[FFmpeg-devel] [PATCH] H.264 fix interlaced flag

[Ivan Schreter]

Michael Niedermayer wrote:
> On Fri, Feb 20, 2009 at 04:04:37PM +0100, Ivan Schreter wrote:
>   
>> Michael Niedermayer wrote:
>>     
>>> On Wed, Feb 18, 2009 at 10:22:34PM +0100, Ivan Schreter wrote:
>>>   
>>>       
>>>> Hi Michael,
>>>>
>>>> Ivan Schreter wrote:
>>>>     
>>>>         
>>>>> Michael Niedermayer wrote:
>>>>>   
>>>>>       
>>>>>           
>>>>>> On Sun, Feb 15, 2009 at 08:23:31PM +0100, Ivan Schreter wrote:
>>>>>>   
>>>>>>     
>>>>>>         
>>>>>>             
>>>>>>> In non-interlaced videos coded with picture struct top/bottom or 
>>>>>>> bottom/top, the picture is marked as interlaced. Attached patch corrects 
>>>>>>> it.
>>>>>>> [...]
>>>>>>>       
>>>>>>>           
>>>>>>>               
>>>>>> please upload a sample that your patch fixes
>>>>>>   
>>>>>>     
>>>>>>         
>>>>>>             
>>>>> Uploaded to dir h264_progressive_deemed_interlaced.
>>>>>
>>>>> However, I cannot upload description, it says "553 Could not create 
>>>>> file". Retrying in new FTP session won't let me even cd to the directory 
>>>>> (550 Failed to change directory). So it's without extra text file :-(
>>>>>
>>>>>   
>>>>>       
>>>>>           
>>>> Did you have a chance to review it and eventually test it with submitted 
>>>> file? I hope the file was uploaded correctly...
>>>>     
>>>>         
>>> the patch is wrong, there seems no relation between the pic_struct and
>>> the interlacing vs progressive.
>>> Maybe the ct_type field could be used and the whole pic_struct->interlaced
>>> code removed
>>>   
>>>       
>> Yes, ct_type could be used. But IMHO, it would be sufficient to say
>>
>> cur->interlaced_frame = FIELD_OR_MBAFF_PICTURE;
>>
>> outside of the switch and not set interlaced_frame at all in the switch 
>> (i.e., same as for missing picture structure). If you are OK with that, 
>> I'll prepare a patch.
>>     
>
> cur->interlaced_frame = FIELD_OR_MBAFF_PICTURE;
> is not correct
>
> progressive frames can be coded as fields and interlaced fields
> can be coded as frames
>
>   
Then only ct_type remains...

Regards,

Ivan