Here’s your problem right here (Looks like #1 from my first post in this thread). No control points included in the results. The “yes” means that your images are geotagged, “no 3D GCP” means that those control points you marked in the raycloud editor were not factored in to your results because you didn’t reoptimize.
It’s also good practice to “generate quality report” after reoptimize, since that is not done automatically like it is after processing each step. If you have an issue after reoptimizing, you want to catch and correct it before you waste all that time processing steps 2 and 3.
edit: I’m confident that the comments above will fix your issue, but other observations from the quality report:
It looks like you did this with a Mavic 2 and Phantom 4 Pro? Did you process these images in separate projects and then merge them?
I see that you’re using a computer with a Xeon processor. I’m not sure if this is still true (and maybe support can chime in here), but the Xeon processor was explicitly discouraged from use by Pix4d ~5 years ago in their support docs. I had a workstation that had one and processing times were definitely longer with it than my current setup.
Your output coordinate system is using EGM96 geoid for vertical. Pix4d recommends that you use “arbitrary” as the vertical option when processing with 3D control points (unless that has changed in the last year or so)
The general consensus of the community is to use “all prior” setting under calibration options for internal camera parameters. I do this with Phantom 4 Pro flights and it definitely has improved results and/or made it easier to catch errors in ground control points. I’m not sure if this is the recommended option for something like a mavic 2 pro.
I see that you’re generating the DTM in pix4d, as well as contour lines. If you haven’t noticed this already, be cautious about using that and trusting the contours. Recommend you purchase something like Global Mapper + LiDAR module as a tool to do some manual surface building, instead of relying on the automated DTM process in Pix4d. (No critism of Pix4D here…I’ve never seen an automated tool like that work in any software for photogrammetry-derived point clouds. It always takes elbow grease).