One of the most frustrating issues when working with a consumer-grade drone is the amount of noise one can encounter at low ISOs. Due to the small size of the drone’s sensor, even ISO 100 can look rather noisy, especially when recovering shadow detail. In this article, we will take a look at a simple averaging technique to significantly reduce noise in your drone photos.
An image of a rainbow from Faroe Islands with minimal noise after averaging
Captured with the DJI Mavic Pro 2 drone
The technique involves taking a series of shots in burst mode using a drone, then combining the images together in post-processing software like Adobe Photoshop in order to create a single “average” stacked image. Since noise patterns in images are random, stacking photos together this way allows to reduce noise levels significantly, while keeping all the details intact.
Below are some of the requirements you need to be aware of:
- Your drone must have a built-in feature to be able to shoot in bursts (5+ image bursts are recommended). Capturing single images might not work well due to potential drone and subject movements.
- The drone should not be physically moved during the process of capturing a burst of images. Any changes in perspective will make it impossible to properly align and average images.
- Ideally, you should not be capturing fast-moving subjects in burst mode. For example, if you have ocean waves or a moving subject, the final averaged image might not look good.
- You must use a post-processing tool that has a built-in capability to average images. I will be using Adobe Photoshop in this article, but you can use other software to achieve similar results.
Please note that while I recommend using this technique to reduce noise in your drone images, the same averaging technique can be used to reduce noise with any digital camera. We will go over this technique in more detail in a future article.
I will use the following image, which I captured in a burst of 5 images for this tutorial:
This single shot shows quite a bit of noise, especially in the right part of the sky
As you can see, there is quite a bit of noise in the image, especially in the sky. With the averaging technique below, we will be able to clean up the noise quite well. So, let’s get started!
Use Drone’s Burst Mode
The first step is to set up your drone in order to capture image bursts. The idea is to take 5 or more images at once, with a very minimum delay in-between, so that the images can be averaged together later in post-processing. To do this, connect your drone controller, fire up the DJI Go app, then under the Photo settings menu, select “Burst”, or “Multiple”. Some consumer DJI drones like the Mavic Pro 2 allow capturing 3-5 image bursts in DNG format – make sure to set it up to 5 bursts.
If your drone does not have a burst mode capability, see how quickly you can take single images. If the speed of capture is reasonable, you might be able to use those images for averaging, but please keep in mind that any differences in perspective or subject movement will make it difficult to properly average the image.
Load the Burst into Post-Processing Software
The next step is to import your images into post-processing software, then load the whole stack into a single image as layers. If you use Lightroom, import your images first. Next, select the group of images, right-click them and select “Edit In -> Open as Layers in Photoshop…”. This will load all images into a single file in Photoshop, with every image appearing as a layer:
Use Median Stack Mode to Average the Images
In order to average the images, you will need to combine them into a Smart Object in Photoshop. Select all images by clicking on the first image, then holding the Shift key and clicking on the last image. Next, go to the Layer menu -> Smart Objects -> Convert to Smart Object. This will combine all layers into a single Smart Object.
Now that we have a Smart Object, we can now average all the images in order to reduce the noise. You can do this by clicking on the Layer menu -> Smart Objects -> Stack Mode -> Median. The result will look like the following:
This means that all the images in the stack have now been averaged into a single photo, which should show very low levels of noise compared to what you had previously seen on a single photo.
Let’s take a look at the image, before and after at 100% crop:
And here is the final image, after I fully cleaned it up using the above technique and post-processed it in Photoshop:
If you have never tried this method of obtaining high quality images with very little to no noise, give it a try – it certainly works!
Please let me know if you have any questions in the comments section below.
#DronePhotography #Noise #NoiseReduction
