Technical Ramblings

In May, I said my next upgrade would be a GoPro. Last week, I confirmed that, as I bought a GoPro Hero 3+ Black. (Thanks to my employer, I even got a 40% discount.) For the first few days, I played with it — did a sunset timelapse, and contributed to the internet cat database by adding pictures of the kitties in a charming 2.7k video. (2.7k probably only available in Flash or Chrome.)

But in part, my reaction was “Enh. What did I think I needed this for?” After all, I already had a camera for my quadcopter (in the form of the A5-powered FC40 camera); I had a DSLR (which can shoot great 720p videos); I had my phone, which could do decent 1080p video, and I always had on me. Why exactly did I need *another* camera?

Today, I finally took the GoPro up on the Phantom, and that particular concern is no longer.

I was recording video, so I didn’t get any particularly high res stills — but that wasn’t really a concern. Shooting 1080p video, I was able to get great shots of both the storm clouds looking out west of town:

And another photo looking East towards Boston.

The colors, the clarity, the overall picture and video quality … this is what I kept seeing in everyone else’s photos, even though I wasn’t getting it in mind. I always wondered if the problem was just my tools… and now I know. As with everything, the user makes a difference… but the tools certainly help.

The FAA is aggressively pursuing those that are flying small unmanned aerial systems counter to the guidelines that it feels it has set. Specifically, the FAA believes that sUAS are aircraft, and are regulable under the rules which apply to aircraft, with specific exceptions carved out by a small number of guidelines and rules.

The legal basis for these claims is dubious at best — the summary at Drone Law Journal covers most of the legal questions attached to this issue relatively well. However, most people flying a remote controlled model aircraft are not in a position to fight the FAA. If you want to stay on the ‘easy’ side of the fight, here are the things that I would recommend, based on what the FAA is currently doing.

Stay under 400′ above ground level. The AC 91-57 model aircraft guidance suggests maintaining a limit of 400 feet AGL. This is well below the 700′ limit which is the typical upper bound for class G airspace, and also lower than the 500′ limit on flight over rural areas for private flights. This is a good idea to ensure that your UAV is out of the way of any aircraft.

Stay more than 5 miles from an airport, or contact the tower if needed to be closer. AC 91-57 suggests 3 miles, but Section 336 of the FAA Modernization and Reform Act of 2012 suggests that the FAA can not promulgate any regulation that prevents flight further than 5 miles from an airport.

Stay out of controlled airspace. The FAA recently issued a fine for an operator flying a drone in downtown Manhattan, even though the operator was more than 5 miles from the nearest airport. The letter he received indicates that he was fined for flying recklessly, but also for flying in a Class B airspace without receiving tower clearance. To stay away from the FAA, stay out of all controlled airspace. (This is also a good idea for safety reasons, in general: controlled airspace is controlled because it is where airplanes can be flying low, close to airports.)

The 2012 Act suggests that the FAA has no right to promulgate regulations that affect model aircraft if they are flown ‘strictly for hobby or recreational use’. Any use beyond this puts you in the sights of the FAA; they have already issued one fine (to Raphael Pirker); although Pirker won in court, the FAA is appealing and has not changed their enforcement approach. This situation is not unlike navigating online gaming regulations, where 카지노사이트 must adhere to strict legal frameworks to ensure fair play and safety for users. Both industries highlight the importance of compliance to avoid legal entanglements and maintain a trustworthy environment.

Stay a ‘sufficient distance from populated areas’. This is, again, from the AC 91-57 guidance to model aircraft operators. A recent case in Denver where a drone was flown over a crowd at a 4/20 rally resulted in the FAA launching an investigation; the FAA’s website claims that “Routine operation of UAS over densely-populated areas is prohibited.” (FAA UAS Fact Sheet)

Don’t fly in a reckless manner. Both the two people who have been issued fines so far have had those fines issued for what the FAA referred to as ‘reckless’ flying.

