358 Pandora's Box

[HerrSchreck]

You see, Schreck, the reason I ask those questions is that it seems VERY unlikely to me that the only reason thsi release didn't get a progressive transfer is because it's low priority or something. This is a high-profile release. There must be a technical reason for this. They didn't just decide to not invest the extra money to do this progressively, especially after years of delaying the release.

I think you're making a distinction here that asks a hypothetical question: should PANDORA-- or shouldn't PANDORA-- be regarded as your run of the mill niche-market-within-a-niche-market silent, pulling the same sort of sales as something like JEANNE NEY or Griffith's BROKEN BLOSSOMS..? both of which are masterpieces, but pull very small sales numbers?

Take this site, which is haunted by dedicated psycho cineastes-- yet, Dent, our discussions regarding silent films pretty much run between the same group of folks... me, you, zedz, the German crew of vogler, ledos, Tommasso, La Cle', and two or three occasional others like dkmb & justelblanc. The 100-200 other dedicated regulars, as well as the approx 200 other semi-regulars, never join in the discussion. Silents just don't do to them what they do to us. We are a tiny minority even among cineastes. This fact has been illustrated by disc co's time & time again via title-sales. Does PANDORA benefit from the hugely agonizing & hyperextended wait illustrated over this and other BBSs' over several years for the disc to be produced, as well as the cultural significance of Brooks? I'm not sure, because beyond this film, and to a lesser degree TAGEBUCH, people are not interested in her films beyond her Pabst rendezvous. It's true that by a film's mere inclusion in the CCollection, interest is automatically generated in the minds of those who've never heard of it before-- it's mere announcement as CC-worthy makes folks want to see it. Will the folks who pick this film up because it's CC-- would they have bought it if HVe put it out with the same interlaced transfer? Do they own PRIX DE BEAUTE & TAGEBUCH? Of course not.

Is this film therefore, despite being a silent, equally interesting in the appetites of the members of this board, as, say BICCICLETE or VERONIQUE because of the lavishness of the presentation (meaning a huge booklet, a fold out pak, and massive amount of extras)? Will the other 95% of the board who are not silent-savvy going to buy this rather than rent, therefore justifying the added expense on CC's part?

I can't answer that. My theory is that this particular release was encoded interlaced because of 1) it's a slightly long(er than usual) film, 2) it's hi-def, 3) there are four scores, three of which are multitracked fullblown orchestral scores 4) plus a commentary, and 5) plus subtitles, all crammed into one RSDL disc. My theory is that in this case it was an issue of space.

As to cost issues of progressive vs. interlaced, this is obvious and old news though I'm incapable of providing percentage margins of cost-savings by "going interlaced". Cheeseball dipshit companies like Alpha et al never encode at hi-bitrates, and never frame by frame progressive. Progressive encoding requires extra labor manhours, and also a dual layer disc vs. single layer. If you multiply the added cost of these premium practices by the number of discs manufactured, you can obviously chop the stereotypically smaller profit margins right off of the whole enterprise. Indulgence in progressive transfers/dual layer discs is rarely seen in any penny-pinching/cost-wary disc enterprise-- not only silents. As to it having something to do with frame rate, or some other variety of "unique to silents" technical issue, I have no doubt that's a whopper of red herring. Kino has many progressive silents, albeit not preconverted, so the frame by frame playback of the film will have each 4th or 5th frame containing two images, and MOC is starting to encode silents progressively, as do co's like MK2 & Gaumont, albeit on certain releases only. Once the tape is created off of telecine, your talking about a process of converting this tape into a data file, which assigns each frame seperate storage (if progressive), and programming which controls rate of playback in strict terms of image data, not organic film. You could tell your program to spool off the frames at blinding speed or dragging slo-mo according to your whims, without any particular challenge to the authoring entity. Once again, 35mm is 35mm-- you run it thru telecine, create a tape, and create it's digital equivalent for storage and duping to disc.

In terms of home video, niche obscurities like silents, old bad films no-one's heard of, bizarro docs, etc, are all well-known for getting short-shrift when it comes to disc-production values. This is very old news, and pretty well known.

EDIT:

Not working in the field, I imagine that restoration work is actually more straightforward on a progressive master.

You're putting a little piece of cart before the horse there. First off I'm sure you mean digital resto with MTI typce cleanup software, as opposed to film resto. Interlacing occurs during the disc authoring process, post digital resto. Digital resto takes place on fullbown hardware with optimal storage on lavish harddrives, prior to the image & sound compression required to fit a film on disc, and the authoring process (i e where the interlacing process occurs, and the reason why it occurs.. i e for home playback).

