Welcome, Autel Pilots!
Join our free Autel drone community today!
Join Us

Starlink Altitude vs exif data off by as much as 40% (173 ft)

ProMapper

Active Member
Joined
Mar 11, 2017
Messages
40
Reaction score
11
Location
Petersburg, TN
I could use a second opinion/test from other users. This is a quick and easy one you can do in 20 minutes from the back yard. straight up. Straight down.

I ran a test consisting of flying at 100 foot elevations and taking a photo at each even 100 foot from 100 to 600 feet from the takeoff point. And, according to the exif data I only violated the 400 foot limit by 25 feet. read on for the explanation.

The reason this is an issue is that I flew several flight lines and ended up with random gaps between lines even though I planned for a 40% sidelap. Many hours during the only clear and calm day and egg on my face!

this test showed me the reason for the gaps. according to the exif information I was off by a maximum of 173 feet or 50 PERCENT OF THE ALTITUDE vertically. the exif, or actual I assume, was always lower than the starlink readout thereby creating gaps in my photos when flying parallel flight lines. I have repeated this test twice. There does not seem to be a constant in this making it impossible to fly a mission at a given altitude. I have given a listing of the data below.

If you are willing to help out. Basically, climb to given 100 foot altitudes, 100,200, 300, 400. Let the aircraft stabilize, take 2 photos at the same altitude writing down the elevation on the starlink app screen. Return to home, load the card up on your computer, right click on the jpg file, select properties, details tab, scroll to the bottom and compare the altitude listed in the details to the one you wrote down! Let us know what you find.

I am really hoping this is a one off machine error or operator error, mine. Either way if it is an issue with all the systems out there then it is a deal breaker for me. I really like this little orange guy. In reading other forums the xstar seems to have fewer rants about quality and I can tell you it WILL TAKE A BEATING! But from a production standpoint the little things are adding up quickly. I am hoping the april release will fix the way point range and elevation issue.

photo-time-exif altitude-starlink readout-difference

108 2:25 267 103 164
109 2:27 297 201 96
110 2:27 297 201 96
111 2:29 329 302 27
112 2:29 329 302 27
113 2:30 363 400 -37
114 2:31 362 400 -38
115 2:31 393 498 -105
116 2:31 394 498 -104
117 2:33 425 597 -172
118 2:32 424 597 -173
 
Wow; not cool! Weather is shutting down here for the next few days, so I won't be able to test this. Hope others will be able to!
 
While I haven't tested at the 100ft intervals like you did, I did have some screen capture videos I could reference. My EXIF data is different than the Starlink-indicated altitude as well. A few examples:

I shot my black, white, gray cards on the ground without the motors running and the EXIF data for that shot says 220ft. But I can't recall if I had been flying prior to that shot (i.e. home point reset or not) as I didn't do a screen capture for the entire session that day.
I have several aerial shots I took at various altitudes (160 - 250ft range) where the EXIF data is 100 - 105ft higher than the actual altitude.

I have some older screen capture videos to reference but haven't had time to look at the EXIF data for those images. I'm curious if this issue is related to the recent Starlink update.
 
I have a couple photos taken at different locations, both at exactly 100 feet as displayed on the app. The EXIF data shows 92 feet altitude on one and the other shows 87 feet altitude. Not as bad as the variances you and GA-XSP are seeing (and I'm glad the app is over-reporting altitude rather than under-reporting) but still a little concerning. Thanks for sharing.
 
With our new Regs here is Canada I have mine set to 300 feet so I'm not breaking the tone of rules we now have. My last video I downloaded the telemetry data and used it with Dashware to display my speed, heading and alt.
I was flying right at 300 feet according to the starlink app but the telemetry data display for alt. was 28 feet lower. I'm guessing it is getting this info from GPS? Correct me if I'm wrong. If I'm right then that would account for not being right on the money. I think they say GPS can be off as much as 32 feet.
 
I shot my black, white, gray cards on the ground without the motors running and the EXIF data for that shot says 220ft.
Does this mean that the EXIF data is wrong and the app is correct? I would hate to be flying lower than I thought.
 
