Hello, I’m getting the same CUDA error. My goal is to run an alignment on the data gained from 360camera video. My current workflow is as following:
Initially i cut equirectangular images into 6 cube sides, align the fronts in Realityscan, export xmps with the draft position and orientation and create missing xmps for the rest of the “cube” with respective orientation values - as described here Reality Scan Image undistortion steps 1-4. Nevertheless the alignment quality is rather low as the cameras treated as separate cameras.
Therefore i’ve added rigs to xmp in hope that it would improve the quality of the alignment as the cameras with the same position will be tied together. Sadly the program crashes with the “position and orientation” alignment settings; with the “locked pose” setting i’m getting the CUDA error.
Do you maybe have some insights about this bug or tips for using workarounds for 360-videos?
Best regards,
Daria