Dense Point Cloud Creation

Can anyone tell me why this dense cloud has turned out the way it did? Has anyone ever seen this before? The first image shows the results of step 1 where it triangulated nicely. I then went to create the dense point cloud I ended up with the results in image 2. Very accurate GCPs were used in the project which were in an arbitrary (local) coordinate system. I have run the process again with a subset of the photos but this time without GCPs and it worked fine so I’m guessing the problem is there somewhere. However, the GCPs were traversed in and I’ve checked for common mistakes such as transposing E & N. Any advice on what went wrong would be appreciated. 

Looks like your model sprang some leaks!

Curious, what the quality report says?

It looks like there was an error in the transformation from the local coordinate system to the source. I loaded the GCPs after step 1 and then optimised. It gave me the warning below in the log. Is this why the dense cloud failed and is there a way to fix the issue?

"[Warning]: Warning w9020: GDAL error <1> No PROJ.4 translation for source SRS, coordinate

[Warning]: transformation initialization has failed"

If you have the time, can you reprocess without using the EXIF meta-data and set arbitrary for datum…but keep your GCP’s.

May need to start a new project just so its a new processing chain and result, and doesn’t affect current job.

I did as you said and it wotked. Many thanks for that. It was clearly something to do with transforming between the arbitrary datum and the native WGS84 geotags. If  anyone knows why there was a problem I’d be interested to hear.