Hello everyone, sorry, I’m using google translate because my english is not really the best. I am having a lot of difficulty in carrying out photogrammetric surveys using DJI drones since the height value extracted by pix4d from the exif data of the photos is the ellipsoid height and not the orthometric one. I state that I use ground targets that are measured by rtk rover.When I upload the photos of the flights, as a first step, I have all in all the ok of the steps (green icon) as a quality report but no 3dg(of course, I haven’t entered them yet ) . When I go to insert the ground control points with orthometric altitude and after confirming them on the raycolud with more than 5 photos each, after having redone the optimization, the error is much higher than the gsd and therefore on the quality report I have a red icon on 3DGCPs. Can anyone give me some suggestions?
Buongiorno Mario,
Thank you for the detailed explanation of your workflow.
Could you ensure that the vertical coordinate system defined for your GCPs and for your outputs is set to arbitrary? You can find more information at the following links:
Pix4Dmapper, unlike Pix4Dmatic, doesn’t support local geoids, therefore one way to define a local geoid as a reference is to set the vertical coordinate system as arbitrary.
Let me know if you have any questions. If you still experience a shift, please attach the quality report of your project. Thank you 
Buongiorno Alice,
sorry for question but…if I use “arbitrary”…wath is the output coordinate system?
Mario
Ciao Alice…
Flight: free hand fly due the building. camera optic:fixed
I’ve follow your tips but…nothing change : red icon on 3dgpc
Have you any ideas?
Thank’s in advance
Mario
Ciao Mario,
arbitrary systems are used for coordinate systems whose origin, scale and orientation are defined by the user. When you use this option no coordinate conversions are applied and the software will try to adjust the results to the GCPs positions to match them. So in this case, we can say that the output coordinate system will be the same as the one of your GCPs.
I would suggest to use this option only for the vertical coordinate system of your GCPs and your outputs. For the horizontal one, you can still choose to define the coordinate system with one of the available methods.
There is a high projection error on GCP4 (more than 5 pixels) and QCP1 (more than 1 pixel), so please ensure these points were marked correctly on the images.
Please also notice that zoom cameras are not suitable for photogrammetric applications, as having a focal length changing during the acquisition is not optimal. Despite you said that you used the option “camera optic:fixed”, Pix4Dmapper recognized 2 different camera models with different focal lengths and optimized parameters:
- EXIF ID: FC2204_4.4_4000x3000
- EXIF ID: FC2204_4.5_4000x3000
I would suggest using a different camera if possible or strictly ensuring the focal length does not vary during the acquisition.
Ciao Alice,
thank’s for replay.
I will try to recalculate the work following your suggest. Regarding the focal length issue, you are right. The flight I had to do in ATTI mode due to the lack of stable GPS signal and inadvertently I changed the focal length of the optics in 3 photos. Normally I do the photogrammetric flight automatically by setting a lens at 30mm so as not to have too much distortion. This time, however, I could only fly freehand and had to keep the drone under control while it flew to avoid crashes.
Thank’s again
Mario
Ciao Alice,
sorry for the inconvenience. I followed you indication:
output system xy wgs84 utm32, z arbitrary (file capture.jpg). GCP system wgs84 geographic / metric (i’ve made two test with same result) and arbitrary for the z axis (capture1.jpg). Once this is done, the problem is that the GCPs, in the raycloud menu (capture2.jpg), are disabled and I have found no possibility to reposition them. Did I skip that step?This is an automatic flight, done with a fixed lens, which I had done the same day and I needed to have the overall view.
Mario
Hi Mario,
Could you please send us the following from the latest project?
▸ The quality report (.pdf format): …\project_name\1_initial\report\project_name_report.pdf
▸ The project log file (text file): …\project_name\project_name.log
▸ The project file (.p4d): …\project_name.p4d
Best,
Nikoleta
Hi Niloleta,
solved the gcp missing problem : I’ve uninstall and then reinstall the system.My be some registry failure. Now,I’ll remake the job.
Mario
Ok, now the icons are green while that of the 3DGCP is yellow but within the limits (unfortunately the GCP4 gives problems but I already knew it at the start). However I have a rms of 0.03 … more than happy! So if I understand correctly, in this way the X and Y coordinates are georeferenced correctly, while on the z axis we refer to the GCP value. Right?
Hi Mario,
I believe your understanding is correct, if you use GCPs in your project the the outputs should align with the GCP position.
Let us know if you have any other questions regarding the project,
I apologize if I reopen this post … but I wanted to continue on this because it is inherent to the problem.
I carried out a survey with a mav 2 pro, with fixed AGL altitude, on mountain terrain. I set the coordinates of the reference system z to ARBRITRARY in order to use the Z coordinates of the GCPs. Everything worked well as it was suggested in the previous posts. I did a second test with this new survey. In the moment in which I import the photos of the drone, I set the reference system of the Z axis of the photos to ARBITRARY, therefore no longer the deafult (EGM96). I got a marked improvement in accuracy: I went from 2 cm to 1 mm. My question is: is it correct to force the Z axis of the drone photographs (wgs84) to arbitrary as well?
thanks and good continuation
Mario
Hi Mario,
Can you attach your quality report after you have set your GCPs? This will help us in understanding what might be going on?
To answer your question, you should not need to set the vertical coordinate system of your photos to arbitrary. As long as you have GCPs and they are correctly marked then your final outputs will match your GCP elevations.
Hi Mike, I’m trying to figure out which is the best setup, but unfortunately the very long processing times don’t allow me to do, as it should be, only one correction at a time.
These are the reports I posted with the other question you are aware of. accuracy has increased by an order of magnitude, especially on the QCPs.
The macroscopic change, putting the z-axis of the photos on arbitrary, is that the position of the initial camera is the same as the final one (blue vs green).
Surely there is no “perfect setup”, I’m trying to understand the behavior of the various options.
thanks again for the support.
Mario
nadirale_old_setup_wrong.pdf (441.9 KB)
nadirale_new_setup_ok.pdf (406.6 KB)