Slashdot Mirror
ILM Now Capable of Realtime CGI
Sandman1971 writes "According to the Sydney Morning Herald, specialFX company ILM is now capable of doing realtime CGI, allowing actors and directors to see rough CGI immediately after a scene is filmed. Actors on the latest Star Wars film watch instant replays of their battles with CG characters. ILM CTO Cliff Plumer attributes this amazing leap to the increase in processing power and a migration from using Silicon Graphics RISC-Unix workstations to Intel-based Dell systems running Linux."
According to the Sydney Morning Herald, specialFX company ILM is now capable of doing realtime CGI, allowing actors and directors to see rough CGI immediately after a scene is filmed.
Wouldn't realtime by WHILE the scene is filmed?
Interested in open source engine management for your Subaru?
...my webserver has been doing realtime CGI for years.
Roses are #FF0000, violets are #0000FF, all my base are belong to you
We have Real-time CGI Porn?
Where ever you go, there you are.
Maybe it IS realtime, but the actors just don't have the skill to watch themselves on a monitor WHILE acting, so they use the obvious 'i'll watch when i'm done method'
"Realtime CGI in Movie Quality" would be impressive, but:
"It's not at full resolution, but at least it gives them something to work with rather than working completely blind after each take."
With all the excitement over ILM using Linux I'm wondering exactly how many Hollywood visual effects studios use Linux.
The way that is worded, it makes it sound as if the processing power of an Intel/Linux combination is superior - whereas it is a matter of the bang for the buck instead.
You can get more processing power with the latter since it is cheaper (I would imagine even moreso with AMD) and easier to maintain. But not because it is inherently special or faster in any way.
I wonder if this will bring Silicon Graphics back into the favor of Intel boxes - for awhile they were okay with WinNT and Intel boxes, but then they dropped all of that - presumably for a higher profit margin and less hassle of maintaining multiple systems (also likely some break in business politics - perhaps someone at MS pissed someone off at SGI).
There are some odd things afoot now, in the Villa Straylight.
that proprietary unix is dying
In the Fellowship of the Ring DVD, Peter jackson can clearly be seen watching golum on a monitor (low poligon, but golum none the less) performing the mo-cap Andy Serkis is performing IN REAL TIME; as it is happening (not after).
So does this make this old news??
I dunno, I feel the ILM have been behind the bleeding edge for sometime now...
alnya
This will probably help on release dates for movies.
We'll get to see Episode III sooner!
I always thought with the current 3d cards coming
:)
out and the horsepower they can throw at things
they would eventually be able to to tv quality
3d animation programs in real time.
Hopefully this is going to lead to alot more 3d
animated series on tv in the near future , and
in time pick up from where final fantasy left
off. I still think it was such a pity that film
didn't get the people into the cinema to watch it.
But I think the advances they made will pave the
way for the future. Mainstream 3d Anime here we
come
Its a pitty they haven't got one of those to write the script!
Steve.
well, I guess they need to get that jar jar binks death scene juuuuust right.
Be you Admins? nay, we are but lusers!
It's hard to tell if this is anything more than a toy at this point. Marginal quality control is now possible. The time from pre-production to release might be a few days difference.
The actors might be able to play their roles slightly better if they know what the final result will be. In movies like EpisodeII they were acting totally blind in front of a screen for most of the movie. Very little of it was actually built.
The biggest question is "When will we have it at home?"
The realtime images aren't -final- renders of the scene. They're just rough drafts. The scene still has to be rendered in full res/texture, which still takes hours per frame.
What ILM has is a supercharged 'preview' button. Just like when you switch to wireframe mode in Lightwave or Maya and see a 'realtime' preview of the animation you're working on. But I'm sure ILM's version looks little bit better.
D
The first, last, and only tech news site on the net
Well, as more and more cgi houses move off of SGI (and on to whatever), they are only really left with their server business. It's really a shame to see a once proud pioneer in the industry reduced to a mere shadow of their former selves, though I guess in this industry, its very common (e.g. DEC, Lotus, Compaq, etc). At this rate it's hard to even see them being around in 4 years, a definite takeover target.
