[FFmpeg-devel] [PATCH] dv: add timecode to metadata

[Matthieu Bouron]

2011/12/23 Reimar Döffinger <Reimar.Doeffinger at gmx.de>:
> On Fri, Dec 23, 2011 at 02:25:19PM +0100, Matthieu Bouron wrote:
>> 2011/12/23 Reimar Döffinger <Reimar.Doeffinger at gmx.de>:
>> > On Fri, Dec 23, 2011 at 01:26:35AM +0100, Matthieu Bouron wrote:
>> >> 2011/12/22 Reimar Döffinger <Reimar.Doeffinger at gmx.de>:
>> >> > On Thu, Dec 22, 2011 at 10:29:52PM +0100, Matthieu Bouron wrote:
>> >> >> +static int dv_extract_timecode(DVDemuxContext* c, uint8_t* frame, char tc[32])
>> >> >> +{
>> >> >> +    int hh, mm, ss, ff, df;
>> >> >> +    const uint8_t* tc_pack;
>> >> >
>> >> > Having the space between type and * and not between * and name is more
>> >> > consistent with the rest of the code (and also more consistent with the
>> >> > actual semantics).
>> >> >
>> >> >> +    ff  = tc_pack[1]         & 0xf;
>> >> >> +    ff += ((tc_pack[1] >> 4) & 0x3) * 10;
>> >> >> +    df  = ((tc_pack[1] >> 6) & 0x1);
>> >> >> +    df  = (tc_pack[1]  >> 6) & 0x1;
>> >> >> +    ss  = tc_pack[2]         & 0xf;
>> >> >> +    ss += ((tc_pack[2] >> 4) & 0x7) * 10;
>> >> >> +    mm  = tc_pack[3]         & 0xf;
>> >> >> +    mm += ((tc_pack[3] >> 4) & 0x7) * 10;
>> >> >> +    hh  = tc_pack[4]         & 0xf;
>> >> >> +    hh += ((tc_pack[4] >> 4) & 0x1) * 10;
>> >> >
>> >> > What is the exact format?
>> >>
>> >> The format is described in the SMPTE 314M.
>> >> For 525/60 Systems:
>> >>
>> >> CF (1b)  DF(1b)         TENS OF FRAME(2b)    UNITS OF FRAME(4b)
>> >> PC (1b)               TENS OF SECONDS (3b)    UNITS OF SECONDS(4b)
>> >> BGF0(1b)              TENS OF MINUTES (3b)    UNITS OF MINUTES(4b)
>> >> BGF2(1b) BGF1(1b)   TENS OF HOURS(2b)    UNITS OF HOURS (4b)
>> >>
>> >> The only difference for PAL system, is that the DF (drop_frame) flag
>> >> is ignored (defined as arbitrary).
>> >
>> > I'd still be in favour of exporting its value as is, or would you
>> > expect any issue with that?
>>
>> If i understand well, you're in favor of exporting the entire value of
>> the timecode as an uint32_t for example and not as its string
>> representation ? So this value should be interpreted in the
>> dv_read_timecode function ?
>
> No, that would be inconsistent with what gxf does.
> If necessary we can do that in addition.
> No, I meant always switching between ':' and ';' depending on the
> DF flag, even if for PAL that flag is not supposed to have a special
> meaning.

For this sample http://samples.mplayerhq.hu/DV-raw/small_test2.dv, the
drop_frame bit is set to 1. if we do not check if it is PAL or not, it
will display a wrong drop-frame timecode. This is why i reset the
drop_frame value to 0 for PAL systems.

> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel