Slashdot Mirror
Rodriguez uses Linux to Edge out ILM
An anonymous reader writes "A Linux device helped legendary independent filmmaker Robert Rodriguez (El Mariachi, Desperado, Spy Kids, Once Upon a Time in Mexico, and others) win the race with ILM (Industrial Light and Magic) to create the first movie ever to use a digital format supporting full-bandwidth RGB. Rodriguez's Sin City, which opens April 1, was shot in Dual Link, or "4:4:4" format, and transferred between tapes and hard drives using SpectSoft's Linux-based RaveHD DDR (digital disk recorder)."
I would bet that a lot of the digital effects used in this film were rendered and perhaps even designed with Linux. If they were done with Windows they would have used a blue screen.
Linux was vital to this project. As we all know, if it were Windows-based, the RGB mode would be 0:0:4, more commonly known to the industry as "blue screen".
Rodriguez's Sin City,
Actually, its Frank Miller's Sin City. IMHO the writer is more important than the director.
"For Great Justice."
Wow, that explains how vibrant the reds, greens, and blues look in black and white!
Les Miserables Volume 1 now up with my reading of
"It's really the same as 1:1:1," explains Howard. "It just means 'take RGB, break it up, send part of it down one wire, and part down the other wire.'"
The compromises in traditional Y'CbCr formats were designed to minimize perceptual loss, keying on the human eye's varying sensitivity to luminescence at various color frequencies. We are most sensitive to brightness in green light, less so in reds, and least with blue. This explains why studios often shoot against a "greenscreen" -- Y'CbCr has most information about green, so it's the easiest color for a software program to identify and replace.
An effective signature identifies a particular user amongst a base of thousands.
I've had unreasonably high hopes for "Sin City" ever since I watched that fabulous trailer. But it's not like graphic novels-turned-movies have impeccable track records on the big screen, so imagine my concern going into this.
l luloid_lane_.html
For lack of a more eloquent explanation, "Sin City" freaking rules.
Remember the first time you saw "Pulp Fiction"? You were unnerved and at times downright repelled, but you admitted that it was the freshest, most original thing to be put into a theater since... ever, and you couldn't wait to talk about how amazing it was with everyone you knew? "Sin City" is kind of like that.
http://sarahlane.typepad.com/sarahword/2005/03/ce
1. Is "Sin City" a family movie?
- Heavens no. It's incredibly graphic and gruesome. I know YOU'RE into that, but don't bring the kids.
6. Does the all-star cast detract from the story at all?
- Refreshingly, no. No one character is the main star, it's more like a bunch of supporting roles. Great supporting roles. These actors are stoked.
8. Are we talking CG animation or live action?
- Almost all the live action was done with green screens and props, then the magic was painted in later. It's amazing.
10. I was pleased to see lots of hot chicks in the trailer. Can I expect more of that?
- You sure can, my friend! But they'll also chop your head off. Literally.
have you read the article?
"But what about "4:4:4 Y'CbCr"?!
Y'CbCr, also known as YUV, is the color space used by film editing equipment. Y represents luminance, while Cb and Cr are color difference signals.
Consumer DV (digital video) cameras typically use a 4:1:1 Y'CbCr format, in which luminance is sampled for each pixel, while Cb and Cr are sampled at every fourth pixel. SD (standard definition) cameras use a 4:2:2 format. HD cameras can use 4:2:2, or a 4:2:0 format based on "spatial" samples of 2x2-pixel squares. Dual Link, however, uses a 4:4:4 technique."
Deliriant isti Americani.
yes, because ILM was much more than just a render farm. Teams of talented artists, model makers and engineers have kept them relevent.
Incidentally, another distinction earned by Rodriguez during the making of Sin City, is that he joined George Lucas and others who have been kicked out of the Director's Guild. Rodriguez's offense, Howard says, was working with a co-director -- Sin City comic book creator Frank Miller -- who doesn't belong to the Guild.
