Andrey The time algorithm seemed quite accurate, and considering the massive amount of deeply nested data I threw at it, calculated in a respectable amount of time. (Not at all like some time prediction algorithms…)
I’m guessing the timer defaults to the “Done!” message when it doesn’t have a prediction? Makes sense, since usually for most users, once there’s no time left, it is done.
To account for the users like me who like to upset algorithms, instead of
IF prediction=null, display "Done!"
you might consider adding a pause and a second check:
If prediction=null, display ""
wait X sec
if prediction=null, display "Done!"
Set X to allow enough time for the algorithm to collect its wits and come up with a new prediction.
On the topic of the option for videos:
I added another library today and the behavior is the same as at initial setup. Once you’ve chosen the folder from the OS prompt, you’re returned to the TF ‘add library’ dialog. The current behavior does not have any of the libraries selected, just as it is when you first open that dialog. But if instead it were changed so that the most recently added directory was automatically selected, the user would see the toggle options for videos and faces, and thus be able to make those selections at the time of choosing the libraries. (It might not hurt to also add a note here that the videos will take considerable additional time!)
An alternative or perhaps supplementary idea might be to index the images first, and then the videos. This would provide “instant gratification” for users eager to get started with faces and such.