[FFmpeg-user] Video filtering for chroma dot crawl and rainbow effect

[Christian Johannesen]

On Monday, January 25, 2016, Andy Furniss <adf.lists at gmail.com> wrote:

> Andy Furniss wrote:
>
>> Andy Furniss wrote:
>>
>>> Carl Eugen Hoyos wrote:
>>>
>>>> Christian Johannesen <chrisjohannesen <at> gmail.com> writes:
>>>>
>>>> ffmpeg.exe  -r 30000/1001 -i
>>>>> D:\media\test_720x486i_dotcrawl.mov
>>>>>
>>>>
>>>> Please remove the input option -r unless you know exactly what
>>>> you are doing.
>>>>
>>>> -filter_complex [0:0]crop=704:480:8:6,pullup
>>>>>
>>>>
>>> Input on ticket is 4:2:0 and you can't crop vertically interlaced
>>> 4:2:0 by 6 without messing up the chroma. AFAIK you need to crop 4
>>> lines at a time.
>>>
>>> Maybe you could crop 6 if you converted (interlace aware) to 4:2:2
>>> crop and then interlace aware back to 4:2:0 (untested!).
>>>
>>
>> Hmm, after looking at the trac sample I think the chroma has already
>> been scaled progressive = broken sample.
>>
>> I may be wrong! only looked quickly.
>>
>
> Looked again and I still think the "master" is broken. As noted in the
> track it's size may be an issue by its self (as 420 interlaced would
> normally be divisible by 4).
>
> Ignoring that it seems that it has been converted/generated incorrectly
> for 420 interlaced. The chroma has bled between the fields so even
> something simple like yadif=1 alone produces corrupt output.
>
> If you don't have access to better/"real" masters I don't think you can
> get good results from this.
>
> What produced this file with height 486?


The file was captured as 720x486 29.97 fps interlaced lower field first
ProRes 422 HQ with an EVS OpenCube server from Beta SP tape. The resolution
is used to retain line 21 data that would not be stored with a 720x480
capture.


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