Really strange pulldown sequence

[MrVideo]

Here's one that I've never seen before...

It is 1080i 29.97, but instead of the normal 2:3 pulldown sequence, this one has the following pattern over 23 frames:

2C/2M/4C/2M/4C/2M/3C/3M/1C where "C" equals a clean frame from the source and "M" means mixed fields from the source. Mapping this out, I get 19 source frames spread over 23 1080i frames. That comes out to ~24.7578 source fps. WTF? That is close to 25fps, but not quite.

A good example of 25fps -> 23.976fps is the recent PBS airings of My Mother and other Strangers, where almost every frame is a mixture of two source frames. It is IVTC'd real easy.

But, during the closing credits, the normal 2:3 pulldown is used.

I'm going to do a test on a section doing:
telecide(mode=3)
Tdecimate(mode=2,rate=25)
Unless someone has a better idea.

[GMJCZP]

Please read all this thread and maybe give you an idea to solve your problem, they are only two pages.
Here (https://forum.doom9.org/showthread.php?p=1772417#post1772417).

[MrVideo]

Please read all this thread and maybe give you an idea to solve your problem, they are only two pages.
Here (https://forum.doom9.org/showthread.php?p=1772417#post1772417).
That was dealing with animation + the file being worked with is progressive. My source is totally different from that discussion.

The test I did of four minutes worth of video resulted in great video @ 25fps. I stepped thru about 50 seconds of the video and it was smooth.

Nothing in the credits even indicates that it was an overseas co-production, with CBS Studios.

[manono]

2C/2M/4C/2M/4C/2M/3C/3M/1C where "C" equals a clean frame from the source and "M" means mixed fields from the source.

The six frame cycles are proof it's from a PAL source and has to be either 24.975fps or 25fps, although 25fps is the safer bet.

You can't just count 23 frames and then declare it to be 24.7578fps as you did. But I guess you got it figured out.

[Sharc]

That comes out to ~24.7578 source fps. WTF? That is close to 25fps, but not quite.

That means slowdown (25.0/1.001) in order to obtain 29.97fps rather than 30.00fps with the telecine pattern. Nothing unusual, I think.
Your pattern is just one possibility to convert 25fps (25/1.001) to 29.97fps, instead of the more common method of slowdown 25 => 23.976 and then apply standard 2:3 telecine for obtaining the 29.97fps.

The advantage of your "strange" pattern is that the audio pitch shift becomes very low (0.999) and usually requires no compensation, compared to the 4% pitch shift one would get with the standard method.

Can you upload a few seconds?

[MrVideo]

The six frame cycles are proof it's from a PAL source and has to be either 24.975fps or 25fps, although 25fps is the safer bet.
First off, a pet peeve of mine... using NTSC/PAL analog terminology in the digital world. Analog broadcasting is dead and gone, so NTSC and PAL are deprecated. In the digital world, the descriptors are the frame rates and resolutions.

That said... I don't normally deal with 25 fps -> 29.97 fps conversions. So, that pattern was unknown to me.
You can't just count 23 frames and then declare it to be 24.7578fps as you did. But I guess you got it figured out.
I didn't just count 23 frames. That is the number of 29.97 fps frames it took before the pattern repeated. But ya, I figured it out based upon how close the pattern comes to 25 fps. So, I figured 25 fps was a safe bet and went with that. Worked out great.

As a side note, someone else took the same material and did a 23.976 IVTC and that video "jerks" about every second as a frame was getting dropped.

[MrVideo]

Your pattern is just one possibility to convert 25fps (25/1.001) to 29.97fps, instead of the more common method of slowdown 25 => 23.976 and then apply standard 2:3 telecine for obtaining the 29.97fps.
No idea why it was done the way it was, but it keeps as close as possible the real time spatial info and the audio not becomming a "slow talker."
Can you upload a few seconds?
30 sec video sample (http://vidiot.com/video_sample_25fps-30fps.ts)

[Sharc]

Yes, it's 25->30 telecine (or 24.975 -> 29.97). I am not sure how it is called, maybe 2-3-2-2-3-2-3 pulldown (with one 2-2-3-2-3 cadence every some frames).
I wonder how far this cadence is supported by HW players & TVs for IVTC, or whether they just play it bobbed with the slight pulldown judder. The slight judder is still much less annoying than some ugly blends I have seen for other 25-> 29.97 conversions.

[StainlessS]

There is more "Really Strange" Telecine stuff described by Scharfis Brain via MediaFire link below in my sig, DATA folder,
web page translated to English from the original Klingon:-
Interlacing_by_Scharfis_Brain, original name "Exotisches Interlacing" @ http://home.arcor.de/scharfis_brain/ExotischesInterlacing
Link is no longer valid and I dont have untranslated page.

EDIT: Just did search on 6 physical hard drives (those currently connected to my system), and not found original Klingon
version ( :) ), although it is possible that I might have a copy lurking on some other hard drive or CD/DVD disk, I would expect not.

Would be a real good idea if someone reliable (where are you Librarian ?), would take a copy of the English version,
just incase my Mediafire account gets zapped and I die in some ghastly accident or something. Might also be prime material for the Wiki
if someone had the mind to add it, is mostly understandable but not being native Klingon, I have likely made a few mistakes.

