DJI Phantom 2 Vision Prop Guards

In my last post I explained how I destroyed several props while testing the Phantom 2 Vision with a heavier camera load and a good wind. Well I finally received my new props as well as a new set of prop guards, these things should be standard issue, and I just finished testing them in the wind. The other day I did a simple hovering test with 23 MPH gusting winds and it handled quite well, however, this type of wind is not conducive to video or photography since the copter was constantly making rapid adjustments to correct for the eddies. Landing is another story in gusty winds, I waited for a lull in the wind and attempted to land but a very strong gust flipped the copter completely over and the props survived the crash perfectly unscathed!

Today there was a relatively mild 15 MPH breeze and I went out and flew it around to test the video and while its still not as stable as it could be it is still better than without the roll axis gimbal, note that most of the video jitter is due to the pitch axis gimbal being so slow. Also note the few seconds of the autopilot trying to gain control in a wind gust when I try to turn it around to fly back to me. Also note the landing gear extensions, scrap PVC pipe with several lightening holes drilled into them, they work like a charm. This entire setup is actually lighter than the GoPro model with the two axis gimbal.

DJI Phantom 2 Vision Thoughts

Now that I’ve had the DJI Phantom 2 Vision for a couple of weeks I thought that I would give my two cents to whoever is interested. First this thing is a blast, its very easy to fly through a battery and takes no time at all to drain it. The Vision line is not marketed as a professional imaging platform and I concur, its a very expensive toy, or at best it is a very good trainer for those thinking of stepping up to a more powerful six or eight motor unit that can cary a much larger payload.

I think that I learned a lot about the performance and capabilities of the Phantom 2 Vision while experimenting with it. I came up with a two axis gimbal version that could have handled my Nikon N1 J1 but with the prototype extended landing gear it was simply too unstable and I wasted five props in the effort. Carbon fiber would have been an ideal material to use for the gear but I found that the extended landing gear had a tendency to topple over much easier, this was despite the fact that I splayed their footprint out to actually be larger than the motors.

The maximum payload is roughly 0.5 to 0.75 lbs. (0.23 – 0.34 kg) especially if there is any wind to speak of. Anything beyond that and the copter can easily become unstable and much of its power is lost to the simple act of stabilizing itself, meaning that there is very little power left over to halt a descent and the resulting crashes.

Bottom line; this thing is fun as all hell to fly in stock form. It takes OK web quality images, videos are a bit shaky and will never rise to a professional quality simply due to a slow pitch servo and the lack of a roll servo. If doing this over again I would have opted for the GoPro version with the two axis gimbal.

As far as eyeballing a much more expensive S800 or S1000 bird, well, I think that I’ll wait a while since there is still a huge controversy over RC drones, in fact just last week a bill was tabled that would have made flying any RC controlled aircraft illegal in Minnesota by anyone except the military and law enforcement. The FAA lost their case with the dude delivering beer to ice-fisherman, but don’t celebrate too long. While it takes a few years for the FAA regulations to flow through their bureaucratic pipeline they still have the emergency rule making powers at their disposal.

If the RC drone flyers out there don’t start self policing themselves then expect to be shut down, i.e. the morons flying their drone that hit a competitive runner. Keep them away from airports, under 400 feet, and for God’s sake AWAY FROM CROWDS! Jeese, these things are little more than high-tech flying cuisinarts.

And when one of  those multi-rotors gets out of control they have NOTHING in common with the rest of aviation, There is very little one can do to regain control once it has been lost, ZERO glide ratio and NO autorotation. And man, that five pounds of very sharp debris falling to the ground can easily MAME or KILL, or at least shred, slice, and dice!

Drone Smashes Into Triathlete During Race Causing Head Injury

DJI Phantom 2 Vision with Roll Axis Gimbal

I found a Futaba coreless servo motor S9402 demo for $39 bucks today and placed it in my updated roll gimbal attached to my DJI Phantom 2 Vision and it works spectacularly! In fact it completely outshines the stock pitch axis gimbal, so, you guessed it, now I’m looking to replace the stock servo with another corless (or brushless BLDC ) servo. I also temporarily solved the landing gear extension by attaching some 3/4″ PVC tubing and elbows. It also has an additional pivot point in the front to stabilize it.

_DSC0846 _DSC0849Here is a video:

As long as I fly it slowly and deliberately (very hard to do since this thing is so much fun to fly) I can keep the landing gear out of the way, also notice that there is about forty degrees travel in the left direction but only twenty degrees in the right due to clearance problems with the servo motor itself. Also note that the stock pitch axis now looks to be quite poor in its performance, you guessed it, now I’m brainstorming how to upgrade it, probably with another servo motor.


DJI Phantom 2 Two-Axis Gimbal Mod

I made a very crude prototype adding a second axis gimbal to the DJI Phantom 2 Vision. Here are some pics, I used an old very cheap servo motor but the proof of concept is in the pudding. I also need to add a second support pivot near the front, currently the entire camera is supported by the servo motor output shaft. A higher quality servo with ball bearings and possibly brushless servo would be ideal.

_DSC0837 _DSC0840And here is a quick video showing the shaky performance, lowering the gain makes it run too slow, also note that the pitch servo is also slow, needs more gain — currently set to fifty but at sixty it overshoots. The roll servo gain is set to thirty. (sorry for the dirty bachelor pad mirror)

There were no modifications to the Phantom 2 itself, other than repositioning the video cable to allow for the lower position of the camera, and also extended landing gear are necessary since the camera hangs below the stock landing gear, I think I saw a pair on the web for about ten bucks. The roll servo was simply plugged into the NAZA-M (V2) Main Controller, plug and play baby! Although if you wish to control the pitch servo gain the NAZA mode must be used, no big deal.

