After reading a couple of threads about peoples cameras being soft and having just had a couple of small crashes and now being worried about the Phantom 3 camera I decided to do a controlled comparison test using a Canon DSLR and my Phantom 3. I was also interested to see if hovering in flight made a difference to the image quality, likewise with shutteer speed.

Three photos were taken in each setup using both RAW and jpg, I chose the best from each set of three RAW where noticable (no real difference tbh) and used Photoshop CS6 to process the RAW images and then tried to match them all in terms of exposure and colour


On my Canon 600D SLR I used a Sigma 10-20 f3.5 lens set at about 12mm (to match as close as possible to the specs of the Phantom), the Canon was on a tripod.
Photos with the Phantom 3 were with it resting on a box with motors off, then hovering in flight at about the same height using both a slowish and then slightly faster shutter speed. The Phantom had a ND32 filter on it.

Settings
Canon : ISO 100, f/22, shutter 1/40th second.
Phantom on a box with motors off : ISO 100, f/2.8, shutter 1/40th second
Phantom hovering, slow shutter : ISO 100, f/2.8, shutter 1/40th second
Phantom hovering, faster shutter : ISO 200, f/2.8, shutter 1/80th second


Images show the four photos side by side, then an extreme crop of both the fence that was about 4m away then of a tree that was about 30m away







here are the originals, in no particular order, spot the diffference





In conclusion, though I could see a difference between the DSLR and the Phantom I am not so sure that there is that much of a difference worth getting excited about
I can post dropbox versions if anyone wants to see the full sized originals.

To me, the top right image looks the best in the four image sets. I'm not sure the order of the images, so not sure which camera/settings that is. How did you focus the 600D, manually using zoomed in live view, or autofocus? Also you used an aperture of f22 on the 600D, at this aperture you are running into diffraction which will effecting sharpness, you should try around f5.6 or f8.

For the four full size image you post, number three looks best to me (600D), even though sharpness looks pretty much the same, the colors look a lot better.

Another good test would in low light using some higher ISO's. In theory the 600D with it's bigger sensor (assuming you used the same f2.8 aperture as the the Phantom) should pull ahead.


Interested to try this myself using a full frame sensor and see if there is much difference. Would also love to see some side by side shots like this of an X5 vs the Phantom 3 camera.

Also, jsut nitpicking, but the title should be Camera Sharpness/Resolution Test (both the sensor and lens effect sharpness)

The top right of the four is the Phantom on a box with the motors off - ISO 100, f/2.8, shutter 1/40th second
The 600D used autofocus then the photo was taken using a 2 second timer to avout hand shake. I went with f22 in order to get the largest DOF to match the Phantom.

The camera and drone were next to each other to try to match conditions.

I would expect to see some differences between a drone and a camera/lens combo, but I was surprised to see so little difference.

Of course, perhaps my Canon and Sigma lens are just soft to start with, LOL

Filters above ND16 tend to introduce noise and "muddy" footage, especially the inexpensive ones. I use SRP filters for my P3P. As the SRP company says : an SRP ND16/CP has the same light transmission as an ND32, with the added benefit of reducing unwanted reflections or glare. The ND/CP combo filters are as if you stacked a Circular Polarizer onto the ND, which adds nearly a full stop. So an ND8/CP and a standard ND16 would have similar light transmission or stop, and the ND16/CP would be very close to an ND32.

i got these - http://taco-rc.com/taco-rc-nd-fi ... ance-free-shipping/

SydneyMark, was the reason for the slowish shutter speeds to take the effectiveness of the gimbals stabilization into account? Also would you mind if I post some comparison shots of my own?

i think the reason it ended up with a slow shutter speed was that I had the Phantom set to 100 ISO, so that kind of set the initial parameters, then I matched the DSLR to that, then once I started taking photos then i thought it perhaps a bit low and it might end up a bit blurry so also took some at 1/80th second.

I would be more than happy for comparison photos to be posted here, I am sure my approach can be improved on.

SydneyMark Posted at 2016-1-18 13:53
i think the reason it ended up with a slow shutter speed was that I had the Phantom set to 100 ISO,  ...

Makes sense.

Couldn't test flying at the moment as it's been raining all day, so all these are taken with the camera on a tripod, and the DJI sitting on a solid box, roughly the same height. I was slightly off on the framing, the DSLR came to 22mm vs the 20mm of the DJI. I was going to use my 24-70 f2.8 so I could use the same aperture as the Phantom, but figured similar framing was more important. Used live view on the DSLR to manually focus on the mesh screen in the garage. Due to the large sensor of the DSLR, Depth of field will be smaller, so objects closer and further away then the mesh will more out of focus then the DJI's small sensor.

Further note, this is about as unfair a test as possible for the DJI, comparing and much larger, heavier and more expensive lens/camera combo. It would require something like the DJI S1000 to even get this thing off the ground. Secondly, the cropped images are extreme pixel peeping, there is almost no example I can think of that you would actually view and image that close up.

DSLR & Lens: Canon 1Dx with Canon 16-35 F4  (Shooting RAW)
DJI Phantom 3 Advanced (shooting RAW)

Raw images where imported into Lightroom CC, the DJI images where cropped to 3:2 aspect ration to match the DSLR (loosing some vertical resolution, but keeping the full width resolution). Images where then exported from Lightroom to 100 quality JPEGS, at resolution of 2048 pixels across and uploaded to flickr.


ISO100 images:

DJI:


1Dx:


DJI Crop:


1Dx Crop:




IMO, the DJI holds up pretty well when viewing images at normal web sizes, it does get a bit muddy in highly detail areas like the leafs of tress. I'm guessing it's largely down to the lens, as there isn't a huge difference in MP of the sensors (12.4 vs 18MP). One area I haven't covered here is Dynamic range, and ability to pull shadows, although I was very impressed the DJI's RAW files abilty to recover highlight and shadow detail, I dought it would hold up the 1Dx's much better sensor, and frankly, I wouldn't expect an $700 NZ (cost of Phantom 3 gimbal/camera module here in NZ) to compare to an $10000 NZ DSLR and lens. Looking at the crops, the difference becomes a lot more apparent.

I've also done an ISO 1600 comparison that I'll post shortly.

Same process as before, using ISO1600. The DJI is at an even bigger disadvantage here, as one of the biggest advantages of larger sensors, is less noise, particaly at high ISO's.
DJI 1600 ISO


1Dx 1600 ISO


DJI 1600 ISO Crop


1Dx 1600 ISO CROP




At 1600 ISO, noise is starting to become apparent, but it isn't terrible. At web sizes the image is still perfectly usable, but large prints are certainly going to suffer. Personly I'm pretty impressed with the results from such a small sensor. The cropped images speak for themself....  Now I just need to get my hands on an X5

I don't see any links... isit my tablet not showing them ? (Android)