Re: GStreamer encoding profile for the Meizu M6 player

Luis Menina wrote:
> Hi Stefan,
> thanks for your quick reply !
>
> Le 10/02/2010 10:58, Stefan Kost a écrit :
>> this means gstreamer is using AVIF_HASINDEX where mencoder used
>> HASINDEX|ISINTERLEAVED|TRUSTCKTYPE. gstreamer should probably set
>> ISINTERLEAVED as well (can you file a bug, I just checked this from
>> http://www.opennet.ru/docs/formats/avi.txt). I could not find what
>> TRUSTCKTYPE is used for (even MS just uses it in their examples
>> http://msdn.microsoft.com/en-us/library/dd388641%28VS.85%29.aspx, maybe
>> we should do that too).
>
> I tried to change the flags (and checked with the logs that I had
> changed the correct bytes), but it didn't work. About the bugs, I
> think I'll open them all at once, once this is resolved and I know
> exactly if this has an impact or not.
>
>> the other changes is that mencoder cleared the avih.bps, but specify
>> strh.bufsize
>
> Couldn't test this one, I have difficulties in finding the right bytes
> to modify.
>
>> finally gstreamer write a dmlh chunk which mencoder does not write.
>
> Replaced the "dmlh" by "JUNK", didn't work either.
> For these modifications, I created the corresponding binary diffs with
> bsdiff, so once I'll fiund the right places to change, I'll easily be
> able to test all the combinations.
for bps (it is in strf_auds):
  guint16 channels;
  guint32 rate;
  guint32 av_bps;
also take the endianess into account when searching.

I don't see any code in gstreamer using bufsize, so having 0 there
should be okay.
>
> For now :
> =========
> bit number 0: "no-dmlh" patch used
> bit number 1: "sameflags" patch used
> bit number 2: "clear-bps" patch used (which I don't have ATM)
> bit number 3: "fill-bufsize" patch used (which I don't have ATM)
>
>
> 0000 => Fail
> 0001 => Fail
> 0010 => Fail
> 0011 => Fail
> 0100 => ?
> 0101 => ?
> 0110 => ?
> 0111 => ?
> 1000 => ?
> 1001 => ?
> 1010 => ?
> 1011 => ?
> 1100 => ?
> 1101 => ?
> 1110 => ?
> 1111 => ?
>
> Do you think "clear-bps" and "fill-bufsize" are just one single patch ?
> Does clearing of filling both of these fields at the same time makes a
> video invalid ?

I would test it separately. Thanks for testing it.

Stefan
>
> Cheers,
> -- 
> Luis
>