Tutorial : Targeted Noise Reduction

I present to you, my second tutorial. Hot on the heels of my “Tutorial : Fix An Out-Of-Focus Blurry Photo (My 3-Step Method)” I delve into Noise Reduction.

What is “noise” in digital photography?

“Noise” is a type visual distortion in digital photography. It comes in two varieties. Luminance Noise is similar to grain which occurs in film photograph; the image has a rough, sandpaper-like texture. Colour Noise manifests in the forms of blotches or bands of colour, typically luminous pink and lime green

How can I shoot to minimise noise?

Noise typically occurs in images captured in low light. At higher ISO settings, noise is higher. This problem is exacerbated in crop sensor cameras. Shooting at a low ISO and using a tripod is an easy way to avoid noise.

---

My Example : Low light, handheld photos of Hungarian planes

I managed to squeeze in a revist to my favourite field of Mig and Sukhoi aircraft in Hungary. Sadly, the light was failing fast. There quite simply was not enough time to use a tripod. Torrential rain also made me feel the need to move faster. I made the decision to shoot handheld. So I racked up the ISO on my Sony A7ii, and got to work. I darted around the field for perhaps a few tens of minutes, before the light completely disappeared.

I got the shots I wanted. Happy days! I knew that I would have to run some Noise Reduction, I had already accepted that. As I ran around that cold wet field, the idea of an evening sat warm infront of my computer wasn’t an upsetting prospect.

You can see the original image below (well, a close up on a representative part of the original image). Yikes! I think you will agree, the version on the right looks better? This is the image after my de-noising attempts.

Let’s see what I did to denoise this beautiful angry-looking Hungarian airplane.

Step 1 – Colour Noise Reduction

The first step is easy, running some Colour Noise reduction. The image will look a LOT better once the Colour Noise is removed.

A great start. Those distracting pink and purple pixels and patches are gone!

---

Step 2 – Luminance Noise Reduction (global)

Open the image in Photoshop. Duplicate the layer. Rename the lower one Original Layer, and the upper layer Noise Reduction Layer.

You want to adjust so the edges are bright white/glowing, and the rest is black/grey.

---

Step 3 – Creating a mask to exclude edges

Time to make our mask to exclude the noise reduction from the edges. Highlight the Noise Reduction layer, and duplicate it again. Name this new layer Filter Mask, and make sure it is the topmost layer. With this Filter Mask layer selected, go to Filter>Convert for Smart Filters. Double click the Filter Mask, which will open this smart layer in a new tab.

You want to adjust so the edges are bright white/glowing, and the rest is black/grey.

Invert the image via Image>Adjustments>Invert

Convert the filter tab image to greyscale with Mode>Greyscale, then save/close this tab and return to the main image.

---

Step 4 – Luminance Noise Reduction (exclude edges)

Now we need to apply this as a layer mask. Highlight Filter Mask layer, Select All and Copy

Select the Noise Reduction layer, and Add Layer Mask (3rd icon on the toolbar at the bottom of the Layers panel)

If you are happy with the results, you could stop here. The next step addressed areas of higher noise and targets them for a more intense effect. The additional step also allows you to selectively mask and preserve areas of detail.

---

Step 5 – Tweaking the final noise reduction

Adjust the Layer Opacity if the effect is too strong for your taste. I used 80% in my image.

In my example;

• Darker areas below the wing and inside the nose cone required further noise reduction. Therefore, I painted these areas with a white brush on the Layer Mask
• I wanted to preserve detail of the red/yellow paint and the bumblebee decal on the nose cone, as well as various nuts/bolts along the body. Therefore I painted these in black on the Layer Mask.

---

Bonus, if you read this far

Download the layered TIFF file, where you can see how these tweaks look like on a medium-resolution demo image: