I recently purchased my DJI Mavic Air which is a great tool. I don’t consider that at that price tag it is no longer a toy. You cannot compare it to its larger brothers like the DJI Mavic Pro or the DJI Mavic Pro 2. By the way, the Mavic Pro 2 will be presented on 18th of July. But I can tell you that the Mavic air has great potential. We all love panorama images. Especially if you can see them printed. But I am a little bit unhappy with the panorama I got directly from Mavic Air. Even if I edited the images there was still noise. Then I questioned myself: what could be done to improve it? This article will show you how you can take better panorama images with your DJI Drone.
The very first time I flew with my drone it was a little bit disappointing. You may ask why. The reason is simple. As a DSLR user I am used to see sharp images. Analyzing everything I had to admit that I cannot expect from 1/2.3 inch sensor to have the same results as a APS-C sensor on my Nikon D3100. To make yourself an idea what does that mean the APS-C sensor on my Nikon D3100 which is a 14Mpx camera sensor is 23x15mm in size, whereas the DJI Mavic Air’s sensor is mear 6.2×4.6mm in size. You can make an idea based on the chart bellow.
If you do the math you will see that area size wise, the Mavic Air’s sensor is 12 times smaller at almost the same pixel count. This means that the pixel size is smaller on the Mavic Air compared to the Nikon D3100. As a consequence you will see less detail in each image and the Mavic Air will be prone to noise.
DJI Mavic Air’s 180 degree panorama size is 12Mpx in size, which is not really bad. In fact the drone is capturing multiple images and stitching them together automatically. Below you can find an example of the 180 degree panorama the DJI is making. I wouldn’t say that it is bad. But considering the 12 Mpx sensor of the Air I would like to see larger images.
How to take better panorama images
The DJI’s software I would say it is really good and it is aimed to ease the process. It is meant specially for people that want great panorama’s straight from their drone. And while this is great for the vast majority of people, it might not be suitable for people that want detail in their images. At ISO 100 the images turn out really good. Once you increase the sensors sensitivity you are cranking the noise as well. Soon enough, even at ISO 200 you will see noise appearing in your images.
So how to take better images and in the same time reducing noise levels? The answer is: technique. I have already presented this topic in this article, but it was tested only on DSLR. I assumed that this technique could be used easily on the Mavic Air.
What do you need?
Using this technique you will not be able to capture a lot of panorama images. Instead the quality will be higher compared to what the Mavic is producing. You will need the following:
- Adobe Photoshop
- Adobe Lightroom
- A DJI drone, but you can apply this technique to any drone with GPS
Taking the images
This technique is really easy to do and to understand. In short the steps are the following:
Step 1 – set the drone’s image format to RAW, this will allow you multiple possibilities in post processing
Step 2 – set your camera to manual mode, never leave it in auto as you might get in trouble and would have different setting from one shot to the other. I suggest the following setting:
- ISO setting – 100
- shutter speed – adjust corresponding to lighting conditions
Step 3 – put your drone in tripod mode, this will try to maintain your drone as still as possible (please consider that you need enough satellites to use tripod mode)
If you don’t have tripod mode you should try to keep the drone as still as possible.
Step 4 – take multiple images of the same frame by switching to Multiple shots
Step 5 – after you have taken 3,5 or 7 shots for the same frame rotate the drone slightly.
By slightly I mean you should have an overlap between images of roughly 30%. This way the software will have a much easier task of aligning the images. The more the images are overlapped, the more images you will need for the panorama.
Repeat steps 4 and 5 until you will have all the images necessary for your panorama.
Once you will have all the images necessary it is time to download the images in the computer and import them in Lightroom. There you will need to adjust the image. But I will not go into details on how to do that. Personally I think that it is a matter of taste the way you adjust the image.
Very important! All the images need to have the same adjustments made. You can adjust one image and synchronize the settings among the other images as well. Additionally you can set a color label for images of the same frame. You can apply different colors. This way it will be easier to import the files into Photoshop.