[Tribe]

Shrek, would it have looked that much different if it has been progressively transferred?

Tribe

[HerrSchreck]

If you have a monstrous big-bucks progressive hidef setup with a primo plasma screen, etc, you'd notice the combing/trailing effect. The stitched-together frames wont be (mostly) invisible as on other equipment. If you have a standard def setup i e typical LCD flatscreen or tube, it'll look beautiful. One of my favorite transfers is Flicker Alley's recent PHANTOM disc, which is interlaced-- most reviews agree with my opinion in that the image produced by this disc is so frigging gorgeous it should be hanging in the fucking L O U V R E...

[Tribe]

But I am under the (mistaken?) belief the good quality progressive and upscaling DVD players DO in fact de-interlace the picture (the best ones in weave mode, not bob and some of them with faroudja chips etc) and in fact can largely get over this problem. But I agree this isnt as good as having a progessively rendered DVD to begin with.

Whoa! Are you suggesting that those nifty DVD players with the HD upconvertors might de-interlace the image?!?

I knew that they improved the image dramatically...but I wasn't aware that that was how it happened. Fantastic.

Tribe

[HerrSchreck]

I always thought the interlaced film frames (esp on single layer discs at low bitrates) did not exist as seperately encoded entities in storage on disc, whereby independent stepping was simply not possible because the individual images were not available as isolated entities. Like a deck of cards where every third card is two cards stitched together, photographed, and the composite replacing the two defaced cards.

And I always thought it was these hi-priced top-tier rigs where interlacing was most heavily felt & resented.

Therefore your "MOTHERFUCKER!!!" a few posts back when hearing the disc was interlaced.

No?

Are you saying you can take, say, BEHIND LOCKED DOORS and have it play back progressively where you can even pause & step thru the film frame by frame? Why should combing/flickering even be a problem then?

Wind up is of course that for the vast bulk (at least 95%) of dvd players out there, progressive playback of interlaced-authored discs is simply not possible.

EDIT: I will say of course that when I switch my unit from Prog to Interlaced, interlaced films look more jiggly & blurry due to trailing, etc. But I still can't step thru the feature frame by frame. In progressive mode, this is the easiest way to tell what a disc is-- if you pause, then can make it continuously stepping frame by frame without a frame appearing to either 1) dupe itself (i e no detectable frame change every two or three frame steps, like the same frame back to back... i e 1, 2, 3,3, 1, 2, 3, 3, etc... meaning it's interlaced), or 2) contain two frames every fourth or 5th frames, like a double exposed photograph... meaning progressive but not preconverted to/from pal/ntsc.... if you can keep stepping and each frame is independent from it's predecessor, then you've got yourself a genuine progressive disc.

[Ledos]

the German crew of vogler, ledos, Tommasso, La Cle'

'Ey! Howcome I have to be lumped into the German group?

[Ledos]

I always thought the interlaced film frames (esp on single layer discs at low bitrates) did not exist as seperately encoded entities in storage on disc, whereby independent stepping was simply not possible because the individual images were not available as isolated entities. Like a deck of cards where every third card is two cards stitched together, photographed, and the composite replacing the two defaced cards.

Frames on DVD's are usually stored individually, and you can step through them with the pause button on your machine. When you sometimes see duplicated frames, or frames that look doubly exposed, it's usually because the DVD has been stored with 30 frames for each second (only applies to NTSC discs), or because a silent film has been given more frames to simulate 20 fps (or whatever) playback while still being played at the regular speeds that an NTSC or PAL tv requires.

[HerrSchreck]

Read above--

I will say of course that when I switch my unit from Prog to Interlaced, interlaced films look more jiggly & blurry due to trailing, etc. But I still can't step thru the feature frame by frame. In progressive mode, this is the easiest way to tell what a disc is-- if you pause, then can make it continuously stepping frame by frame without a frame appearing to either 1) dupe itself (i e no detectable frame change every two or three frame steps, like the same frame back to back... i e 1, 2, 3,3, 1, 2, 3, 3, etc... meaning it's interlaced), or 2) contain two frames every fourth or 5th frames, like a double exposed photograph... meaning progressive but not preconverted to/from pal/ntsc.... if you can keep stepping and each frame is independent from it's predecessor, then you've got yourself a genuine progressive disc.

As for what you're saying about frame speed playback determining the ability to step thru each frame individually during pause, that's ... going to require some footnoting, and certainly has little to no bearing on the interlacing vs. progressive discussion. Playback speed has no bearing on genuine frame by frame stepping when the film is paused.

