I noticed that save_movie() was creating movies slightly shorter than the desired length -- a 6 second movie would become a 5.8 seconds movie, for instance.
I think this is primarily an issue in ffmpeg which would be almost impossible to solve, but I also noticed a minor floating point instability in the calculation of the number of samples in my data.
Coverage remained the same at 72.45% when pulling a8ee442622feada19c9e9d9dd944a7c9df0f4d9c on aestrivex:fix_sample_floatingpoint_offby1 into 330cbe1aec70219a1f903bb6936263e888aeb174 on nipy:master.
I noticed that
save_movie()
was creating movies slightly shorter than the desired length -- a 6 second movie would become a 5.8 seconds movie, for instance.I think this is primarily an issue in ffmpeg which would be almost impossible to solve, but I also noticed a minor floating point instability in the calculation of the number of samples in my data.