Horizon tilt is my biggest issue with the H3-3D, assuming the ND filters en route from PolarPro solve the jello issues I am seeing. Here are my observations on horizon tilt, with one big asterisk on my experiences: I crashed my new P2+H3-3D from ~30' up when TBE took it into an oak tree while I was adjusting my Lilliput on its tripod. That crash bent the main gimbal arm substantially, and I am pleased and amazed I was able to bend it back and get the gimbal back in the air. Point being my behaviors *might* be related to the crash and/or post-crash re-alignment.
1) After the above crash, I installed the repaired H3-3D on my older P2, without cutting into the shell, by the following cheap hacks: (a) used long screws from P1 prop guards to hold gimbal to shell; (b) used four plumbing faucet washers between gimbal plate and shell; these have a center hole in them for the screw, and in theory provide some additional vibration absorption. Yes the gimbal and camera hang very low to the ground, but I always launch from and land to my hand, not terra firma. The benefit of this cheap hack is that you can EXPERIMENT with the alignment of the gimbal plate on all four corners, by tightening or loosening the screws, and because the faucet washers are somewhat compressible. I stumbled on this "feature" yesterday, which allowed me to quickly confirm that adjusting level of gimbal plate relative to underside of P2 shell does make a difference.
2) This morning at first light with very little wind I flew ~60min / four batteries. Immediately prior to those flights, I did a full IMU calibration on a very level and stable surface, then a compass calibration. My camera setup is H3+ with Ragecams 5.4mm lens (lighter than stock lens) and RageCams carbonite ND filter (solves jello, but don't like the visual effect on the image). I balanced that setup by putting double-sided tape and a length of 1/8" solder on the opposite end of the gimbal (where some of you are putting bottle cap + wire clip + coin). On each new battery, I insure the lid of my box is level (small bubble level), and start up the P2 and gimbal on that case, allowing the gimbal to figure out where/what level is.
3) With above configuration, and it's ease of adjustability on the gimbal plate level relative to shell level, I determined that the corrective action is counter-intuitive (at least to me). So in my case the camera was rotated several degrees clockwise of level-horizon, so it seemed I should tighten the gimbal screws on the right side of the ship and loosen the gimbal screws on the left side of the ship in order to rotate the camera counterclockwise, bringing the horizon closer to level. Wrong. Just the opposite provided the corrective action. If right side of image is low, further loosen right side screws, and tighten left side screws.
4) I am seeing "banking" (roll) by the gimbal when I yaw the aircraft. I think many have noticed that the gimbal -- powered on but sitting on a flat surface -- yaws when you yaw the aircraft. I find that curious, and am not sure if that's related to the banking I am seeing during in flight yaw. I am also seeing that even after my adjustments above to the gimbal mounting screws, the horizon is still tilted somewhat (right horizon low, left horizon high, aka clockwise rotation of the camera). As a result, when I yaw left, I get a nice level horizon, and when I yaw right, my horizon tilt becomes worse. No yaw gives some lesser amount of horizon tilt, but still noticeable tilt. I have seen the "FPV" option in the Assistant, and understand it is to provide a "banking" effect in gimbal operations, but have never enabled it in the Assistant. Nonetheless, my gimbal is clearly banking.
I am curious how my experiences above compare to those of others who have been chasing the "horizon tilt bug".
Kelly
PS: I may post some pics later to show the faucet setup, and a video of my gimbal's startup sequence.