Up til yesterday, I'd been shooting video with the camera settings in the default 16:9 aspect. When I shot stills, I noticed that they were 16MP and not 20MP. So I figured it must be because of the sensor crop. So I changed it to 4:3 to shoot some larger stills. I found those to be 17.7MP, still not 20MP, but that's a different story...

Prior to the change in aspect settings, the drone always shot rock stable video with not a hint of lateral jitter on descents nor any "bumping" of the camera on forward flight.

However, after those settings, and in much calmer air (all prior flights were on windy days), I noticed about 3 "bumps" that caused the camera to jolt left/right, affecting about 3 frames of video at 24FPS in a ten minute flight, and any time I descended, the camera was shaking laterally just a bit.

So that being the only change in settings, I am wondering if changing aspect for the stills to 4:3 reduces the range of the in camera image stabilization?

No it doesn't and you don't need to open multiple threads on the same topic.

May I know what device that you are using?

DJI Natalia Posted at 2018-5-29 07:40
May I know what device that you are using?

Phantom 4 Pro v2.0 with iPad Retina on both flights.

The ONLY setting I changed was camera aspect ratio between flights. I'll hopefully have a video showing just the anomalies uploaded shortly.

Below are excerpted spots where excessive jarring of camera is noted on this last flight.

Points of interest are listed in the description:

Mark Weiss Posted at 2018-5-29 08:30
Phantom 4 Pro v2.0 with iPad Retina on both flights.

The ONLY setting I changed was camera aspect ratio between flights. I'll hopefully have a video showing just the anomalies uploaded shortly.

Yes, the lower the aspect ratio, it will also lower the image stability.

DJI Natalia Posted at 2018-5-29 09:25
Yes, the lower the aspect ratio, it will also lower the image stability.

A user told me this camera doesn't have digital image stabilization and that it relies solely on the gimbal correction. Is that a false assumption?

I can't imagine that 20MP imager displaying an 8.8MP image would NOT use the extra imager area for stabilization to clean up any jitter that the gimbal didn't correct completely.

Mark Weiss Posted at 2018-5-29 10:31
A user told me this camera doesn't have digital image stabilization and that it relies solely on the gimbal correction. Is that a false assumption?

I can't imagine that 20MP imager displaying an 8.8MP image would NOT use the extra imager area for stabilization to clean up any jitter that the gimbal didn't correct completely.

Let me get back to you. I just needed to check this information.

Up til yesterday, I'd been shooting video with the camera settings in the default 16:9 aspect. When I shot stills, I noticed that they were 16MP and not 20MP. So I figured it must be because of the sensor crop.
Changing the aspect ratio for stills makes no change to video resolution.
Changing video resolution is a different setting.

The options available are:
Stills - Image Size   
3:2 Aspect Ratio: 5472 × 3648
4:3 Aspect Ratio: 4864 × 3648
16:9 Aspect Ratio: 5472 × 3078

Video Image Size   
4096×2160
3840×2160
2720×1530
1920×1080
1280×720

So I changed it to 4:3 to shoot some larger stills. I found those to be 17.7MP, still not 20MP, but that's a different story...
If you want max resolution, you have to select 3:2 which will give you 20MP images.

The movement you saw is down to the gimbal movement

"Prior to the change in aspect settings, the drone always shot rock stable video with not a hint of lateral jitter on descents nor any "bumping" of the camera on forward flight."

Image ratio for photos and video issues cannot be related.
But if you did encounter video issues and you are wondering if the ratio change might be the reason:
Why don't you go back to your initial ratio settings and see if the video issue persists?

You have a gimbal problem...

Mark Weiss Posted at 2018-5-29 10:31
A user told me this camera doesn't have digital image stabilization and that it relies solely on the gimbal correction. Is that a false assumption?

I can't imagine that 20MP imager displaying an 8.8MP image would NOT use the extra imager area for stabilization to clean up any jitter that the gimbal didn't correct completely.

'A user told me this camera doesn't have digital image stabilization and that it relies solely on the gimbal correction. Is that a false assumption? '

The user was correct.

Eric13 Posted at 2018-5-29 14:30
"Prior to the change in aspect settings, the drone always shot rock stable video with not a hint of lateral jitter on descents nor any "bumping" of the camera on forward flight."

Image ratio for photos and video issues cannot be related.

I set it back to 16:9 and took the drone up to do a sunset time lapse. On the way down, I watched closely for any side to side jitter. Not a hint of jitter or wobble. I guess I'm leaving it at 16:9 unless I need to explicitly take photos and not video.