Image sample from web page:-
https://s20.postimg.org/pbpwea16l/Exotic_Interlacing_zps6oz4kuek.jpg (https://postimg.org/image/afrd6oprt/)

[manono]

I didn't just count 23 frames. That is the number of 29.97 fps frames it took before the pattern repeated.
A better indicator of what you have is often found by putting on TFM by itself and then counting duplicate frames. This one's interesting as it's been time compressed, I think, because the 'source' framerate is greater than 23.976fps.

With TFM(pp=0) alone I get this pattern:

665 6665 665 6665 6665

There are some slight variations but that's pretty much it. Each number represents a 'mini-cycle' of a sequence of unique frames ending in a dupe frame. There are a total of 103 frames in the 'grand cycle' of which 18 are dupe frames. Using:

TFM()
TDecimate(cycle=103,cycler=18,MaxNDL=6,m2PA=True)

You get an actual framerate of 24.733fps, not very far from your initial guess. I noticed a couple of dupe frames at scene changes, but no missing frames that I could see. I think I was wrong about it being from a PAL source (or a 25fps source, if you prefer), but the patterns we saw were the result of time compression - speeding it up to fit a time slot exactly. And the version you saw that had been IVTC'd to 23.976fps would be missing frames, as you said. And yours at 25fps would have some dupe frames every so often, which is way better than missing unique frames. An interesting sample, thanks.

[Sharc]

... This one's interesting as it's been time compressed, I think, because the 'source' framerate is greater than 23.976fps.
Not sure if I got this right: The pattern indicates a speedup (time compression) from 23.976fps (or 24fps) to 24.733fps in order to achieve a certain target playback time @29.97fps. Does it really exclude a 25fps source which was slowed down to 24.733, then apply the same pattern in order to telecine it to 29.97fps? (Although one might prefer a different pattern for 25 -> 29.97fps telecining).

[MrVideo]

Thank you for the report. I keep learning a lot of stuff here recently. I'm wondering if they'll speed up next week's episode as well. If so, I'll have the tool to investigate it.

Deep down inside it just seemed wrong that it was 25 fps. Never thought that it would be a speed up issue. I've seen the ABC network do that, but after they've received the 720p59.94 air master. That resulted in blended frames.

This one was a lot better as the speed up was being done with the 23.976 progressive master, using telecine to get there. No blended frames or fields.

Thanks again.

BTW, I'm recoding using your filter settings.

[MrVideo]

Does it really exclude a 25fps source which was slowed down to 24.733, then apply the same pattern in order to telecine it to 29.97fps?
This was a 100% U.S. production, so shooting at 25 fps would make no sense. It would be a first in U.S. history (that I know of).

[manono]

Not sure if I got this right...
You do, but some of what I said was just informed speculation. I don't know what show it is or its origins. The speaker has a British accent (I think) but by itself that's not much proof of anything. MrVideo did say the end credits were hard telecined 29.97fps, and that's a clue I should have noticed earlier before blithely declaring it to have originally been 25fps.

Edit: I now read MrVideo says it's a US show.

Coincidentally, a friend recently sent me another time compressed show to study. Both the opening and end credits were 23.976fps hard telecined to 29.97fps, but the body of the episode had been time compressed/speeded up to 24.xxx something.

So, maybe the production company made it for a certain length and maybe the network, wanting more ads in it after ordering the show, speeded it up to what we see. Just guesses.

Does it really exclude a 25fps source which was slowed down to 24.733

I suppose not, except I've never heard of a show being slowed, whereas these days there are countless examples of shows being speeded up. And MrVideo's latest comment seems to contradict that theory.

[MrVideo]

Edit: I now read MrVideo says it's a US show.
I'm surprised that no one has figured it out.

[FranceBB]

Interesting, really.
In the Sat TV I work for, we are speeding up promo ads (not commercials from external companies, but the ones we use to promote shows that are going to air) in order to get a fixed length (00:00:05:00, 00:00:15:00, 00:00:30:00 (hours: minutes:seconds:frames), but we would never modify the length of a show, except for the usual 23.976 -> 25 speed up with pitch adjustment, of course.

[Sharc]

This was a 100% U.S. production,.....
Oh I see. I missed that point and was hoping to see a decent (and novel) 25 to 29.97 conversion. :)
(For 25->29.97 telecining the simpler and more common solution would probably be 2-2-3-2-3 pulldown (or a cyclic shift of it) resulting in 30.00fps or 29.97 with very little slowdown of 0.999 i.e. hardly noticable audio pitch shift)

[Katie Boundary]

This was a 100% U.S. production, so shooting at 25 fps would make no sense. It would be a first in U.S. history (that I know of).

Mainframe was a Canadian company, but Canada is on the NTSC standard just like the US, so it made no sense for them to animate their stuff at 25 FPS, but that's what they did. And they had a crazy pulldown pattern for converting their stuff to 29.97 without slowing it down by 0.1%.

Short version: North American productions made for PAL with crazy pulldown patterns to convert them to NTSC are not a new phenomenon.

[MrVideo]

First off. as pointed out previously, NTSC and PAL are deprecated. They no longer exist.

Animation is not the same as producing scripted dramas. I've not heard of it being done before, especially when it is 100% produced by an American production company.