So far, the FAA has only been fought in court once, and they lost. Another lawsuit is ongoing, over non-commercial search for missing persons. (A petition to encourage the FAA to change its policies allow these flights is up at the White House petition site, if you believe this is an important issue to fix.) But they have sent cease and desist letters to many, and it seems unlikely that this will stop anytime soon.

By following the above recommendations, it is possible that you may be able to avoid the wrath of the FAA, until they change their policies to pursue people even more aggressively. I am not a lawyer, I am definitely not your lawyer. But if you want to enjoy your new Phantom, these are the recommendations I would make to you, to enjoy your hobby with the least risk of persecution by the FAA.

So, I’ve been an amateur photographer for years. I started helping to take pictures of family vacations with my dad’s AE-1 back when I was in grade school. I took photo classes in college, and I got my first digital camera in 2003. (I still have it, by the way.)

I’ve had a Digital Rebel for the past 9 years. (I got my first in December of 2005.) I’ve had my most recent camera for another 3 or so. I’ve brought these cameras to a dozen conferences and events; I’ve used them in bars, on skating rinks, at the beach. I have uploaded over 6000 public pictures to Flickr — which means that in that time, I’ve probably *thrown away* another 12000 or so. (My rate of success has drifted over time, from as low as 10% to as high as 75%.)

I won’t claim I’m an expert photographer — I’m good at taking candid portraits, but a lot of other things continue to evade my skills — but I’ve taken a lot of pictures, and I like a lot of what I take these days.

And now that I’ve started looking at video, that all changes.

I have no meaningful experience with video. I helped run the camcorder at a couple of events for my dad (who has been doing video for years), but that’s it. Now I’m faced with an entirely different medium, with a different set of requirements, and a different set of tools, and I find myself feeling completely frustrated by my efforts.

The number one thing that I learned with digital photography is you have to prepared to throw away 90% of what you take, especially when you get started. (It took me a while to get good at this.) Most of the pictures you take will be crap; even if you got the frame and content you tried to capture, you might have caught a person blinking, or have a photo that’s more out of focus, or with the wrong color balance to the point you can’t adjust it any more.

With video, the same is true (at least, I feel that way): most shots don’t come out the way you want them to. Even worse: a minor edit to a shot can change it from being perfectly reasonable to being silly (or from being perfectly silly to looking just reasonable). So in addition to finding the perfect shot, you’ve got to edit it right.

Attention spans are short. I once read an article from someone complaining about ‘quick cuts’ in modern TV. Since then I’ve paid careful attention, and learned that it is rare for most TV to have single visual cuts that are longer than about 3-5 seconds. (If you watch fast-paced shows like The Amazing Race, this drops by half; you have tons of super-fast clips.) Even for a relatively short, 3.5 minute video, my Ohanami video, on average people only watch one minute of it before they stop. (And I consider that video one of my better / more entertaining efforts so far!)

When I started editing video from my quadcopter, Jess jokingly sent me a link to the Friend Who Sent Link To 8-Minute YouTube Video Must Be Fucking Delusional Onion article, in response to my first Phantom FC40 edited video. I defended my position, indicating that I had put a lot of work into trimming out the boring parts… but since then, I’ve become much more ruthless. (In the case of the Ohanami video, I edited approximately 1.5 hours of total video footage into a 3 minute clip reel, and it was relatively well-received.) As usual, my lovely wife was right: people don’t care about 90% of the footage I’m filming, a lesson I should have learned from photography.

Even beyond editing, I’ve got new equipment problems. While the DSLR is great for shooting photos, shooting video with it means holding it out far from me, which makes steady shooting hard. You get jumpy, jerky footage (as you can see in the Us at the Zoo video I put together from this weekend); zooming with a lens is great for photos, but crappy for videos.

In part, this is my choice of tool. The DSLR is a fine camera now that I’ve learned to use it, but it’s not clear that it’s the best tool for shooting video, but it’s what I’ve got for the moment. But even ignoring that, I’ve got the bigger problem: Shooting candid, entertaining video is an entirely different ballgame.

