terrestrial photos produce point cloud rotated 45 degrees from horizon

Hello-
I used an iphone 12 to manually capture photos of a rocky area that happens to have a survey marker embedded in it. GPS is turned on. I walked around the rocks and took photos with estimated 80% overlap.
I imported the photos into pix4d mapper and processed the ray cloud & point cloud. The point cloud is rotated appr. 45 degrees from the horizon.
When I open the image display to mark MTPs, all the photos display in landscape orientation rather than portrait orientation the way I captured them.
I created an orthophoto to check the GPS accuracy of the photos. They are within 1.4 m of the published coordinates of that survey marker. I do not have GCPs.
I need to create a 3d model with these photos. I’ve created the 3D model and the texture and resolution look great, but I also need it to be rotated correctly so I can export it into other apps.
I’ve tried capturing imagery with the camera in landscape orientation, and adding MTPs but that didn’t help. I also tried adding an orientation constraint with no success.
Any suggestions?
I’m attaching the report and a screenshot of the 3D model.

Thank you.

Elk_Rock_Island_Survey_Marker_report.pdf (241.2 KB)

Hi akhillyer,
Can you provide the quality report for this project? We can take a look.

I might also suggest looking at our PIX4Dcatch mobile application. It is free and runs on iOS and Android. It will automate much of the data collection and ensure that you have proper overlap. If you do not have a PIX4Dcloud license, you can extract the images either with a dongle or upload to a shared drive.

I added the attachments.

Thanks.

You might want to play around with the calibration method. Try using the Accurate Geolocation and Orientation method.

Are you able to return to the site and scan it with PIX4Dcatch?

I tried several combinations of calibration parameters. The Accurate Geolocation and Orientation Method is not available. It gives the message below.

I switched the lat and long coordinates, which produces a level point cloud. Only problem is, it’s upside down. Pix4D doesn’t recognize the vertical coordinate system.

I might try adding some rough image point GCPs.

I will try going back to the site and collecting with Pix4Dcatch.

I do believe PIX4Dcatch will make this process much easier.

However, for this project, you might just need to incorporate scale and orientation constraints.

Thank you.

Will let you know when I refly. (Waiting for a break in the rain.)

image001.jpg

image002.jpg

image003.jpg

image004.jpg

image005.jpg

image006.jpg

image003.jpg

I collected the data again using Pix4Dcatch and it worked great!
That’s a really nice product.
Thanks!