Mavic Air, Reviewing the camera
1867 17 2018-3-11
Uploading and Loding Picture ...(0/1)
o(^-^)o
Viridis
Second Officer
Flight distance : 2383556 ft
  • >>>
United Kingdom
Offline

Now i've reviewed many things in my time, mostly mobile phones and other tech that ive purchased over the years, but one thing ive never actually gotten around to reviewing is a drone. I can't actually think why i havent yet if honest, as most reviews on phones and the like always end up focusing on the camera. It's part of us now as a species really isnt it, the ability to capture a moment digitally to share, to print, to stick on the wall at home etc etc. It's now almost the case, that each one of us has that extension in our pockets, ready to capture a memory, in seconds.
As mobile technology accelerates beyond what we consider ''normaility'' each manufacturer throws innovation after innovation into the camera field. We now are at a time when mobile sensors are really, really getting interesting. Just this year we've seen multiple apetures on smartphones, alongside previous years rediculous fps slow motion, PDAF, laser focus, RGB colour sensors and many more.
The really interesting part though, is all these sensors are just 1/2.3'' small... they are tiny.
The push for bigger sensors in aerial platforms is more than justified, as we take our photography to the skies, the ability to lose noise and get greater low light clarity is just one of the reasons larger sensors just work better than smaller ones. But of course, there is a major drawback... size.
We compare the photography of today through social media, it is the interactive gallery of the generation we are, not defined by age, but by what we see, what we capture, what we share. To have that ability in our pockets through the use of smartphones is what has created the interactive world we live, size matters more than we know, and the ability for us to reside in this super portable world is of utmost importance to dji. Not everyone can haul an Inspire case or Phantom backpack around whilst trekking or climbing, and so the super portables were made.
There are three super portable drones that dji have made, the Mavic Pro, the Spark and the new Mavic Air. All three of these craft contain imaging sensors straight out of the smartphone industry and have been tweaked and tuned for us to take the best possible imaging, skywards. As the sensors also move forward in advancement, so do image signal processors, where the limitation of 1080p 60fps was a very genuine limit just a few short years ago, we now have 4k video in devices a quarter the size.
As with any tech, there are always some comparisons that cannot be avoided, its part and parcel of any developing technology, in the case of 1/2.3'' sensors, its noise.
But has the Mavic Air managed to deal with this consistant problem for such small sensors? lets find out.
(You will have to excuse the lack of inclusion of Mavic Pro and Spark in the examples posted, I dont have them and can only use what i have to compare.)

Pattern test
So the idea is simple, force a lower than usual light scenario but not too dark, and then at different iso's shoot a picture of a test pattern. easy hey?
Well, no. because drones are generally fixed focus much further away we have to do a few things to make the experiment even work.
1. download a huge resolution test image then print off onto multiple sheets allowing us to make a much larger target.
2. make sure each craft has roughly the same fov for the shoot
3. realise very early on that doing this whilst flying is definitely not going to work inside, so also find a nice tall stool.

For the picture comparison I will be using Mavic Air, Phantom 4 pro, and an Osmo as technically its the closest i have to the Mavic Air camera spec.
the full size image they are shooting is the following, i will be cropping slightly to the right in order to not be in the ''perfect zone''


ISO 100

Mavic Air


Phantom 4 Pro


Osmo

At iso100, we can already see that the Focus of the Phantom 4 pro is just another level, but thats ok, what we're more concerned with is noise and its nice to see both the P4P and Mavic Air are producing really solid shapes with very little noise. the Osmo however is I think not going to be that great in lower light for this test.

ISO 400


We can see the Mavic is still producing some nice clean shapes and still no sign of massive amounts of noise, a few of the very darker areas are starting to get a bit of grain, but not anything thats unmanageable. lets bump it up to 800.

ISO 800

Mavic air


Phantom 4 pro

We're definitely seeing some grain appear now, at ISO 800 you're going to see this on almost ALL 1/2.3'' sensors, the ISP on Mavic is dealing with lots of this in the solid areas as best it can, but you can see that the contrasting lines are starting to become blurred as more noise disrupts what should be a clear contrast, but his is still very much fixable in post relatively easy. So Now we're seeing the noise we'll pull back the Phantom to compare. overall you can see the Phantom is cleaner but still has the noise appearing too, the extra megapixels and Pixel size are certainly working at the moment for the phantom. Considering the noise on both craft at ISO800 are quite comparably close, the mavic is really holding its own against the 1'' Exmor RS on the P4P.