Mark Weiss Posted at 2018-5-29 17:06
I set it back to 16:9 and took the drone up to do a sunset time lapse. On the way down, I watched closely for any side to side jitter. Not a hint of jitter or wobble. I guess I'm leaving it at 16:9 unless I need to explicitly take photos and not video.

I'm leaving it at 16:9 unless I need to explicitly take photos and not video.
Setting the still image aspect ratio only affects stills.
It  doesn't do anything to your video aspect ratio which is a separate setting.

But whatever still or video aspect ratio you set will have absolutely no impact on the gimbal and whether it stays steady or glitches.

Do you take photos while shooting video?
If so be aware that the camera can lock on target for a moment witch will show up on the video when the camera catch up afterwards. Also be aware that images taken in that condition will have the resolution of the video not the settings for photos.

KedDK Posted at 2018-5-29 21:09
Do you take photos while shooting video?
If so be aware that the camera can lock on target for a moment witch will show up on the video when the camera catch up afterwards. Also be aware that images taken in that condition will have the resolution of the video not the settings for photos.

I did take three photos, but I stopped video recording before each photo. Why did I stop? Because when I didn't stop recording, the camera only saved JPEGs and I could not find any DNG files on the SD card. So after that, I changed my procedure to stopping video, take the photo, then restart video. Now it saves both DNG and JPEG files.

Last night I took the drone up to catch the sunset. I just let it hover for 17 minutes while the video recorded. I didn't even use Tripod mode, which probably would have made it even more steady, but it was pretty darned steady, with no rapid movements and only barely perceptable slower drift. But it did not ruin the shot. I was only at 152' above my rooftop and my wife said "why didn't you go higher so you'd see the lake in the foreground?" So that's my next sunset mission.

I'm going to play with 3:2 aspect next and then see if I get any wobble on descent. I'll likely do several trials to rule out any random behavior.

Hi Mark, the image resolution will not affect the image stabilization. If the camera issue appears again, please try to reset the camera settings, then choose a proper resolution to check whether it helps.

DJI Susan Posted at 2018-5-30 23:34
Hi Mark, the image resolution will not affect the image stabilization. If the camera issue appears again, please try to reset the camera settings, then choose a proper resolution to check whether it helps.

I will do a few test runs with both aspect ratios and see if there's no correlation. It sure seems that it should not matter what mode the camera aspect is set to for stills.

Could you also confirm or deny whether the camera is doing in-camera stabilization? If only a third of the 20 megapixels are in use for video, it would seem that the extra pixels could be used for stabilizing any jitter that gets past the gimbal compensation.

Mark Weiss Posted at 2018-5-29 09:14
Below are excerpted spots where excessive jarring of camera is noted on this last flight.

Points of interest are listed in the description:

The image resolution will not affect the image stabilization. Kindly reset the camera settings on your DJI Go 4.

DJI Natalia Posted at 2018-5-31 07:01
The image resolution will not affect the image stabilization. Kindly reset the camera settings on your DJI Go 4.

[view_image]

Thanks for your other answers, but can you research my other question about in-camera stabilization?
The sensor in this camera is very similar to the sensor in my Sony RX10 Mk III, which has 4.5 stops of image stabilization capability. While I can understand that there would be no optical stabilization due to the package size, there can be ample digital stabilization with a 20 megapixel sensor when only 8 megapixels are in use for video recording.

Mark Weiss Posted at 2018-5-31 08:01
Thanks for your other answers, but can you research my other question about in-camera stabilization?
The sensor in this camera is very similar to the sensor in my Sony RX10 Mk III, which has 4.5 stops of image stabilization capability. While I can understand that there would be no optical stabilization due to the package size, there can be ample digital stabilization with a 20 megapixel sensor when only 8 megapixels are in use for video recording.

Thanks for your other answers, but can you research my other question about in-camera stabilization?
You can research it all you like but there still is no in-camera digital stabilisation.
A camera on a gimbal has all the stabilisation it needs.

Labroides Posted at 2018-5-31 14:07
Thanks for your other answers, but can you research my other question about in-camera stabilization?
You can research it all you like but there still is no in-camera digital stabilisation.
A camera on a gimbal has all the stabilisation it needs.

I know if that's the same Sony sensor used in the RX10, that it has the capability. And why wouldn't they use it, especially where a gimbal has no immunity to lateral shake? A gimbal only corrects for pitch and yaw, but does nothing about lateral movements.  The gimbal is mounted on a shock mount, but that has limitations.

The DJI folks ought to know the answer to this question absolutely.

Mark Weiss Posted at 2018-5-31 14:40
I know if that's the same Sony sensor used in the RX10, that it has the capability. And why wouldn't they use it, especially where a gimbal has no immunity to lateral shake? A gimbal only corrects for pitch and yaw, but does nothing about lateral movements.  The gimbal is mounted on a shock mount, but that has limitations.

