Understanding bad_frames.npy usage

[darioringach]

Using:
Name: suite2p
Version: 0.14.4

I am crating an int64 np array shaped as (#nframes,) in bad_frames.npy
Suite2p seems to read and report the number of bad frames correctly after registration.
So it seems all is working Ok...

However, the data in Fall.mat shows very similar data (fluorescence and spikes) to what I obtain without the bad_frames (which comprise about 20% of the frames). Morever, ops.meanImg is exactly the same in both cases.

What am I missing?

Thank you!

Dario