The Soviet Influence: Battleship Potemkin/Drifters

[Calvin]

Any word on the frame rate issue?

[MichaelB]

There's no way of reliably telling until my copy arrives, though I should get it well in advance of the release date (contributor's privilege!).

But I imagine the retention of the full orchestral recording of the Meisel score didn't give them much room to manoeuvre.

[antnield]

From the booklet:

"Battleship Potemkin is presented in High Definition from a 4k restoration undertaken by Transit Film, Germany with Omnimago Laboratories. The film runs at its original speed of 18fps within a progressive 24fps presentation."

[MichaelB]

That's good news - I knew they'd closely analysed the Kino encode when preparing this one, but I didn't know whether the retention of the Meisel score had restricted what they could do. But it sounds as though they've pulled it off.

Incidentally, this will probably be bad news for some, but I understand that this release is definitely Region B.

[antnield]

I've just popped the disc in for a quick glance at the transfers...

Potemkin runs for 68 minutes and 51 seconds, with the opening seconds devoted to the Deutsche Kinematek logo. The opening titles and subsequent intertitles are all in Russian with optional English subtitles and the framing is such that we can see the curvature on all corners. Damage is to be expected, of course, but detail and contrast levels look to be superb. The Meisel score is presented in either DTS-HD Master Audio 5.1 or LPCM Stereo.

Drifters is a touch softer, no doubt inherent in the materials. It is presented in a new 2k scan made from an original 35mm tinted print held by the BFI National Archive and runs at 24fps. Jason Singh has composed the new score. (The previous UK disc, issued by Panamint Cinema, came with piano accompaniment from Mike Nolan.)

The three shorts - North Sea, Granton Trawler and Trade Tattoo - are derived from the same masters used when they were included in the BFI's GPO sets. Needless to say, Len Lye in high definition is a wonderful thing! (See also The Birth of a Robot on the first Humphrey Jennings Blu.)

Oh yes, and definitely Region B.

[isakborg]

Looking for some bottom line information. To wit: how does the BFI BD Potemkin stack up against the Kino? A frame rate of 18 fps is mentioned for the BFI, with a running time just shy of 69 minutes; my Kino BD's running time (per the back cover) is 71 minutes. I don't believe the Kino is 24 fps, but with Eisenstein's fast cutting, it's often hard to tell. If the So: any ideas concerning the running time discrepancies? I'm always trying to find that ideal edition of certain silent films and am wondering if the BFI is an improvement on the Kino. Thanks.

[Calvin]

Off the top of my head, the Kino release is from a 25fps master that they slowed down to 24fps. The BFI release presents the film at its original speed - 18fps - within a 24fps transfer so it's the one to go for.

[MichaelB]

The Kino and BFI releases are sourced from the same restoration, so imagewise they should be all but indistinguishable (encoding issues aside) - but the technical producer of the BFI disc had the advantage of being able to examine the Kino disc and work out ways of improving the framerate presentation. Antnield's post suggests that this is indeed what's happened, so it should be the best Blu-ray version to date - but I haven't seen it myself yet.

[swo17]

Not to mention that at least a couple of the shorts included with the BFI are bona fide masterpieces.

[andyli]

I don't understand the maths here. If BFI=18 fps and Kino=24fps then how come the difference is only a little more than 3 minutes (Kino transfer clocks in at 72:01 according to DVDBeaver)?

EDIT: On second thought this is more like a 24/25fps discrepancy. It is possible that BFI simply went with the 25fps master and didn't slow it down.

[hearthesilence]

A little confused, if the BFI frame rate is 18fps within a 24fps transfer, is Kino's rate actually 24fps within a 25fps transfer?

[Calvin]

I think Kino's is just 25fps slowed down to 24fps.

EDIT:

Relevant posts from the Kino thread:

I don't have a purely scientific answer for this, but Battleship Potemkin is 18fps and their Blu is progressive. This means that they have some way of tricking the format.

I watched the BD of Kino's Battleship Potemkin, and it seems they did this: The 18 frames per second were transfered to 24 frames per second. every 3rd frame is repeated as the 4th, so of the 24 frames on the BD, 6 are repeated.

That is also why the BD is a few minutes longer than the DVD, which was interlaced from the 18fps source.

