HD For Indies

SOME COMMENTS ON DAY ONE OF THE DIGITAL CINEMA SUMMIT:

On the spec itself:


It's good that the spec is media independent -

now that the specs are here, the tricky part -

what's it all mean?

It's lumpy, complex and expensive

it'll keep smaller players out or at bay...right?

you'll need technically sophisticated help

Off the hip thoughts/conjecture:

HEY! IS 2048X1080 SQUARE PIXELS OR NOT? HOW TO HANDLE 2.4? (I think it's square and you'd crop vertically? Dunno...)

250 Mbit was found acceptable for 2K @ 24fps...not 4K, not 2K@48fps. 4K @ 48fps isn't even in the spec. The industry may be shooting itself in the foot by choosing what is doable now rather than what it needs to be. 2K@24fps is great for now..but what about 4K? What about 2K 3D (48fps), what about 48fps 2K? It just smelled wrong to claim this'll last for 40 or 50 years with this standard. If you want a 50 year technology lifespan, you then need some really, really mature stable tech to hope it'll last that long. Otherwise, you're too financially constrained up front to bootstrap it to start. Like 4K projectors for now...they're barely doable. They did well to specify wide color gamut, 2K & 4K resolutions, 12 bits/channel non-linear...but this one detail, 250Mbit JPEG2000 may be the achilles heel of the whole system. The good news is that it's supposed to be pretty easy to make a good JPEG2000 encoder. The bad news MIGHT be that also implies there's not a whole lot of optimization to be done on JPEG2000 encoders, that unlike MPEG, it won't get much better over time.

Talking to someone, it's NOT the chipsets for decoding that are the limitations, it's not the drive speed (250 Mbit is about 30 MB/sec...totally doable on a single drive, even across the entire capacity). Conjecture: perhaps it's because the current servers would require too significant a retooling to handle it? What a shoot yourself in the foot gesture...to have got so much else so right and shortsheet it on this one crucial gesture....

I curious/saddened that they declared 24fps it, with a 48fps alternative. This doesn't play nice with existing broadcast standards...still. 60 fps would be idea, but they aren't doing it, because the bandwidth was quoted as a problem. However...playback at 250 Mbit is about 30 megabytes/sec....single modern hard drives do that no problem. Perhaps he meant network bandwidth? Where's the pinch point?


DCI is a brittle system - you can afford the spec or not....2K projector vs 4K is only point of flexibility I see so far. With film, you can keep handing it down - big markets get first prints, as those are finished send to smaller markets, etc. Smaller markets can afford the basics - a 35mm projector and some kind of sound system. DCI spec calls for projector, server, all kinds of infrastructure which is likely to be pricey.

ON JPEG2000 FORMAT

Good choice, not interframe nonsense (no frame to frame compression, each frame is it's own freestanding entity). You sacrifice some bandwidth, but the quality stays up during harsh transitions.

The quality is good, the flexibility is good, I'm just dissapointed at the bandwidth limitations in place.


ON AUDIO

LOTS of headroom here to do whatever they want. 16 channels of 48 or 96 Khz audio is TONS of room. 24 bit audio, too. Far better than CD (which is 2 channels of 16 bit 48 Khz)

ON PACKAGING

Sounds like they've got it under control, I like the metadata open endedness of this with MXF and XML. I like that the movie itself insists on being seen correctly - the unauthorized Mormon recutting of the topless scene in Titanic couldn't happen again. (Heard about this from Wendy.)

The guy discussing the spec talked about how human readable it was as an XML file....but that realllllllly doesn't jive well with the "just hire some high school/college kids" to run theaters. I should get friendly with Alamo Drafthouse folks some more and see who runs what...

ON POST WORKFLOW FOR THIS

nobody (at the high end) was too worried about the XYZ stuff. You get a box, you plug it in...no big deal. For indies, perhaps you take things as far as you want with your system and hand it over to the XYZ house...they'd need a profile of your color authoring environment to make sure it carried over correctly....doable in theory, hassle in practice. How close would you get with "Yeah, I was using a SuperBigCo projector model # 1138, white point 5000, gamma 2.6, Xenon bulb" and that gets them going....

As a practical matter, you just do your thing and take it to their place, say "no no no...mine was lighter/darker/whiter/etc" and spend some time checking color on their setup. Perhaps a generalized adjustment and you're off...



ON SECURITY VS PIRACY

The DRM they want to put in place feels similar to military levels of paranoia on distribution of military intelligence. Failure to communicate is sometimes desirable if communication isn't guaranteed secure.

They talked about only a very few circumstances where you'd get "dark screen" - if you messed with the content, tried to play it on wrong server or outside allowed dates, etc. But that's only if everything is working right....and there's a loooooooooot of pieces that need to play right here.

In current practice, the film is a point of vulnerability, but it's fairly well controlled since it's a single piece of physical media that is awkward to read into digital form and has forensic marks on it.

