What's this pattern called 1234456788

[Alwyn]

Came across this pattern in The King and I post:

1234456788

Can someone please tell me what it's name is? I can't find it on the Telecine wiki page... maybe it's not telecine.

Thanks!

[JN-]

Its mentioned in this post … https://www.avforums.com/threads/canon-hv20-24p-3-2-pulldown-removal.2441749/

“ Films historically transferred for video use in NTSC-land were telecine'd marginally slower at 23.976 fps (which is close enough to 24 as makes no material difference, yet which converts readily to 29.97) and every fourth frame was duplicated. So 4 becomes 5. Or, the frame sequence becomes 1234456788....... (The term 3:2 pulldown comes from this sequence; alternate pairs of 2 original frames are made into 3).”

[Sharc]

Came across this pattern in The King and I post:

1234456788

Can someone please tell me what it's name is? I can't find it on the Telecine wiki page... maybe it's not telecine.

Thanks!

Incomplete IVTC, decimation missing (decoding)
or
Unprofessional (dumb) film to NTSC framerate conversion by just repeating every 4th frame (encoding).

[Alwyn]

Thanks JN, what intrigues me is that the "other" Telecine, 3:2 pulldown, has 3 Progressive and 2 interlaced frames PPPIIPPPII. This is not the pattern with this duplicated 5th frame video.

Sharc, you may recall we had a similar thing with a flying movie: 123455, being a PAL movie.

I understand it's telecine, but is there a particular name for it like "3:2 pulldown"?

It's not a biggy, it seems as though TFM TDecimate fixes it, I was just curious if it had a name.

[jagabo]

When you single rate deinterlace (ie make 29.97p from 29.97i) a film source that has been through 3:2 pulldown (ie 23.976p to 29.97i, 59.94 fields per second) you get that pattern -- one duplicate out of every 5 frames. You see this quite often in streaming videos like Pluto.tv.

[Sharc]

Thanks JN, what intrigues me is that the "other" Telecine, 3:2 pulldown, has 3 Progressive and 2 interlaced frames PPPIIPPPII. This is not the pattern with this duplicated 5th frame video.

Sharc, you may recall we had a similar thing with a flying movie: 123455, being a PAL movie.

I understand it's telecine, but is there a particular name for it like "3:2 pulldown"?

It's not a biggy, it seems as though TFM TDecimate fixes it, I was just curious if it had a name.

Not as I know. Call it "primitive telecine" or whatever. It's what simple or basic framerate converters (and some NLEs) do: just repeating (or skipping for downconversion) frames creating judder.
(We have discussed this frame repetition for @chollo's horse soft-telecined DVD as well.)
PAL sources (25fps) will first be slowed down to 23.97fps and then converted to 29.97fps NTSC using this frame repetition process instead of proper 2-3 telecine using field-based interleaving.
Or you get this pattern as per jagabo's explanation.

See also here:
https://avisynthplus.readthedocs.io/en/latest/avisynthdoc/corefilters/fps.html#changefps

[Alwyn]

@Jagabo, OK, maybe that is where The King video came from. So it has been TCed with 3:2 Pulldown originally, just messed-with after that.

[jagabo]

That King and I clip is encoded progressive at 23.976 fps with pulldown flags. Some decoders will perform the pulldown and deliver 29.97i, some will ignore the pulldown and deliver 23.976 progressive, some will ignore the pulldown and deliver 29.97p (with the duplicate frames).

[JN-]

What a mess.

[hello_hello]

Edit: I guess I took too long to post. Everyone else beat me to it.

If you're referring to frames rather than fields, it's what happens after field matching to reverse the telecine process.

There's two repeated fields in each cycle. Usually the field matching process matches one of the duplicate fields twice. If you look at this telecine pattern:
https://upload.wikimedia.org/wikipedia/commons/8/8c/32pulldown.svg
and if the field matching was checking each even field for a match with the odd field of the same frame and the odd field of the next frame and keeping the best match.