ISO 1600

Mavic Air


Phantom 4 Pro

..and just because we can..

Osmo

Well, aside from the Osmo falling out its own backside at ISO 1600, the Phantom is really showing what the extra Megapixels are capable of, they are keeping the noise under control quite well, and keeping the image as sharp as possible. However, I am taking into account the focus distance here and will point out that even though the mavic is now showing a lot of noise at ISO1600, youll notice the bands in the middle horizontal section 9 are still defined enough that the noise can be corrected in post, with not too much overall image quality loss, a quick luminance noise reduction, clarity increase and sharpness can pretty much clean up the ISO1600


ISO 3200


Mavic Air


Phantom 4 Pro

Pretty much both craft are now struggling with the noise, The Phantom is now also losing clarity in contrast, and is gaining some colour artifacts in the left upright pattern, the pattern is still defined in most places but you can see there is an obvious quality drop as we shot beyond iso1600. the Mavic held out quite well but i would do everything possible not to shoot in iso 3200 as the quality is very much comprimised, it can be of course cleaned up, but at the cost of overall quality. If i must be faced in a situation where iso 3200 was a necessity, i would probably lower resolution in post after de-noise to try and bin the odd bad pixels.

All in all, the Mavic did in fact hold up quite well against the Phantom 4 Pro, the control of noise in photographic scenarios is really quite remarkable, only really struggling as the iso levels hit quad digits. The complete opposite was seen with the Osmo, which just couldnt compete with the level of clarity that the fliers could produce, this in itself is interesting as
1, regardless of Osmo noise in low light, it wasnt really that low light and..
2. the Osmo was the closer specced device to the mavic, I was expecting similar results.
I would consider these results as a win for Mavic

4K Video Test
So we know that Mavic can record 4k at 100mbps, same as the Phantom. Excellent, perfect, direct comparison is possible. Well, it would be if that burning ball of gas in the sky would come out from behind those crappy clouds that seem to be hanging around like a wet fart, for a literal view of UK weather at the moment you can go to google images and just type ''England weather'' or ''British weather'' it's uncannily close.
Anyway, today i got a semi-cloudy day so went off with the drones ready to go and obviously when i arrived, the clouds came back.. brilliant.
Actually, that worked quite well as i can show the video quality in murkier than normal conditions.

Whilst the Phantom showed more clarity and sharpness, I honestly dont think the difference was relative to the price difference of these craft, the Mavic was capable of picking up subtle differences the phantom actually didnt, adding clarity to the blades as they transitioned from the ground background, to the sky. The Mavic also, believe it or not, was more stable whilst recording than the Phantom, the zoomed in crops show definite wobble in places from the Phantom, whilst the Mavic was solidly in place.
On the whole, i think that the Mavic Air, and even the Mavic Range as a whole, represents a real opportunity for ultra portable drones to become as mainstay as such travel companions as selfie sticks or even in some regards, phone cameras themselves. As more and more people travel around our non-flat earth, they take with them as little as possible, to do as much as they can. Mavic air represents just that, a tool so small that it can capture your memories, and let you put them in your pocket.
Thanks for taking the time to read this quite legnthy post, as you can tell i havent actually got to use the Air much, so will reserve a post to add to this if the sun ever decides to come out when im not working.


MAVISO100.jpg
2018-3-11
Use props
Viridis
Second Officer
Flight distance : 2383556 ft
  • >>>
United Kingdom
Offline

reserved   
2018-3-11
Use props
ggeorgis
lvl.4
Greece
Offline

Hi, Thanks for taking the time to do the comparison....did you use any correction on the footage in order to compensate for the lens distortions in the MA?
Im mostly referring to this kind of warping as shown in the following video. I did not notice any of these distortions in your video when panning horizontally....



