“Oh no, he’s gone and done it now…”
For a long time, I resisted. For two reasons: firstly, the flight technology hasn’t quite matured to the point that I’m comfortable enough with my own flying abilities to not crash or injure something or destroy the aircraft; secondly, the camera quality really wasn’t worth bothering with – especially when you’re used to something with let’s say, a little larger sensor. The first problem has recently been solved; the current generation of consumer-level drones packs so much guidance technology in (GPS positioning, radar, obstacle avoidance cameras, subject tracking and recognition cameras etc., inertial navigation gyros) that it’s really quite difficult to crash or hit something: it just won’t let you, unless you decide to turn all of the aids off. The second has also been solved to some degree, though not in the Mavic Pro I’ve started flying recently.
As you know, Hasselblad was recently acquired by DJI; two weeks ago, I visited DJI HQ in Shenzhen for a look around, some planning discussions, and of course – a crash course by their best instructors in how to fly. I’ve flown a) R/C aircraft; b) real aircraft; c) crappy toy helicopters – and that’s about it. Nothing made in the last five years, that’s for sure. They brought out three aircraft: a Mavic, a Phantom 4 Pro, and an Inspire 2. The Mavic is the smallest, easiest, most automated model; the ‘entry level’ – but really quite sophisticated in its own right, and superseding the Phantom 3. The Phantom 4 is a larger, faster, heavier model with a camera that has a 1″ sensor (the Mavic is 1/2.3″, similar to what’s found in an iPhone), and the Inspire 2 is the largest of the lot before you go to custom builds – it flies a M4/3 camera on a special gimbal which can be detached and attached to a handheld handle to shoot separately, which shoots 5.2K30p and 4K60p video – in raw. It of course flies the fastest, furthest, and most stable in winds. All of them were equally easy to fly, but if I had my pick – it’d be the Inspire 2, because it has the better camera and the larger size contributes considerably to stability.
Surprise number one: the Mavic is a lot smaller than you’d expect – as you can see from the first image; the top object is of course a loaded H6D-100c with HTS and 24mm. The Mavic has a similar packing volume to a D810 or 5D-series body, with a bit more required for the collapsible controller and spare batteries. The controller is a clever affair that folds up for transport but expands to hold your phone or tablet between its jaws, with data plugged in via the socket of your choice – thereby saving space and cost (no need to provide a screen). The controller itself does the processing, and has its own battery and display – along with the usual control sticks.
Load the app, unfold the props, remove the camera gimbal protector (I keep forgetting to do this), update all firmware, get a green light across the board and you’re ready to go. Takeoff is a remarkably anticlimactic affair – sticks down and to the centre, and the props start; a light touch on whichever one you’ve assigned to altitude and the drone climbs. There are three other important controls: forward/backward, bank left/right (think of this as fly sideways in plane) and pan left/right (yaw, or rudder). It’s important to remember two things when flying a drone: firstly, it’s not dynamically unstable like a helicopter, and will continue to hold position and hover if you let go of the sticks; this is taken care of by GPS and other sensors. It’s important to remember because smooth motion requires that you actually slowly release the sticks rather than follow your natural instinct of letting them snap back to centre and let the inertia of the aircraft dampen motion.
Secondly, the camera doesn’t have a separate pan gimbal; the gimbal moves in that axis but is used to compensate for the motion of the aircraft so the camera itself is a stable platform. You only have pitch control (on a finger wheel) – panning must be accomplished by turning the drone, which is why you still have banking control so you can bank/turn to aim the camera and track, and/or fly backwards/sideways etc. In practice, it’s somewhere halfway between an aircraft and a helicopter. I find I fly better when I think of it as an aircraft; the motion is smoother – which of course translates to better footage.
There are some gotchas, though: the vision sensors can get confused with big panes of glass (i.e. if it can’t ‘see’ edges) – and that can cause issues compounded with ground effects and airflow next to big buildings. I’ve been flying off my balcony a lot which probably isn’t at all ideal for this reason; that, and GPS doesn’t tend to lock until I’ve cleared the building a little. One particularly bad afternoon – reflections off glass, perhaps no GPS signal – and the following carnage ensued as the result of optimism and inexperience:
Yes, I destroyed all of the blades in the same accident. And those things are pretty tough. The Mavic survived, though the damage could have been limited if I’d had some sort of instant motor-kill switch; it was grinding itself into the wall and tile for a good few seconds after initial contact.
To be honest, the lack of images in this post is due to one main reason: I’ve been spending my time learning to master the flight part to the point it’s intuitive before splitting my brain into two looking for things to shoot. My ultimate objective remains shooting stills, in addition to simply enjoying flying – which I think will get much better once the VR goggles are available. The little fiddling I have done with the capture part has made a few things clear, though: it’s works great for video, and footage is pretty good; the camera is nothing to shout about, though. Unfortunately, payload limitations restrict size which in turn affect the largest optics/sensor etc. that can be carried; my Mavic’s camera is a bit soft towards the corners, which is to be expected (you don’t see this so much in video mode). There are also severe limitations on resolution and dynamic range: what were you expecting from a cameraphone sensor?
Those are easily solved by the old ‘more better’ method; the most annoying thing for me is that I can’t rotate the gimbal 90deg to shoot vertical orientation images; all compositions must be landscape, which of course does not happen in still photography! Basically: I’d like full roll control over the gimbal. And this isn’t going to be solved by going to the bigger drone; its gimbal can be operated independently off a second controller, but the camera still cannot turn to portrait orientation.
A few more practical observations to close: it doesn’t work so well indoors, because of air currents bouncing around, because of the anti-collision systems limiting how close you can fly to objects, and because there’s obviously no GPS signal which means no holding position (something that the computer does incredibly well, making a lot of thrust changes to hover even in surprisingly heavy winds). Batteries last about 25min or a little more; I’ve been limiting my flights to 15min to have some return to home reserve in case it has to fly upwind. The drone also has a lot of other features including flying a predetermined path; auto return-to home; object tracking; tripod mode for long exposures. I’ve yet to try these, but it’s in the plan. Stay tuned for future instalments in this series…one day I might even fly the M600 combination carrying the H6D-100… MT
Visit the Teaching Store to up your photographic game – including workshop videos, and the individual Email School of Photography. You can also support the site by purchasing from B&H and Amazon – thanks!
Images and content copyright Ming Thein | mingthein.com 2012 onwards. All rights reserved