I’m enjoying learning new tricks, and producing slightly more compelling footage and editing. I’m enjoying this new medium; I’ve been watching lots of other people’s videos, to learn how to do it right — or at least better. (My favorite candid family gathering video is this Easter video that someone posted in one of the quadcopter forums due to its use of a quad for a couple of the shots. This has actually been a primary motivator for me to get out there and see what I can do.) So far what I’m learning more than anything else is “This is hard work!”

My next upgrade for the quadcopter is going to be a GoPro. Specifically, I’m looking to pick up a Hero 3+ Black (though if I find a great deal on the Silver, I might take that instead.)

A friend of mine joked the other day that he was disappointed in the quality of the shots from my quad; linking the video of flight in DC as what my video should look like. He was clearly just making a snarky comment, but I wanted to establish that there are two huge differences here: a gimbal mount and a GoPro.

A gimbal mount offers in-flight stabilization using a gyroscope, and typically also offers pitch control (looking up and down) via a 7th channel on the transmitter. A good gimbal (like the ZenMuse H3-3D) costs about $350 — and is also balanced around a particular type of camera (in this case, the GoPro), for both balance and placement. A gimbal is key for doing stable shots: a quadcopter stays in the air primarily through brute force, constantly adjusting its four motors to stay level, and shifting in moving air is constant. Without a gimbal, you will always have shake, especially when hovering or moving slowly. (With a strong lateral or forward movement, there tends to be less drift.) This YouTube video demonstrates two different types of gimbals, and is a good video of a gimbal in action.

But currently I’m shooting with a cheap 720p camera, and the more I shoot, the more I notice it (and see how it suffers). Upgrading to a GoPro — in addition to making the gimbal usable without hacky mounts — seems like it will have the following effects:

1. Better color. Everything taken from a modern GoPro is just significantly more consistent in color, and that’s a huge problem, especially in low light, from the cheap FC40 cam.
2. Better compression (or maybe none? I don’t know.) The FC40 cam does a shitton of compression on-device before you ever get to the video, which makes getting some types of video out of it (like my forest flight) impossible to get good quality out of. This is a real PITA for flying in a lot of places — even the river gets affected.
3. Better consistency in light. The FC40 does do exposure adjustement, but it’s super jerky and ugly, especially in the evening. The GoPro footage I’ve seen doesn’t seem to have the same problems.
4. Better base resolution. The black offers up to 2.7k video at 30 fps, or various different sizes (like the 1440p super-wide) that give you plenty of room to crop crap out while doing post-processing for things like image stabilization.

The reasons for choosing the Black over the Silver aren’t huge — the Silver doesn’t have the higher end resolutions (4k, 2k, 1440p), but is otherwise pretty similar. However, when you look at eBay, the price differential is < $50, rather than the $100 difference retail, and I think that the $50 is worth it. I know that the video I'm producing isn't great yet, but that's okay. As time goes on and I can invest some more in my tools, and also learn better video editing techniques, I feel confident I'll be able to continue to improve in this area. But step one is definitely a GoPro.

So, the Phantom FC40 comes with a wireless camera, controllable from an iPhone or Android phone. As adumbrated at hotrate.com, it’s a small device — about two inches by two inches, similar to the size of a GoPro if you cut off the ‘lens’ part. I think the camera is a stripped down version of a another wifi camera; at least, the specs, shape, etc. seem to match.

It records 720p video to a built-in SD Card, and is controllable via an App on the iPhone or Android. However, there’s no way to control it from a laptop, which sort of annoys me. I’d also like to experiment with linking the camera and doing live-streaming straight from the camera to the web — I think that would rock.

After doing some research, I found that it looks like the software on this camera is very similar to the wifi support on the GoPro: It uses an “Ambarella streaming” web-app hosted by Apache Coyote on the device. The device provides an RTSP stream, and in theory allows you to browse the files on the device and get access to them.

However, it appears that DJI stripped most of this functionality out — in a pretty hacky way, from what I can tell, leaving half-complete stuff and stubs. It has instead implemented its own hacky interface that you can use to control the camera, though it still does RTSP streaming to the phone application.

