[FFmpeg-devel] [PATCH] avdevice/decklink: adjust for timecode lag

[Ilinca Tudose]

Hi Marton,

The issue with the out of sync TC was reproducible on all tapes and decks
that we tested. I don't have the exact number now, but a few dozens, less
than 100. They all had between 7 and 17 frames out of sync. We were not
able to obtain anything more in sync than 7 frames.

The TC sync was tested by setting up the deck to "burn" the TC with the
image while capturing the content with TC through ffmpeg. We then play the
file in a player that supports timecodes and compare the burned-in TC with
the one captured in the metadata.

We used Decklink quad 2 and several Sony decks: J30, J3, JH3. We tested on
multiple decks from each model and confirmed the issue was present + that
Gyan's patch seemed to fix it. We have used several types of Betacam tapes
and HDCAM tapes. I can not comment on whether this is the best solution,
but can confirm it works.

Let me know if you need more info.

Thanks,
ilinca

On Fri, Aug 9, 2019 at 1:08 PM Gyan <ffmpeg at gyani.pro> wrote:

>
> > On 28-07-2019 01:37 AM, Marton Balint wrote:
> >>
> >>
> >> On Fri, 26 Jul 2019, Gyan wrote:
> >>
> >>>
> >>> Patch supported by and tested at Google.
> >>
> >> Seems like a huge hack for a decklink issue... ;)
> >>
> >> I wonder if there are better approaches to work around this. Like
> >> dropping all frames until we have a valid signal.
> >
> > Let's say the scenario is like this,
> >
> > Frame #            TC                 Status
> >      1                   no               dropped
> >      2                   no               demuxed
> >      3                   no               demuxed
> >      4                   no               dropped
> >      5          04:11:21:04       dropped
> >      6          04:11:21:05       demuxed
> >
> >
> > We don't want to drop frames before 6; they still should be digitized
> > (sometimes the TC appears after ~15 frames). But the stored TC needs
> > to be 04:11:21:04 + (2 - 5) =  04:11:21:01
> >
> >> How is this repoducible? do you have an exact model / signal type /
> >> TC source requirement?
>
> (Anchor for reply)
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".