/. comment:
ob
SGI (aka Silicon Graphics Inc.) was found dead today at the age of 20. After being a high flyer in his youth, often seen hobnobbing with Hollywoods power elite, the latter years were not so kind and saw him in the throes of an identity crisis. Eventually his reliance on a small circle of friends was his undoing, as he was slowly replaced by more mainstream competitors. He will be sorely missed, as while he was at the top, he was a role model for "cool" in the industry, and helped to usher in one of the most exciting (and abused) technology shifts in the motion picture/video entertainment industry since the advent of talkies and color.
would be to develop a program that re-writes Lucas's inane dialogue in real time...
This is, largely, nonsense.
These images are *not* realtime! A PC is not capable of rendering a CGI screen, in realtime, and merging that, in realtime, with a video feed, and then displaying that, in *realtime*.
Say what you like about Linux, or high speed CPUs, or XXX vendor's high end GFX card - the architecture and the tools are physically incapable of this.
If you look at the extras on the LOTR:FOTR DVD set, you'll see people walking around, with a camera on a stick. This *is* displaying real time camera images, merged into a low res, non final rendered, scene of the Cave Troll fight in Moria.
A point of reference - the machine's they are using for this are SGI Octanes. Not Octane2s, but Octanes.
They did that work around, what, 3 years ago? And the Octane, at that time, was only 3-4 years old.
Can anyone show me a PC from 1997 that can manage that? Anyone?
Despite the fact that the Octane is an ancient piece of kit, there is nothing from the PC world that can match it's capabilities.
SGI have always been, and always will be, a niche player.
You would be a fool to buy expensive SGI kit for a renderfarm - buy Intel PCs with Linux. Similarly, you would be fool to try and do realtime CGI with that same kit - that's a specialist task that calls for specialist skills.
This article does not show that SGI is dying, or that they're being thrown out of the GFX workstation market.
This article *does* confirm what is widely known - the once cutting edge ILM are now many years behind people like Weta Digital.
Throwing around "Linux" and "Intel replacing SGI" sound bytes to try and get some news coverage for a dated effects house isn't going to change that.
Pah - Jim Henson's Creature Shop, Weta and Framestore have been doing this sort of thing long before ILM. Framestore did this for Dinotopia, Weta for Golum, and JHC for a variety of different things - all too numerous to mention here.
SGI laughed at the unassuming threat of the video chipsets, thinking that they would never be as fast as brute force. Even Pixar thought the same. Boy, were they wrong though. You can set up a cheap-ass render farm for about $250k, taking up minimal space that can do the same job as a SGI render farm that costs a cool $2 million (Shuttle SFF PC w/ 3 gig CPU + ATI 9700). Of course, there's still the software side.
The Nvidia's GeForceFX and ATI's Radeon 9800 both contain features that even through the marketing-hype has some real value to programmers out there. Just look at Doom 3. It will run well on some computers that are just 6 months old. Now, imagine taking 250 of them, as a Beowulf cluster!!1
Post your server's url to slashdot, and we'll see just how realtime it is.
ILM developed its proprietary file format, OpenEXR
Hmm.. i sense a trend in calling things open when they are actually closed. This is eroding the intended meaning of "Open" in front of fileformats or products.
Dark Helmet - "What the hell am I looking at? When does this happen in the movie?"
Col Sandurz - "Now. You're looking at now, sir. Everything that happens now, is happening now."
Dark Helmet - "What happened to then?"
Col Sandurz - "We passed then?"
Dark Helmet - "When?"
Col Sandurz - "Just now. We're at now, now."
Dark Helmet - "Go back to then."
Col Sandurz - "When?"
Dark Helmet - "Now."
Col Sandurz - "Now?"
Dark Helmet - "Now."
Col Sandurz - "I can't."
Dark Helmet - "Why?"
Col Sandurz - "We missed it."
Dark Helmet - "When?"
Col Sandurz - "Just now."
Dark Helmet - "When will then be now?"
Col Sandurz - "Soon."
Facts do not cease to exist because they are ignored. - Aldous Huxley
Intel-based Dell systems running Linux
So conflicted...Intel bad...Linux good...Dell ambivalent...
"In a 32-bit world, you're a 2-bit user. You've got your own newsgroup, alt.total.loser." -Weird Al
Now that's Cost Savings!