The DJI folks ought to know the answer to this question absolutely.

Mark, you are not going to get an answer from DJI on this. The people on here are forum moderators, their technical people do not participate in the forum at all. Electronic image stabilisation is a not a function of the camera sensor, it is in the image processor chip, and the DJI cameras do not use electronic image stabilisation, as you have been told several times. DJI were one of the first companies to offer gimbals for picture stabilisation with their Ronin range, and the Phantoms have inherited the technology from there. A good gimbal in an aircraft is perfectly capable of stabilising the picture by itself.

'I did take three photos, but I stopped video recording before each photo. Why did I stop? Because when I didn't stop recording, the camera only saved JPEGs and I could not find any DNG files on the SD card.'

The camera will not shoot DNG if you take a still while the camera is recording video. In fact the camera cannot change its resolution or any other settings, because it is meant to be shooting video at the time. To record a DNG shot, it takes too long and would miss frames, so you are lmited to a simple frame-grab, and it also is subject to any movement in the camera at the time because it uses the same shutter speed the video is set up for.

Mark Weiss Posted at 2018-5-30 05:55
I did take three photos, but I stopped video recording before each photo. Why did I stop? Because when I didn't stop recording, the camera only saved JPEGs and I could not find any DNG files on the SD card. So after that, I changed my procedure to stopping video, take the photo, then restart video. Now it saves both DNG and JPEG files.

Last night I took the drone up to catch the sunset. I just let it hover for 17 minutes while the video recorded. I didn't even use Tripod mode, which probably would have made it even more steady, but it was pretty darned steady, with no rapid movements and only barely perceptable slower drift. But it did not ruin the shot. I was only at 152' above my rooftop and my wife said "why didn't you go higher so you'd see the lake in the foreground?" So that's my next sunset mission.

I did take three photos, but I stopped video recording before each photo. Why did I stop? Because when I didn't stop recording, the camera only saved JPEGs and I could not find any DNG files on the SD card. So after that, I changed my procedure to stopping video, take the photo, then restart video. Now it saves both DNG and JPEG files.


Erm, surely this would account for the erratic movement, you've tried joining the clips and badly, hence the judder

The clue is in this

Up til yesterday, I'd been shooting video with the camera settings in the default 16:9 aspect. When I shot stills, I noticed that they were 16MP and not 20MP. So I figured it must be because of the sensor crop. So I changed it to 4:3 to shoot some larger stills. I found those to be 17.7MP, still not 20MP, but that's a different story

But hats not all you chaged was it cause you started to stop the video during a shoot to get a still image too, then restarted the video again

Bashy Posted at 2018-5-31 20:03
I did take three photos, but I stopped video recording before each photo. Why did I stop? Because when I didn't stop recording, the camera only saved JPEGs and I could not find any DNG files on the SD card. So after that, I changed my procedure to stopping video, take the photo, then restart video. Now it saves both DNG and JPEG files.

You misunderstand my statement. I know that there will be a jump in time if I joined the clips. But I'm not talking about wobble between clips. I'm talking about the wobble WHILE VIDEO IS RECORDING.

Now I have formulated a theory that may explain this, after reading a very interesting thread elsewhere on this forum about prop balancing. Each time we put the props on our P4P, there is a random chance that the props will end up on different motors. If one prop has less than perfect balance out of the factory and that prop ends up in a location where its vibration has a bigger influence on the camera gimbal, then it may exceed the ability of the isolation to do its job. If that prop ends up in a different location where it has less influence on the camera, footage may appear smooth. That may account for the randomness of 'gimbal wobble' that many folks are reporting here.

Fedex has my Crystal Sky Ultrabright out for delivery today, so I hope to be able to actually SEE the quality of what I'm shooting and be able to notice if there is wobble soon enough to land the drone and swap the rotors.

I have another theory that bug parts that get stuck to the rotors in the warmer season may be slightly unbalancing them and causing vibration levels to increase. There have been some flights where after landing, the rotors are a bloody mess. I keep baby wipes with me now for cleaning my drone and rotors after every flight.


As far as DJI is concerned, if the mods forward my questions to the engineering team, they should be able to forward me an answer, as any company with competent staff are capable of.

You are theorizing too much, Mark. You should be a scientist and not a drone pilot ;-)
I wouldn't expect that your wild theories make it up the ladder at DJI.

There are all kinds of stabilization techniques these days:
Lens/Image/Optical/Digital/Sensor.
You are not gonna make DJI justify to you why they have chosen the Gimbal.

