[HELP] (un)Sharp filter

[DaGoose]

What they are doing is mixing EdgeCleaner, DeRinging und FFmpegs (un-)Sharpen.
I think mixing FFmpeg with Avisynth/Vapoursynth filters is in general bad and often stupid:
a. it easily can cause color issues
b. you easily loose control about the bit depth&co you work in (at least last I checked figuring out what color formats FFmpeg really is using is a real pain. Also controling how rgb<>yuv conversions are also rather complicated to control.

If you use Avisynth, instead of using FFmpegs (Un-)Sharpen, one could use one of the Avisynth filters for unsharpening http://avisynth.nl/index.php/External_filters through the custom addition.
If you use Vapoursynth there are also (un-)sharpen filters, see: https://vsdb.top/
Personally I'm no fan of unsharpen filters. (which is why in Hybrid only FFmpegs unsharpen filter is supported out of the box) Also unsharpen filters often have problems with noise content.

Cu Selur

RGB to YUV conversions in VapourSynth appears more straightforward than in AviSynth, at least from my experiences. All you have have to do is call the resize node in vapoursynth.core, then a resize algo method, and lastly define the color format plus matrix/transfer/primaries in the formal parameters of the method header. I don't understand how this process can be complicated to control, assuming that you aren't doing this process each time you call a filter, which shouldn't be happening unless each filter has different color format requirements. 

Looking at AviSynth specifically, I think your argument for mixing AviSynth and FFMpeg filters is meaningless because the amount color conversions that happens with AviSynth alone is already really bad. Furthermore, the majority of FFMpeg filters use a large variety of color/pixel formats.
I have copied the settings shown in the video guide (I was the one who created the video) and this partial AviSynth+ script is generated by Hybrid. A 10bit video is being put in a 16bit container (instead of YUV420P10?) and is dithered down to 8bit:

# color sampling YV12@10, matrix: bt2020nc, scantyp: progressive, luminance scale: limited
LWLibavVideoSource("C:\Users\Goose\DOCUME~1\VAPOUR~1\DAYOFT~1.MKV",cache=false,format="YUV420P16", prefer_hw=0,repeat=true)
# current resolution: 1920x1080
# filtering
# deringing using MosquitoNR
MosquitoNR()
EdgeCleaner()
# Dithering from 16 to 8bit for encoder
ConvertBits(8)
PreFetch(8)
# setting output fps to 23.976fps
AssumeFPS(24000,1001)
#  output: color sampling YV12@8, matrix: bt2020nc, scantyp: progressive, luminance scale: limited
return last

This doesn't appear to be a problem in VapourSynth, but there's no EdgeCleaner port. Which is why I stuck to AviSynth in the video guide.

Regarding FFMpeg (un)sharpen, figuring out what color formats FFMpeg uses should be no struggle. Just go to libavfilter in FFMpeg's source code and  there's usually an enum in each filter's source containing the color/pixel formats used by FFMpeg. Looking at the unsharp filter, there appears to be wide support for color/pixel formats (see line 200: https://github.com/FFmpeg/FFmpeg/blob/ma..._unsharp.c). 10 Bit support for unsharp was also released over a year ago so there's essentially no loss of colors/conversion for the majority of videos.

[Selur]

I agree, the problem isn't the exact example but more of a general nature.
a. there is no real control over the filter order when mixing ffmpeg and Vapoursynth/Avisynth filters
b. there is no real control over the filter order when using ffmpeg in itself
Filter in Vapoursynth and deinterlace/ivtc in ffmpeg usually causes problems, but isn't possible in newer Hybrid versions and I tried to workaround some problems of the older versions.
-> I still think it's really bad to combine ffmpeg and Avisynth/Vapoursynth filtering and not hint that this could be problematic.

This doesn't appear to be a problem in VapourSynth, but there's no EdgeCleaner port. Which is why I stuck to AviSynth in the video guide.

Only in Hybrids frontend. HAVSFunc (https://github.com/Selur/VapoursynthScri...avsfunc.py) does have EdgeCleaner in it. So one could use it in a custom section.

Cu Selur

Ps.: I hope we agree that the title "How to Upscale Video Resolution in Hybrid CORRECTLY | Upscale Resolution Without Loss of Qualiy" is just a marketing gag since that filtering is not lossless (as basically no filtering operation is). To get a good uscaling you should always try different methods, there is no general 'correct' way.

[DaGoose]

I hope we agree that the title "How to Upscale Video Resolution in Hybrid CORRECTLY | Upscale Resolution Without Loss of Quality" is just a marketing gag since that filtering is not lossless (as basically no filtering operation is). To get a good upscaling you should always try different methods, there is no general 'correct' way.

An overwhelming number of video titles for guides like this are worded as such. There's a 100 character limit for YouTube video titles, and I wanted the guide to differentiate from others by using different words in the title, while also it not being "clickbait". 
I can agree that there's no "correct" way for doing this process, as every video is different and may require some other filters. 
However, I don't agree with the process not being lossless in quality. I'm NOT explaining this because you're wrong. You're definitely right that no filtering is lossless, I understand that too. This is where the word "quality" is very vague. The definition of "without loss of quality" is different to the average consumer compared to what you and I would think. Over my past experiences with talking to thousands of regular consumers on YouTube, "without loss of quality" would mean no bitrate "degradation", or anything that wouldn't result in a blocky video, not so much regarding the colors of the video.

[Selur]

Yes, quality is always in the eye of the beholder.