I know it's slightly offtopic, but Robert Rodriguez wasn't kicked out of the DGA. He quit because they wouldn't allow him to credit Frank Miller as a co-director.
Kudos to him, I say.
Whatever. Maya + RaveHD does not make you a visual effects facility capable of turning out 300 shots in a post production schedule of 7 months.
Just like having a copy of the GIMP doesn't make you into one of the leading creative ad agencies in the world. It takes a mix of talent, skill, experience and tools to be the top of your field.
- sarcasm is just one more service we offer -
It doesn't mean you get better color, but it does mean you get a better movie. The guys who digitally drop the characters into the backgrounds have more freedom to create what they want, since they can more easily make distinctions between foreground and background based on color.
Which means that they spend less time chroma-keying (picking out the background colors) and more time making movies. As with any big project, the finished product is filled with flaws that only the actual creator can put his finger on, but the overall sense of polish makes a big difference to the feeling you get when you watch the movie.
You get the same effect writing software: all those little hacks you had to do to get it out the door aren't immediately visible to the user, but they'll piss you off every time you look at them.
The changes aren't even necessarily subtle: they may have to substantially alter a shot if they can't get the background to drop out properly. You wouldn't notice without being in the editing booth, but you'll probably like the movie that much more for getting more of the director's vision onto the screen.
I'm a director [for the stage] myself, and though it's very different from film, we're constantly asking "how much can I get away with?" rather than "what can I create?" You tell yourself that the audience won't notice that you couldn't find the right prop, or that you didn't have time to get rid of the dim spot in the lights, but it pisses you the hell off and looks unprofessional even if the audience couldn't elucidate the difference.
It would be interesting to have a director go into detail on a commentary track to say, "Well, we would have done X, Y, and Z, but we couldn't because the technology was too limited." The closest you get is the re-released Star Wars movies. Well, maybe it's not such a good idea after all.
I think you're a bit misguided...
today i can create on a home PC in hours what ILM would have taken years
Technically, yes, but you need a lot of really skilled writers and animators to create the movie in the first place. Merely having the capability to render photorealistic 3D images does not by itself make a good movie -- look at the difference between Final Fantasy: The Sprits Within and any Pixar production to see that.
with the rise of even schools having massive renderfarms (like the g5 one)
Again, I think you're mistaken. I assume you're referring to the Terascale cluster at Virginia Tech, but it's not a "renderfarm", it's a supercomputer cluster designated for scientific research. Check out their web page here, they have details about what kind of jobs they'll accept.
ILM has an advantage in designing *new* machines, building on their own past achievements, as well as the newcomers', like anyone else. But the most important geniuses at ILM were those designing and using their machines. Organizational culture, rooted in its executives' vision, distinguishes winners from losers. Jumps in tech are opportunities to be exploited, not guarantees of success, for the big innovators as well as the later exploiters.
A more relevant factor in ILM's relevance is budget. ILM has big budgets, to attract talent and explore more opportunities, more of which they can afford to lose before hitting a winner. But their budget is so high that they can only be hired by big budget projects. Which are run by people who fear any risk, and which tend to make effects budgets "show their value" by featuring the effects, rather than using effects solely to support the rest of the picture (characters, story, etc). So we get ILM working big, bombastic, boring projects. Meanwhile, cheaper (Linux, etc) effects houses can spring up, try stuff, experiment with both effects and other risky, unproven parts of the picture. Again, the bottleneck is brains: if ILM supports the vision of a visionary film, it has an advantage. If ILM's execs apply it to the deadend of mere "special effects extravaganzas", it will be as relevant as fireworks exhibitions.
--
make install -not war
I didn't know I could get DDR for linux. Is there a USB Dance Pad I can get? Oh, wait, this must be some othe kind of DDR.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
I just want to point out that during my tenure at a certain Linux company, the name of which rhymes with "VA Linux".... OK it was VA Linux, back in the heady days of year 2000 I was telling certain key members of upper management there that if VA is going to sell high proced Linux boxen then they ought to consider building and selling boxen specifically for FILM PRODUCTION. I repeated myself more than once. I was told by certain key players in upper management who no longer work there that "We're not interested in going after niche markets."
