Watch the above video for a live footage guide of how to set up your drone for stills photography and see a variety of examples for its use—including low light long exposures, HDR bracketing and panoramas.
Choosing a drone
Consumer drones are ever increasing in popularity, and they’re a brilliant filmmaking tool for obtaining dramatic and powerful shots that enhance video content—what about their photographic capabilities though?
For consumer drones, 1/2.3” sensors have been the norm, usually coupled with a resolution of 12 megapixels. Whilst this sensor and resolution combination is suitable for video, it leaves a lot to be desired for stills photography. Typically, users have had to look at bigger and more expensive drone models in order to graduate to a larger sensor yielding better quality.
Thankfully, we’re reaching a point where that technology is trickling down into the consumer and prosumer drone models—DJI’s Mavic 2 Pro (store page), which I’ve been using for this article and video, has a 1” sensor co-engineered by Hasselblad with a resolution of 20 megapixels, making it far more suitable for stills photography. The lens itself is also a Hasselblad model and gives an equivalent 28mm field of view, which is suitably wide angle without veering too far into ultra-wide territory (with the 1” sensor crop factor, it makes the lens’s actual focal length around 10mm).
When deciding on a drone model, however, you should be aware of the concessions you may have to make. Larger pixel resolutions have some implications for video shooters. A 12-megapixel resolution is very close to the resolution of 4K video—for example, with a 3:2 aspect ratio, the pixel resolution of the DJI Spark drone is 3968x2976. The resolution of 4K video is 3840x2160. This means the camera only has to read almost the full width of the sensor, resulting in very little cropping.
However, with a 20-megapixel sensor, we’re looking at a 3:2 pixel resolution of 5472x3648. The camera then has to reach a 3840x2160 resolution for 4K video.
Options for reaching this resolution include:
- ‘Windowing’ the image sensor—this takes a 3840x2160 crop of the 5472x3648 resolution.
- ‘Line skipping’, which discards lines of resolution to reach the desired resolution and often results in artefacting and degraded picture quality.
- ‘Pixel binning’—this combines sensor values pre-demosaicing, generally regarded as better than line skipping.
- ‘Subsampling’, which involves resampling the image signal. Results vary based on the method and implementation.
Windowing (taking a 3840x2160 crop of the sensor) means the resulting video doesn’t make use of the full field of view that the lens offers. Line skipping, pixel binning or subsampling the sensor output does provide a full field of view, but the results can often be sub-optimal, with softness, moiré and false colour being common issues.
DJI’s Mavic 2 Pro offers two different 4K video modes: HQ and FOV. HQ mode takes a window of the sensor and therefore has a crop factor, whereas FOV records the full field of view the lens offers at the expense of some finer detail.
All of this has no implication for stills photography, however—if photography is your primary concern for a drone, then definitely look at models with larger sensors and greater pixel resolutions. 1” sensors are a sweet spot for good image quality whilst keeping the weight and bulk of the drone to a minimum.
If you are more concerned over video flexibility and your stills photography needs are secondary, however, you may want to look at drones that offer a 12-megapixel resolution since they could better fit your needs.
Drone stills photography
The majority of the actual photography is done through the drone’s smartphone/tablet app, since it allows you to customise the settings—a crucial step for getting the most out of both your photo and video sessions.
The app typically lets you choose between shooting modes as you would a traditional camera—so you have full auto, aperture priority, shutter priority and manual modes.
Flying a drone in the air opens up the opportunity to completely change your perspective of a scene. The best example of this was when I took the drone to the north of Norfolk to photograph the coast. Ordinarily, Norfolk is very ‘flat’ to photograph, and I often struggle to become inspired by what is on offer compositionally.
Flying a drone allowed me to observe the textures and patterns of the coastline above, and suddenly my view of the Norfolk coast was changed—crashing waves, sandy dunes and marshland all formed interesting compositions and allowed me to capture something unique and very different to my usual photography.
Another benefit of piloting the drone was the ability to reach areas that otherwise may have been inaccessible—there’s a shipwreck off the coast whose accessibility is highly dependent on the tide, and it’s very easy to get cut off and be stranded by the tide that loops inland. The drone allowed me to fly out and capture some shots of the shipwreck as the sun was setting without having to risk calling the coastguard!
