[FFmpeg-user] zimg colorspace yuv -> rgb possible?

Andy Furniss adf.lists at gmail.com
Thu Jan 28 00:32:01 CET 2016


Carl Eugen Hoyos wrote:
> Andy Furniss <adf.lists <at> gmail.com> writes:
>
>> ffmpeg -loglevel debug -i snk-420i.y4m -vf scale=interl=1
>> ffmpeg.png
>
> Do you understand this message to add the missing options?
> http://thread.gmane.org/gmane.comp.video.ffmpeg.devel/181474/focus=181547

Hmm
>
I wouldn't be too confident of getting that working, plus the follow
up posts seem to say it's not enough anyway.

Interesting what is said elsewhere in the thread by kierank like -

http://article.gmane.org/gmane.comp.video.ffmpeg.devel/181527

Seems interested in implementing something -

http://article.gmane.org/gmane.comp.video.ffmpeg.devel/181524

and 6 months later it is he that says

13:44:12 < <at> ubitux> so mmh... -vf format=yuv422p will work as expected?
13:44:27 <+kierank> with forced interlaced scaling yes
13:45:41 < <at> ubitux> kierank: -vf scale=interl=1,format=yuv422p ?
13:45:52 <+kierank> ubitux: i believe yes

so did he implement something?

TBH reviewing the many pngs I've made I don't think he did - or it
didn't work!

True that 420 -> 422 is better looking than
420 -> rgb  but if I use the sws flags on the latter it actually looks
the same as doing 420 -> 422 first. So I think that full_chroma_int is
able to mask the worst case and by luck starting from 422 does similar.


More information about the ffmpeg-user mailing list