Sharpener for YUY2 source

[yup]

Hi folk!
Please advice sharpener for YUY2 source (analog capture). I try use FFT3D and not very satisfy. I search filter in space domain (not frequency) like LSF but for YUY2.
With kind regards yup.

[Didée]

For progressive content:

yuy2clip = last
yuy2clip .ConvertToYV12()
LSF(...)
ConvertToYUY2() .MergeChroma(yuy2clip)

[Adub]

Won't the color conversion decrease the quality, if only a little bit?
Or am I totally wrong on this?

[Didée]

1) There is zero impact on quality. YUY2<-->YV12 does not affect luma at all. Chroma is affected by the conversion, so note that the original not-converted chroma is copied back at the end.

2) The posted code is superfluous for LS/F, since LS/F already does exactly that when the input is YUY2.


@ yup: what's the actual problem? Finding a sharpener that supports YUY2, or finding a sharpener that gives a "nice looking" result?

[Blue_MiSfit]

Didee beat me to the punch ;)


Note that he's merging the original chroma back in.

I wonder, does LSF primarily operate on the luma plane?

I haven't got much experience with it...

~MiSfit

[Adub]

Ah, okay. Thanks for the explanation.

[yup]

Thanks to all!
Didée recomend filter which work only luma plane, tomorrow I testing it, also my source interlaced, 2 approach,
first

AVISource("x.avi")
source=last
yuy2=SeparateFields(source)
ConvertToYV12(source, interlaced=true)
SeaparateFields()
LSF()
ConvertToYUY2() .MergeChroma(yuy2)
Weave()

second

AVISource("x.avi")
yuy2=nnedi2(fields=-2)
ConvertToYV12(yuy2)
LSF()
ConvertToYUY2() .MergeChroma(yuy2)
SeparateFields()
SelectEvery(4,0,3)
Weave()

Please advise which approach better, my source preliminary filtered MVDEgrain2 or MVDEgrain3.
yup.

[Didée]

That's how rumours are born. "Didée recomend filter which..." -- Nope. I did not recommend anything. Just tried to focus your vague question.

In the OP you mention YUY2 and spatial-domain filtering ... but nothing about progressive/interlaced.

Now we know it's interlaced, which makes the point about YUY2 rather unimportant. The more important question is "how to sharpen an interlaced source" ...

... and there we are again, digging up the basic question of spatial filtering of interlaced sources for the x-th time. Chewing the cud.