Please create an account to participate in the Slashdot moderation system

 


Forgot your password?
Close
typodupeerror
×
Movies Data Storage Hardware

Animation Sophistication: The Croods Required 80 Million Compute Hours 196

Posted by Soulskill from the that-is-incroodible dept.
Lucas123 writes "It may be a movie about a stone age family, but DreamWorks said its latest 3D animated movie The Croods took more compute cycles to create than any other movie they've made. The movie required a whopping 80 million compute hours to render, 15 million more hours than DreamWorks' last record holder, The Rise of the Guardians. The production studio said between 300 and 400 animators worked on The Croods over the past three years. The images they created, from raw sketches to stereoscopic high-definition shots, required about 250TB of data storage capacity. When the movie industry moved from producing 2D to 3D high-definition movies over the past decade, the data required to produce the films increased tremendously. For DreamWorks, the amount of data needed to create a stereoscopic film leaped by 30%."
This discussion has been archived. No new comments can be posted.

Animation Sophistication: The Croods Required 80 Million Compute Hours More

Animation Sophistication: The Croods Required 80 Million Compute Hours

Comments Filter:

  • But... (Score:5, Insightful)

    by Ixtl ( 1022043 ) on Tuesday March 26, 2013 @07:04PM (#43286991)
    ... it's still terrible. They could have made it with construction paper cut-outs and hired some decent writers instead of spending 70 million on fancy CGI and celebrity voices, and then making the same cliched shitpile we see every two or three months. Also, as is traditional on Slashdot, I am basing my vociferous opinion exclusively on the obnoxious 30-second trailers I've seen, and have not actually seen the movie.

    • Re: (Score:3)

      by Nemyst ( 1383049 )
      The trailer was so incredibly obnoxious (and that's coming from someone who usually doesn't mind most trailers, even the dumb ones) that I don't think the rest of the movie matters. They thought the trailer was a good representation of the movie: imagine the rest.
      • As far as I can see, the absolute minimum plot necessary to justify the stringing together of some ok-but-not-great gas.

      • Re: (Score:3)

        by fatphil ( 181876 )
        The trailers contain what they think is (at least teasers towards) the best bits of the movie. The movie will, on average, be even worse.

        It looks like they've got a new hair model that they wanted to show off, and possibly a new particle simulation model. Woo, freaking woo. Apparently, they have no interest of having the characters mouths say anything apart from "ah-ooooh-oooh-awwww-eeeeeee-ah" when the dialogue actually does something like "why do you always say that?". Fuck the particle model, fuck the ha

        • In other words, they can only manage to do one half of what Pixar can do most of the time.

          Then again, Brave wasn't that good aside from the CGI. Went through so many rewrites there wasn't much story left.

    • Re:But... (Score:5, Interesting)

      by Dahamma ( 304068 ) on Tuesday March 26, 2013 @07:50PM (#43287299)

      Actually, two of the three writers (John Cleese and Chris Sanders) are more than decent. Must have been the 3rd guy who screwed it up.

    • Prejudice... (Score:5, Interesting)

      by mapuche ( 41699 ) on Tuesday March 26, 2013 @08:10PM (#43287389) Homepage

      I guess your comment is the variant of "haven't read the article, just the summary".

      The script is very well done in terms of human relationships and interactions. It's not a movie about fart jokes, the characters are fairly complex (for an animated movie). It is worth to watch it before forming any opinion.

      • Maybe. But they're ugly as a rhinoceros's butt. And that turns me off completely. Plus they act like they're on dope all the time.
        Aliens on dope trying (and failing) to mimic human beings, that's what I got from it.

      • Yeah! Give it a shot, they've been working on this since 7120 BCE (at least the computer has been running that long!)

        • Re: (Score:2)

          by gl4ss ( 559668 )

          yep..

          it looks like crap, took a long time to compute and it's debatable how many "computer hours" it took to complete because "compute hour" isn't an actual unit in any way.

  • All I wonder (Score:3)

    by onyxruby ( 118189 ) <onyxruby@ c o m c a s t . net> on Tuesday March 26, 2013 @07:10PM (#43287035)

    How many animation studios were forced out of business? That seems to be Hollywood's favorite metric for Fx and computer animation.