Adjustments in Photoshop
After we have downloaded the images in the PC it is advised to launch Lightroom and import the images. Select multiple images shot for 1 section and import them into Photoshop as Layers.
Our images will be loaded into Photoshop as separate layers. Let’s proceed to the next step which is aligning the images.
After your images were loaded on a separate layers, we will need to align the images. This can be easily done. Select all the images and from the Edit menu select Auto-Align layers.
You can leave it on Auto as the software will handle it. Depending on your PC this process can take a while. You will need to be patient. Especially if you are using a quite slow machine like I do.
After your images are aligned we can proceed to the next step which will be converting our images into a Smart Object. From the Layer menu from Smart Objects select Convert to Smart Object.
The next step is where the magic happens.
Once all our layers were converted to a smart object, this is where the magic will happen. From the Layer menu from Smart Object/Stack Mode we will select Mean.
What will this do our Smart Object? It will take the mean value of each pixel. As a result our images will look cleaner, you will see an improvement in the noise and will obtain more details. If you want to find out more about stacking modes I recommend this article.
We should save the resulted image preferably in TIFF format. Although it takes up more space than JPEG, you will have more data available in the image. For the next step we will repeat this process for all sections of the panorama. We will repeat the process from Step 7 till Step 9 until we will have all image sections averaged.
Once we finished stacking the images we will have them ready to be stitched.
Stitching images in Photoshop
You will have by this time all the images available for stitching. We will need to import all the averaged images into a stack. How to do that? You need to select from File/Automate the option Photomerge.
You should browse for your files. As a suggestion I recommend to tick Blend Images Together and Geometric Distortion Correction. The option for Vignette Removal should be used only if your lens is casting a vignette on the photo. The other option of Content Aware Fill Transparent Areas tries to guess the missing pixels and fill the content with something. This is not always working. Hit OK and after a little time we should have the end result.
The below image was is the end result of 10 images (5 images for 2 frames) which was afterwards cropped and adjusted in Adobe Photoshop using Camera Raw Filter. The original image was 13.8 Mpx instead of the original 12 Mpx. But please note that this is not a 180 degrees panorama. For 180 degrees the image size would have been larger.
If you don’t want to use Photoshop for stitching the images I recommend another alternative, Image Composite Editor. This a a free alternative to Adobe Photoshop. The funny thing is that it works even better than Photoshop.
There is only one issue with it. It can available only by Windows OS users. You can download it from the following link.
Stitching a panorama in Image Composite Editor
The software has a really slick design. You will be able to make Panorama images, Panorama from video or to open an existing panorama. Let’s hit New Panorama and select all the images that we have processed. Once the images are loaded in the app you either can create a simple panorama or a structured panorama.
The simple panorama will make more or less everything by its own. If you want a structured panorama then you can set a lot of parameters. If you have a simple panorama with only one row of images it is better to make it simple. Just hit the NEXT button.
The software will automatically align, overlap the images and create the panorama. But before giving us the end result we can adjust any distortion that is in the image. This is where the software is performing better than Photoshop.
You can choose from multiple types of projections and with the help of your mouse’s cursor you can move the image to adjust it in the best way. The final steps are cropping the image and saving it into a file format. I recommend to use the TIFF format. It contains a lot of useful data. This can be helpful in the post processing of the file.
The above image I imported into Photoshop and adjusted the colors to pop a little bit more.
DJI Mavic Air’s panorama feature is really nice. It can easily create stunning panoramas. While it does this in an automatic way there is still more potential to it. You only need to work a little bit more. But I can promise that the end result will be much better. This is a perfect technique especially if you have high noise images.
Another important thing to consider that this technique is applicable for both Spark, Mavic Pro users or any other drone with GPS and camera.
The Mean stacking mode will remove any unwanted noise and still preserve much of the detail.