The reasons one sees two frames in a given step, or the appearance of a duplicated frame (i e no progress to a new frame between two steps when paused advancing), are for improper conversion from standard to standard, or interlacing, respectively. This is basic stuff. See the links I provided on the previous page, or review the definitions of Interlacing vs Progressive to understand the discussion at hands here.

a silent film has been given more frames to simulate 20 fps (or whatever) playback while still being played at the regular speeds that an NTSC or PAL tv requires.

Utter quackery!

This is all wildly OT so if you want to posit pet theories about stepping and phenomena you see on your machine, go to the Technical Discussion section.

[vogler]

the German crew of vogler, ledos, Tommasso, La Cle'

'Ey! Howcome I have to be lumped into the German group?

and me....I'm as English as cheese rolling!

The point is true though. I'd not really thought about it but there are only a few of us with a major interest in silents. More people take an interest occasionally in the odd film here and there but generally there's only the few of us who get wildly excited at the prospect of every new silent release.

[thomega]

this site, which is haunted by dedicated psycho cineastes

=D>

My theory is that this particular release was encoded interlaced [...] as an issue of space.

Interesting. I had thought that tweaking the compression would result in less artifacts at the same bandwidth than the rather crude method of interlacing that was only ever useful for analog devices. Also, if 160min of Ran with a commentary fits progessively on a DVD9, the 130min of Pandora's Box should fit, even with four scores. In addition, compression could be more efficient with b/w material than with color material.
But I admit that I'm arguing theoretically, never having used high-end video hard and software.

As to cost issues of progressive vs. interlaced, this is obvious

Not to me.

Cheeseball dipshit companies like Alpha et al never encode at hi-bitrates, and never frame by frame progressive.

I thought that they're encoding interlaced, because they're recycling old masters, which were interlaced from the telecine stage on. Redoing a transfer from scratch will be too expensive for Alpha et al.'s business model.

Not working in the field, I imagine that restoration work is actually more straightforward on a progressive master.

You're putting a little piece of cart before the horse there.

That's what I'm famous for ...

First off I'm sure you mean digital resto with MTI typce cleanup software, as opposed to film resto.

Of course.

Interlacing occurs during the disc authoring process, post digital resto.

Are you sure? As I said above, I can not image that you gain more than you loose by interlacing, unless you started from an interlaced master.

[tryavna]

Interlacing occurs during the disc authoring process, post digital resto.

Are you sure? As I said above, I can not image that you gain more than you loose by interlacing, unless you started from an interlaced master.

Where's Peerpee? Perhaps he can explain. I've always got the sense from his posts that the difference between getting a progressive or an interlaced transfer was almost as easy as flipping a switch. (Obviously, I'd think that a progressive transfer would take more time, but is it really that easy?)

[peerpee]

It depends on the telecine, and the ensuing master (usually digibeta). There are MANY ways of encoding the digibeta to MPEG2 (DVD).

Sometimes a master will have 3:2 pulldown (which is added, I believe, for something to do with showing on TV). If this master with pulldown is encoded progressively without removing the pulldown it can introduce allsorts of minor motion problems and there are many discs on the market that have been encoded this way.

To properly progressify, often, a number of renders are required. Remove the pulldown, deinterlace, then encode progressively. Many DVD publishers/authoring houses don't understand these particular issues, or are too pressed to do more than one render (which is usually interlaced), as each step costs money and ties powerful computers up for a day or two.

Obviously, I'm speaking generally here, and there are many other problems that can arise, especially if the digibeta master has been standards converted from one format to another before you get it.

[Gigi M.]

BACK COVER

[lady wakasa]

The point is true though. I'd not really thought about it but there are only a few of us with a major interest in silents. More people take an interest occasionally in the odd film here and there but generally there's only the few of us who get wildly excited at the prospect of every new silent release.

I definitely have the interest - I just never thought there was too much happening here (i.e. Criterion boards) specifically with silents. (Plus I don't have the technical background you guys do.)

I came here to see what was being said about the Pandora's Box release. But I'll try to drop in more often and hopefully learn something.

[davebert]

Obviously, I'm speaking generally here, and there are many other problems that can arise, especially if the digibeta master has been standards converted from one format to another before you get it.

Hey peerpee, is their a website or a book or anything about these processes? I've been hoping to find a resource that might lay out the technical requirements, steps etc. that would be undertaken on the elements. I've been hired to work at a company that loves to use non-progressive PAL-to-NTSC transfers (I got the feeling interviewing it's because they don't know any better) , and it might be a service to us all if I knew what specifically to suggest. Whether they'll take the advice, who knows...