    • The bigger sacrifice would seem to be that of cellulose film.

      • The bigger sacrifice would seem to be that of cellulose film.

        ...and a media that sacrifices itself automatically is of *what* value? (You are aware that there is a very expensive rush on to save the last century of cellulose film archives that are fading away into oblivion simply becuase they are cellulose...)

        • Re:All I wonder (Score:4, Informative)

          by Charliemopps ( 1157495 ) on Tuesday March 26, 2013 @08:04PM (#43287361)

          Also make the point that, films that were pirated, will never be lost. The entire Tom and Jerry archive was lost in a vault fire. We have none of the originals left. All they had after the fire was the film that had been cut down for TV viewing. That's why all the Tom and Jerry episodes are in 4:3 instead of their original wide screen format.

          • Re:All I wonder (Score:5, Informative)

            by NoMaster ( 142776 ) on Wednesday March 27, 2013 @12:06AM (#43288427) Homepage Journal

            Negatives for all the Tom and Jerry shorts prior to 1951 were lost in the 1967 MGM fire. Up until 1954, T&J was produced in Academy ratio (1.37:1), which is almost indistinguishable from 4:3 (1.33:1). Later ones were produced in a variety of formats from straight Academy ratio, to widescreen 1.75:1 on Academy ratio negative, to Cinemascope.

            The only real difference between initial theatrical and current TV/DVD releases of the pre-1951 cartoons (apart from the obnoxious habit of whitewashing out the culturally-insensitive bits) is loss of the original titles on some, and the odd 'lost' sequence.

          • That's why all the Tom and Jerry episodes are in 4:3 instead of their original wide screen format.

            I

            Hanna and Barbera ultimately wrote, produced and directed 114 Tom and Jerry shorts at MGM cartoon studios in Hollywood from 1940 to 1957.

            Tom and Jerry [wikipedia.org]

            These are the shorts that people remember and all but the very last were produced 4:3, not widescreen.

            There have been many attempts to revive Tom & Jerry, none showing any great sympathy or understanding of the characters. Chuck Jones struggled here and it shows.

  • You would think that for stereoscopic imagery instead of single-viewpoint imagery that the data-storage requirements would double rather than increasing by 30%. Maybe there's compression of imagery involved to save that space?
    Regular 2-d imagery = one viewpoint = $K$ amount of storage.
    3-d stereo imagery = two viewpoints = $2 \times K$ amount of storage

    What's wrong with what I'm thinking?

    • Just a guess, but... 1 input (scene), 2 outputs (renders) ?
      Perhaps 30% is 33.3333333% :)

    • Re:Shouldn't it double? (Score:5, Insightful)

      by Nemyst ( 1383049 ) on Tuesday March 26, 2013 @07:19PM (#43287099) Homepage
      Things like 3D assets, textures, etc. don't suddenly need to be duplicated. In fact, the 3D scene itself needs very little changes, just having two cameras instead of one. It's once the movie's rendered that things double in size, but that's only a subset of the total movie's required space.
      • Yeah, I was just talking about the "rendered stereo product" storage space, but you're right about the other assets. So that almost tells us that 70% of the original storage requirements were for the assets (textures, skeletons, linkages, etc) and sequence data and that only 30% of the original storage requirements were for the rendered raw frame images. Thanks for pointing it out to me.

      • Things like 3D assets, textures, etc. don't suddenly need to be duplicated. In fact, the 3D scene itself needs very little changes, just having two cameras instead of one. It's once the movie's rendered that things double in size, but that's only a subset of the total movie's required space.

        While it's true that most 3D scenes don't need major changes, and the camera data is very small, this isn't always true. In 2D, artists will occasionally use horribly nonphysical hacks to make something look the way the

      • It is still twice the rendering and disk space for the rendered files and 4X the headaches. I've worked 3D back in the 80s and 90s and it's a pain in the ass. I was doing effects rigging as well as some camera work. The problem is in film you get used to hiding things where as 3D has a nasty habit of seeing around the object or person. The cameras weighed a ton and we had one tear up a crane arm late one night. I was approached by a group that wanted to do a 3D film a couple of years ago. I tried to talk th

    • The 2 rendered viewpoints for 3D vision are not the only stored data. It's all the 3D models and textures and animation sequences AND the rendered scenes. So 30% seems pretty reasonable.

