Altitude issue with GCP

@Andrew. I launch the test later in the afternoon (I have to make a copy, install new version, ecc.). Hope you’re right however I do not explain the different behaviour using two different software versions.

I let you know later

Dear pix4d. We ran into the same exact problem.
We figured out that 4.0.25 and 4.1.21 versions have some issue with using GCPs.

I got a linear area, a highway. About 6km long and 300m wide. I have 6 GCPs along the full length. Based on APM log i ran the initial processing so i got a geolocated project but the accuracy is low. I calibrated the area with GCP-s and i saw that i got around 30m height difference.
After running the ‘Reoptimize’ or ‘Rematch and optimize’ steps i got a result that my GCPs are merged with the surface, but the height data is wrong.
For example, i know that my GCP height is 90m but the next point right by the gcp got 52. With zero visible height difference.
Even the GCP has weird data. It says in  X Y Z data that my heigt is 90m, below i can see: Initial position with Z 52m, and computed position with 52m.

So as i see, pix4d is bringing the GCP to the surface, not the surface to the GCP.
I re-ran the project on 4.0.21 version with exact the same steps and everithing ran as would expected.

We got a commercial licence by the way.

Dear csomy,

Out of pure curiosity would you share your loacation of the survey (country/state/city) and the vertical reference systems you used for your GCPs and the corrosponding geodetic settings in Pix4D (coordinate systems)?

Best regards,

Boris

1 Like

Dear Boris
The location of my survey is approx:
N 46.771140°
E 20.713026°

I did a conversion on the standard wgs ardupilot coordinates to the hungarian standard EOV. 
I set my image properties to use this koordinate system. My GCPs are in EOV too.

I think that the koordinate system is not the real problem here. I got a calibrated height that is not used, but moved by pix4d.

Same problem that Massimiliano had. Sorry, right now i cannot present any screenshots. Maybe in the first days of the week.

I did all the same steps on the 4.0.21 version and that calibrated my surface right to the GCP.

Regards: Gabor

@Massimiliano, did you have a chance to set the Ground Control Point Vertical Coordinate System and Output Vertical Coordinate System to Geoid Height Above Bessel 1841 Ellipsoid [m]: 0.000? If you did, please let us know if updating the vertical coordinate systems fixed the problem. If updating the vertical coordinate system did not fix the problem, please share a screenshot of one of your ground control point’s properties.

@Csomy, please verify that your Ground Control Point Vertical Coordinate System and Output Vertical Coordinate System are correct. If you do not know which vertical coordinate system to choose, I recommend applying the second option, with zero meters above the ellipsoid. You can learn more about managing your project’s Ground Control Point Coordinate System and Output Coordinate System at https://support.pix4d.com/hc/en-us/articles/202560029#label2. Don’t forget to Reoptimize after updating both coordinate systems.

If a geoid offset is applied to either your ground control points or your results, I expect that your elevations will be incorrect by approximately 42 meters in the Z axis because of your project’s location. If changing your Ground Control Point Vertical Coordinate System and Output Vertical Coordinate System do not fix the problem, please share a screenshot of a ground control point that is similar to Massimiliano shared on December 05, 2017.

Hi Gabor,

Can you check if you get the same results as below when converting your WGS84 coordinates to EOV (HD72, GRS67) coordinates:

Did you consider the vertical difference between the systems caused by the different ellipsoids?

Best regards,

Boris

1 Like