Also, just thought about this, but how is everyone pronouncing KINO?
Kee-No? Ky-No? I always thought it was Kee-No but then again, never heard anyone say it.

Kino's BD of POTEMKIN is not as you describe.

If it were done properly, as you describe (3 unique / repeat / 3 unique / repeat) then it would be the same running time as 18fps 35mm. It does not use this method, I've analysed it, and the pattern replicates that of a 25p HD master (which was created by the Germans for the Transit PAL DVD release). To get to 24p (for 1080p) they simply slowed down the 25p master to 24p (like Criterion's ANTICHRIST or BERLIN ALEXANDERPLATZ). So what you end up with is one extra unnecessary repeat frame every second, and an unnecessarily longer running time. From the top of my head, the Transit PAL DVD is 83 minutes and the Kino BD is 87 minutes.

The result is visually not too bad, but if they'd gone the extra mile, it would have been worth it. Stripping away all the repeat frames from the 25p master and then extrapolating your own 3/repeat/3/repeat pattern to go from 18 to 24 would have resulted in slightly better motion and a proper running time of 83 minutes (or whatever the 18fps 35mm and PAL DVD runs at, I think it's 83 minutes).

(btw. Pretty much everyone says "Kee-no" as in "the German for cinema")

[TMDaines]

A little confused, if the BFI frame rate is 18fps within a 24fps transfer, is Kino's rate actually 24fps within a 25fps transfer?

No. Read the posts above. Unless I'm misunderstanding something Kino's speed is (18/25) * 24, thus the film is at 17.28 fps. If you played the Kino video at 25fps and the BFI video at 24fps, they would both be showing the film at its correct 18fps. As 1080p footage is at 24fps on a Blu-ray, the Kino has to be slowed down.

[MichaelB]

A little confused, if the BFI frame rate is 18fps within a 24fps transfer, is Kino's rate actually 24fps within a 25fps transfer?

No. Read the posts above. Unless I'm misunderstanding something Kino's speed is (18/25) * 24, thus the film is at 17.28 fps. If you played the Kino video at 25fps and the BFI video at 24fps, they would both being showing the film at its correct 18fps. As 1080p footage is at 24fps on a Blu-ray the Kino has to be slowed down.

That finally makes sense! And if you do some further maths, you end up with...

BFI: 68 mins 51 secs (source: antnield above)
Kino: 72 mins 01 sec (source: Beaver)

BFI/24*25 = 71 mins 43 secs
Kino/25*24 = 69 mins 8 secs

So there's a discrepancy of less than twenty seconds, and I'm willing to bet that this is explained by distributor logos of differing length or one version given a few more seconds of black leader at the end, or something equally obvious. Since they're both from the same restoration, there shouldn't be any differences at all in the content of the actual film.

[andyli]

Nothing made sense yet. That's not a discrepancy of 20s. 71:43-69:08=2:35 which is too long to be explained by distributor logs and such.

[MichaelB]

Nothing made sense yet. That's not a discrepancy of 20s. 71:43-69:08=2:35 which is too long to be explained by distributor logs and such.

You misunderstand me - I'm talking about the comparative running times of the Kino and BFI versions after making the 24/25fps adjustment. The BFI disc would run 71 mins 43 secs (versus the Kino's 72 mins 1 sec) after such an adjustment, and the Kino would run 69 mins 8 secs going the other way (versus the BFI's 68 mins 51 secs). Hence a difference between them of approximately eighteen seconds.

We already know that the film itself is sourced from the same HD master, so such a discrepancy is almost certain to be down to differing distributor logos or something similar.

[andyli]

Now I see how you compare them. Yeah that makes sense. So the conclusion is while Kino slowed down from 25fps to 24fps, BFI stuck to the original master and did some magical pull-down to 18fps without altering the running time?

[MichaelB]

Based on Peerpee's analysis:

If it were done properly, as you describe (3 unique / repeat / 3 unique / repeat) then it would be the same running time as 18fps 35mm. It does not use this method, I've analysed it, and the pattern replicates that of a 25p HD master (which was created by the Germans for the Transit PAL DVD release). To get to 24p (for 1080p) they simply slowed down the 25p master to 24p (like Criterion's ANTICHRIST or BERLIN ALEXANDERPLATZ). So what you end up with is one extra unnecessary repeat frame every second, and an unnecessarily longer running time.

