[FFmpeg-devel] [PATCH 7/7] Add check in av_write_header() which validates the rawvideo codec tag.
Michael Niedermayer
michaelni
Wed Jun 2 03:06:31 CEST 2010
On Tue, Jun 01, 2010 at 03:11:30PM -0700, Baptiste Coudurier wrote:
> On 06/01/2010 10:59 AM, Michael Niedermayer wrote:
>> On Tue, Jun 01, 2010 at 07:29:34PM +0200, Stefano Sabatini wrote:
>>> On date Tuesday 2010-06-01 17:53:36 +0200, Michael Niedermayer encoded:
>>>> On Tue, Jun 01, 2010 at 12:39:41PM +0200, Stefano Sabatini wrote:
>>>>> On date Thursday 2010-05-27 21:59:52 +0200, Michael Niedermayer
>>>>> encoded:
>>>>>> On Thu, May 27, 2010 at 12:28:28AM +0200, Stefano Sabatini wrote:
>>>>>>> Make the function fail if there is no codec tag associated to that
>>>>>>> pixel format.
>>>>>>> ---
>>>>>>> libavformat/utils.c | 9 +++++++++
>>>>>>> 1 files changed, 9 insertions(+), 0 deletions(-)
>>>>>>
>>>>>> this patch requires as prereqesite avienc/riff to override codec_tag
>>>>>> so its 0 for rgb formats
>>>>>
>>>>> The problem with the current code is that if
>>>>> avcodec_pix_fmt_to_codec_tag(avctx->pix_fmt) in rawenc.c returns 0,
>>>>> then av_write_header() does:
>>>>> st->codec->codec_tag= av_codec_get_tag(s->oformat->codec_tag,
>>>>> st->codec->codec_id);
>>>>>
>>>>> which sets the codec tag to the first entry with codec id == RAWVIDEO,
>>>>> which may have nothing to do with the pixel format of the output
>>>>> stream.
>>>>
>>>> you are confused.
>>>> codec_tag must be 0 for rgb in avi. and the first entry is 0
>>>> if rgb in avi is stored with non 0 codec_tag currently then you
>>>> introduced a very serious bug. And the fix is to override it in
>>>> riff.c as i said multiple times.
>>>
>>> I have not the faintest idea about what "rgb in avi" is, there are
>>> tons of RGB pixel formats which can fit in rawvideo, also I have no
>>> idea what you mean by "override it in riff.c" and I'm quite tired of
>>> playing at mind-guessing.
>>
>> what is unclear on
>> rgb in avi must use codec_tag=0 ?
>> the raw rgb format in avi is indicated by the number of bits per pixel
>> not by the fourcc which must be set to 0.
>>
>>
>>>
>>> What the code currently implements is just wrong for both AVI and NUT,
>>> in the case of AVI if the pixel format cannot be mapped to any codec
>>> tag, codec_tag is set to 0,
>>
>> it must be 0 this is as it should be
>>
>>
>>> which when reading it is interpreted as
>>> YUV420P,
>>
>> if so thats a bug and i would suspect a new one
>>
>
> Please don't break .mov
does fate test mov/avi/nut yuv & rgb?
if not, it should
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
I am the wisest man alive, for I know one thing, and that is that I know
nothing. -- Socrates
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20100602/915e62d9/attachment.pgp>
More information about the ffmpeg-devel
mailing list