[MEncoder-users] ATSC closed captioning

[Reimar Döffinger]

On Wed, Nov 10, 2010 at 11:26:51PM -0800, David Liontooth wrote:
> On 11/09/2010 12:41 PM, Reimar Döffinger wrote:
> >On Tue, Nov 09, 2010 at 09:24:47AM -0800, Scott W. Larson wrote:
> >>Reimar Döffinger<Reimar.Doeffinger at gmx.de>  wrote:
> >>>We have three samples, two showed only half the characters, one showed
> >>>a mixture of two languages. All are working now, but it could of course
> >>>have broken something...
> >>Did you find an ATSC samples to test with?
> >Yes. And I'd like to implement the reordering by reusing the pts reordering
> >stuff we already have (see the buffered_pts stuff in libmpcodecs/dec_video.c)
> >instead of having yet another reordering code.
> >I just don't know when I'll have time for it.
> The problem with ATSC closed captioning is that it's time-consuming
> to figure out the reordering.

Not really. It is time-consuming to do it properly and resiliently,
but pts-based reordering isn't "doing it properly" anyway.

> It's true it's not optimal to have yet
> another reordering code, but the alternative -- realistically,
> judging from the past five years -- is to have no support for closed
> captioning.

The "yet another reordering code" also means keeping all the reordering
bugs that are already fixed in the other one.
Also, at that point it will _never_ be possible to implement reordering
properly (namely based on the actual reordering the decoder does).
Which also means that e.g. there is no chance that CC will _ever_ work
with files remuxed into e.g. AVI.

> Merge soon, merge often. Nobody has time to write all the code.

Nobody has time to maintain all that merged code unless you ensure
that it requires minimal maintainance in the first place, and you might
neither believe nor like it but maintainers are in even shorter
supply than developers.
Either way I now have a proof-of-concept using the other reordering
code and also updated it to allow selecting each of the 4 possible
subcc channels individually, but it still needs a lot of clean-up
to be acceptable.
Also it can't work with -correct-pts since that bypasses the
MPEG header parsing, so at some point we'll need some way to get
the user data out of FFmpeg - I guess the question is at which point
though (after parsing, or after decoding or ...?)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: atsc.diff
Type: text/x-diff
Size: 8496 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/mencoder-users/attachments/20101127/32e10bfd/attachment.diff>