[FFmpeg-user] Decombing via screening

Paul B Mahol onemda at gmail.com
Tue Apr 14 12:26:25 EEST 2020

On 4/14/20, Mark Filipak <markfilipak.windows+ffmpeg at gmail.com> wrote:
> On 04/14/2020 04:23 AM, Paul B Mahol wrote:
>> On 4/14/20, Mark Filipak <markfilipak.windows+ffmpeg at gmail.com> wrote:
>>> How do I combine two combed half-pictures so that the resulting picture
>>> is
>>> screened?
>>> Combed input frame:
>>> 11111...
>>> 22222...
>>> Desired output frame:
>>> 12121...
>>> 21212...
>>> It occurs to me that I need to unweave the half-pictures, line-double
>>> both,
>>> screen both via
>>> pixel-sized screens that are offset by 1 pix, then reweave them.
>> Look at weave, doubleweave, separatefields and so on....
> Thanks. I'll run them through my tester.
>>> I've not found a line-doubling filter and I've not found a screen method
>>> that actually works.
> Actually, after I posted, I thought of bob. I think bob line doubles.

I think you can double lines with separatefields,scale combination.
You just need to set right size args to scale filter and use nearest
neigbour interpolation
for doubling lines.

I still think that what you do is wrong thing to do.
Why? mpv player already have various interpolation algorithms for
X->60 progressive upsampling.
Note that they are not motion ones, as that one is very fragile and slow.

>>> Also, a method to zoom pixels in the output so I can actually see them
>>> individually without a
>>> magnifying glass would be very helpful.
>> datascope or pixscope filter.
> Thanks again.
>>> I've tried every filter I could find that has anything to do with
>>> blending,
>>> screening, blurring,
>>> etc., but I may have missed something.
>> You are just a big troll.
> Yup! That's me! Mark, the ffmpeg troll. The troll of your dreams. Just keep
> your hands to yourself,
> Paul.
>>> A really ambitious method would be: Replace the combed frame with a
>>> synthesized frame that's a
>>> motion-vector meld of the preceding and subsequent progressive frames,
>>> but
>>> that seems pretty wild
>>> and probably beyond ffmpeg's capabilities, but I simply don't know.
>>> Thanks,
>>> Mark.
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-user
> To unsubscribe, visit link above, or email
> ffmpeg-user-request at ffmpeg.org with subject "unsubscribe".

More information about the ffmpeg-user mailing list