P3A Altitudes in EXIF Data..?

[Bad Andy]

I have a Phantom 3 Advanced and I'm trying to get a grip on how the altitude numbers in the JPG EXIF data are derived. Does anyone know for certain? Everywhere I read I'm seeing different answers. And yeah, I know GPS derived altitude can be shaky...

As an experiment I flew the drone at dusk with a laser measuring "tape", brought it up to an indicated 10m altitude on the tablet, and measured about 11m. Similar results at 20m. PS, not surprisingly, the aircraft freaks out a bit if you accidentally hit the bottom sensors with the laser. ;-)

Then I took a series of pics starting from on the ground (Flushing Meadow Park, supposedly 2m ASL), then 5m altitude increments up to 50m as indicated on the tablet. The EXIF data showed the following.

6.141 (on the ground)
2.241
3.258
8.058
13.358
18.258
22.858
28.458
33.258
37.958
43.558

Thoughts??

Thanks!

PS, something a bit scary... some apparent GPS longitude drift in the last shot I took. First pic was showing 73;50;11.00300000002, the numbers stayed pretty close until the last pic which was 73;50;13.09710000001. I know it's not a lot, but still, a reminder about consumer GPS I guess.

 

[Oso]

I have a Phantom 3 Advanced and I'm trying to get a grip on how the altitude numbers in the JPG EXIF data are derived. Does anyone know for certain? Everywhere I read I'm seeing different answers. And yeah, I know GPS derived altitude can be shaky...

As an experiment I flew the drone at dusk with a laser measuring "tape", brought it up to an indicated 10m altitude on the tablet, and measured about 11m. Similar results at 20m. PS, not surprisingly, the aircraft freaks out a bit if you accidentally hit the bottom sensors with the laser. ;-)

Then I took a series of pics starting from on the ground (Flushing Meadow Park, supposedly 2m ASL), then 5m altitude increments up to 50m as indicated on the tablet. The EXIF data showed the following.

6.141 (on the ground)
2.241
3.258
8.058
13.358
18.258
22.858
28.458
33.258
37.958
43.558

Thoughts??

Thanks!

PS, something a bit scary... some apparent GPS longitude drift in the last shot I took. First pic was showing 73;50;11.00300000002, the numbers stayed pretty close until the last pic which was 73;50;13.09710000001. I know it's not a lot, but still, a reminder about consumer GPS I guess.

I know the way it has been derived has changed since the P3 was first launched. That may be why you are seeing different answers. It seems that @Meta4 has always had a good understanding of this, so I copied him. Cheers!

 

[sar104]

The altitude specified in the EXIF data is the absolute GPS altitude, but the relative altitude (barometric relative to takeoff point) is in the XMP data.

 

[sar104]

Here is an example of the XMP data from a jpeg file:

xmlns:tiff="http://ns.adobe.com/tiff/1.0/"
xmlns:exif="http://ns.adobe.com/exif/1.0/"
xmlns:xmp="http://ns.adobe.com/xap/1.0/"
xmlns:xmpMM="http://ns.adobe.com/xap/1.0/mm/"
xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:crs="http://ns.adobe.com/camera-raw-settings/1.0/"
xmlns:drone-dji="http://www.dji.com/drone-dji/1.0/"
xmp:ModifyDate="2017-10-15"
xmp:CreateDate="2017-10-15"
tiff:Make="DJI"
tiff:Model="FC220"
dc:format="image/jpg"
drone-dji:AbsoluteAltitude="+3233.03"
drone-dji:RelativeAltitude="+248.10"
drone-dji:GimbalRollDegree="+0.00"
drone-dji:GimbalYawDegree="-46.10"
drone-dji:GimbalPitchDegree="-18.00"
drone-dji:FlightRollDegree="-1.50"
drone-dji:FlightYawDegree="-43.50"
drone-dji:FlightPitchDegree="+2.00"
crs:Version="7.0"
crs:HasSettings="False"
crs:HasCrop="False"
crs:AlreadyApplied="False">

The absolute altitude reference is defined in the EXIF:

GPS Altitude Ref : Above Sea Level

 

[Meta4]

I have a Phantom 3 Advanced and I'm trying to get a grip on how the altitude numbers in the JPG EXIF data are derived. Does anyone know for certain? Everywhere I read I'm seeing different answers. And yeah, I know GPS derived altitude can be shaky...

Someone was asking about this a few days ago.
Here's my reply from then: What's The Consensus on EXIF Elevation Data Craziness?
Similar photos taken on the same place as that ship photo on different daysshow altitudes that are + or - values.

Here's a graph for the values you got in your testing (including the 2m AGL)
It shows a fairly consistent difference of about 5 metres

PS, something a bit scary... some apparent GPS longitude drift in the last shot I took. First pic was showing 73;50;11.00300000002, the numbers stayed pretty close until the last pic which was 73;50;13.09710000001. I know it's not a lot, but still, a reminder about consumer GPS I guess.

I can't make sense of the position numbers you gave - I don't think you were at 73 degrees latitude 11 longitude.
But if you can give them again, I can calculate the distance between them just for interest

 

[sar104]

Someone was asking about this a few days ago.
Here's my reply from then: What's The Consensus on EXIF Elevation Data Craziness?
Here's a graph for the values you got in your testing (including the 2m AGL)
It shows a fairly consistent difference of about 5 metres

I can't make sense of the position numbers you gave - I don't think you were at 73 degrees latitude 11 longitude.
But if you can give them again, I can calculate the distance between them just for interest

Some of the ambiguity here is caused by the fact that the takeoff point is at, or near, sea level, which makes it less obvious what is being measured. In my example above, the difference is very clear.

