No clue, what the 'best' is.
How I would approach such a source:
- throw Retinex at it (while moving it to the end of the filter order) to better see the problems
![[Image: grafik.png]](https://i.ibb.co/XfPMV4tv/grafik.png)
- check the interlacing flag (tff seems fine)
- crop what should be inactive borders of the video
- check whether the luma scaling is correctly signaled (pc vs. tv scale); seems fine
Due to the severance of the artifacts in this source (multiple times compressed, badly captured) and the general darkness of the source, I would think whether the source luma levels should be adjusted.
Then I would try how Deblocking and some other heavy filters (MCTemporalDenoise, BasicVSR++, SpotLess) influence the output.
Due to the bad shape of the source, there are no details any normal sharpener could emphasize.
![[Image: grafik.png]](https://i.ibb.co/ks2vdMQ4/grafik.png)
Assuming you don't plan to change the luma levels (make the whole thing brighter), I would probably use different filters&co on dark and bright areas of the image (= use some luma masking when using filters).
For the chroma you can try using HQDN3D, set luma strength to 0 and just use the chroma denoising.
=> The only thing one can do with such a source is trying to reduce compression artifacts while trying to keep some of the existing details.
Cu Selur