Props:
If you think that a few bugs throw the system off balance - I want to see the bugs you are hitting.
My drone and props are sometimes covered with bugs and blood big time and I don't bother to clean this.
Never had an issue, never saw my drone hovering unnormal.
If you think that props being on different motors each time you put them on being a problem:
Remember there are two silver and two black ones. The relation is always symmetrical. You can't put a silver prop on a black motor.

Prop Balancing:
I bought a prop balancer only to find out that all the DJI props I ever used were perfectly balanced.
Disappointing - I wanted something to justify the prop balancer purchase ;-)

Happy flying!

Eric13 Posted at 2018-6-1 11:04
You are theorizing too much, Mark. You should be a scientist and not a drone pilot ;-)
I wouldn't expect that your wild theories make it up the ladder at DJI.

I'm a retired engineer, actually. ;)
I'm theorizing as to why the random gimbal vibration occurs and disappears on different flights. I don't have enough flight time to form a reliable trend however. but hope to accumulate enough data to spot if there is a trend relating to prop balance. It has to be that, if it's not the change in aspect ratio.

One fellow that did a video about a Du Bros (sp?) prop balancer, showed several sets of props he bought. The DJI were fairly close, but a little off. The counterfeit ones were awful. I think it can't hurt to have balanced props as it relieves a strain on the motor bearings.

I've set the camera to 3:2 now and am awaiting my delivery and some better weather, so I can do some more testing.

Mark Weiss Posted at 2018-6-1 09:54
You misunderstand my statement. I know that there will be a jump in time if I joined the clips. But I'm not talking about wobble between clips. I'm talking about the wobble WHILE VIDEO IS RECORDING.

Now I have formulated a theory that may explain this, after reading a very interesting thread elsewhere on this forum about prop balancing. Each time we put the props on our P4P, there is a random chance that the props will end up on different motors. If one prop has less than perfect balance out of the factory and that prop ends up in a location where its vibration has a bigger influence on the camera gimbal, then it may exceed the ability of the isolation to do its job. If that prop ends up in a different location where it has less influence on the camera, footage may appear smooth. That may account for the randomness of 'gimbal wobble' that many folks are reporting here.

'As far as DJI is concerned, if the mods forward my questions to the engineering team, they should be able to forward me an answer, as any company with competent staff are capable of.'

From time to time the moderators say they will pass on comments to the development team. They very well may, or may not, but you don't get an answer back. This forum is primarily a self-help forum, you get more help or answers on here from the users than DJI. And, as I said once before, the DJI development or technical people do not participate in the forum. You are wasting your time addressing questions or comments to them.

The reason the image stabilisation is not used and the gimbal is, is this, lets  look at hovering as an example, in effect its like sticking the camera on to a tripod, now, if you have a DSLR and ever used a tripod and left the image stabilisation on, you will find all your images are rubbish (blurry), this is cause the IS is working all the time, even when there is no movement, so what should be a steady shot now turns in to an unsteady shot, this is where the gimbal is superior to IS, it works with movement, the more the movement the harder it works, i know its not the best explanation but i aint good at explaining stuff but felt the need to try.

As for the issue at hand, i take it you mean the wobble at about 1 second in? my apologises, i didnt notice that at 1st and thought you was on about the issues further in.....
There is deffo something amiss and i feel that the issue has been overlooked due to talk of IS

Bashy Posted at 2018-6-1 21:12
The reason the image stabilisation is not used and the gimbal is, is this, lets  look at hovering as an example, in effect its like sticking the camera on to a tripod, now, if you have a DSLR and ever used a tripod and left the image stabilisation on, you will find all your images are rubbish (blurry), this is cause the IS is working all the time, even when there is no movement, so what should be a steady shot now turns in to an unsteady shot, this is where the gimbal is superior to IS, it works with movement, the more the movement the harder it works, i know its not the best explanation but i aint good at explaining stuff but felt the need to try.

As for the issue at hand, i take it you mean the wobble at about 1 second in? my apologises, i didnt notice that at 1st and thought you was on about the issues further in.....

I have not used a DSLR on a tripod with an IS lens, but I have used a cinema camera on a tripod with image stabilized lenses and I can report that the only 'anomaly' that IS causes is that when panning and stopping, the image will 'swim' backwards for a couple of seconds as the IS re centers itself. It has absolutely NO other degrading effects on image quality.

But as I am finding, and have reported on in another post, apparently the craft's gimbal is not compensating for roll and fails to keep a level horizon, but only when in flight. Manually rolling the aircraft shows that the gimbal keeps the camera locked to horizon. So something's up with software and flight mode.

As for communicating with engineering at DJI, if the mods can pass info to them, that's good. But I will contact them directly about questions I need answers to.