De-telecine 4:2:2 10bit content with AviSynth?

[embis2003]

I'm working with SD ProRes content in AviSynth and I want to de-telecine this ProRes master, however, the TIVTC package dosen't support Stacked16 input. I can convert to YUY2 to maintain 4:2:2, but loose 10bit. I can convert to YV12, but loose 4:2:2. It seems to be one or the other which really sucks. Anyone have any ideas?

[poisondeathray]

Is the pattern consistent ? (no edits prior to pulldown removal , or cadence breaks) . If so , you can use a combination of separatefields and selectevery to keep 10bit422

[embis2003]

Is the pattern consistent ? (no edits prior to pulldown removal , or cadence breaks) . If so , you can use a combination of separatefields and selectevery to keep 10bit422

If I convert to 8bit yv12, tfm().tdecimate() works flawlessly, so I would assume so, besides residual combing in the chroma channel, which can be fixed If I use this script for interlaced chroma instead: https://forum.doom9.org/showthread.php?p=1582950#post1582950

[poisondeathray]

TFM is an adaptive field matcher - so it will change "on the fly" and adapt to the changing patterns. It it only supports 8bit . That script uses TFM for some parts, so 8bit only too. srestore is used for deblending and only supports 8bit as well. Even if they supported 10bit the script would have to be adjusted to 0-1023 range for the masktools parts

I was asking if pattern is the same. If it was, you could keep 10bit422 without using TFM. There are only 5 combinations for 3:2 pulldown, it just depends where in the cycle you start. It sounds like you probably can't do this with your source

[embis2003]

TFM is an adaptive field matcher - so it will change "on the fly" and adapt to the changing patterns. It it only supports 8bit . That script uses TFM for some parts, so 8bit only too. srestore is used for deblending and only supports 8bit as well. Even if they supported 10bit the script would have to be adjusted to 0-1023 range for the masktools parts

I was asking if pattern is the same. If it was, you could keep 10bit422 without using TFM. There are only 5 combinations for 3:2 pulldown, it just depends where in the cycle you start. It sounds like you probably can't do this with your source

The entire file maintains three progressive frames followed by two combed frames. And 8bit restrictions of plugins are fine because i can convert to stacked16 but the only issue at that point is the yv16 output, which tfm only support yv12. If all else fails, I'll just go with 10bit 420.

[poisondeathray]

The entire file maintains three progressive frames followed by two combed frames. And 8bit restrictions of plugins are fine because i can convert to stacked16 but the only issue at that point is the yv16 output, which tfm only support yv12. If all else fails, I'll just go with 10bit 420.

Here a wrapper function in this post for separatefields and selectevery. Just test 0-4 and it works in 10bit422 with avs+ if your pattern is consistent
https://forum.videohelp.com/threads/294901-Avisynth-manual-IVTC-trouble#post1856153

I question what is causing the combed chroma; I'm wondering if your source filter is returning the frames correctly, or if you're converting correctly, or if it's in the actual source (can you see problem on separate fields?) . If you go down to YV12 it has to be done in an interlaced manner (interlaced=true) if you don't remove pulldown first . That is one common cause of chroma combing (422=>420 done in a progressive manner ,instead of interlaced manner, before removing pulldown).

[embis2003]

The entire file maintains three progressive frames followed by two combed frames. And 8bit restrictions of plugins are fine because i can convert to stacked16 but the only issue at that point is the yv16 output, which tfm only support yv12. If all else fails, I'll just go with 10bit 420.

Here a wrapper function in this post for separatefields and selectevery. Just test 0-4 and it works in 10bit422 with avs+ if your pattern is consistent
https://forum.videohelp.com/threads/294901-Avisynth-manual-IVTC-trouble#post1856153

I question what is causing the combed chroma; I'm wondering if your source filter is returning the frames correctly, or if you're converting correctly, or if it's in the actual source (can you see problem on separate fields?) . If you go down to YV12 it has to be done in an interlaced manner (interlaced=true) if you don't remove pulldown first . That is one common cause of chroma combing (422=>420 done in a progressive manner ,instead of interlaced manner, before removing pulldown).

WOW i feel extremely stupid haha. turns out no interlaced chroma, didn't know about the interlaced parameter, no more combing now.

[embis2003]

The entire file maintains three progressive frames followed by two combed frames. And 8bit restrictions of plugins are fine because i can convert to stacked16 but the only issue at that point is the yv16 output, which tfm only support yv12. If all else fails, I'll just go with 10bit 420.

Here a wrapper function in this post for separatefields and selectevery. Just test 0-4 and it works in 10bit422 with avs+ if your pattern is consistent
https://forum.videohelp.com/threads/294901-Avisynth-manual-IVTC-trouble#post1856153

I question what is causing the combed chroma; I'm wondering if your source filter is returning the frames correctly, or if you're converting correctly, or if it's in the actual source (can you see problem on separate fields?) . If you go down to YV12 it has to be done in an interlaced manner (interlaced=true) if you don't remove pulldown first . That is one common cause of chroma combing (422=>420 done in a progressive manner ,instead of interlaced manner, before removing pulldown).

Its not working so I guess the pattern isn't consistent. i put the correct number of frames and it shifts over time.