    • Dunno. Napkin: 250000000000000 / (120*60*48*2); ~2 hour movie (120 minutes), 60 seconds per minute, 48 frames per second, one for left-right eye (2); or ~360MB per-frame. Perhaps a dozen or so layers per frame (different lighting models, shadow models, etc.,) leaves ~30MB per ``frame layer'' in super-duper-master resolution losslessly compressed. Animation paths/models/textures/voices, etc., also probably take up quite a bit, but likely not nearly as much as the raw image data.

      • Dunno. Napkin: 250000000000000 / (120*60*48*2); ~2 hour movie (120 minutes), 60 seconds per minute, 48 frames per second, one for left-right eye (2); or ~360MB per-frame. Perhaps a dozen or so layers per frame (different lighting models, shadow models, etc.,) leaves ~30MB per ``frame layer'' in super-duper-master resolution losslessly compressed. Animation paths/models/textures/voices, etc., also probably take up quite a bit, but likely not nearly as much as the raw image data.

        Imagine... All of that just to render a napkin.

    • If they were just storing the stereoscopic movie it'd be under 20gig. Clearly they are storing the entire data set including models, textures, etc... In fact, I'm rather surprised that the space increased at all. The rendered version of the movie should be tiny compared to all those assets stored for re-use in McDonalds ads, and sequels.

      • My guess would be they have a version of the rendering that is stored at high resoloution with no compression (or maybe minimal intraframe compression) so that they can convert to whatever form the market demands for decades to come. This version may also contain material that was unused in the final film.

        I dunno what resoloution movies are done at nowadays but if we assume 3 bytes per pixel, 24 frames per second and 8 million pixels per frame that is about 2 terabytes per hour.

        • The images will be OpenEXR, with 16bit floats per colour channel, uncompressed. RGB is a bit optimistic, it's more likely to be RGBA+depth. Typical film footage is a 2k image, or slightly above 1080p.

          • These days, even small theaters show in digital 4K. If you're rendering on a computer you have no reason not to go for 4K as well. Times 2 for the second eye.

      • No, the rendered images will consume the most storage. The textures will all be procedural, and geometry data doesn't actually consume as much as you might think (since most of it will be static, and a large amount will be procedural). Version control repositories (to put it in programmer friendly terms) for the asset revisions may be quite a bit resource hog. 20Gb is extremely optimistic for uncompressed open exr's.

    • Re: (Score:2)

      by mill3d ( 1647417 )

      I can't be certain that this is the technique Dreamworks used, but it makes sense and would save disk space:

      http://en.wikipedia.org/wiki/Deep_image_compositing [wikipedia.org]

      Instead of rendering images for both eyes, you render a "deep" image that contains depth information. The images for each eye are then written out of post production rather than out of the CG software.

      • Hey thanks! I'd never heard of that technique before. It must require extra "hidden behind this edge" info too, since each eye can see things that the other eye can't see obscured behind the edges of foreground objects.

      • I can't be certain that this is the technique Dreamworks used, but it makes sense and would save disk space:

        http://en.wikipedia.org/wiki/Deep_image_compositing [wikipedia.org]

        No idea how much this was used on this particular film, but for the record, deep images don't save disk space. They churn through it like nobody's business! The idea is driven around storing many samples per pixel instead of just one, so you have a *lot* more data than with a normal "shallow" render and compositing pipeline. It is extremely useful

    • You would think that for stereoscopic imagery instead of single-viewpoint imagery that the data-storage requirements would double rather than increasing by 30%. Maybe there's compression of imagery involved to save that space?

      Generally speaking, nothing clever happens for stereo storage. It's just that the actual rendered frames are only a small part of the total data involved in making a film. I've never worked at Dreamworks so I can't speak in detail about their pipeline, but I wouldn't be surprised i

    • The two data sets are intimately related in predictable and purposefully engineered ways. They are a 'known transform' away from each other. Not exactly the right terminology, but you get the idea.

