Are any of the VLC deinterlacing algorithms designed for deinterlacing a film (progressive) source?
“Inverse telecine” usually means reversing the 3:2 pulldown on 60Hz sources, though; will that work correctly for 1080i/50?
and at 1x while trying to acquire lock-on.
implies it might do the right thing on a 50Hz signal as it wouldn’t find a cadence. Perhaps.
Again, the filters in AVISynth take progressive input and de-interlace it. The only hint they get is which field comes first, and 99% of the time, it’s the top, so you almost never have to change it. Your TV could do the same thing by examining the incoming video and de-interlacing only if necessary (again, as the filters in AVISynth do). This sort of de-interlace without changing the frame rate is easy to do, but not available in most displays (unless the incoming signal is flagged as interlaced).
What most displays have instead is a “motion smoother”, which will often do a very similar thing as a true de-interlacer, as long as they are not set too high (where they add interpolated frames and generate the “soap opera effect”).
So, again, if your TV had a good enough de-interlacer, we wouldn’t be having this conversation.
It has a good enough deinterlacer - for 1080i/60 anyway. I don’t know if it can handle 1080i/50 properly - quite likely not.
But unless I can get the Vero 4K+ to output a native 1080i signal, the ability of the TV to deinterlace it is irrelevant.
@grahamh I’ve just noticed that, on my 1080 test clip, the Vero actually plays it correctly if I turn off hardware acceleration. That should help narrow down the problem a bit. (Will also give you a visual reference as to what it should look like).
Same thing happens on my Nvidia Shield TV - gets it right if I turn off hardware acceleration, but not with it on. Curiously, my phone plays it correctly with hardware acceleration active (Kodi v18.1 under Android).
Proves my point. Vero does its deinterlace in hardware. Let me try…
Made some steps to bypassing DI as it resolves some Live TV issues; but I’m not sure we are completely avoiding this (which is good); with the possible exception of this use case.
Your 1080 S&M clip plays OK without hw acceleration on Krypton here but not on Leia (different TV - needs more investigation).
It looks like the deinterlacing method can be changed.
There are also attempts to identify and handle progressive frames as interlaced; but they may only be applying for 4:2:2 content currently.
Investigating. Hopefully this can just be done as part of pre-di; as the full deinterlace pathway on AMLogic seems to cause playback problems with VC-1 content. With any luck the issue is caused by the deinterlacing occurring now; rather than a lack of post-processing.
Can you please not refer to it as an S&M clip? People will think I’m sending you kinky porn. 
Anyway, that sounds interesting - I’ll leave you guys to it…
Would explain why Sam and graham are so active in a a thread about deinterlacing its a topic hard to get excited about
If this is fixed in 4.9; it might be worth waiting for that.
Sam