There are two common scenarios where noise creeps into our photos. One is when shooting at high ISO, typically indoors or at night. The other is when we try to lighten and extract detail from the shadows. In both cases, the problem is a lack of light and the solution is to either gather more light when shooting or use noise reduction during post-processing. In this tutorial, you’ll learn how to reduce noise in post with Adobe’s solutions.
There is a lot of debate about which software does the best job of reducing noise. There are many great options, and some of them can outdo Adobe in some scenarios. But I still prefer using Lightroom or ACR (Adobe Camera RAW in Photoshop) most of the time for a few reasons. First, while I’ve seen some results that are better, the ones I’ve tested are only slightly better (sometimes worse) and not very compelling in my opinion. I typically find that LR / ACR provide results that are good enough (indistinguishable from other options in 40×60″ print sizes I use most). Second, the Adobe tools are generally much simpler to learn and use than other options (which sometimes have dozens of confusing sliders). Third, it is typically much faster to adjust a few sliders when you’re already working in LR or ACR. And fourth, there is some convenience in being able to adjust the settings in a RAW Smart Object (though you can apply many 3rd party filters to a Smart Object as well).
Lightroom / ACR offer several tools that can affect and control noise in the “details” tab. These tools fall into a few bucks including sharpening (the first 4 sliders), luminance noise reduction (the next 3 sliders), and color noise reduction (the last 3 sliders). While this tutorial briefly covers sharpening (because it affects noise), you should definitely check out my tutorial on deconvolution sharpening to learn how to make the most of sharpening. Also, while this tutorial is demonstrated using ACR (Adobe Camera RAW) in Photoshop, the sliders work exactly the same in Lightroom.
Sharpening is very important to set correctly, as sharpening adds noise and therefore has a strong impact on noise reduction. (You may <alt/option>-click any of these sliders while sliding for an enhanced grayscale visualization).
Amount: Controls the overall amount of sharpening, per the next three sliders.
Radius: Controls the size of the sharpening effect.
Detail: Controls the sharpening algorithm used (unsharp mask when set to 0, deconvolution when set to 100, and a blend of the two in between).
Masking: Creates an invisible mask that limits sharpening to areas of detail when set to a value greater than 0. This is intended to help avoid sharpening noise, but tends to create strange artifacts/transitions. This is usually best left at 0.
- Recommended workflow: Set radius to its minimum (0.5) and detail to its maximum (100) for deconvolution sharpening. Set masking to its minimum (0), as this slider tends to produce artifacts when used. Then adjust amount to whatever final value looks best (when viewed at 100% or closer).
Luminance Noise Reduction
These sliders are the critical tools for noise reduction and where you should pay the most attention. (You may <alt/option>-click any of these sliders while sliding to visualize in black and white).
Luminance: Controls the overall amount of luminance noise reduction, per the next two sliders.
Luminance Detail: This is like “masking” for sharpness. It controls the pixels that should NOT get noise reduction. Slide to the left to get maximum reduction, and slide to the right to preserve fine details (such as secondary stars, or the edges of the brightest stars).
Luminance Contrast: This helps restore contrast lost to noise reduction, such as the softer gas clouds in the night sky. Try increas
- Recommended workflow: Adjust sharpening first per the above workflow (or temporarily set to zero if the image is extremely noisy). Then, set luminance temporarily to a high value so that you may more easily visualize while tweaking detail and then contrast (in that order). Once you’ve optimized detail/contrast, adjust luminance to whatever final value looks best (when viewed at 100% or closer).
Color Noise Reduction
These sliders can be important in certain niche scenarios, but are generally fine at defaults. If you want to keep things simple, you can generally ignore these.
Color: This controls the overall amount of color noise reduction, per the next two sliders. Most cameras have a “Bayer filter” to capture color and need some color noise reduction, so decreasing below the default 25% is generally a bad idea. Increasing towards 50 may be helpful in some high ISO images, though very high amounts tend to remove too much color at edges.
Color Detail: This is also like “masking”. It controls the pixels that should NOT get color noise reduction. Slide to the left to get maximum reduction, and slide to the right to preserve color at edges. The default 50 is generally very good. Try sliding to lower values if you want to eliminate color on edges (such as around stars). You should avoid high values, as this is prone to showing color noise (anything over 70 is typically a risk). So 25-50 is generally a good range.
Color Smoothness: This helps smooth color over larger areas of the image. Very low numbers can improve color glow around small objects, but can also result in local blotchiness. Larger numbers can create more uniform color, but may also dull finer details. The default 50 is generally great, but you might want to experiment.
- Recommended workflow: Adjust sharpening and luminance noise reduction first per the above workflows. Then, set color temporarily to a high value so that you may more easily visualize while tweaking detail and then smoothness (in that order). Once you’ve optimized detail/smoothness, adjust color to whatever final value looks best (when viewed at 100% or closer).
How to further improve noise reduction
While this tutorial covers global noise reduction in a single RAW, there are some tricks you can use to push the results even further for high ISO shots of the night sky, including:
You can use the workflow shown above with either of these techniques to help get the best overall results.