  • Okay, so everyone's doing 3D now, fair enough. I have no idea what high-definition means in this context though - have movies moved beyond 4k?

    Lets hope their next title will be in high-frame-rate too. This should be a no-brainer particularly for animated titles. Double the processing requirements again!

    Unless of course they go for a motion-interpolation to generate every second frame but the end result wouldn't be nearly as good.

  • what's interesting is that even though their storage requirements have been increasing, the cost of the needed storage has probably been dropping drastically along the way. I bet they are spending less on storage now than they did in the 90s, even though they are probably storing a few orders of magnitude more data. so from a costs perspective...which is how DreamWorks looks at it...it's a non-issue

    • Re: (Score:2)

      by Kjella ( 173770 )

      That's what I was thinking too, if the simplest solution is to just get a bigger disk, then let's just do that. Doing otherwise is like a company rationing office supplies [thedailywtf.com]. Personally I just bought 2x4TB drives that'll give me 5TB more HDD space (I'm retiring two 1.5TB drives) because I'm too lazy to sort through it all. Hell, I can't even keep my downloading in pace with technology, at one point I had ten HDDs operational now I'm down to six and if I wasn't looking for room to expand I could go down to fou

    • Yeah, I took a tour of Wavefront Technologies back in the early 90s and they were still measuring storage in gigabytes. They'd just unpacked their first HDTV setup and, if I remember right, the demo system had 10 gigabytes of storage and it was ridiculously expensive because it had to be able to read/write crazy-fast to handle the HD content. I think that was about the time I paid $600 for a 212 meg drive at a computer show and it was a great price. A couple years later, I met a guy who was working on a

  • from producing 2D to 3D high-definition movies... (Score:3)

    by Osgeld ( 1900440 ) on Tuesday March 26, 2013 @07:36PM (#43287205)

    ..."over the past decade, the data required to produce the films increased tremendously."

    but yet the quality and entertainment of such movies at best remains about the same.

    Besides I always get a kick out of the HD systems at a cinema, sure brag about your HD flickering mirrors and your THX Super surround, doesnt mean squat when the picture is still fuzzy and the speakers sound like a cheap set of computer speakers with too much bass running though 2 metal 1 watt tweeters and a 8 inch floppy as shit "sub"

  • But... Why? (Score:2, Interesting)

    by Anonymous Coward

    How does the efficiency of this compare to some of Pixar's movies (like Brave or WALL-E)?

    I only say this because sometimes long render times and exuberant technical requirements are not signs of proper craftsmanship. When they are, you often point out why they are- maybe you've developed a new BRDF shader that takes a bit longer to render but offers results closer to an unbiased renderer that nobody else can achieve. Or maybe you've written a new global illumination system that, once again, takes a bit long

    • Re: (Score:3)

      by dbIII ( 701233 )

      over-the-top lighting rig

      Then you compare it to the first "Toy Story" which had hardly any shadows and a pile of other shortcuts but is still going to keep a lot of people glued to the screen for a couple of hours.

  • It seems that while the graphics and data requirements increase dramatically the actual quality of the stories and movies are increasing just as rapidly but in the other direction. I would much prefer less focus on graphics and 3D and them spending some money of some bloody writters that don't just rehash the same shit over and over.

    • Re: (Score:2)

      by Krymzn ( 1812686 )
      Modern animated movies don't all consist of clichés, double entendres and wise-cracking sidekicks, nor even have much swish CGI. I am, of course, talking about Studio Ghibli.
      • yes and they have made some of my favourite animated movies, however I still wish there were more animated studios up to their standard, their are a couple but I wish their were quite a few more like them, perhaps then we would see more substance it what is being produced.

  • Easy way to save time (Score:3)

    by GrahamCox ( 741991 ) on Tuesday March 26, 2013 @07:40PM (#43287239) Homepage
    Here's an optimisation they missed, which could have sped render time up dramatically. Take the script, and get some people, dressed up as the characters, to read it out in rooms designed to look exactly like the rendered backgrounds. Of course, the people would need to just go beyond just 'reading' the script, they'd have to sort of pretend they were those characters, like it was some sort of act. Then film it.
  • This is a worthless statistic. Maybe they assests were all bloated and inefficient and the movie took way longer to render than it should have. Maybe they were incredibly efficient and every texture and model was optimized and the movie actually took half as long to render as it would have for anyone else. Maybe the hardware was a couple of pentiums in a warm basement in southern california.

