Not really a bug, merely an imperfection.
While examining the EXIF data on pictures taken by the mini4pro, I noticed that the GPS altitude showed an altitude that wasn't quite realistic.
A picture taken at ground level should indicate the ASL - which was actually about 3m - showed up in the EXIF data 'GPS altitude' as 177m.
A picture shot at about 85m - as indicated on the RC - was shown as being 262m high (impossible with the 120m limit and a local restriction of 90m)
As I intend to use these pictures for a research project, it is important to get the exact altitude.

You must convert the datum to go from WGS to ASL.  
This is Belgium?  the difference should be between 40m and 47m.  The number you mentioned are more indicative of a GNSS signal fault.

LV_Forestry Posted at 11-2 06:47
You must convert the datum to go from WGS to ASL.  
This is Belgium?  the difference should be between 40m and 47m.  The number you mentioned are more indicative of a GNSS signal fault.

That's a nice suggestion, but how do I do that ?? ;-)

WVdE Posted at 11-2 08:19
That's a nice suggestion, but how do I do that ?? ;-)

You download the geoid and with your GIS you resquest the altitude offset for each picture coordinates.
Then you remove this offset from the altitude recorded in EXIF.

Or else more simple you use a photogrametry software that has this function.

geoid for Belgium :

https://www.isgeoid.polimi.it/Ge ... um/public/hBG18.dat

Not really a bug, merely an imperfection.
While examining the EXIF data on pictures taken by the mini4pro, I noticed that the GPS altitude showed an altitude that wasn't quite realistic.

Unless DJI have changed the way they calculate what they call "GPS Altitude", it's a serious flaw.
In previous models, what they call GPS altitude is not derived from GPS at all.
It's been extrapolated from barometric data without any allowance for normal changes in air pressure.

Do some checking by photographing from the same location and at the same height in different weather conditions and see how consistent the results are.
I commonly get metadata indicating that photos taken close to sea level were shot from 100 metres or more below sea level.
But others taken at other times in different weather systems are recorded as being well above sea level.

DJI's "GPS Altitude" is not to be relied on at all.

Labroides Posted at 11-2 13:30
Not really a bug, merely an imperfection.
While examining the EXIF data on pictures taken by the mini4pro, I noticed that the GPS altitude showed an altitude that wasn't quite realistic.

Can you detail what you mean by extrapolating baro data?

The GPS altitude is the altitude that the GPS receiver returns and that's it.  

It is not ASL it is WGS.  and the altitude displayed on the screen is above take off point.  

An error of 100m I ask to see.

LV_Forestry Posted at 11-2 13:56
Can you detail what you mean by extrapolating baro data?

The GPS altitude is the altitude that the GPS receiver returns and that's it.  

I'm talking about DJI consumer drones like the OP asked about.
Your RTK-equipped  Matrice might be quite different.

LV_Forestry Posted at 11-2 13:56
Can you detail what you mean by extrapolating baro data?

The GPS altitude is the altitude that the GPS receiver returns and that's it.  

See the examples I pm'd you.

DJI's "GPS altitude" is a poor attempt to derive altitude above sea level from the barometric data.
However they have assumed standard atmospheric pressure and make no allowance for normal changes in air pressure as weather changes.

Metadata from photos taken at the same height, same location, but different days with different weather, can show wild variation in "GPS altitude"

LV_Forestry Posted at 11-2 13:56
Can you detail what you mean by extrapolating baro data?

The GPS altitude is the altitude that the GPS receiver returns and that's it.  

Here's the biggest variation I've spotted so far.
Much more than 100 metres.
Same spot, different day.

https://drive.google.com/file/d/ ... fh/view?usp=sharing

Now tell me that "GPS altitude" comes from GPS?

LV_Forestry Posted at 11-2 13:56
Can you detail what you mean by extrapolating baro data?

The GPS altitude is the altitude that the GPS receiver returns and that's it.  

The GPS altitude is the altitude that the GPS receiver returns and that's it.  
It is not ASL it is WGS.  
It doesn't matter how it's expressed.
If it was from GPS, it shouldn't be varying by 100 metres or more on different days.

and the altitude displayed on the screen is above take off point.  
Is it really?  
Most DJI flyers manage to work this out from on first flight.
How is this relevant to the discussion?

Labroides Posted at 11-2 14:24
I'm talking about DJI consumer drones like the OP asked about.
Your RTK-equipped  Matrice might be quite different.

