Great question! It wasn't very straightforward tho I'm afraid:

• Cut all power traces going to the front of frame where the core usually sits. (to avoid any short circuits after the next steps)

• Drilled the discovery plates for a new standoff mounting point (about 74mm away from the original).

• Cut away most of the bottom plate's action camera "seat" to allow maximum gimbal movement.

• Took the tiny2's stock alu mount and bent it at 90° in 4 places. (not a trivial task, bending near the tiny bolts while keeping the distance to a minimum was tricky to say the least and chances of ruining the parts are high).

• Isolating mount goes as follows: standoff -- silicone fuel tubing -- small FC dampers -- big gimbal dampers -- tiny2 alu frame. (silicone tube could probably be omitted making assembly easier). Once (finally!) together and tweaked to fit (bending alu arms), bolted into frame as one piece.

Another option I considered that doesn't involve the tiny2's alu bracket was to use http://www.goodluckbuy.com/8mm-tube-clip-fixed-rubber-anti-vibration-adapter-for-quadcopter-multicopter-landing-skid-4pcs.html and make a plate from FR4 or carbon fibre plate to join the goodluckbuy mounts to the tiny2's four little bolt holes.

Added a few pics to give a better idea of how it went together. There is little room for error, about 1mm top and bottom and right where the bends are the alu is pretty much brushing the PCB plates (4 square holes could solve this).
I'm happy with the result, it's quite neat and works beautifully.

Hope that helps!

Just finished the re-build and thought she was worthy of a few snaps before the re-maiden :)
Propulsion system, 110% satisfied, slightly better numbers than eclac suggested! Regarding video, you were right, was never happy with the two axis system, no amount of gimbal balancing, PID tuning or software stabilisation yielded results I was happy with, phantom style flying was pretty good tho.. Enter the tiny2!

Yeah I've seen some amazing footage with the tiny2. Got anything up to show how it performs?

Haven't flown the tiny2 yet! Should be more than capable! I just hope its performs well without the stock suspension mount..
If not, you might see a XuGong appearing here xD

Finally got a flight in with the tiny2, and I must say i'm very impressed!! Bottom plate needs more trimming to get it out of the way during faster turns but here's a look:

Almost done!
I suspect the dampers are too soft, might have to stuff 'em. Failing that, an order to fox tech for stiffer dampers.
Looking at the landing gear left over, I could bolt the CF tubes to the top frame, and hang a 3axis out the front...

A wise man once said: People think you can just research and buy "the right things" and be done, when in reality you'll just end up slowly buying ALL things.

These are what came on my gimbal, Anti-Vibration Rubber Damper Damping Ball for Gimbal Gopro DJI Camera BLK. They're fairly firm and offer a good amount of space between the plates to accommodate most boards.

Thanks I must be doing something right if thats your only suggestion!
What I love about the "Eachine Light-2D" gimbal is its only 160g

I´ll try perfect yaw behaviour in the APM first, maybe program a switch for really slow yaw/pan.
I can see small 3 axis working well out the front of a qav500 type frame, anything else and your looking at more weight in landing gear, retracts etc, especially with the wide angle on these action cams.

Listed most parts as many are easily overlooked, including some bench equipment.
Only ran autotune and few test flights on this one so far, still in experimental stage. Currently fine tuning the radio data rates vs range and waiting for all the 3D printed parts to arrive. Should have some pics up in the next week or so.
This build never ends!

The 3rd axis isn't necessarily about yaw speed. It's more about eliminating micro movements in the yaw as you move. This is the video that convinced me,

I agree, very convincing! I reckon there is a lot to be said for correct tweaking of yaw PID's in the FC in the first place.
If your final video is going out as 1080p then recording at 2k or 4k resolution gives you room to eliminate wobbles in post production too. Here is one example from adobe:

When everybody's going to have a retina or equivalent display in the near future I'd like my final video to be in 2k or 4k. That being said I'd rather spend a few extra dollars on a better gimbal than several hundred for software like Adobe Premiere.

True. Don't get me wrong, if I were to buy one of these cheap gimbals today I would probably go for the StoM32 3axis. Even tho it does add an extra layer of complexity and weight.
Check out DaVinci Resolve Lite software, its free and has "automatic clip stabilizer with camera pan smoothing".

Relatively speaking, we only just got 1080p streaming and TV. And my internet connection can't even handle that!
I see 4k right now as almost exclusively for film work aimed at the big screen or new TV stuff. In that case you may have to lift more than an action camera anyway.
1080p will do for me on this build! The retina displays will upscale it anyway ;)

I'm waiting for a Xiaomi Yi that can do 4k. I can't imagine we're very far off. The GoPro 4 has been doing 4k for some time now. Upscaled 1080p can't hold a candle to 4k on my monitor.

You may have to wait until well after the release of a "hero 5" with the H1 chip. Then we might see the A9 in a Xiaomi doing 4k@30fps. But then you will be waiting for a Xiaomi with a H1..

I'm not holding my breath. For now my Yi is great and the footage is quite acceptable, but I don't think I'd want to crop it.