[mythtv] [mythtv-commits] Ticket #8522: Include only half the blank frames between commercial breaks

[Jim Stichnoth]

On Mon, Sep 13, 2010 at 10:25 AM, Chris Pinkham <cpinkham at bc2va.org> wrote:
> * On Mon Sep 13, 2010 at 10:14:12AM -0700, Jim Stichnoth wrote:
>> I'm not sure exactly how the (let's say) 10-second upper bound should
>> be applied.  Suppose the commflagger analyzes a recording like this:
>>
>> 0:00:00 Start of program segment
>> 0:10:00 End of program segment, start of blank interval
>> 0:11:00 End of blank interval, start of commercial
>> 0:15:00 End of commercial, start of blank interval
>> 0:16:00 End of blank interval, start of program segment
>> 0:30:00 End of program segment
>>
>> Using the 1/2 blank interval, the commflagger would set a commercial
>> break from 0:10:30 through 0:15:30.  What is the intention if the
>
> With the 10-second max, I'd say the commercial break would be flagged as:
>
> 0:10:10-0:15:50
>
> So we include 10 seconds at the beginning of the blank space that starts
> the commercial break and 10 seconds a the end of the blank space that ends
> the commercial break.

Good, that would be my preference as well.

By the way, I'm thinking of defining the maximum in terms of number of
frames, rather than seconds.  Something like 240 frames would give
4-10 seconds at 24-60 fps.  I recall discussion of keeping track in
the DB of frame rate changes, but since those probably happen at
commercial boundaries, it's probably better to play it safe until the
code stabilizes.

>> I'm happy to update the #8522 patch to eliminate the setting and use a
>> single approach.  But first I'd like another 2-3 weeks to test, since
>> the Fall TV season is about to hit full-force, and the summer TV
>> season didn't really allow for adequate testing.
>
> Sure, take your time.  Unfortunately we got at this a little too late for
> a feature freeze, and I can't justify this as a bug fix, so it will have
> to wait until after 0.24 is released in a few weeks.  Thanks for looking
> into this and for being accomodating.

No worries.  In any case, I still think it needs more time for
adequate testing.  I wouldn't mind getting more eyes on it, especially
by people who use detection methods besides BuildAllMethodsCommList().

Jim