UK to Build Network of 150 Digital Cinemas
mikael writes "According to this article at the BBC, a network of 250 digital screens in 150 cinemas across the country is being planned. Each film is losslessly compressed from 1 Terabyte down to 100 Gigabytes and encrypted onto a portable hard disk drive with a key unique to each cinema, which is then delivered to the cinema. Each cinema projector will be capable of showing films at resolutions of 2048 x 1080 pixels. "The key benefit is the distribution and screening of documentaries, British and foreign language films, as making a digital copy is considerably cheaper than spending over £1500 pounds to make a copy of a single film". Other benefits include better picture quality and the ability to show more films each day." The UK Film Council has a brief overview of the project as well.
Isn't 2048 x 1080 significantly less than regular movie film as far as resolution goes?
Subject says it all. There's something fishy about a feature film at 1080p24 compressed "losslessly" down to 100GB. That's 573GB (yes, bytes) per hour uncomrpessed, assuming 24 bits per pixel. Even D5 compression isn't lossless, and that's 5:1.
To get anywhere near that much, you have to at least convert it to the sum of cosines using Fast Fourier Transformation, which, since it distorts the data by converting it to not the exact amounts but the nearest amounts, is inherently lossy.
Any programmers in the UK want to start a lawsuit for false adverts?
- - - - - Fear not the reaper, but my shiny white teeth.
Perhaps the Media Companies really DO get it, but dont want to lets us know they do..
Still, I wonder exactly what scheme they use to play these.. And, if I work out the numbers...
100 GB for 2 hours. Thats 7200 seconds.
We dont know if thats GB or GiB, so lets assume its GB. 100GB/7200sec or 1 GB per 72 seconds. Thats about 13.9MB per second for all sound channels and video.
If they really do spend THAT much on making vinal film, why not instead hook up to a fiber optic network and transmit ALL films to a server at the theater?
Granted, that's a pretty high resolution by most people's standards, but take into account that it's being projected onto a 30 ft or larger screen, and it seems (to me) that it's not a high enough resolution.
Someone once mentioned to me that the frames that Pixar renders out for it's films are something on the order of 4000 x Something resolution, which sounds a bit more comperable to film.
Presumably each file is digitally watermarked, so that rips could be identified as having originated in one particular projection booth.
So... what do you do, as a theater owner, knowing that your ass is personally on the line if pirate copies of your copy of the movie appear?
If it were my theater, I'd have the only key to the server room, which would be the only place that the hard drive would do any good.
Since the data has to flow from the server room to the projector when the movie is being shown, I'd enforce access logs on the server, so I could tell if the file had been read at times other than showtime.
But that still doesn't stop the $8/hour projectionist from installing a device that intercepts the data, copies it, and then passes it along to the projector.
Are there such things as video projectors that accept an encrypted stream of data?
Never attribute to malice that which can be explained by mere idiocy.
If its found to be watermarked, the scheme becomes slightly more complex. If the watermark is undetectable, or not easily able to be found (which they aren't supposed to be, correct?) You need someone else at another unrelated theater to grab a digital copy as well. Assuming both are digital copies, then a frame by frame comparison should point out any watermarks. A little manual (or scripted) touching up will take care of it.
Regards,
Steve
35mm digital pixel equivalent
There's been lots of fuss about this digital cinema system. Appearently the projectors last 3-5 years before the technique is "outdated". Sure it cost much less to get a digital cinema projector. But the analog last for 15 years or more.
Consider the screen size, for a rather small 15x45 screen the pixels will be 1/4inch x 1/6inch tall. It might seem little but its not, you see the pixels if you start looking at the picture quality, plus you need one very powerfull projector (6000 lumens and up) to get decent contrast. Using systems like Watchout, Blend Pro or whatever else you choose you can have resolution 4 time higher as source, make each projector project 1/16 of the source at native resolution of 1280x1024 (to date no projector have a native resolution of 2048x1080), 16 relatively cheap projector of 1500 lumens later you end up with a projected resolution of 5120x4096 in which you may fit as many lower resolutions as you wish, each pixel is damn small and you have a very well spread 9000 lumens projection. Cheaper and better...
Anshutz is rather intersting. He is major stockholder in Qwest (until he finally gets thrown in jail for all the qwest stuff that pulled). As part of that, he was trying to figure out how to fill the pipelines that he has all over the country. and the Answer: send digital movies. So now, he is busy buying movie houses under the name of regal and getting ready to turn them digital. All of them will be filled via qwest lines (or some local if qwest is not in area).
I prefer the "u" in honour as it seems to be missing these days.
Assuming both are digital copies, then a frame by frame comparison should point out any watermarks.
Not necessarily!
Assume the pixels of the movie come in a stream M. Now assume you have a pseudorandom sequence P1 that tells you which pixels to add your watermark data to. That gives you a pixel stream M' which is mostly zeros except for the watermark. You distribute M + M' = M1 to the first theatre.
Now do the same with another copy of the movie, with a different pseudorandom sequence P2. This sequence is uncorrelated with P1. You get another set of pixels M'' which have watermark data. M' and M'' are "mostly disjoint". Now do M+M'' and get M2.
There is no operation you could do with two movies that would remove the watermarks. You could only add a second watermark to the movie.. Example:
Subtract the two movies: M1-M2 = M + M' - M - M'' = M'-M'' .. you get a stream of random bits which represents the two watermarks, but with one "inverted".
Average the movies: (M1+M2)/2 = M + (M' + M'')/2 .. now you've added a second watermark to the movie. The MPAA will put the smack down on BOTH of you.
However, there's something in that last one.. what if you did this with all 150 movies? (M1 + M2 + .. M150)/150 = M + a tiny bit of noise representing all the watermarks averaged together.
Would this work? I don't know. They could easily thwart this too, by making sure the frames of the movies weren't identical. They could film at a higher frame rate, and then downsample to a slightly lower framerate, which would make sure each movie's frames were completely different. I don't know enough to contemplate the effectiveness of this. They could even downsample non-uniformly across frames with another pseudorandom sequence and it would be extremely difficult to reconstruct the original M, let alone remove the watermark. So if you tried the trick you would just get noise.
I suspect that digital projection will not provide anywhere near the rich brighness gradient we have come to expect from film.
2048x1080 resolution is BETTER QUALITY than analog film? Not likely. They are likely referring to the absence of scratches and whatnot that build up over time, but I have a hard time believing that 2048x1080 projected on a large screen will not look pixelated.