VA no longer sells heavy Linux boxes but obviously someone is, and they're selling them to Hollywood.
*sigh*
I know you guys have to frame everything in terms of "LINUX WINS!" but you know what? It's not a race. Nobody was sitting around a table going, "I wanna be the first to make a feature with 4:4:4 dual-link RGB!" In fact, just the opposite: Everybody was sitting around going, "Let somebody else try 4:4:4 HD video. I don't want to take a chance on it with millions of dollars of somebody else's money."
Besides that, this whole thing is completely wrong. We've been using 4:4:4 for years in film production with a device called a "datacine." Go out and shoot 35mm film, which by the way has more color sensitivity than any video camera on the market, then run it through a device that scans each frame at high bit depth and high resolution in (you guessed it) 4:4:4 RGB.
Seriously, these machines have been around for more than a decade. RGB production is nothing new. You guys are making it sound like it's revolutionary, or worse, like it COULDN'T BE DONE WITHOUT LINUX. Inferno has done 4:4:4 since the mid-90s, and that runs on SGI gear.
See: chroma subsampling. It's even got diagrams. Though it could use a bit of cleanup.
Wikipedia to the rescue again!
--grendel drago
Laws do not persuade just because they threaten. --Seneca
During ingestion, the RaveHD wrote sequential DPX files for each shot to a standard Linux JFS filesystem on a fiber-channel disk array, Howard says. When all required shots had been ingested, the entire JFS filesystem was made available via Samba and gigabit Ethernet to the studio's production workers.
JFS isn't one of the high profile filesystems on Linux; People usually talk about Reiser, EXT3, or XFS. I wonder what lead the developers to choose JFS.
noah
For Spy Kids three that is !
... well as others have commented few Graphic Novels progress well to film but that does not preclude the opportunity that it can occur !
There is a extra on the dvd for the film where RR ( heavens forbid i even attempt to spell his name ! ) explains how to create some impressive visual and audio affects for your own home movies ( the family friendly ones people ! ) . He gives a very clear and engaging discussion with example film of how to include visual affects , editing and audio additions to make the films more interesting. Id say he understands how open source benefits every on e since he is so willing to share his expertise and experiences.
As for Sin City
And thats why Firecrackers and kittens don't mix.
"It's unlikely you would be able to see any difference between a 4:2:2 video, and a 4:4:4 video."
And there is a difference, at least when you're picking out stills and doing CMYK separations on them. Look how blocky and crapulent the yellow channel looks when you separate out an MPEG still or a JPEG image, and how sharp the black channel looks.
The real question is "Is size a consideration?". If it is not, do an uncompressed 4:4:4 AVI. I believe it will take around 100GB/hr in SDTV, and something like 600GB/hr in full HDTV. Don't quote me on those, but something like that. You can do lossless compression, but it will still be *huge* videos.
Now, if we assume that it is, the real question is, are the bits better put to use compressing the vid, or improving the color depth? Personally, I'd rather take two separate 4:2:2 pixels than two 4:4:4 pixels mixed up to save space (a gross oversimplification, but you get the idea).
Depending on what we want, it might be more effective to increase resolution, decrease compression or increase the frame rate than it is to improve the color clarity.
Resolution: SDTV is enough if you are more than 10x the screen size away. HDTV is enoug hif you are more than 3x the screen size away. In front of my PC, or if I could get a video wall, I'd be maybe 1x away. You'd need a super-HDTV that is to HDTV that which HDTV is to SDTV.
Compression: Difficult to say. Trained eyes can spot artifacts (blocking, shearing etc.) in almost any vid. Lossless vids would mean much bigger vids.
Framerate: We can easily move to 60p. That should put us near the "flicker rate" of the eye at 72Hz, perhaps even 90p for perfection.
So yes... the colors aren't perfect. But nothing else is either.
Kjella
Live today, because you never know what tomorrow brings