 

[Bad Andy]

I can't make sense of the position numbers you gave - I don't think you were at 73 degrees latitude 11 longitude. But if you can give them again, I can calculate the distance between them just for interest

Thanks for the cool graph. That's just the bizarre way that Windows 7 was showing the longitude number...

I'm kind of assuming that's minutes;seconds;hundredths.bazillionths. Or something like that. But we all know what happens when we assume.

So... I uploaded the same pic (last in the altitude sequence) to metapicz.com and got the following (along with 43.56 for "absolute altitude" and 50.70 for "relative altitude"):
Latitude 40.732529 North
Longitude 73.836971 West

If you want to tell me the drift, the first pic in the sequence showed:
Latitude 40.732526 North
Longitude 73.83639 West

Oh, and the kicker:
That web site sez the first pic was actually 6.141m BELOW sea level!!! ;-) Then the second was 2.241 BSL. Finally the third was 3.258 ASL. Darn Windows...

 

[Meta4]

So... I uploaded the same pic (last in the altitude sequence) to metapicz.com and got the following (along with 43.56 for "absolute altitude" and 50.70 for "relative altitude"):
Latitude 40.732529 North
Longitude 73.836971 West

If you want to tell me the drift, the first pic in the sequence showed:
Latitude 40.732526 North
Longitude 73.83639 West

The distance between those points is 160 ft - 49 metres which is too big an error for the normal variability of GPS
Can you confirm the last 3 digits of the longitude numbers?
Are they 971 & 639?

 

[Bad Andy]

The distance between those points is 160 ft - 49 metres which is too big an error for the normal variability of GPS
Can you confirm the last 3 digits of the longitude numbers?
Are they 971 & 639?

Wow, that's more than I thought it would be. Yikes.

Those numbers were copy-and-pasted right from the metapicz.com screen, albeit with a brief stop in Notepad to make sure no formatting code got pasted into the forum post. I'll check again later (gotta do a little work) but I doubt either are wrong unless the web site screwed 'em up. If you want to see for yourself I'll put up links to the original pics when I get home.

FWIW most of my pics from the confines of that airfield are pretty close to the numbers from that first pic in the altitude checking sequence.

 

[Meta4]

Wow, that's more than I thought it would be. Yikes.

Those numbers were copy-and-pasted right from the metapicz.com screen, albeit with a brief stop in Notepad to make sure no formatting code got pasted into the forum post. I'll check again later (gotta do a little work) but I doubt either are wrong unless the web site screwed 'em up. If you want to see for yourself I'll put up links to the original pics when I get home.

OK thanks.
If you have any other testing pictures from the same area taken on different days, I'd be interested to know how they compare to the actual height.
Is the difference similar or do you get larger/smaller numbers and are they above/below sea level ?

 

[Bad Andy]

OK thanks.
If you have any other testing pictures from the same area taken on different days, I'd be interested to know how they compare to the actual height.
Is the difference similar or do you get larger/smaller numbers and are they above/below sea level ?

It seems that Windows was showing me degrees;minutes;seconds. The semicolons make for weird phrasing. And I think I'd like to take back everything I said about GPS longitude drift. Looking again at the pics, the aircraft definitely did go closer to the Flushing Meadow lake, and by roughly the distance you describe. I have no idea why I did that. D'OH!

So {embarrassed eye-roll} getting back to the topic, altitudes...

Except for the day I took those 5m steps I never went out of my way to note altitudes when I was taking pics. But going through the ground level pics I have:

A ground pic at the Forest Park airfield shows the absolute altitude as 1.968 BSL.

A pic from atop the trunk of my car at the Kissena Park velodrome parking lot shows a whopping 36.115 BSL.

And a ground pic from another day at the Flushing Meadow airfield shows the altitude as a ridiculous 46.615 m BSL.

In all cases the relative altitude was fairly sane. Zero in most cases, 1m in a couple of others.

The way I'm reading it all, the standard EXIF data is GPS -based and the XMP subdata contains GPS data characterized as "absolute". So I guess the "relative" number in the XMP subdata is just the mathematical difference of GPS data from the take-off altitude?

Thanks for making me look deeper than the EXIF data!

 

[Meta4]

The way I'm reading it all, the standard EXIF data is GPS -based and the XMP subdata contains GPS data characterized as "absolute".
So I guess the "relative" number in the XMP subdata is just the mathematical difference of GPS data from the take-off altitude?

It's most likely to be the barometer altitude that you see in the app.
As you've seen the GPS altitude can be out by a substantial amount and the amount can vary a lot from day to day.
Your numbers help me get a better idea of what the Phantom does/doesn't do.
Now I'm sure that it isn't WAAS enabled (which would make the altitude more accurate).

 

[Bad Andy]

It's most likely to be the barometer altitude that you see in the app.
As you've seen the GPS altitude can be out by a substantial amount and the amount can vary a lot from day to day.
Your numbers help me get a better idea of what the Phantom does/doesn't do.
Now I'm sure that it isn't WAAS enabled (which would make the altitude more accurate).

Get a load of this pic... from almost 55m up (AGL) and the data says the absolute altitude is almost 71 BSL.