...I'm assuming that the BFI simply dropped the unnecessary repeat frame every second, and presented the film as 3 unique / repeat / 3 unique / repeat throughout. Which would allow them to present an 18fps film as a 24fps encode while maintaining the correct speed and also avoiding the need for an interlaced transfer - and it should also run smoother than the Kino because of the total consistency of approach.

And this makes perfect sense once you compare the running times, because the Kino's extra length can indeed be mostly explained by the presence of a single extra frame every second.

But I can't stress enough that this is speculation - I think antnield is the only person in this thread who actually has a copy of the disc!

[andyli]

Wait... where does "the presence of a single extra frame every second" come from? I thought Kino simply slowed down the 25fps master, which leads to the longer time? If you slow down from 25fps to 24fps shouldn't you get one fewer frame as opposed to one extra every second?

[MichaelB]

Peerpee's analysis. Based on first-hand examination of the Kino disc frame by frame, he reckons that Kino took a 25fps encode with an extra duplicate frame every second, and slowed it down to 24fps, which accounts for the slightly longer running time.

By contrast, it looks pretty likely that even if the BFI was working from the same original master, they stripped out the extra duplicate frames so that the video master ran natively at 24fps, and the even distribution of original and repeat frames (3-1-3-1 throughout) created a perfect simulation of 18fps within that.

I'm reasonably sure that this is in fact what the BFI did, because I remember discussing precisely this process with the transfer's first technical supervisor early last year - and the comparative running times suggest that his successor did something similar.

[andyli]

So it's the 25fps master that has an extra duplicate frame per second! That's a crappy way to make a master other people have to rely on.

[hearthesilence]

Thanks for the answers!

Jesus, I can picture some poor schlep at the BFI sitting in front of a computer for hours on end, moving though that master frame-by-frame and cutting out each duplicate one-by-one. It would be a simple but tedious, painstaking way of doing it.

[MichaelB]

So it's the 25fps master that has an extra duplicate frame per second! That's a crappy way to make a master other people have to rely on.

Not if it was primarily intended to fuel European HD broadcasts - which may well have been the case.

Jesus, I can picture some poor schlep at the BFI sitting in front of a computer for hours on end, moving though that master frame-by-frame and cutting out each duplicate one-by-one. It would be a simple but tedious, painstaking way of doing it.

I strongly suspect it's an automated process! Stripping out one specific, regularly-spaced frame every second doesn't sound as though it would be a huge challenge to automate - you just have to make sure that you get the right starting point. But I daresay they checked this.

[peerpee]

So it's the 25fps master that has an extra duplicate frame per second! That's a crappy way to make a master other people have to rely on.

Not if it was primarily intended to fuel European HD broadcasts - which may well have been the case.

Dr Anna Bohn – who was part of the original German restoration team – asked my opinion (circa 2007?) about what master format a DVD label would prefer to receive for such a project. Considering the framerate of the film, I advised HDCAM SR, definitely progressive *not* interlaced, and thankfully they did that.

They aimed for 25p though, because at the time (this is 4-5 years ago or something?) their primary concern was a German PAL DVD.

Because they did all the work progressively, we have individual frames captured as individual frames, and it's super-useful for Blu-ray. Slowing the 25p master down to 24p (like Kino did), instead of removing the repeat frames and getting to 24p more elegantly, didn't really show a very good understanding of the technical side of what's gone on. I'm really glad the BFI sound to have nailed it (haven't seen the disc yet).

[peerpee]

Jesus, I can picture some poor schlep at the BFI sitting in front of a computer for hours on end, moving though that master frame-by-frame and cutting out each duplicate one-by-one. It would be a simple but tedious, painstaking way of doing it.

I strongly suspect it's an automated process! Stripping out one specific, regularly-spaced frame every second doesn't sound as though it would be a huge challenge to automate - you just have to make sure that you get the right starting point. But I daresay they checked this.

There's a plugin for After Effects which does this most neatly. We went through tons of options to perfect the MoC COEUR FIDELE Blu-ray, also taken from a frame-repeated 25p HD master, and rendered properly by MoC on Blu-ray in exactly the same way as this POTEMKIN issue. All thanks to encoding wizard Leroy Moore at IBF, one of the best in the business!