For most people who commercially operate drones the new DJI X5 is a very exciting product. Not only is it the first time DJI have produced a camera with interchangeable lenses, but it has an integrated 3-axis gimbal in a very compact and lightweight package.
I desperately want this camera to perform. It is clearly targeted at the huge Panasonic GH4 community that DJI service with their Zenmuse gimbal. It has the same sized micro four thirds sensor as the GH4, uses the same MFT lens mount system and can also shoot 4K, but it has been designed to work with the compact Inspire 1 and the OSMO.
In the box you get the Zenmuse X5 Camera, the new DJI 15mm F1.7 lens, a lens hood, a weighted lens ring to help balance, a 16gb Micro SD card, a carry case and some paperwork.
The reason we are so excited about this camera is it will allow us to use the much more convenient, efficient and portable Inspire base rather than a full blown S900 or larger setup. This means everything it one case or two, rather than a boot full of kit. It’s other key selling point is you can control everything from the DJI Go App. This includes ISO, Shutter, Aperture and even Focus. In fact every setting on the camera is changeable in the air, it even supports autofocus.
To fit it to the DJI Inspire you need to purchase the separate mounting kit, which includes a new Vibration Absorbing Mount and some high heels for the craft to give a bit more ground clearance. It is a 10 minute job to fit, but you will need to find your own solution (and scalpel) to get it back in the Inspire case. Check out the above video by DJI on how to mount the X5 onto your Inspire 1
With the standard DJI 15mm lens the whole things weighs in at a shade over 500g, which is just under the body only weight of the GH4 (excluding battery, lens and gimbal). It has a 4/3 CMOS sensor with 16M effective pixels and a max resolution of 4608 x 3456. The ISO range is 100 – 25,600, the electronic shutter supports 8s to 1/8000s and it takes MicroSD cards up to 64GB. We are predominately looking at the video capability of the X5 in this review, but it also supports a number of photo functions including burst shooting, auto bracketing and time-lapse in both JPEG and DNG. For video it supports 1080 up to 60p, UHD up to 30p and full 4k up to 25p. All of which it can wrapper in a .mp4 or .mov format at up to 60mbps. There is an ever growing list of supported lenses, from DJI’s new 15mm F1.7 right up to the Olympus 45mm F1.8. Don’t forget we are talking about MFT lenses so you can essentially double the focal length to get the full frame equivalent. That means the 45mm is just over 90mm on a full frame camera – pretty serious for a drone.
We had limited time with the camera as it is in high demand to be demoed by some really interesting companies. So the focus for the review sits in two parts. Firstly how well does it fly and secondly how does it stack up against the helicam go to GH4 as well as the (GoPro sensor based) DJI X3 and the Sony A7S.
Unfortunately our flying time was completely sabotaged by the British weather and we only managed to squeeze in a couple of flights before the rain came. To be honest though it allowed us to quickly ascertain how the craft and camera performed in the air. The Inspire managed the extra weight well and flight times were only slightly reduced. We would probably adjust a few flight settings as the additional weight made the craft a bit sluggish at times. The gimbal performed very well and keeps DJI’s excellent gimbal track record intact. As a camera operator it was a delight to be able to change all the camera settings whilst in the air. For us the most exciting test at this stage was how well the 45mm lens would work, we managed to squeeze in a very quick test, but didn’t have time to balance the gimbal properly which lead to a bit of jello in some of the footage. Even with a bodged test we did get some results and love how the extended length produces big background shifts. The movement of the gimbal is limited like the X3 due to all the cabling going back to the craft, but being mindful of its limitation meant it never caused any issues.
Having escaped the rain I headed to Bournemouth University, with the talented Clemens Majunke, to run a workshop for the MA Cinematographers on basic benchmarking for cameras. With a test chart, light meter, a single HMI and some objet d’art we had a basic test environment.
Our first test was to figure out a native ISO for the DJI X5 as we couldn’t find any in the manufacture specs. As we moved up the ISOs from 100 to 800 we saw very slight improvement in dynamic range, but strobing artefacts in the blacks and heavy fixed pattern noise made any other ISO higher than 400 unusable with the current firmware version and we got the best results at 100 ISO in D-LOG picture profile. We also found that the ISO rating seemed a little off and should be slightly higher. We compared 4k over 1080p for an HD output, in essence we wanted to see if it was better to shoot 4K for a 1080 deliverable and downscale in post rather than shooting 1080 on the camera and letting the camera handle the downscaling. The 1080 in camera output performed best and we found less compression artefacts and no significant disadvantages in overall sharpness in the 1080 footage. This is most likely due to the limited 60mbps bitrate, it might just be too slim for 4k in this camera.
Next we tried to rate the DJI X3 to its best ISO and settled on 200 in D-LOG picture profile with a slightly smaller dynamic range but more detail in the blacks. We repeated the 4K vs 1080 in an HD environment test and found 4K was cleaner than 1080. Once again the ISO rating seemed wrong and should be lower.
Although we had tested the GH4 before we quickly re-ran the tests. We set the GH4 to 400 ISO and Cine D picture profile, slightly less dynamic range than native 800 but much cleaner. We also re-tested the 1080 vs the 4K in an HD environment and found the 4K to perform best.
With the A7S there wasn’t really much choice as it doesn’t do 4K internally and the lowest ISO in Slog-2 is 3200.
For the exposure tests, we set the cameras to their best ISOs and resolutions we found out in the previous test and over and underexposed each camera.
Without going into all the detail of our analysis we found the best performers for Colour Accuracy as follows:
With our limited test environment, we found it difficult to precisely measure the dynamic range of the X5 but found it very similar to GH4, with more reserve in the highlights at the tested lower ISO. The GH4 latitude shifter more towards the blacks, because of the higher ISO used in the test. The A7S was difficult to compare because the lighting setup wasn’t perfect for high ISO (lens stopped down and net pattern in the background), none the less it demonstrated the best dynamic range, although most of it is in the shadows at 3200 ISO. We possibly should have tried lower ISO’s and a different picture profile to see how much more we could have achieved in the highlights. The results are:
- 2.X5 (very close)
The A7s was tricky to expose and to grade and we found best results between 1 and 2 stops over. It demonstrates the most accurate colours and best dynamic range, especially in the shadows. Due to the higher ISO and less in camera noise reduction we of course found more noise with the A7S. When shooting with the A7S and Slog-2 we typically over expose our shots and stop down.
A quick rolling shutter test followed were the X5 performed well, although rolling shutter was visible it wasn’t too ugly.
We also ran a few quick tests on the new DJI 15mm F1.7 lens and found performed ok in the higher apertures (5.6 and above). As it opened up, things began to soften quite quickly. We will probably be sticking with our Olympus glass for the time being.
Overall the A7S performed best followed by the X5 and GH4, although both look a bit over sharpened and the X5 has major noise issues at higher ISO. The GH4 has a warm colour shift and the most accurate ISO rating. The X5 seems to have a more accurate colour processing and as expected is optimised for exterior work, the manufacturer ISO ratings are too low by ~1stop.The X3 performed surprisingly well in difficult light situations. Although it had the worst sharpness, worst colour accuracy and DR we were really surprised how well it performed.
So there we have it. It is not a GH4 killer but a very worthy adversary. It allows the Inspire to become a more serious aerial platform and the interchangeable lenses and complete remote operation are a couple of very serious features. The noise strobing at the higher ISOs needs to be fixed and it will be interesting to see if DJI can increase the 4K bit rate, I imagine that the hardware may be the limiting factor here. Our Inspires will be getting the upgrade at some point and as the firmware evolves we might find the GH4 staying at home a little more.