Closed AfterJourney00 closed 1 year ago
Hi,
Thanks for pointing out this. Yes, you can just ignore this registration. Other frames are synchronized well. It is due to a small difference on frame extraction. Currently, the last frame time is set as the minimum ending time of all kinects. But when I did the registration, the last frame time a bit longer (probably the longest or average ending time). Hence I got one more frame and obtained registration for that as well. Again, the motion is usually slow towards the end of the sequence. Therefore registration is still possible with tiny synchronization problem between kinects.
Best,
Hi,
I found that the following sequences have one frame more registration result than corresponding rgbd data(e.g. both of them start at frame
t0003.000
, then registration result ends at framet0050.433
but rgbd data ends at framet0050.400
). Should I just ignore the registration result att0050.433
or any special operations should be taken to address this problem?And I wonder how the "one frame more" registration result is fitted without corresponding rgbd data?
Thanks a lot!