Other than that, I won't have a chance to actually watch Pandora's Box for two weeks, but the fat booklet is quite solid.

[HerrSchreck]

Interlacing occurs during the disc authoring process, post digital resto.

Are you sure? As I said above, I can not image that you gain more than you loose by interlacing, unless you started from an interlaced master.

You don't truly think they restore the image in the shrunk down files created specifically for disc storage do you?

In other words-- there are two general (and I mean generally) storage points or repositories for the film after creating the d5 or digibeta post telecine: hard drives i e in digital resto labs on huge studio workstations with hi rez screens for very highly detailed examination far beyond what's visible at home, for proper restoration, music & video editing, etc where the frame by frame files are much larger, and where the film can be worked on in terms of image, soundtrack, etc. In other words the film in it's highest possible quality digital state-- this is where digital restoration takes place because this provides the team the highest possible quality image for detecting flaws, marks, deterioration, etc... same goes for sound, prior to it's compression to disc. Think in terms of fidelity. Think of the digital lab's hard drives as a large reel to reel, then think of the dvd as a little cassette that you used to put in your answering machines. If you wanted to edit a symphony, you would want to do so where you could hear it in it's highest possible fidelity, so you could truly hear what you were editing, no?

Labs don't work off of DVD's-- discs (the other general medium of long term storage) are a cost effective commercial medium to deliver to the consumer a very good, though compromised, representation of what was sitting on a digital videotape post-telecine, that has been subsequently dumped to hard drive, examined (in cases if the digital resto you mentioned) carefully, cleaned up with software on hardware, compressed into transportable files that can fit on a disc, and shipped to stores, all at a reasonable price.

[thomega]

You don't truly think they restore the image in the shrunk down files created specifically for disc storage do you?

Of course not.

But I believe that the world of video (and to a lesser extend audio too) is full of legacy formats that owe their existence solely to the limitations of now obsolete hardware. That's why I was wondering, if there might be still toolchains in production use that employ interlaced formats.

However, I concede the point that interlacing in the authoring step is indeed an effective measure to save bandwidth. Given the appropriate hardware (CRT displays or good deinterlacers and upscalers) it appears to produce less artifacts than higher compression to get at the same bandwidth.

Thanks for the education.

[denti alligator]

The Beaver suggests this is not interlaced. Combing is the result of another problem:

Secondly - this image exhibits 'combing or 'trailing' - which I understand is a process of the frame rate conversion (from European silent to current NTSC)

[miless]

The Beaver suggests this is not interlaced. Combing is the result of another problem:

Secondly - this image exhibits 'combing or 'trailing' - which I understand is a process of the frame rate conversion (from European silent to current NTSC)

maybe it's the same problem that occured with Haxan because it was shot at (roughly) 18 frames per second

[HerrSchreck]

It's not progressive guys-- whatever you wanna call it, the frames are interlaced. He's even complained on the SYMBIOTAXIPLASM (his next CC review following PANDORA):

I wish I could phrase this film experiment's appeal with one word... but I can't. 'Unconventional' seems to leave out so much. Criterion offer this long-unreleased auteur gem in a special two-disc edition, along with its sequel, Take 2 1/2, made thirty-five years later with executive producers Steven Soderbergh and Steve Buscemi. Take One is progressive and probably looks about as good as it ever has. It has a few blemishes and some visible digital noise - I suspect the screen captures give a fair representation. Criterion offer optional subtitles, which is appreciated as the audio can be as 'independent' as the image production appearance. Unfortunately Take 2 1/2, on the 2nd disc, is not transferred progressively and shows visible 'combing' (come on Criterion - cut this out). It maintains a healthy pragmatic fun feel.

He's probably just assuming you know what combing/trailing means. Or twitching to be hard on CC in a positive review for the same thing that causes him to condemn other co's.

It's not neccessary to interlace guys just because something originated in the PAL standard... They started with reels didn't they? So that goes out the window: they did their own telecine so that stuff goes out the window.

I mean come on-- have we forgotten why we've been beating up on Kino all these years? Absence of preconversion causes GHOSTING (extra frame within a frame, like a double exposed photo, not interlacing which is computer/digital generated artifact of stitching artifically).

[Ledos]

I mean come on-- have we forgotten why we've been beating up on Kino all these years?

That's for their PAL-to-NTSC conversions. Criterion don't do those (at least I don't know of any titles where they did it).