Yes sure iam also talking about consumer drone.
I looked at the images you sent me, I had the same kind of weirdness once with a P4P. With the M2 Z and P, no problem.

The altitude displayed on the screen is relevant here because i affraid that the OP misundertood that one and ASL.

LV_Forestry Posted at 11-2 22:04
Yes sure iam also talking about consumer drone.
I looked at the images you sent me, I had the same kind of weirdness once with a P4P. With the M2 Z and P, no problem.

Those photos were all taken with a Mavic 3 pro in the last few months, so DJI is still confusing people with their false "GPS Altitude"
It didn't seem that the OP was confused about the screen height.

I'd suspect that the Mavic 2 also gave false "GPS Altitude".
If you go back and check old photos taken around the same place and height over time, you'll probably find they are all over the place as well.

Are the images associated with the altitudes mentioned taken somewhere in the pink zone below?

Labroides Posted at 11-2 23:00
Those photos were all taken with a Mavic 3 pro in the last few months, so DJI is still confusing people with their false "GPS Altitude"
It didn't seem that the OP was confused about the screen height.

Yes, that's what I'm looking at, and honestly no. The biggest variation I have is 15m but this is not abnormal for a GNSS receiver without correction.. On the other hand I have a P4P dataset where yes there is a hundred meters difference, Only in some images, not for the whole flight.

Hi there. Thank you for reaching out. The altitude seen from the image information is derived from the GPS property. The GPS altitude of the aircraft is an altitude relative to sea level, which is calculated by the GPS and barometer rather than by measurements from the aircraft to the ground, which cannot serve as an altitude reference. Please check the information details from the image with professional software (Exif viewer). The image’s GPS altitude is consistent with the one checked from the direct property in the computer. When checking the actual altitude from the aircraft to the ground, the relative altitude shall prevail (RelativeAltitude in the software). Please note that DJI consumer aircraft are built with an aerial photography function, but not all aircraft support accurate topographic mapping. To meet higher positioning accuracy and altitude accuracy requirements for mapping, DJI has released the Phantom 4 RTK and Matrice RTK, which are more suitable for the surveying and mapping industry. The altitude precision of Consumer products is limited. Thank you for your understanding.

DJI Paladin Posted at 11-2 23:13
Hi there. Thank you for reaching out. The altitude seen from the image information is derived from the GPS property. The GPS altitude of the aircraft is an altitude relative to sea level, which is calculated by the GPS and barometer rather than by measurements from the aircraft to the ground, which cannot serve as an altitude reference. Please check the information details from the image with professional software (Exif viewer). The image’s GPS altitude is consistent with the one checked from the direct property in the computer. When checking the actual altitude from the aircraft to the ground, the relative altitude shall prevail (RelativeAltitude in the software). Please note that DJI consumer aircraft are built with an aerial photography function, but not all aircraft support accurate topographic mapping. To meet higher positioning accuracy and altitude accuracy requirements for mapping, DJI has released the Phantom 4 RTK and Matrice RTK, which are more suitable for the surveying and mapping industry. The altitude precision of Consumer products is limited. Thank you for your understanding.

"The GPS altitude of the aircraft is an altitude relative to sea level"

This is totally false

LV_Forestry Posted at 11-2 23:15
"The GPS altitude of the aircraft is an altitude relative to sea level"

This is totally false

He means relative to the geoid .... close enough for someone who has no experience in the subject.

DJI Paladin Posted at 11-2 23:13
Hi there. Thank you for reaching out. The altitude seen from the image information is derived from the GPS property. The GPS altitude of the aircraft is an altitude relative to sea level, which is calculated by the GPS and barometer rather than by measurements from the aircraft to the ground, which cannot serve as an altitude reference. Please check the information details from the image with professional software (Exif viewer). The image’s GPS altitude is consistent with the one checked from the direct property in the computer. When checking the actual altitude from the aircraft to the ground, the relative altitude shall prevail (RelativeAltitude in the software). Please note that DJI consumer aircraft are built with an aerial photography function, but not all aircraft support accurate topographic mapping. To meet higher positioning accuracy and altitude accuracy requirements for mapping, DJI has released the Phantom 4 RTK and Matrice RTK, which are more suitable for the surveying and mapping industry. The altitude precision of Consumer products is limited. Thank you for your understanding.