Progressive frame 1 - Even A only matches odd A.
Progressive frame 2 - Even B in frame 2 can match odd B in frames 2 and 3. The best match is kept.
Progressive frame 3 - Even C in frame 3 doesn't match the odd field in frame 3 but it matches the odd field in frame 4.
Progressive frame 4 - Even D in frame 4 doesn't match the odd field in frame 4 but it matches the odd field in frame 5.
Progressive frame 4 - Even D in frame 5 matches the odd field in frame 5. The two even D matches result in a duplicate frame.

mpeg2source("O:\The King and I_edit.d2v")
TFM()
TDecimate()
CropResize(704,0, 8,114,-8,-118, InDAR=15.0/11.0)

Of course you can convert 23.976fps progressive to 29.97fps progressive by simply duplicating every 4th frame, but it doesn't look very smooth.

I didn't bother upscaling much. After cropping the top and bottom borders from the DVD sample you're only left with 248 lines of resolution so there's not much in the way of picture detail.

[Alwyn]

Edit: synchronised post with Hello Hello.

@Jagabo, Aha! That explains a lot. When I opened it in VDub, I spotted the 5th frame. When I opened it in AVIDemux, there was no 5th frame, and doing a Copy Copy export resulted in the frame disappearing with no setting changes.

JN... agree!

Related issue: with this frame pattern, in VDub, I can set the ITVC to Single Field Only:


[Attachment 81566 - Click to enlarge]

It is leaving every 24th frame (that's a duped frame). It's removing all the other dupes.

Autodetect doesn't work well; sometimes it misses a dupe, on other frames it has bad interlacing.

Any ideas?

Edit 2: Hello Hello, your file doesn't have any dupes. Was that only with VDub's IVTC filter, and if so, what version VDub are you using?

[Selur]

That King and I clip is encoded progressive at 23.976 fps with pulldown flags. Some decoders will perform the pulldown and deliver 29.97i, some will ignore the pulldown and deliver 23.976 progressive, some will ignore the pulldown and deliver 29.97p (with the duplicate frames).

It's normal 'soft telecine', which DGSource (form DGDecNV) with fieldop=1 works fine on.

[Sharc]

@Alwyn: Also remember what has been discussed here and in following posts about hard- or soft telecine, decoding, source filters, patterns, repeat patterns etc.
https://forum.videohelp.com/threads/399508-Conversion-4-3-to-16-9/page3#post2741370

or see the last sentence in the post here:
https://forum.videohelp.com/threads/415608-Uncompressed-AVI-shows-jagged-edges-on-play...e2#post2746780
When you baserate deinterlace - like QTGMC(FPSDivisor=2) - a standard 2-3 telecined source you get abcddefghhijkll.
That's why one shouldn't just QTGMC or deinterlace telecined stuff, but rather IVTC it correctly.

and in similar context
https://forum.videohelp.com/threads/415608-Uncompressed-AVI-shows-jagged-edges-on-play...e2#post2746887

[jagabo]

As to why deinterlacing NTSC telecined video delivers one duplicate in every group of five frames, consider four film frames, A, B, C, D. After 2:3 pulldown you get

A a B b B c C d D d

where capitols represents top fields, lower case bottom fields. Notice that the fields always alternate between top and bottom. And each film frame is repeated for the duration of 2 fields, then 3 fields, then 2 fields, then 3 fields, etc.

When woven into interlaced video frames you get:

Aa  Bb  Bc  Cd  Dd

That's the expected three progressive frames two interlaced frames pattern get from NTSC telecined film.

A single rate deinterlacer will start with the top (or bottom) field of each frame and fill in the other field with whatever it technique it uses. You get:

A?  B?  B?  C?  D?

The question mark represents whatever the deinterlacer filled in to complete the video frame. As you can see, one of the four original film frames is duplicated. Which makes sense because four progressive film frames were converted into five progressive video frames (24 film frames into 30 video frames over each second).

[Alwyn]

Thanks very much, Jagabo. I think getting there.

Sharc, my eyes were glazing over on that thread so please forgive me for letting a couple of lines of yours slip through the cracks.

[Sharc]

Sharc, my eyes were glazing over on that thread ....

I feel with you....