[mythtv-users] mythcommflag in 0.17 better and worse

[John Patrick Poet]

On Wed, 16 Mar 2005, Mark Matula wrote:

> I'm using the "All" method of detection and have sine like .11 or so.
>
> Commercial detection seems better at avoiding false-positives in the middle
> of shows and no longer seems to cut up shows like Law and Order with a lot
> of dark scenes that seemed to throw off detection in the past.
>
> Two thumbs up.
>
> But I noticed that on sitcoms that have that last minute or two of show
> after squeezing in one last set of commercials, (e.g. Everybody Loves
> Raymond), Myth is skipping right off the end of the show at the last
> commercial break.
>
> One thumb down.
>
> I've tried setting 'strict' detection both ways (offhand I don't recall the
> wording  - if 'enabled' means more aggressive or less - but I've tried
> both).  I'm still finding myself missing that last bit of show.  Is anyone
> else seeing this?  Would one of the other detection methods avoid this?
>
> Thanks.  (It's great when my only complaint is about a tiny fraction of a
> show).
>
> --
> Mark Matula
> Mikey, you will be missed:
> http://eblo.ratula.net/mikey.html


When the "ALL" option was first added, commercial detection was the best it
has every been.  It was amazing how accurate it was for me with content from
my HD-3000 cards.

Since then it has gone down hill a bit.  I believe that the concept of
"black" had to be changed to help it work better for some poeple.  It still
works very well, but it is not as good as it briefly was.

I have thought about backtracking that part of the code to the way it was
when the ALL option was first added, but have not had the time.


John