blue and green dots are not together after processing

 Hi there, 

I have run image in large scale >2000 images for a river survey. I have loaded all the images all at once with the flight plan using Pix4D capture. and DJI Phantom 4 advance. All the elevation I set up to 80 meters height. after the processing I have some sections jumped with some of the blue and green dots are not in the same position. is there any alternatives to bring back blue and green dots together into its “correct” position? how would I solve this?

Dear Jose

That looks like a very interesting project! My initial suggestion would be to set the Internal Parameter Optimization to AllPrior (Processing Options > 1. Initial Processing > Calibration > Camera Optimisation >  Internal Parameter Optimization) and re-run initial processing.

We have noticed this issue in the past with perfectly nadir flights and flat terrains. The AllPrior option relies more on the initial camera positions (blue dots) during optimization and hence it would set the computed positions (green dots) to where they should be i.e. the blue dots. 

Hope this helps. We would be keen to know if this fixes your issue.

Best regards

Amritha

3 Likes

Dear Amritha,

Thanks for the information and suggestion. I follow the instruction and it works perfectly the noisy images. 

 

appreciate your help

Hi, I have got the same problem but I could not solve it in the same way?
could you help me?

Assignment_report.pdf (1.4 MB)

Hello, as far as the model is aligned with the GCPs, most probably, there is no reason to worry.

Most probably, your images’ geolocation has a vertical error that is fixed adding the GCPs.
Another possible cause of the shift is that the vertical coordinate system is not set correctly in Pix4Dmapper. What is the images’ and GCPs’ vertical coordinate system?

From the quality report, it seems that the vertical coordinate system both for the images and the GCPs is the EGM96 Geoid. Do you confirm that this is the case in reality?

1 Like

Dear Christina, thank you for your response and help.
The GCPs coordinate system is EPSG:28992 RD- NEW Netherlands, also I sit the coordinates of the images as default WGS 84 (EGM 96 Geoid), however, the images are not geolocated so I added the GCPs manually with well-measured and accurate x,y and z.

I’ve tried different coordinates but still the same error, and also I continued the process till the end I could not see goo result.

Hello,
Thank you for sending the additional information.
Your images are geolocated (you can see if from the quality report, the rayCloud and the properties of the images).

Unfortunately, I do know know whether you need to select the EGM 96, the EGM 2008, another geoid or an ellipsoid for your images and your GCPs. This information needs to come from you. You need to “tell” Pix4Dmapper what is the geoid or ellipsoid of your the images and GCPs.

If the information about the coordinate system is not inserted correctly in Pix4Dmapper, you may see this vertical shift. However, as I mentioned in my first email, when the model is ajusted well to the GCPs, you most probably do not need to worry.

I continued the process till the end I could not see goo result.

What do you mean by “goo result”?
Thank you

Thank you for your quick response I appreciate it, the coordinates for the GCPs are EPSG:28992 RD- NEW Netherlands, I’ve tried EGM 96, EGM 84 and EGM 2008 as well but I got the same vertical shifts with different values also the finial result and output like the orthomosaic there was some distortion as you can see in the screenshot


I’m trying to see where is the source of the error to handle it somehow but I can not find it out.
with regard to " I continued the process till the end I could not see goo result." I meant that I’ve done the 3 steps 1-initial process, 2- point cloud and mesh, 3- DSM.
but nothing changed. still the same error in all the steps.

Hello,
It may be that your images are in a different geoid or the WGS84 ellipsoid:

The important thing is that teh GCPs are defined correctly and the model alligns with the GCPs.

The biggest distortions seem to happen at the borders of your project, something expected as the image overlap is lower. The trees are also expected to give some noise:

I hope the above information helps :slight_smile:

I got this error after completing initial step, these images from Rededge P sensor