See posts https://forum.dji.com/forum.php?mod=viewthread&tid=133180
2018-3-12
Use props
Viridis
Second Officer
Flight distance : 2383556 ft
  • >>>
United Kingdom
Offline

Hi,
The footage in the video is direct from the cameras.
The only modification was the zoom and crop in the side by side and close up comparisons
2018-3-12
Use props
ggeorgis
lvl.4
Greece
Offline

Viridis Posted at 2018-3-12 04:33
Hi,
The footage in the video is direct from the cameras.
The only modification was the zoom and crop in the side by side and close up comparisons

Thanks .... seems that not all MA are created the same then.....your footage is impressive...
Although to my eye the P4P is more contrasty the only place that I could see obvious differences is at the building roof...Not bad when considering the price differential.
2018-3-12
Use props
Viridis
Second Officer
Flight distance : 2383556 ft
  • >>>
United Kingdom
Offline

Yeah, on the whole it's very close, but yeah the finer details such as the barn roof definitely had more clarity on the phantom.
Does it really equate to $500 difference though? In my opinion no, the Mavic certainly can perform above its price point.
2018-3-12
Use props
Wachtberger
Captain
Flight distance : 261509 ft
Germany
Offline

Thank you very much for this excellent comparison!
2018-3-12
Use props
ggeorgis
lvl.4
Greece
Offline

Viridis Posted at 2018-3-12 04:48
Yeah, on the whole it's very close, but yeah the finer details such as the barn roof definitely had more clarity on the phantom.
Does it really equate to $500 difference though? In my opinion no, the Mavic certainly can perform above its price point.

The pictures during the ISO test were JPG or RAW?
2018-3-12
Use props
Viridis
Second Officer
Flight distance : 2383556 ft
  • >>>
United Kingdom
Offline

Jpeg, I did consider shooting raw but wanted to show "direct" images.
I will be following up with different shots in raw.
2018-3-12
Use props
A CW
Captain
Flight distance : 13926112 ft
  • >>>
United Kingdom
Offline

Thats a serious comparison and with excellent information - thanks very much
2018-3-12
Use props
hallmark007
Captain
Flight distance : 9827923 ft
  • >>>
Ireland
Offline


Great review thank you for taking the time, look forward to your next one.
2018-3-12
Use props
ImTravis
lvl.3
Flight distance : 109734 ft
United States
Offline

Thanks for the good information. Was hoping to read someone comparing cameras.
2018-3-12
Use props
ggeorgis
lvl.4
Greece
Offline

Viridis Posted at 2018-3-12 05:16
Jpeg, I did consider shooting raw but wanted to show "direct" images.
I will be following up with different shots in raw.

Amen to that....if you could also post the raws to dropbox/gdrive...that would be super...
2018-3-12
Use props
Mullheliflier
First Officer
Flight distance : 12305135 ft
  • >>>
United Kingdom
Offline

Thank you for the comparison. I discounted the Mavic Air originally, but maybe you have me thinking about one....
2018-3-12
Use props
Viridis
Second Officer
Flight distance : 2383556 ft
  • >>>
United Kingdom
Offline

If anything it shows just how advanced Dji optics have progressed over a few short years, I was genuinely surprised how comparable the Air is to its bigger cousin, both for images but even more so for video.

2018-3-12
Use props
Montfrooij
Captain
Flight distance : 2560453 ft
  • >>>
Netherlands
Offline

Thanks a lot.
It does make me wonder where the distortion on the MA footage went with your example.
2018-3-12
Use props
Viridis
Second Officer
Flight distance : 2383556 ft
  • >>>
United Kingdom
Offline

Probably would be more visible if I had horizon higher, on the photos themselves, it was mentioned that they were cropped to the right so are not showing the full frame.
I do plan on adding to this thread with brickwork patterns and will happily show full frame,I do expect to see some pincushion distortion.
2018-3-12
Use props
Viridis
Second Officer
Flight distance : 2383556 ft
  • >>>
United Kingdom
Offline

I went back to my full set of footage captured and have indeed found a sample that contains the distortion when turning.

It definitely seems to be visible more, if turning fast, in this video I was lining up in position above the tree to get the same location as the phantom shot and once there, turned quickly to face the turbines.
The distortion is visible in this situation, in fact if you watch the original video posted earlier at 2x speed its also visible there.
However, with slow steady turns it does not seem to be an issue at least in this scenario.
2018-3-12
Use props
Advanced
You need to log in before you can reply Login | Register now

Credit Rules