Flying your drone up and pointing its camera straight down can reveal some great abstract compositions. Normally, I wouldn’t look twice at the marshlands, but their small streams and ruts provide some inspiring compositions.
Sometimes you don’t even have to fly the drone at a high altitude—being able to shoot from a slightly higher perspective allowed me to catch the reflection of the sunset in the water, something that would have been lost if I was shooting from my height.
What better way to represent the sweeping vistas an aerial drone can capture than to stitch some panoramas?
You can make use of automated panoramic shooting modes which will automatically position the drone and take multiple shots. You can even shoot spherical (360x180) images.
However, as shown in the video, I prefer to shoot panoramas manually, and I’ll often shoot two or more ‘rows’ to increase the vertical resolution. This is easy to achieve too—simply nudge the drone left or right whilst taking shots, ensuring there is 20 to 30% overlap between each image. The resulting JPEG or RAW files can then be stitched in post-production.
If you’re shooting with automated panorama modes, be sure to double check the image format—it may change back to JPEGs only, and you may want to shoot RAW files too. You may also want to check options such as ‘Save original panorama’, as by default only the stitched file might be written to storage, whereas ideally you would want to keep the unstitched images too.
Shooting and flight modes
As detailed in the video, I tend to stick to aperture priority mode for convenience. The camera will automatically determine the shutter speed and ISO, although I disable Auto ISO and specify the ISO value manually.
Drones also feature different flight modes, such as the Mavic’s positioning, sports and tripod modes. It’s worth enabling and experimenting with them, since tripod mode is one way to achieve sharp results at lower shutter speeds—with suitable wind conditions, it’s possible to get sharp photos all the way down to a one second exposure. This helps immensely in low light situations where the only alternative is raising the ISO and compromising the image quality.
Having a drone that shoots RAW is hugely important for stills photography, since you can take advantage of the full sensor data as opposed to compressed 8-bit JPEGs. All of the images in this article and video were processed from their RAW counterparts enabling me to push the tonal detail further without banding and apply more advanced noise reduction. Whatever you choose to edit with, shooting RAW (or indeed RAW+JPEG) is highly recommended.
The majority of drones will write out RAW files in Adobe’s DNG format, which is fantastic because it’s an open standard and is widely supported by RAW development software.
Whilst drones can autofocus reasonably well, it may comfort you to know that they also have manual focusing options and additional features to support this choice, such as tappable icons to rack between infinity and close focus, and focus peaking to help check that the shot is in focus. Being able to tap to rack the focus to infinity is especially useful in low contrast scenes where the autofocus may struggle.
On the DJI GO app, you can set the focus peaking threshold—I tend to set it to high, which gives a much thinner tolerance for highlighting sharp details and gives a more accurate representation compared to using a normal or low threshold. In low light, a high peaking threshold is sometimes harder to see, in which case I’ll also review the images using the playback feature to double check sharpness.
HDR exposure bracketing and merging
Sometimes the dynamic range of a scene may exceed the capabilities of a camera sensor—drone photography is no exception in this regard. Exposure bracketing is used to capture the full dynamic range of a scene, and can also be used to enhance SNR (signal to noise ratio) by merging pixels at different exposure values for more precision.
Auto exposure bracketing can be done within the drone’s camera app, but may be slightly limited—for example, I could only capture 5 exposures with 1EV spacing, which wasn’t enough to capture the brightest areas of the sunsets I was trying to photograph. To get around this, I switched to manual mode (M) and adjusted the shutter speed to capture a range of exposures which I could then HDR merge in post-production. It’s a good idea to switch the drone to Tripod mode in order to keep it still between shots so that minimal image alignment is required.
Drones have always been an exciting proposition, and technology is moving at such a fast pace that they’re becoming ever more attractive to serious photographers who have demanding technical standards. The combination of larger sensors, high megapixel counts and RAW support (along with control over shooting parameters) means that we’re truly in a great time for drone photography.
I hope the video and article have given you some inspiration to try drone photography for yourself. Don’t forget to check out the other Shooting Series episodes and let me know your thoughts on Twitter @JamesR_Affinity.