    At any rate, the time it takes to render a movie is about as interesting as the average histogram from all the fram

  • I'm amazed that a full-on Hollywood production can fit in 250TB.

    That's really not all that expensive any more. Unless my math is wrong that's well within the budget of a medium-sized post-production facility.

  • Progress (Score:3)

    by sootman ( 158191 ) on Tuesday March 26, 2013 @08:47PM (#43287581) Homepage Journal

    I've heard that the average time to render a frame has stayed around 3 hours, from "Andre and Wally B" to now. May or not be true, but it's probably close. It's amazing to look at the differences between "Toy Story" and "Toy Story 3", which makes a particularly good test case because they have the same characters but they're 15 years apart. I remember being amazed at how things looked in Toy Story, from Rex's bumpy texture to the messed-up paint at the bottom of Andy's door, but if you watch the first right after watching the third, you'll be amazed at the differences. I'd say it's most noticeable in the human characters but if you look closely you'll see it everywhere.

  • ... a good chunk of that was the animators surfing the web while their rendering ran.

  • ...if they still only have 2D, low development characters?

    Seriously, could they maybe start spending a few of those millions they pump into special effects and more shiny into scripts that actually, you know, make me WANT to watch the movie? Or at least make me want to stay longer than the 10 minutes it takes to know how it is going to progress and end?

  • What happens to all of the assets, models, and 3D work that was put into the film? Does it go into some massive archive for the studio, or some kind of common repository? Projects like that, it's easy for files to get lost and lose all of the work put into it. I'm curious to know what the life-cycle of the digital assets is once a movie is completed. Anyone got any light to shed on the subject?
    • I have been thinking about this a lot lately. At some point in the not too distant future, games and moves are going to SHARE art assets. Will we be able to remix the assets on our computers and make new adventures with cinema quality props??

  • "The Croods took more compute cycles to create than any other movie they've made."

    Does this mean they've finally managed to change the expression on character faces from their usual open-mouthed stupidity?

  • So many computations, so much emptiness. (Score:3)

    by master_p ( 608214 ) on Wednesday March 27, 2013 @05:53AM (#43289797)

    I saw the preview trailer...excellent graphics, top-notch animation, very good voice acting...but it failed to grab my attention. There was a void.

    On the other hand, the movie 'Up' was a lot cruder, in terms of technical aspects, but so much more moving than 'the croods'.

  • It's been said that as we grow older we lose our sense of wonder about the world. That's certainly true of the slashdot crowd. Here we have an article talking about the technology behind creating an animated movie, and I see a bunch of comments bashing the script and progression of the movie. Face it folks, this movie isnt for you. Dreamworks wasnt trying to make it for you, and as long as youre older than 14 Dreamworks wont try to cater to you with this genre ever again. Let the kids have their fun with fa

    • When you get to the point of calling an animated film "a kid's movie" you're part of the problem. Dreamworks has a big tendency to put out pretty-looking drivel. Sometimes as a kid or even as an adult you want to turn your brain off and watch something mindless. I enjoy Hannah Montana now and then. But when you get to the point of taking all the art away from a movie just because it's a kid's movie, you're breeding a generation of dumb people.

      Just compare something like Kung Fu Panda with Finding Nemo o

  • Maybe it's just me, but the fact it's not a doubling is actually a little surprising. I can picture it - the models are 3D all the way though to the final rendering, where you need to pick between 2D and 3D (2D is probably a single eye view of 3D). But at some point those shared models need to become pixels, and a 3D rendering seems a 2x increase in storage. Or are the models so complex now that the end rendering filesizes don't dominate storage needs?

    That said, the trailers don't make me want to watch this

  • It would have been completed in half that time. LONG LIVE EIAS!

Slashdot Top Deals

If all the world's economists were laid end to end, we wouldn't reach a conclusion. -- William Baumol

Close