With our new Regs here is Canada I have mine set to 300 feet so I'm not breaking the tone of rules we now have. My last video I downloaded the telemetry data and used it with Dashware to display my speed, heading and alt.
I was flying right at 300 feet according to the starlink app but the telemetry data display for alt. was 28 feet lower. I'm guessing it is getting this info from GPS? Correct me if I'm wrong. If I'm right then that would account for not being right on the money. I think they say GPS can be off as much as 32 feet.

From what I have just read on a aviation site the difference between GPS and BAROMETRIC altitude can run 5% depending on temperature. I believe that is what you are referring to which does not apply here. We are seeing a difference in display between starlink and exif. I have requested and received the sensor specs on the camera and I believe I can calculate the image scale at the point of the sensor and thereby determine the actual altitude. Not sure I can get that done this evening or not.

My problem is that I need to be able to calculate a given altitude which relates to the amount of ground covered in one photo. Specifically across the flight path. I then use that distance to determine flight path spacing to include overlap and thereby not having any gaps. If I think I am flying higher than I really am then I will have gaps and be very embarrassed with friends and those that asked for the photos.

Each of these examples should be forwarded to support via email so they know the problem is not random and gives them more info to use in the problem.
 
While I haven't tested at the 100ft intervals like you did, I did have some screen capture videos I could reference. My EXIF data is different than the Starlink-indicated altitude as well. A few examples:

I shot my black, white, gray cards on the ground without the motors running and the EXIF data for that shot says 220ft. But I can't recall if I had been flying prior to that shot (i.e. home point reset or not) as I didn't do a screen capture for the entire session that day.
I have several aerial shots I took at various altitudes (160 - 250ft range) where the EXIF data is 100 - 105ft higher than the actual altitude.

I have some older screen capture videos to reference but haven't had time to look at the EXIF data for those images. I'm curious if this issue is related to the recent Starlink update.

WOW, this is the most disturbing. Sitting on the ground and the aircraft thinks it is at 220 feet? or does it?
GPS elevations are based on ellipsoid height which are different all over the world. BUT, we are not dealing in absolute elevations in reference to sea level but merely a reference point elevation.

Questions for anyone to answer

1. I wonder how long it is taking the aircraft to calculate a good solution? It is well known that the longer you sit on a single point the more accurate the elevation is going to be and the horizontal accuracy will be WAY good! with 20 satellites the [position should not be an issue!

2. Could we be taking off before the gps is getting a good solution?

3 where is the ground elevation coming from. The aircraft gps or the starlink tablet et al?

New test. with the aircraft on the ground and motors running take a photo every 30 seconds BUT DO NOT move the aircraft. We want to know if the elevation in the exif changes over time even if on the ground.

This test I think I can do this evening. will report if I do.
 
EXIF data is not the XSP's strong point. If nothing changed, 4K footage doesn't even include any. Either way, GPS does not tell the quad it's altitude. Altitude on all prosumer UAV's is determined by the barometer. It may also be showing height above sea level..
 
Here are some of mine taken a few weeks ago. For me the starlink is correct and EXIF is wrong.
These were at lower elevations. No way was I at 200 plus feet.
exif altitude starlink readout
218 11
219 12
223 27
227 41
231 53
229 50
220 19
 
This is what I have been able to dig up from my screen grabs...

EXIF / Starlink Altitude
277 / 79
277 / 79
259 / 9
258 / 9
291 / 123

282 / 180
283 / 181
282 / 181
281 / 181
280 / 181
280 / 180
280 / 181
279 / 176
279 / 176
279 / 176
280 / 175
279 / 188
279 / 189
220 / 0 (motors not running)
278 / 173
341 / 374
342 / 375
342 / 370
268 / 145

Altitude @ home points where this data was collected was between 730 and 840 feet MSL.
Blue data is from January; the remainder is from the beginning of April.
 
Last edited:
EXIF data is not the XSP's strong point. If nothing changed, 4K footage doesn't even include any. Either way, GPS does not tell the quad it's altitude. Altitude on all prosumer UAV's is determined by the barometer. It may also be showing height above sea level..

good information to know but how is the barometric altimeter adjusted for pressure which changes constantly. You are correct in that most barometric altimeters are adjusted for the local pressure and thus read out in sea level datum.

If memory serves me the native gps data is in meters and should be adjusted by the local geoid to get a reference to sea level or for that mater any local datum you care to use.

Keep in mind that instead of using sea level for all elevations they elected to use a datum at the home takeoff point. So whatever method they are using it has to be adjusted yet again to local home elevation.

