the single-window mode, which lays out all its toolbars and canvases in an Adobe-style layout.
The photo editing toolkit is breathtaking, and features layers, masks, curves, and levels. You can eliminate flaws easily with the excellent clone stamp and healing tools, create custom brushes, apply perspective changes, and apply changes to isolated areas with smart selection tools.
GIMP is an open source free photo editor, and its community of users and developers have created a huge collection of plugins to extend its utility even further. Many of these come pre-installed, and you can download more from the official glossary. If that’s not enough, you can even install Photoshop plugins.
This combination of power and flexibility make GIMP the best free photo editor you can download – whether you’re using Windows, macOS or Linux.
Fine manual controls packaged in an accessible interface
More is not, believe it or not, always better. Paint.NET‘s simplicity is one of its main selling points; it’s a quick, easy to operate free photo editor that’s ideal for trivial tasks that don’t necessarily justify the sheer power of GIMP.
Don’t let the name fool you, though. This isn’t just a cheap copy of Microsoft’s ultra-basic Paint – even if it was originally meant to replace it. It’s a proper photo editor, just one that lands on the basic side of the curve.
Paint.NET’s interface will remind you of its namesake, but over the years, they’ve added advanced editing tools like layers, an undo history, a ton of filters, myriad community-created plugins, and a brilliant 3D rotate/zoom function that’s handy for recomposing images.
It might not have every feature you can dream of, but if your machine is a little underpowered we can’t think of a better free photo editor.
A superb photo editor that takes many of its cues from Photoshop
Photo Pos Pro isn’t as well known as Paint.net and GIMP, but it’s another top-quality free photo editor that’s packed with advanced image-enhancing tools.
This free photo editor’s interface is smarter and more accessible than GIMP’s array of menus and toolbars, with everything arranged in a logical and consistent way. If it’s still too intimidating, there’s also an optional ‘novice’ layout that resembles Fotor’s filter-based approach. The choice is yours.
The ‘expert’ layout offers both layers and layer masks for sophisticated editing, as well as tools for adjusting curves and levels manually. You can still access the one-click filters via the main menu, but the focus is much more on fine editing.
Photo Pos Pro also includes a clone brush for erasing unwanted blemishes, and there’s extra support for batch-editing and scripts to help you save time when refining a whole folder of photos.
The free edition of Photo Pos Pro only has one drawback: files can only be saved at a maximum resolution of 1,024 x 2,014 pixels, which might be too small if you’re planning to have them printed professionally. If you want to remove this restriction, Photo Pos Pro Premium is available for a license free of £24.50/US$24.90/AU$41.89.
Advanced tools and filters that are easy for beginners to master
PhotoScape might look like a rather simple free photo editor, but take a look at its main menu and you’ll find a wealth of features: raw conversion, photo splitting and merging, animated GIF creation, and even a rather odd (but useful) function with which you can print lined, graph or sheet music paper.
The meat, of course, is in the photo editing. PhotoScape’s interface is among the most esoteric of all the apps we’ve looked at here, with tools grouped into pages in odd configurations. It certainly doesn’t attempt to ape Photoshop, and includes fewer features.
We’d definitely point this towards the beginner, but that doesn’t mean you can’t get some solid results. PhotoScape’s filters are functional and not at all beginner-like, so it’s if good choice if you need to quickly level, sharpen or add mild filtering to pictures in a snap.
One-click photo enhancement tools and advanced manual controls
Fotor is more a photo enhancer than a full-fat manual editing tool. If there’s specific area of retouching you need doing with, say, the clone brush or healing tool, you’re out of luck. However, if your needs are simple, its stack of high-end filters that really do shine.
Fotor’s most brilliant function, and one that’s sorely lacking in many free photo editors, is its batch processing tool – feed it a pile of pics and it’ll filter the lot of them in one go, perfect if you have a memory card full of holiday snaps and need to cover up the results of a dodgy camera or shaky hand.
Professional-grade filters that work alone, or as plugins
Google’s unending determination to corner just about every market sometimes pays dividends for the pincher of pennies. Take its purchase of German developer Nik in 2012, for example – its Nik Collection photo editor plugin range retailed for US$500 at the time, and in early 2016 Google decided to do away with the price tag and release the powerful collection for free.
These are perfect free plugins if you’re already using Photoshop, and you can add them to compatible host applications when you install them, but they can also be run as standalone free photo editors if you hunt down their executable files. They won’t appear in your list of Windows apps – you need to look in C:\Program Files\Google\Nik Collection. To edit a photo, drag it onto the EXE file of your chosen editor. It’s a strange system, but it works!
Apply special filters to selected areas for stunning results
The ‘free’ suffix offers some indication of what you’re getting here: On1 Effects 10.5 Free is a cut-down version of On1 Effects 10.5 proper, pulling out just a limited selection of its filters. But we’re still happy to recommend it, mainly because of its methodology.
Essentially this is an taster for the full version, but its diminished filter range – HDR, vignette, vintage, glow etc – is still useful and worth trying if you’re after vibrant effects; you’ll have to try another program for sharpening, blurring and noise reduction, so On1 Effects Free isn’t great if you want to preserve the honesty of your photos.
Simple, but with the high quality you’d expect from Adobe
Adobe Photoshop Express is a very different beast to the full version of Adobe’s mighty industry-standard image editor, but it’s useful for giving well-composed pictures the boost they need to become stunning.
There are no advanced editing tools here, so you won’t be able to paint out blemishes, adjust lighting, or even crop your images. Instead, you’re given a selection of good-looking Instagram-style filters and a quick link so you can send the results directly to Facebook, or save them to your PC.
Adobe Photoshop Express is simple, but the filters are excellent and more are available as in-app purchases if you want more choice.
- On a mobile device? Check out our list of the best photo editing apps.