In trying to work with the camera, I learned the following things:

– The camera sets up a wifi access point, using 192.168.1.1 as its IP, and serves DHCP addresses starting with 192.168.1.100 to clients. It serves an HTTP server on port 80. (This is different than the GoPro which defaults to a different IP setup and runs on port 8080.)
– Most interaction to the camera is through posting of XML to /CGI/ calls.
– Most HTTP calls/functions in the camera require a “Cookie” called “Session” with a value in it. If they don’t have it, they simply return an Error code.
– To get a session, you post to “/CGI/CameraLogin?Device=Mobile&Stream=RTP_H264_WQVGA” with a ‘password=’ (blank) form value. This returns a set of data in XML (‘Login OKFC40_S7NO4621102U40_NO_CODECA06‘) and also has a Set-Cookie header: ‘ Set-Cookie: Session=750997680’. However, the Set-Cookie header is preceded by a space, and therefore is not recognized by regular browsers; it seems the FC40 app knows to look for this, but this prevents trivial use of forms in browsers to replicate the functionality. (No idea if this is intentional or not.)
– There is a CGI for RemoteControl
– There is a CGI for status.

I’ve put together some documentation for these and put them into a fc40 camera github repo. I’ll probably end up expanding it a bit more.

The most interesting thing, of course, is the live streaming of video. After poking it a bit, I did find out that it does RTSP streaming, and I was able to discover that VLC does, in theory, play RTSP streams. However, although the setup mostly works, when it tries to PLAY a stream via VLC, the camera immediately closes the connection. Having inspected packets, I believe the only difference is that the User-Agent is VLC instead of being blank/not included, which is apparently a not-unheard of trick for security-by-obscurity for RTSP streams. I have not yet gotten to the point where I can test this theory, but I’m working towards it.

I wish that DJI was a bit more open about these things, but maybe the reason they’re not is because they took the hardware from some OEM who didn’t want people to get this instead of a more expensive model… or instead of upgrading to a GoPro, which does have a pretty open setup out of the box where all these things ‘just work’. In any case, I figured I’d share what I learned for others to play with.

So, I briefly mentioned this in another post about YouTube, but I bought myself a Phantom FC40 Quadcopter for my birthday.

It’s an all-in-one easy-to-use out of the box flying platform. It comes with a camera, which has wifi support for remote operation (“FPV”). It requires no experience with flying any type of aircraft — it’s pretty much all automatic, and driving it is more like a video game than anything else. It normally retails for $500; I picked it up for $430 during a sale at B&H. (Since then, they’ve been almost constantly sold out, so either supply is tiny or demand has picked up a lot.)

It is the most exciting toy I have ever bought. Flying it is super neat, and the videos that it takes are brilliant. For a long time I’ve considered that I’d like to get my pilot’s license, but I had never really had the ability to set aside the money it would take to do it. The Phantom is certainly no pilot’s license, but it still lets me see the neighborhood I live in in a different way, which is part of my intent.

I would like to do some things with mapping using the quad: doing super-local aerial imagery stitching. But so far, I’ve been having too much fun just flying the thing around.

The FC40 — which is the low end, basic model — doesn’t have any remote telemetry, so I can’t tell how far, fast, or high I’m going. The transmitter, in theory, goes to about 1000 feet, and remote video goes to about 300 feet. I don’t think I’ve hit the 1000′ limit, but I’m pretty sure I’ve passed the 300 foot limit.

I need to get more practice on video editing: creating a compelling story is hard. (More on that in another post sometime.) But the hardware itself is terrific. It’s simple enough that even little kids can do it. The copter is resilient — even when I accidentally got turned around and flew the thing into a fence at full speed, it came out basically unscathed (minus a few replaceable prop guards). It’s stable — even in 10-15mph winds, it holds steady, and if the transmitter goes on the blink, it will return to its start position, for example.