[HerrSchreck]

Maybe I'm confused about what you're saying. Isn't that what I said-- that Kino has been getting beat up for lack of preconversion? I only brought up preconversion because it's the strange explanation Gary gave for the combing/interlaced frames on PANDORA. Preconversion, like I said in the above post, is irrelevant, because combing/interlacing is not a result of improper conversion from standard to standard-- ghosting is.

Lack of preconversion of a digital videotape, for a distribution company that doesn't do their own telecine i e Kino w their FWMS silents (the primary reason that a company would want to preconvert a videotape prior to striking a disc from it is because they themselves did not do the telecine that created the tape... which CC always does for the most part, which makes the whole NTSC standard issue bizarre in this case), results in ghosting. Combing has to do with interlacing.

Is the combing/interlacing on SYMBIOTAXIPLASM or BEALES OF GREY GARDENS because of PAL/NTSC? How about KOKO... NANOOK?

No-- it's just because they're cutting corners. (Or just couldn't fit it on the disc, at least with their desired bitrate for definition.) That's why Gary's complaining-- he feels they're being cheap, cutting corners.

Are we clear now? I didn't make the disc, I can't answer for Gary for raising the standards issue... nor can I make the disc progressive. The fact remains, I'm reporting FYI and only FYI-- the disc is not progressive. And if you think the disc has to be interlaced owing to some super-mystery-europe-silent-thingy... I don't know how to combat that. I think you're just reflexively forgiving CC. FLOATING WEEDS is from Japan and is progressive, as are a goodly number of silents regardless of whether or not they cross standards.

[Ledos]

Maybe I'm confused about what you're saying. Isn't that what I said-- that Kino has been getting beat up for lack of preconversion?

Maybe I misunderstood you, but I thought you were making the point that it was unfair Kino gets bashed while Criterion does not.

And if you think the disc has to be interlaced owing to some super-mystery-europe-silent-thingy... I don't know how to combat that. I think you're just reflexively forgiving CC.

I don't know if you're addressing me, but I don't think something has to be interlaced. However, the point someone else was making is relevant - if a silent movie is supposed to play at, say, 18 fps then something must be done (such as repeating some frames) to make it look like it plays at 18 fps on a PAL or NTSC system.

[MichaelB]

Also, if 160min of Ran with a commentary fits progessively on a DVD9, the 130min of Pandora's Box should fit, even with four scores. In addition, compression could be more efficient with b/w material than with color material.

...though silent films often present their own encoding difficulties, because unless the print is in immaculate condition there's a distinct possibility that there may be greater differences between individual frames than there are with a more recent film simply because of print damage, debris, unstable framelines, etc - which of course is going to be far more demanding on a lossy difference-based system like MPEG-2.

I don't know whether this applies to a late silent like Pandora's Box (and I haven't seen this particular DVD), but some of the turn-of-the-(20th)-century material put out by the BFI presented encoding challenges very much along the lines faced by the creators of Criterion's Brakhage DVD collection, and for very similar reasons.

[HerrSchreck]

I don't know if you're addressing me, but I don't think something has to be interlaced. However, the point someone else was making is relevant - if a silent movie is supposed to play at, say, 18 fps then something must be done (such as repeating some frames) to make it look like it plays at 18 fps on a PAL or NTSC system.

Ugh.. I give up.

That's called pulldown, and on a progressive dvd shouldn't or create combing & trailing.

The pulldown factor is in effect for any film, not just silents, as the frame rates for PAL or NTSC video frame rate standards are far higher than the 24/25 fps for even modern sound film. On videotape you're talking about timed electronic tape fields vs. actual film frames (film), so fitting the film frames has always been the mathematical challenge getting any film to tape, silent or talkie.

From Answers.com

The technique for converting movie film to TV/video. Pronounced "tel-uh-sin-ee," "tel-uh-sin-uh" or "tel-uh-scene." Because film runs at 24 frames per second (fps), and NTSC video runs at 30 fps, telecine inserts duplicate frames into the video to make up the difference. Telecine has been used to convert countless movies to videotape for ultimate distribution via TV, cable and satellite networks.

In the video editing studio, when a TV monitor is used to observe the changes being made to film-based material, telecine creates a compatible format for the TV (the process may be reversed when the editing is finished). Lastly, the telecine function is built into DVD players to convert 24 fps source material to the TV format.

Because it is time consuming and costly to do the telecine conversion, videotaped movies that have already undergone the telecine process are sometimes used as the source for DVDs. As a result, DVD content may be comprised of 30 interlaced fps or 24 progressive fps.

Again, this is wildly OT... as to the reasons this disc is displaying the alternating stictched combing of interlacing, I can't answer.