We have one major issue. The aircraft and starlink and exif do not agree.

I am hoping others follow suit and run some good testing.

thanks for everyone's input.
 
Been doing some google surfing and found this page. Some good reading on it. Did not know that even prop wash can have a affect on the readings.
What’s Your Altitude?

All good stuff but so far no one has been able to tell why the exif data from the photo file does not match the readout on the starlink display.

The ONLY place that the starlink can receive the altitude of the aircraft is from the aircraft but it is irrelevant to this issue what system is being used. We have seen several reports on this thread that this is not an isolated issue so I for one would like to know if this is a fix that will be coming in the late April droid release.

Since Autel is so adamant on us not breaking the 400 foot elevation barrier they have hard coded the max elevation into the waypoints. I would think they would be as adamant about getting it right. We have read examples of the error being lower than reported to the starlink display and have read the opposite. So I guess there is really no way of knowing how high we are flying.

The ramifications of this should be self evident. Perhaps users of the Xstar Premium have already experienced the negative effects of this by watching the aircraft crash when they thought they were doing everything right. Of course the pilot in command has the ultimate responsibility but ...

I'm confident there is a fix and sometimes the fixes are quite complex so I guess I like others will have to just wait it out but ...Please hurry with the fix. :)
 
  • Like
Reactions: Wrsdas
I could use a second opinion/test from other users. This is a quick and easy one you can do in 20 minutes from the back yard. straight up. Straight down.

I ran a test consisting of flying at 100 foot elevations and taking a photo at each even 100 foot from 100 to 600 feet from the takeoff point. And, according to the exif data I only violated the 400 foot limit by 25 feet. read on for the explanation.

The reason this is an issue is that I flew several flight lines and ended up with random gaps between lines even though I planned for a 40% sidelap. Many hours during the only clear and calm day and egg on my face!

this test showed me the reason for the gaps. according to the exif information I was off by a maximum of 173 feet or 50 PERCENT OF THE ALTITUDE vertically. the exif, or actual I assume, was always lower than the starlink readout thereby creating gaps in my photos when flying parallel flight lines. I have repeated this test twice. There does not seem to be a constant in this making it impossible to fly a mission at a given altitude. I have given a listing of the data below.

If you are willing to help out. Basically, climb to given 100 foot altitudes, 100,200, 300, 400. Let the aircraft stabilize, take 2 photos at the same altitude writing down the elevation on the starlink app screen. Return to home, load the card up on your computer, right click on the jpg file, select properties, details tab, scroll to the bottom and compare the altitude listed in the details to the one you wrote down! Let us know what you find.

I am really hoping this is a one off machine error or operator error, mine. Either way if it is an issue with all the systems out there then it is a deal breaker for me. I really like this little orange guy. In reading other forums the xstar seems to have fewer rants about quality and I can tell you it WILL TAKE A BEATING! But from a production standpoint the little things are adding up quickly. I am hoping the april release will fix the way point range and elevation issue.

photo-time-exif altitude-starlink readout-difference

108 2:25 267 103 164
109 2:27 297 201 96
110 2:27 297 201 96
111 2:29 329 302 27
112 2:29 329 302 27
113 2:30 363 400 -37
114 2:31 362 400 -38
115 2:31 393 498 -105
116 2:31 394 498 -104
117 2:33 425 597 -172
118 2:32 424 597 -173

A Google search using "Altitude vs. exif data" quickly shows that this problem is not exclusive to the X-Star, as the DJI Phantom Forums have been hotly debating it for years. I will be keeping my X-Star and just fly it as sensibly as possible.
 
  • Like
Reactions: Wrsdas and kgam9456
Interesting issue I've not paid any attention to. From my real flying days we would set the altimeter on the runway so the reading was MSL not AGL. There is an answer somewhere in all of this. From what I've seen there does not seem to be a way to reset the barometer for a good local reference. Between a good barometer reference and GPS data we should be real close as to what the real flying AGL is within the drones system. Really doesn't mater that much as the flying rules are still VFR, but it would be nice to use a reference closer to what other aircraft use and maybe even more accurate.
 

Latest threads

Members online

Forum statistics

Threads
11,228
Messages
102,655
Members
9,818
Latest member
redwingaerials