[FFmpeg-user] telecine pattern 5555 - Judder-free, 60 FPS telecine (?)

Mark Filipak markfilipak.windows+ffmpeg at gmail.com
Sun Apr 5 11:30:47 EEST 2020

Ooops again! Sorry, maybe I'll get it right this time.

Hi Ted,

On 04/05/2020 03:04 AM, Ted Park wrote:
>>> Even if that were true (which I have no idea to be honest) are you sure you’re not assuming a raw uncompressed source and no compression afterwards either?
>> Frames 3, 8, 13, 18, etc. are combed. The rest of the frames are progressive. The progressive frames display the effects of decombing.
>> The source is a 24fps progressive H.264 AVC video. I'm transcoding it to, among other test transcodes, 5-5-5-5 pull-down @ 60fps progressive, H.264 AVC. I'm not sure what you mean by "assuming a raw uncompressed source and no compression afterwards", but I hope I've responded appropriately.
> I mean I know what combing artifacts look like, but I wouldn’t know how to produce combing artifacts in an encode deliberately ...

Whenever you telecine you are deliberately creating combing. The combing is a consequence of the 
standard, 2-3-2-3 pull-down telecine. Any 24fps video displayed on a 60Hz TV must be telecined 
somehow. The question is, will it be telecined by the player? by the TV? or in advance by the user 
via a transcode?

If telecine is performed by the player or the TV, it is 2-3-2-3 pull-down telecined. That telecine 
exhibits so-called telecine judder because of the 2-3-2-3 cadence.

If telecine is performed in advance by the user and it's a 5-5-5-5 pull-down telecine, there is no 
judder on playback on a 60Hz TV because there is no cadence and because the video is already 60fps.

A 5-5-5-5 pull-down telecine does have combing, but the combing is 2 frames out of every 10 frames 
(i.e., 20%). The combing (C) versus progressive (P) frames looks like this:
[A/a__________][B/b__________][C/c__________][D/d__________]   ...original p24 video
[A/a_][A/a_][A/b_][B/b_][B/b_][C/c_][C/c_][C/d_][D/d_][D/d_]   ...5-5-5-5 telecine
P     P     C     P     P     P     P     C     P     P           20% combing @ 12Hz
A-------------AB-------------BC-------------CD-------------D      judder free

A 2-3-2-3 pull-down telecine also has combing, and the combing is 2 frames out of every 5 
frames(i.e., 40%). The combing (C) versus progressive (P) frames looks like this:
[A/a__________][B/b__________][C/c__________][D/d__________]   ...original p24 video
[A/a_______][B/b_______][B/c_______][C/d_______][D/d_______]   ...2-3-2-3 telecine (player)
P           P           C           C           P
[A/a_][A/a_][B/b_][B/b_][B/c_][B/c_][C/d_][C/d_][D/d_][D/d_]   ...frame rate doubled (TV)
P     P     P     P     C     C     C     C     P     P           40% combing @ 6Hz
A----------AB----------------BC----------CD----------------D      10% judder @ 12Hz

A 4-6-4-6 pull-down telecine is comb free but still has judder.
[A/a__________][B/b__________][C/c__________][D/d__________]   ...original p24 video
[A/a_][A/a_][B/b_][B/b_][B/b_][C/c_][C/d_][D/d_][D/d_][D/d_]   ...4-6-4-6 telecine (player)
P     P     P     P     P     P     P     P     P     P           comb free
A----------AB----------------BC----------CD----------------D      10% judder @ 12Hz

Those are the facts. I don't think 5-5-5-5 telecine to 60fps has ever been done before. Some people 
are opposed to it as though it is religious heresy. That's not going to stop me.

Earlier today I posted a request for help with crafting a special minterpolate filter that operates 
solely on the combed frames (frames 3, 8, 13, 18, etc.). That post is titled
"minterpolate only frames 3, 8, 13, 18, etc.".
I hope that someone will respond to it. But even without decombing, I consider the 60fps video to be 
superior to any 2-3-2-3 pull-down telecine because I hate judder.

>... if asked if I knew how so take this with a grain of salt but I think the general “every encode reduces information” caveat applies, especially as you’re doing some image processing in between.
> You might say that technically, it is a reversible process, but multiple iterations of de/compression with an efficiency-focused codec such as H.264 will quickly show diminishing returns despite any novel techniques to enhance perceived quality that you use in between decompression and compression.

Once transcoded to 60fps, why would I ever reverse it? If I ever get a 120Hz TV, I'll simply drag 
out the original disc and remux to a p24 MKV container with zero loss. In the meantime I will have a 
60fps video to watch that's as good as the original and is free of telecine judder.

More information about the ffmpeg-user mailing list