BUT....camcorder theft is still a wide open field. They can put forensic marks on it to backtrack after the fact. This will help cut down on "from the booth" theft, and will assist in investigating viewer theft, but won't stop it beforehand. "Spiking" films with the brown dots is distracting to the audience and creates difficulty for the camcorderists, creates a recording problem for them, but doesn't stop them altogether. They'll have a very well documented base of "Yessir, the barn door was in fact left open at 0837 hours, and an unknown quanitity of cows got out." But no idea who opened it, or how much value got away...

Good quote: "We've created the most popular online game, it's very challenging: it's called piracy."

The studios are very, very worried about digital piracy. Notice large chunks of time spent discussing piracy as well as the technical capabilities of these systems.

One other thought for the long term - once they go this route, if there is a problem with the DSM, there are no longer tons of readily readable digital copies in the field.


KDM (key delivery message) is the unlocker for content - only good for a given device in a given time window

Clearly, the studios are taking security VERY seriously. They are scared to death someone is going to make perfect digital copies of their stuff, to the point that they're worried about someone breaking open a D-Cinema projector and inserting probes into the guts of the machine between the on-the-fly decryption chipset and the part that feeds the projector. Clearly, at some point, a plain, raw, unencrypted signal will have to exist in some machine to be projected....but they're trying to protect against everything they can. Because even if someone did this...the plan is to have watermarks and forensic fingerprints in it to know which exact projector (down to the serial #, who it was sold to, and where it is supposed to be) and AT WHAT TIME OF DAY it was being projected. If the encrypted keys are decrypted, it's just for the copy for that one projector and that one movie, NOT for all the movies everywhere as was the case with DVDs. They are upping the ante to military levels of paranoia.

ON DCI IMAGING SPEC

2K/4K flexibilty - great. Adequate to more than so.
12 bits/channel - Wow. People can't tell the difference. The valid point was raised that using 12 bits to cover a color space that goes WELL beyond what the human eye can perceive is a bit pointless; there's a lot of wasted bandwidth in there. The presenter agreed, saying only about 1/2 was used. But they found 11 bits adequate, 12 bits gave double the space.
audio - all the room you'll ever need. 16 channels, and can have alternate sets of 16 channels for foreign languages. Wow.
JPEG2000 - good compression choice, meets the needs
250 Mbit/sec - DAMMIT. Good enough for 2K@24fps. 4K @ 24fps WAS NOT EVALUATED from what I was told by someone who supposedly knows. 2K @ 48fps wasn't evaluated, I don't think. I need to find out more, sounds like this could be the weakest point in the whole structure.
Security - Secure? Very. Complex? Much moreso than presently. Don't expect $8/hr employees to understand it all. "Uh, she broke, boss." might be a common refrain. The level of technology being brought to bear, compared to the skill level of those expected to use it is SCARY. The ability to fully automate a theater is a good thing....because it'll have to be fully automatable to run in this user environment.

Overall - it's ambitious at this point. It'll take some more time to nail down the practical business execution side of things - what logging data gets reported to whom, how often do dark screens happen (miss a screening, $4000 lost on a Friday night), and all of the technical infrastructure for a multiplex can be daunting. There's another aspect of it - might this add yet MORE pressure for theaters to have more screens, if the core data management infrastructure is a fixed cost to get started? It's gonna cost BIG to get into this game at the level they're asking.

Smaller markets would be in trouble for awhile, since they wouldn't be able to afford this level of gear. A thought - at exactly what size of community could you cost justify an install? Since theaters only really service folks within a 15 mile radius (that's the furthest I drive anyway) at best in cities, and sometimes just a few miles, what kind of population density is required to cost justify one of these suckers? There's a crucial graph - how big the local community in "convenient" driving range has to be to cost justify placing one of these....then cross that with what percentage of US population lives below that threshold. Would we be seeing another cultural ghettoization, like us poor souls who were late to get cable? Growing up in Austin, TX (state capitol, many tens of thousands of university students), we were "hearing about the MTV" for a couple of years before we got it...and Austin while not huge was not exactly small. It's always big markets and then trickle down, but how far the trickle down goes and how long it takes is the pertinent question. Even SMALL communities have movie theaters for 35mm. And Oh Yeah - you can't really charge more for digitally projected movies, either. The only thing better is the quality - if done right, it's better than answer print, and never degrades.

A big server, network, digital projectors...just to take a completely wild stab at it, I'll guess (don't know real #'s) it's $100K of infrastructure, and digital projectors (I'll be nice) at $75K apiece with all the goodies these guys want (that's below current market price for such things...at least based on last NAB's pricing and the recent Sony 4K projector pricing). Plus all the audio stuff, but that's largely the same. I heard somewhere along that way that a good 35mm projector was $30K...still true? This theoretical 10 plex would require $850K of digital gear, or $300K of analog projector gear. (This is pointless, I so don't know the numbers involved). It'll take some time to make $1/2M in difference up in saved costs. Then again, I hear Hollywood spends $600M to $700M on prints each year. Assuming for the moment that once you have a DI the prep costs to compress and package up the stuff and put on a recycable hard drive is about $100, vs $300 (heard that # somewhere, I could be so wrong), it would take
3.5M prints to....oh never mind. I don't know the real numbers here so this is bullshit wasted time. Gotta eat.

-mike