Stability of the M2P

[noosaguy]

Hi all, I've done a couple of short Hyperlapse sequences to practice and on both I've noticed the drone moving around in the air quite a bit. To be fair, there has been a bit of wind around for both and both were shot in reverse (Waypoint mode).
I processed the DNG files in Lightroom then imported into Final Cut Pro and stabilised them but still the shots are unstable?
Any ideas?

 

[sar104]

When you say "a bit of wind", what, exactly, do you mean? Did you check the pitch/roll data or upload the flight log to AirData to get the wind calculations?

 

[noosaguy]

When you say "a bit of wind", what, exactly, do you mean? Did you check the pitch/roll data or upload the flight log to AirData to get the wind calculations?

Thanks for the reply sar104. I'm embarrassed to say I have no idea how to do that... I've looked in the User Guide but that is useless. But, the wind certainly wasn't too bad or i wouldn't have flown. I'm wondering if flying the sequence in reverse and over water might have had something to do with it?

 

[sar104]

Thanks for the reply sar104. I'm embarrassed to say I have no idea how to do that... I've looked in the User Guide but that is useless. But, the wind certainly wasn't too bad or i wouldn't have flown. I'm wondering if flying the sequence in reverse and over water might have had something to do with it?

My guess would be the problem was wind rather than anything else.

Airdata UAV - Flight Data Analysis for Drones

You can also just retrieve the flight log and post it here if you don't want to try to interpret it yourself.

 

[Trees138]

I experience that as well to varying degrees out here. It is not unusual to get VERY strong winds in the front range of the Rockies, so it's easy for me to write off. I learned stabilization in davinci resolve and chalked it up to **** happens. Luckily, when rendering a hyperlapse from the original RAW files you have a LOT more wiggle room, even to wind up with 4k.

Were you using Go4 or Litchi? If you've used both, do you see any difference in quality?

 

[noosaguy]

I experience that as well to varying degrees out here. It is not unusual to get VERY strong winds in the front range of the Rockies, so it's easy for me to write off. I learned stabilization in davinci resolve and chalked it up to **** happens. Luckily, when rendering a hyperlapse from the original RAW files you have a LOT more wiggle room, even to wind up with 4k.

Were you using Go4 or Litchi? If you've used both, do you see any difference in quality?

I'm a real newbie with this so I'm only using Go4. So you're suggesting that the drone doesn't cope with wind all that well? Even if I use FCP10 and stabilise, it is still wobbly.

 

[Downunder]

It's actually changing altitude and DJI recommend, when doing time lapses, is to try and keep high as any object close will look worse than those far away. All DJI drones do not maintain stable height as it uses a combination of GPS and barometric pressure to determine height so not really super accurate. Just above the ground it'll use it's vision sensors which lock the drone in. Just have to live with it and fly higher. I've done many low altitude lapses and you'd be amazed how much these things move around even in zero wind conditions.

 

[noosaguy]

It's actually changing altitude and DJI recommend, when doing time lapses, is to try and keep high as any object close will look worse than those far away. All DJI drones do not maintain stable height as it uses a combination of GPS and barometric pressure to determine height so not really super accurate. Just above the ground it'll use it's vision sensors which lock the drone in. Just have to live with it and fly higher. I've done many low altitude lapses and you'd be amazed how much these things move around even in zero wind conditions.

Thanks for the feedback. I was quite low because the run was quite short 2-3 seconds. I'll use greater altitude next time.

 

[Meta4]

All DJI drones do not maintain stable height as it uses a combination of GPS and barometric pressure to determine height so not really super accurate.

GPS is not usable for accurate height and DJI drones only use barometer for altitude.
The barometer is reasonably accurate for keeping stable height.
The jerkiness in the video is due to horizontal position inaccuracy.
GPS will only be good for 1-2 metre accuracy which isn't very important when the subject matter is further from the camera but get in close and you really notice it.

 

[noosaguy]

The jerkiness in the video is due to horizontal position inaccuracy.
GPS will only be good for 1-2 metre accuracy which isn't very important when the subject matter is further from the camera but get in close and you really notice it.

