EN CN
Curiosity, why does horizon drift appear?
  • I see a lot of movies shot by (any type of) gimbal where the horizon does not stay, or is level. Sometimes a bit, sometimes a lot... (If you ask about correcting this the usual answer is calibration)

    I am wondering why this drift appears. I somewhat think it is inevitable, in the SimpleBGC manual, Gyro thrust, its even mentioned, more chance of horizon drift if high gyro thrust....

    Now in flight controllers, using the same type of sensors and software I guess, horizon drift would be kinda deadly.

    So why is it that the gimbal software is able to lose its sense of level, whereas my Phantom (or Hubsan x4) doesn't?

    Just interested, that's all....
  • I guess as there is a pilot or GPS/Compass assisting the multi copter staying at level*.

    Using the FC_PITCH and FC_ROLL should help the gimbal staying at level, because they provide what the flight controller thinks is the level.

    *But probably you have heard of Naza fly-a-ways :-)
  • Well, the gps or compass have little to do with staying level, and the pilot, that's me, even less.

    So, how to connect the naza to the SimpleBGC, to tell it what's level? That would be kinda nice.
  • Ok, the naza has F1 and F2 outputs for controlling a gimbal. Research needs to be done :)

    Anyone tried this maybe?
  • Just put the multicopter on air on manual or ATTI and hands of sticks and lets see how long it stays level ;-).

    I will test the NAZA gimbal outputs in few weeks or so.

  • ATTI will remain level, but float with the wind. Manual I only try on simulators for now. :)

    If just the roll could be assisted a bit by NAZA output to remain horizontal all the time, that would be great.

    (I normally work with tilt shift lenses and bubble levels, you know ;-) )

    Even companies like xaircraft with their stella gimbal dare to publish demo videos with some enormous horizon errors, its incredible....
  • the boards are only 8 bit. When the 32bit will be released it will be totally different due to the speed of the corrections. I had horizon drifts before, but the secret is extremely well balanced gimbals. The performance is hugely affected by unbalanced gimbals.
  • And it has the option for second IMU that will help even more. But it is not only about the processor speed, but how the processor is used and it is pretty amazing that everything we have now is running on the current processor, and fits into its memory.
  • So is everyone saying horizon drift (reference points drifting) is to be expected with the current hardware? I experience drifts in any one of the 3 axes at different times, mostly only one axis at a time.

    And is the only solution for now manual joystick compensation?
  • With the 2.4b6 I have perfectly steady horizon....
  • ok guys here's what I've found..................

    horizon drift depends on the speed of rotation and the distance for the normal axis of rotation (either of the copter with 2axis setup or the yaw axis on the gimbal+/- the axis of rotation of the copter on three axis).

    I get no horizon drift if I fly around my camera as the centre point but if I just yaw then its horizon drifts in correlation to the distance from the axis of rotation and speed.

    I have managed to solve the problem but at a cost. The roll motor must be counter balanced perfectly behind as in front effectively like carrying two cameras one in front and on behind. The issue with this in addition to the weight is the size of motor on the roll axis to move double the mass quickly enough is large. all of this must be centralised in the normal axis of the copter, so forget pushing the camera out to loose the rotors and countering it with the batteries at the back. the only way is down.....drop all in the normal axis and fly steady to avoid starting inertia drift from the G generated in pendulum.

    one advantage is the thing is now so heavy it kills the high frequency vibration!

    There is no perfect solution.....I'm even planning to shoot in 4k with the GH4 to give the crop room if the shots are particularly dynamic but then you loose the frame rates.

    best compromise I have found after all the hardware balancing is shooting at least 50p 1080 and run it out at 2.35:1 110% cine view. not perfect but short of that fly straight and pan slow...

    best of luck to all