[FFmpeg-devel] [PATCH] avcodec/dv_profile: dv files with wrong dsf flag - detect via buf_size

[Marton Balint]

On Tue, 16 Mar 2021, Mark Plomer wrote:

> Unfortunately it is not possible to move the hack to the bottom, because 
> the "normal processing" (dv_profiles lookup loop) will catch it 
> otherwise (with a wrong profile), because it does not check the frame_size:

Then at least check the 50/60 flag as well in the VAUX source pack, e.g.:

pal   = !!(frame[80 * 5 + 48 + 3] & 0x20);

Regards,
Marton


>
>     for (i = 0; i < FF_ARRAY_ELEMS(dv_profiles); i++)
>         if (dsf == dv_profiles[i].dsf && stype == 
> dv_profiles[i].video_stype)
>             return &dv_profiles[i];
>
> But as the new "if" is very specific and catches an "invalid" 
> combination only (with the frame_size check), it should not break 
> anything, I think:
>
>     if (dsf == 0 && stype == dv_profiles[1].video_stype && buf_size == 
> dv_profiles[1].frame_size)
>
> At least, I put the new hack after all other "pre-normal-processing" 
> hacks ;-)
>
> Regards
> Mark
>
> Am 13.03.21 um 17:48 schrieb Marton Balint:
>>
>>
>> On Sat, 16 Jan 2021, Mark Plomer wrote:
>>
>>> Trying to fix https://trac.ffmpeg.org/ticket/8333
>>>
>>> Some of my old DV AVI files have the DSF-Flag of frames set to 0, 
>>> although it is PAL (I think they were rendered with Ulead Media 
>>> Studio Pro) ... this causes ffmpeg/VLC-player to produce/play 
>>> corrupted video.
>>>
>>> In other players/editors it works fine including:
>>>
>>> - VirtualDub
>>> - Windows Media Player
>>> - AVCutty
>>> - Ulead Media Studio Pro (very old)
>>>
>>> I had a look at VirtualDub ... there the PAL/NTSC detection is based 
>>> on the frame rate from AVISTREAMINFO header (dwRate/dwScale) - see 
>>> 
> https://github.com/Pavuucek/VirtualDub/blob/f47ebd2536f0034b048180d0b9cb9bde0ab10c73/src/VirtualDub/source/VideoSource.cpp#L1211
>>>
>>> As I don't know, how to access the AVI header info inside 
>>> dvvideo_decode_frame()/ff_dv_frame_profile(), I tried another 
>>> workaround by checking the buf_size against the dv_profile.
>>>
>>> It works fine now, but I don't know, if this is really the best 
>>> solution ...
>>
>> --- a/libavcodec/dv_profile.c
>> +++ b/libavcodec/dv_profile.c
>> @@ -281,6 +281,10 @@ const AVDVProfile* 
>> ff_dv_frame_profile(AVCodecContext* codec, const AVDVProfile
>>         && codec->coded_height==576)
>>          return &dv_profiles[1];
>>
>> +    /* hack for trac issue #8333, dv files with wrong dsf flag - 
>> detect via buf_size */
>> +    if (dsf == 0 && stype == dv_profiles[1].video_stype && buf_size 
>> == dv_profiles[1].frame_size)
>> +        return &dv_profiles[1];
>> +
>>
>> If possible, then it is probably better to move this fallback to later 
>> in the code, right after the hack for trac issue #217, so previous 
>> hacks won't get broken...
>>
>> Regards,
>> Marton
>> _______________________________________________
>> 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".
> _______________________________________________
> 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".