For those of you who wish to know I measured the pulse width of the pitch servo control signal, centered and level was 1.53 ms, tilted right 90 degrees was 2.10 ms, tilted full left was 0.991 ms.

DJI Phantom 2 Vision Capitol Shot

Here is a shot of the State Capitol Building in St. Paul Minnesota. Today there was a bit of wind which makes even taking still shots with the Phantom 2 Vision very challenging without a two axis gyro stabilized gimbal. Even when hovering in a wind the copter has a pronounced tilt into the wind, luckily this is easy to correct but requires some cropping.


It did not turn out all that bad, that is for a web photo. I would never even consider printing this image since there too much noise, at ISO 650, and the lack of resolution. The image also  required a lot of post manipulations in Photoshop including in Nik’s Dfine 2 and  layering  to make adjustments for the differing light sources. I also liked the image better than the version where I used the Adobe lens profile. But Damn! It sure is a hell of a lot of fun to fly! And there is no other way to get a shot like this one.

DJI Phantom 2 with Nikon 1 J1

Here is a video and some pics from the DJI Phantom 2 Vision quadracopter. The main point is that with no gyro stabilized gimbals the video is not very good, the actual quality is much better and there is quite a bit less distortion than the Vision FC200 camera, mainly due to the 140 degree wide angle lens of the Vision camera.

The Phantom 2 flew fairly well, however there was a little shimmying while hovering and climbing, I think the gains may need some fine tuning with the extra load.

Here a re a few pictures using the Nikon 1 J1, note that there is little distortion and that both downtowns are visible, Minneapolis at the far right horizon. They are taken at Battle Creek Park, McKnight and Upper Afton roads intersect near the left of the frame. Also note that there is a world of difference taking a single still image from a video camera stream and a still image from a camera actually designed for still images.


3M’s main world headquarters is the building on the right._DSC0768

Here is an an example video using the Vision camera which has a single pitch axis gimbal, its overlooking St Paul Minnesota from the Mounds Park Bluffs. Note the video is a little more stable but still not at all acceptable.

And here is an RAW image taken with the Vision camera, note the large amount of distortion that would require a huge amount of manipulation in order to correct. The Adobe lens profile file does not seem to be available on the DJI site any longer, looks as if I’ll have to make my own — this should be no problem. Project for the evening, l can post them here if there is any interest, I’ll be doing both the RAW and standard profiles using Adobe’s Lens Profile Creator.


Bottom line at least semiprofessional still images can be made with the DJI Phantom 2 using a higher quality camera, unfortunately the only two axis gyro stabilized gimbal that is available is only compatible with the (also) very wide angle GoPro camera. The other annoying characteristic of the Vision camera is that when capturing a RAW still image it takes about TEN seconds per image, this is a huge lag time PER IMAGE with a complete loss of FPV. Disappointing :-(

******* added 20:00 ********

Here are corrected images using Adobe lens profile correction, this one uses the RAW image but the jpeg version works similarly, although some additional adjustments may be necessary in the customs settings menu after being applied. Norte that there is still a pronounced curvature at the horizon and there is a very pronounced blur at the right side of the frame due to the amount of manipulations necessary, very common with fisheye lenses. Also I had a very difficult time shooting a set of images that Adobe Lens Profile Creator could actually use, I have found that this is actually common when profiling very cheap glass.

DJI00002_raw_corr DJI00002_rawIf you wish to use these here is the link to the zip file: phantom_vision_fc200_lens_profiles

In several forums I have seen posts that recommend changing the parameters in the lens profile file to allow the profile to be applied to RAW files. This is a very bad idea in general since many cameras do some distortion correction in-camera so the profiles for the RAW images are very often very different from the JPEG or TIFF profiles. However this does not seem to be the case for the files created by the Vision FC200 camera, in fact the DNG RAW images produced by the Vision camera are also only 8 bits deep so I’m not exactly sure why they chose to use a RAW format in the first place (marketing?) when a lossless TIFF format would have been entirely adequate.


Nikon 1 J1 on the DJI Phantom 2

Here is a pic with my Nikon 1 J1 mounted, for now there is no gimbal so I’m stuck with one position, but it does use the stock Vision camera vibration isolators, although they may be a bit overloaded. I was not able to flight test it since it is snowing and raining but I did manage to get it off the ground inside my apartment. It took a bit of extra power but it seemed to hover OK and was fairly well balanced. Hopefully the weather will cooperate soon and I will get some photos and videos to post.

_DSC2910I also opened up the body to inspect the innards and found everything to have been well placed and all the boards were of good quality, boards were very clean and coated with glyptal varnish for protection. I was a little disappointed in that the Power Management Unit (PMU) was Phantom specific, I was hoping to maybe have hobbled together a six rotor bird using these innards but it does not look good for now.

Also the ESC’s or BLDC motor controllers use a proprietary micro controller so reverse engineering and hacking the firmware may pose some problems. The BLDC commutation appears to be trapezoidal, a definite plus, but I was hoping to apply a sinusoidal commutation scheme to reduce vibrations, looks like I will have to design completely from scratch for that. The same thing with the GPS unit, it uses an off the shelf GPS module but the micro controller is also proprietary, rats!

_DSC2885 _DSC2904I am already finding myself eyeballing the professional models, the s1000 looks awesome but man, probably upwards of $7,000 to get a really good setup with first person perspective and video downlinks. Some of the videos on the web taken with this thing are truly amazing, but then again I’m not really a video guy, I’m only looking for a stable platform for stills, at least for now.