We’re a 3D scanning pipelined studio and with the latest 2.1.1 release, reality scan has either a bug in our pipeline or there’s something I’m missing with how RS is reading XMP camera priors.
We make XMPs from a “master” project, copy and rename those XMPs to match the new image files for processing in Reality Scan. With camera priors set to be used in the alignment settings, we align and get great alignment results. But since this update from 2.1.0 to 2.1.1 I’ve noticed that all of the scans are consistently orientated wrongly.
We have not changed anything in our RS CLI for this update, and as far as I understand even with certain CLI commands being changed or added in this update, it does not effect our scripts.
In the images below you can see the difference in orientation from RS 2.0.1 and 2.1.1.
Hello, @ryanwellence1 thank you for reporting this. There were changes made to the XMP exporter, but it shouldn’t affect older scripts. We did not experience this behavior ourselves during testing and were not able to reproduce it now on our sample full-body scan dataset.
Would you be ok if I send you a private Box link where you can upload a sample of your data, project, scripts, and any XML parameters you are running?
Perhaps if you just share the global config file, that would help us find which remembered setting is causing this. You export the Global config file from the Application seetings or by using the exportGlobalSettings command
@ryanwellence1 We plan to push a hotfix, probably next week. We already did change some of the XMP code but we have no way of veryfing if it solves the issue you experienced. If you or anyone else could provide us with a dataset and 100 reproduction steps, we could verify it.
Sorry, we’ve just been so busy this week at work I haven’t had time to change our pipeline back to get you your datasets. Soonest I could get them back is next Thursday 4/23.
