Short reviewIt's pretty stable out of the box requires a bit of trimming. Plenty of power and does [assisted] flips with ease. Unfortunately the remote is pretty crap (compared to my DX6i) probably from a reused USB gamepad.
The yaw controls affects the throttle, not sure if because of the combined stick usage or something else is happening there.
It's pretty hard to keep a stable hover because of the sensitive throttle. Combined with the issue above it means it's pretty hard to do panning shots - rotating around an object to be filmed.
Camera is of low/mediocre quality, nothing to write home about. It's probably reusing a "spy" camera, similar to the one I wrote about: http://hackcorrelation.blogspot.de/2014/03/inside-stuff-teardown-and-repair-of-spy.html
The sound is mostly from the engines, to get a decent sound you will need to take the complete unit apart and reposition the microphone by soldering new wires.
The microSD card has frequent failures and probably needs to be replaced. I have not run a "true" capacity test on it but it might be 512MB instead of the advertised 2GB.
Flight times are shorter than advertised, probably around 4 minutes until the battery is discharged to a still safe level (3.7V). You do get 5-7 minutes of flight until the automatic voltage shutdown kicks in, but after that the battery only has ~3.3V.
I'm using a 500mA charging adapter instead of the supplied USB "charger", reducing the charge time to ~15 minutes. RC-style batteries are pretty safe to charge at 2C - the included battery is 350mAh - so any charger under 700mA should be ok.
TeardownIt's a two-board construction, with the quadcopter control board and the camera board sandwiching the battery holder.
Microphone is on the lower-right side in this picture, next to the microSD metal shield.
The quadcopter board is based on Invense MPU-6050C 3-axis accelerometer and gyro combination.
Driving all this is a chip marked HL004 436AB 242B014-80 ARM. I assume an 80MHz ARM chip similar to the one used on Arduino Due. This enables leveraging the existing Arducopter code.
Communication is provided by Beken BK2425 chip: http://www.sunrisedigit.com/en/Show.asp?id=38 which is in itself a chip similar to nRF24L01+.
The camera board is connected by 3 wires so I assume power (red) is running all the time while the signal wire (either white or green) is being put to ground whenever a picture or video is required.
The spy pen cameras I have use a short press of the button to take a picture and a long press to start/stop video recording (or viceversa). This means that in case of video it will take at least the long-press duration (3 seconds?) until the recording is really started.