Errors in Pix4D Mapper processing X, Y & Z camera orientation

I’ve come to learn that Pix4D mapper does not correctly process X, Y & Z coordinates in geotag data for Omega, Phi and Kappa values.

I first noticed this when I used a logging device with an IMU. The resulting camera orientations from the geotags were about at right angles to the calibrated orientations. I thought that there might have been a problem with the IMU or the logging device.

On my earlier Pix4D projects, my geotags were using MSL altitudes instead of Ellipsoid altitudes. The MSL altitudes were about 30 meters off from the calibrated altitudes. Perhaps for that reason Pix4D ignored the X, Y & Z values in the geotag.

But when I loaded geotags with proper Ellipsoid geotags, the X, Y & Z were processed - and were again about 90 degrees off from the calibrated camera orientations.

I’m afraid to run steps 2 and 3 with these erroneous camera orientation values. (takes over 24 hours to process.)

All I can think to do is simply remove the X, Y & Z values from the geotags so that Pix4D doesn’t get confused by them.

The best result would be to have Pix4D properly convert the X, Y & Z coordinates to the Omega, Phi and Kappa values.

Is this a known bug? A new bug? Or am I simply making a mistake in my procedures.

I’ve attached two screen shots that illustrate the problem. The first is where the geotag used the MSL altitude - and the X, Y & Z values were ignored.

The second is where the Ellipsoid altitudes were used - and the Omega, Phi and Kappa values are wrong.

Hi @joe.stroup,

Could you please send us the following?

▸ 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


Hello @Nikoleta_Dimopoulou -

I’m sorry that I’m so late at replying to your response.

Since I posted my question I’ve come to learn the answer to this phenomenon.

I’ve learned that Pix4D will only convert X, Y and Z camera orientation to Omega, Pi, Kappa values if they are included on the image EXIF. In my case example, I provided X, Y and Z camera orientation values on a CSV file - read after the image files were identified.

It’s not easy to insert X, Y and Z values into a batch of Image files - but I’ve found one way that I’m told will work.

The best solution of course, is for Pix4D to not require these values in the EXIF. Other software, Agisoft Metashape for example, handles this properly.

Once again - I apologize for my late response.

Hi Joe,

Thank you for your response. No worries at all about the late reply.

There are a few points I hope to shine some light on and clarify if I may. :slight_smile:

The way PIX4Dmapper works with the geotag data is how much should PIX4Dmapper take into consideration geotag’s accuracy to determine how to best run step 1 keypoints matching and camera calibration. Under normal circumstances, It would not ignore the geotag by default.

X, Y & Z coordinates represent the position, whereas Omega, Phi, and Kappa values represent the camera rotation.

I am not quite sure what you meant in the first post by "converting the X, Y, & Z coordinates to the Omega, Phi and Kappa values. For more information about this, please feel free to give it a read on this article: Yaw, Pitch, Roll and Omega, Phi, Kappa angles.

In terms second post about not requiring the values in the EXIF data
Although not all information in the EXIF is necessary, Pix4D products read the EXIF information to create a photogrammetric project that can successfully process the calibration, certain EXIF metadata is required in the images. Other metadata are optional but improve the results. Feel free to refer to this when you have a chance.

EXIF and XMP tag information for project creation

Thank you for your feedback and for sharing your data with us. It is always valuable for us to have the opportunity to hear from our users and receive feedback. :pray: Have a good day. :slight_smile:

Best regards,