The altitude seen from the image information is derived from the GPS property. The GPS altitude of the aircraft is an altitude relative to sea level, which is calculated by the GPS and barometer.
If you look at the metadata on image file I posted a link to and read what I posted, that's clearly not true.
The image metadata shows that an image shot at close to sea level was >150 metres below sea level.
Other photos taken on different days show completely different numbers with a large variation.

The Absolute Altitude or "GPS Altitude" in DJI metadata is completely useless and varies with atmospheric pressure.

Labroides Posted at 11-3 02:50
He means relative to the geoid .... close enough for someone who has no experience in the subject.

this is also false.  A GPS gives an ellipsoidal altitude.  It's always been like this for DJI drones.  

Now if they did some weird things with the new Mini4 and Mavic 3 consumer series by mixing baro + GPS in order to handicap them even more, and prevent the user from doing mapping without products from the enterprise range, I won't be surprised.

I think he has no idea what he is writing about, he is just copying and pasting a text that DJI gave him.  The difference between us and the admins is that we use drones.

LV_Forestry Posted at 11-3 03:57
this is also false.  A GPS gives an ellipsoidal altitude.  It's always been like this for DJI drones.  

Now if they did some weird things with the new Mini4 and Mavic 3 consumer series by mixing baro + GPS in order to handicap them even more, and prevent the user from doing mapping without products from the enterprise range, I won't be surprised.

Now if they did some weird things with the new Mini4 and Mavic 3 consumer series by mixing baro + GPS in order to handicap them even more, and prevent the user from doing mapping without products from the enterprise range, I won't be surprised.
It's always been the same with Phantom 3 and 4 series that I've had and I've seen this question put by owners of all the other consumer models..

Why would they blend GPS and (varying) atmospheric pressure
It's what DJI has always done.
Can you think of any other explanation for the variations I demonstrated ?

I think he has no idea what he is writing about, he is just copying and pasting a text that DJI gave him.  
That's how things work around here.

Labroides Posted at 2023-11-2 13:30
Not really a bug, merely an imperfection.
While examining the EXIF data on pictures taken by the mini4pro, I noticed that the GPS altitude showed an altitude that wasn't quite realistic.

Besides that I also have multiple (maybe all) photosets where the altitude is inverted, so photos high in the sky get a low altitude and photos low to the ground get a high altitude number. Maybe it is because the value is negative and the tagging process can't handle negative values.

After that many years of market domination, I would think they could handle a simple thing like this. Just give me relative altitude as altitude which is fine on all images I checked. Or use the f#cking gps doesn't matter if it is inaccurate because this is even worse. Also the app could just give a warning to wait a little longer before takeoff, if I want (more) accurate altitude data

djiuser_mHMZ20T2b2EJ Posted at 3-9 14:07
Besides that I also have multiple (maybe all) photosets where the altitude is inverted, so photos high in the sky get a low altitude and photos low to the ground get a high altitude number. Maybe it is because the value is negative and the tagging process can't handle negative values.

After that many years of market domination, I would think they could handle a simple thing like this. Just give me relative altitude as altitude which is fine on all images I checked. Or use the f#cking gps doesn't matter if it is inaccurate because this is even worse. Also the app could just give a warning to wait a little longer before takeoff, if I want (more) accurate altitude data

the altitude is inverted, so photos high in the sky get a low altitude and photos low to the ground get a high altitude number. Maybe it is because the value is negative and the tagging process can't handle negative values.
The issue has nothing to do with inability to handle negative values.
And images shot close to the ground will always have a lower altitude in their metadata than images from greater heights.
There are two different heights in your image metadata.
You need to use the Relative Altitude, not the Absolute Altitude data.

After that many years of market domination, I would think they could handle a simple thing like this. Just give me relative altitude as altitude which is fine on all images I checked. Or use the f#cking gps doesn't matter if it is inaccurate because this is even worse. Also the app could just give a warning to wait a little longer before takeoff, if I want (more) accurate altitude data
The data you are having a problem with is not derived from GPS at all.
The Absolute Altitude shown in metadata from photos of the same subject, shot from the same spot, but a few weeks apart can show differences of 100 metres or more.
Your issue is that you are imagining that DJI's Absolute Altitude (sometimes labelled GPS Altitude), is an actual useful number.
It's not ... it's a garbage number that cannot be relied on for any purpose.