What do you mean The jerkiness in the video is due to horizontal position inaccuracy."?

 

[Turn11]

May I suggest you get UAF Forecast from the Play Store or App Store. You may be very surprised at the wind velocity/gust speed between the ground speed and 100 meters.

Another one that is good to have is Ventusky Weather Maps.

 

[yysc]

What do you mean The jerkiness in the video is due to horizontal position inaccuracy."?

The Mavic 2 has the following position hold parameters:

Hovering Accuracy Range
Vertical:
± 0.1 m (when vision positioning is active)
± 0.5 m (with GPS positioning)
Horizontal:
± 0.3 m (when vision positioning is active)
± 1.5 m (with GPS positioning)

This means as it does the hyperlapse flying very slow it will be affected by wind, and these smalls deviations from trajectory will be shown in the video. All those fantastic hyperlapses that you see have been stabilized in post with Adobe Premiere or similar.

The in-drone software for hyperlapse creation is very limited, it exports only at 1080p with a huge center crop and doesn't do stabilization. For best results ask the drone to save to your SD the original JPG/DNG files, dismiss the in-drone video file and do the proper Hyperlapse with a good NLE video software.

For best results you'll need ND filters.

 

[Meta4]

What do you mean The jerkiness in the video is due to horizontal position inaccuracy."?

GPS is not pinpoint accurate and from second to second, the positions your Mavic's GPS gets can have a horizontal error of +/- 1-2 metres that the flight controller will work to smooth out to some degree.
If the camera is moving a half metre or so horizontally during your flight, the difference between shots will be much more noticeable for close subject matter.

 

[noosaguy]

The Mavic 2 has the following position hold parameters:

Hovering Accuracy Range
Vertical:
± 0.1 m (when vision positioning is active)
± 0.5 m (with GPS positioning)
Horizontal:
± 0.3 m (when vision positioning is active)
± 1.5 m (with GPS positioning)

This means as it does the hyperlapse flying very slow it will be affected by wind, and these smalls deviations from trajectory will be shown in the video. All those fantastic hyperlapses that you see have been stabilized in post with Adobe Premiere or similar.

The in-drone software for hyperlapse creation is very limited, it exports only at 1080p with a huge center crop and doesn't do stabilization. For best results ask the drone to save to your SD the original JPG/DNG files, dismiss the in-drone video file and do the proper Hyperlapse with a good NLE video software.

Thanks for that I'm discovering that stability is difficult. That video was created in LRTimelapse 5 and stabilised in Final Cut Pro X. It was created using full frame RAW files.

 

[noosaguy]

GPS is not pinpoint accurate and from second to second, the positions your Mavic's GPS gets can have a horizontal error of +/- 1-2 metres that the flight controller will work to smooth out to some degree.
If the camera is moving a half metre or so horizontally during your flight, the difference between shots will be much more noticeable for close subject matter.

Thanks I m going to have to pull back.

 

[yysc]

Thanks for that I'm discovering that stability is difficult. That video was created in LRTimelapse 5 and stabilised in Final Cut Pro X. It was created using full frame RAW files.

Yep also your video is flying at low altitude with close subjects that makes the whole process harder. In the video I posted previously objects are at 500m+ so image shifting is way smaller.

 

[noosaguy]

Is it because the M2P is moving so quickly in this video that you don't notice any movement even close to the water?

 

[Meta4]

Is it because the M2P is moving so quickly in this video that you don't notice any movement even close to the water?

It's just that the gimbal is doing its job properly, keeping things nice and stable and the active tracking is very smooth.

 

[noosaguy]

It's just that the gimbal is doing its job properly, keeping things nice and stable and the active tracking is very smooth.

Why doesn't it stay 'nice and stable' when doing timelapse?

 

[Meta4]

Why doesn't it stay 'nice and stable' when doing timelapse?

Because in timelapse, your camera wasn't locked on to and tracking a specific target.
Also, there is movement in the video but because you are seeing a frame every 24th or 30th of a second, the motion is smooth.
While you don't see the inbetween frames in the timelapse.