I bought the Phantom — rather than doing something more “DIY”/open source — because the price for what you get was terrific. By the time you buy motors, electronics, and a camera, you’re already looking at $350-$400 of retail parts; getting all of that, in a pre-assembled ready to fly package was brilliant.

The support story on the Phantom is a bit weird; especially with the FC40, which is among the newer models, there’s a bit of an issue around replacement parts (I’ve lost one of my vibration dampers for the camera, and I can’t figure out how to get a new one, because no one sells that as a part for the FC40), but I think that’s likely to slowly go away as the FC40 becomes as widespread as the other Phantom models.

Of course, with any new technology, there are a set of FAQs that you should expect. The Quad is no exception.

Is that a drone?
It really depends on what you mean by the word drone. Many people who ask this think of drones like the US Government uses them — far-flung remote-operated bombing or long-term surveillance machines. This isn’t that. On the other hand, if what you mean is “Can this thing fly and take video, possibly without me seeing the person operating it?” the answer is yes — in that sense it is a drone. My usual answer is “I just call it my quadcopter.”

Is that a camera on the front?
Yep! It’s a 720p video camera which records to a memory card inside the camera. I can also hook it up to an iPhone or Android — but realistically, I don’t, because so far I’m not good enough to fly my quad without actually staring at it the whole time. It’s something I want to work on.

How high can it go?
Higher than I can see it. Under the most conservative rules regarding model aircraft — which is the most straightforward way to look at the Phantom — you can fly legally up to 400 feet, as long as it isn’t being done for commercial purposes and as long as it is more than 5 miles from an airport. I don’t have good data, but I believe I have flown higher than that when flying in the middle of nowhere, though not much — when I get that high, I can’t see the copter, so an errant gust of wind can mess things up if I’m not careful, so I generally try to stay pretty close.

How far can it go?
The transmitter is rated for up to 980 feet (I usually say ‘a quarter mile’), but some people have reported having it work up to twice that without any kind of booster. It runs off the same frequencies as some wifi routers, so it isn’t as long distance in a city as in the middle of nowhere.

Is that legal?
So long as I’m 5 miles from an airport, staying low, and not bothering people, the commonly held belief is that it falls under the FAA’s model airplane rules, and is legal to fly. Though the FAA makes contradictory statements all the time, so the “and not bothering people” becomes really key. 🙂

Do you go to MIT?
Nope.

Between this, and the cost ($500 is generally viewed as “a lot cheaper than I would have thought it would be”), this forms the majority of the common questions I get about flying the quad around Boston — and I *always* get questions. I’ve gone out and flown approximately 20 different occasions now, and I’ve never *not* gotten questions from *someone* 🙂

My most ‘interesting’ photo

One of the things that has always confused me is how Flickr’s ‘interestingness’ score worked. It’s clearly not based directly on views — my most ‘interesting’ photos have ~500 views, while my most popular ones have ~5000. The same is true of comments, tags, groups, etc: Some of my most interesting photos are, all in all, the least obviously ‘popular’ ones.

However, reading through the patent on Interestingness, I see:

[0027] The statistics engine generates statistics and other metrics based upon aggregated metadata. In one embodiment, the statistics engine determines the popularity of metadata (e.g., tags) within a grouping of media objects over a predetermined time period. For example, the statistics engine may determine the number of different users that have assigned a particular tag to one or more media objects within all groups on the system, within a single group, or within a set of media objects, over the last 24 hours. The aggregation engine may determine (and display) a histogram of the tags, and may determine the most frequently assigned tags (at any point in time or over a predetermined time period) by determining those tags either having a frequency exceeding a minimum threshold frequency or belonging to a predetermined number of the most popular tags.

The patent application doesn’t directly claim that this is used in the interestingness calculation, but this type of time-period based tag aggregation/valuation is clearly a non-obvious metric that can’t be calculated by looking directly at user-visible information on a single photo.

Still, none of this really changes the fact that I wish that flickr had a per-person ‘most interesting this week’ sort: my ‘eastern bunny’ photo above has been at the top of the interesting list for ages, and I want new interesting photos, damnit!