Behind the Image: Post Processing for Bird Photography

Published: 10/15/2018

I often get questions on social media about how I post process my avian photography. Typically, folks understand the options provided by Lightroom, Capture NX-D, DPP, etc, regarding white balance, saturation, and contrast, but will get stuck on how sharpening (referred to here as SH) and noise reduction (NR) work together to produce high quality images, especially at higher ISOs. This is something I struggled with (and still do!) for a while as well, but have worked out a system that appears to work for both web and prints.

I've learned pieces of this material through various folks throughout the web, some who have more complete tutorials on certain topics (which I will reference here). 

For this tutorial, we're going to be editing a photo of an Atlantic Puffin I took back in Summer 2018. I actually hadn't seen this photo in my files until I got a recent email about implementing NR and SH, so that was a nice treat. In particular, I chose this file as it was a decent (albeit not crazy high) ISO at 1,250, and there was a range of light and dark colors in the background with can present issues for noise reduction. This image was taken with a Nikon D500 and 300 PF at 1/4000 f/4.5 ISO 1,250. You can read my 300 PF lens review here. I will use Nikon's own Capture NX-D RAW converter, which I find relatively simple to use in order to obtain clean files. If you use Lightroom, you will likely be unable to use these values explicitly, but this guide should still be helpful in terms of setting workflow and expectations.                

So to start, let's look at the final, edited image. I liked the wing position here, and especially the head turn. The face is sharp, and although the wings are out of depth of field, I liked the sense of movement (and boy these birds can move!). For each image shown, click it to view larger.

Okay so that's where we want to go. But where we start is actually the straight out of camera (SOOC) image. This file has no sharpening applied yet, no noise reduction, and indeed no edits.The only thing I've applied here is the standard camera profile in order to get a grasp on Nikon's colors and white balance. There are many guides available in how to set white balance, but I find with Capture NX-D that the auto mode is generally pretty accurate. You can see noise especially in the dark regions in the right side and on the wings. There is also noise on the bird which will be shown further down.       

Now let's apply some edits to the file just to get baseline 'depth' and and saturation. I apply some saturation, and bring the highlights down and shadows up a little, however detail and noise are relatively unaffected.

Now onto the bulk. In general there are two kinds of sharpening: input and output sharpening, and there are two kinds of noise reduction (and any adjustment): global and selective. This short tutorial will mostly focus on input SH and global NR. 

To that end, we first need to specify how much SH to add to the file. Too little, and the image won't show any detail, and will even appear soft, too much and the image will appear crunchy and you will introduce even more noise into the image. Indeed, you can see this in the image below. Each image is a 100% crop (i.e. each pixel shown is a pixel from the original file) centered on the bird's face. Note that the focus point landed on the bird's cheek. For each crop here, NR is set to be zero.

On the left, we have SH = 0, resulting in a soft image. The middle two, at SH = 3 and = 6, you see more detail being brought out, although SH = 6 shows noise and the details are beginning to look crunchy and harsh (note under the beak). On the far right at SH = 9, the file is over sharpened and quite noisy. In general, I like SH = 3. 

Once we've settled on how much to input sharpen, we must now specify a global NR parameter. The aim here is to remove the coarse grain noise without impacting the detail on the bird.

At ISO 1,250 on the D500, some amount of NR will be necessary. At lower values or using a flagship full frame model, perhaps not. Additionally, the amount of NR required will change with the file, any under exposure, and shadow adjustments.

For each image below (again a 100% crop), SH is set to be 3. On the left, we have NR = 1 which is insufficient to remove the large grain present in the file. On the far right, we have NR = 100. At this maximum NR level we see we have eliminated the noise but at the expense of detail! The file now looks too smooth and plastic. In the middle, at NR = 30 and NR = 50, we see minimal loss of detail, and substantial reduction in noise. I like NR = 30 here. In general, look at both the background and in areas on the bird that don't show detail (e.g. the far underwing).

After applying a global adjustment of NR = 30 and SH = 3 on top of the edits we already have, we have the following image. Here you can see an improvement in detail and a reduction in noise when compared to the SOOC image. However, you can still see some noise in the background, again in the dark areas. The noise handling on the bird however has been improved and that's what we're really after here.

To address the noise in the background, some additional NR can be applied to just the background (no longer globally). Arash Hazeghi and Artie Morris have an excellent guide on this step in their Professional Photographers Post Processing guide which can be found on either of their websites (here).

Finally, after some clean up of various specular highlights as well as working to tame the highlights on the sand eels to bring out more detail, we now move onto the final step: output sharpening. How much output sharpening depends on your use. If you're aiming to make a large print, you will likely need more sharpening. For web, you can generally just eyeball it in Photoshop to find values that look good. Again this may depend on the file. After resizing the image, sharpening just the bird yields the final image, shown below.

I hope this was helpful, and feel free to email me or comment with any questions.

(C) Alex Becker, 2018, no reproduction allowed without explicit permission.