[FFmpeg-devel] [PATCH] avdevice/decklink_dec: Extract 1080i and NTSC VANC

Devin Heitmueller dheitmueller at ltnglobal.com
Tue Jan 23 06:12:38 EET 2018

Hi Ray,

> On Jan 22, 2018, at 10:47 PM, Ray Tiley <raytiley at gmail.com> wrote:
>> Could you confirm whether you are capturing via SDI or analog (i.e.
>> composite/component)?  Also what is the capturing device and SDK version
>> you are using?  I’ve found various bugs in the numbering of VANC lines in
>> some cards, particularly with interlaced formats, and it would be good to
>> understand if perhaps this is a card-specific issue.
>> I'm capturing SDI, the only decklink card I have access to is the Decklink
> mini recorder, and I was using the latest sdk ( I can check exact version
> tomorrow during work.) It's definitely possible I expanded the search lines
> a bit too far, I was just hacking until I found the data.

Ok, so it’s possible that in the three cases you’ve changed, the start line was actually fine to begin with and the problem was just the end line was wrong.  I could totally believe that.

I can say that all three start lines are correct according to RP 168, but I would have to do the math to check the end lines (there was a time I could tell you from memory where the active video starts/ends for each video standard, but those days are long gone).
> The source equipment was again a Matrox LE4. The source file was 1080i
> while I was testing the NTSC output, so it was perhaps a scaling issue.
> However the spec says that VANC data will be on the luma channel for high
> definition sources, so that is why I tried extracting it from the
> interleaved and was able to successfully parse the VANC packets.

Right.  At least in libklvanc I don’t even bother looking at the chroma channel, since I’ve never seen a piece of equipment that puts any VANC there.  For CEA-608 payloads and CEA-708 CDP packets, SMPTE ST 334-1:2015 Sec 4 clearly states that it must be in the luma channel for HD sources.

> I chatted with someone in the video developers slack community and they
> suggested that for NTSC the data is indeed interleaved.

Hmmm, this is a pretty ambiguous explanation.  Got a link to the conversation?

I also wouldn’t rule out a decklink bug, but it’s too early to make such an assertion.

If you can get a dump of the raw VANC line data that I can review, that would be helpful.

> I have access to some SD captioned files that I can play out my test
> hardware to eliminate scaling.

Great.  I think you’ve definitely found some issues - it’s just a question of whether we really have to deviate from the spec in order to work properly for your use case.


More information about the ffmpeg-devel mailing list