I was friends with several SGI employees when SGI decided to ditch their Intel/WinNT support. Two of my friends were directly involved with the NT-related operations. The decision was mainly related to a series of falling-outs with Microsoft over things like the Fahrenheit relationship. Officially it was attributed to cost-cutting and re-focusing on core competencies, though (SGI has been in a bad way financially for quite a long time).
These days a large part of their revenue stream depends heavily on service contracts for their custom hardware. It would take a seriously impressive balancing act for them to support commodity hardware and remain afloat...
(As a side note, since their custom hardware is so heavily graphics oriented, when things go wrong it often does really interesting things... like an entire render-job where everything ends up with the same bump map, but is otherwise normal...)
Slashdot quality declines as the number of hot grits posts decreases. - Provolt's Law, Apr-09-2005
A moden dual proc Xeon can come very very close to what an Octane was able to do in 1997. It's not the same thing, but it's close enough to do the job. Octane2 (with the right software) would be overkill, so here are the differences between a used Octane and a dual Xeon:
The Xeon is new. That means you can get a good warranty and not have to worry about using used equipment.
The wiz-bang factor. These days most SFX software runs on both IRIX and Linux. Even Apple's Shake does. So does all of the latest Linux utilities. It's "cooler" to many people to use a Linux workstation over an Octane.
The CPU. Granted, the Octane's torque came from it's architecture, not its CPU... but this alone does not make up for the raw power of those Xeons. It's like racing an 18 wheeler with a F1 race car. The 18 wheeler can haul a lot more, but the F1 race car will get you to the local Wal-Mart a lot faster. For small tasks, the Xeon will feel a lot faster.
This is why you'll still see a lot of existing Octanes, Octane2s, Onyx/Onyx2/Onyx3000 systems in use by hollywood. They work fine. But for new employees, and for replacement hardware, you'll almost certainly see a dual proc PC running Linux. There are, of course, come artists that prefer one over the other.
As for render farms, you're right. It only makes sense to use Intel or AMD. Using SGI (or Sun) big iron for rendering would be insane. The render software isn't even optimized for IRIX or the SGI Origin architecture anymore. I think the very last holdout was ILM, who still had their three huge Origin 2000s. (Running 1999 R10K 250MHz processors [about the equiv of a PIII/550]..... no wonder they find their new renderfarm to be faster.....).
You're right on the money about ILM being behind Weta (and possibly others). ILM is still a cool shop, but not as current as many of the others. Hell, ILM did most of the work for Episode 1 on SGI O2s. O2s! The O2 was an ok video editor, by no means a 3D or CPU powerhouse! I can't belive they got it done at all. There are HUGE differences between the O2 and Octane.
You need to be able to render something in realtime, combining rendered scenes and a live video feed, and then to display that on a monitor or writing it out to digital storage - and all in real time.
This is not a function of the CPU of the GFX card - they both play a part, but not as much as you seem to assume.
The main thing here is *bandwidth*. You have to have a large amount of non-contentious bandwidth to chuck all that data around.
Point to point bandwidth between multiple devices - CPU to memory, CPU to GFX, GFX to memory, CPU to output pipe, input pipe to GFX, etc. etc.
One approach was UMA, which SGI pioneered in the O2. A 7year old SGI O2, even a low end one, can handle 800mb textures in realtime with ease. There is little available today that can even approach that.
Another approach is some sort of IO switch. SGI have the crossbar in the Octane/Octane2, which is old tech based on the IO infrastructure of the Origin 2000.
The Octane's crossbar switch gives you a large amount of non-contentious bandwith, which is why, for instance, a low end 195mhz SGI Octane from 1998 can apply in incoming digital feed as a texture to an object, and display that on digital output, in real time.
Remember the Anubis warriors in The Mummy Returns? Their skin was a texture from a feed that was applied in real time - this was so the animators could get a good feel for how their animation changes effected the scene.
Sun have a similar sort of setup with their UPA crossbar on their Ultrasparc kit. IBM and DEC^W^Compaq^WHP a similar deal.
The reason UNIX vendors can charge lots for their kit is the years of R&D they've put into solving problems like this, which just don't appear on low end commodity kit.
The Origin2000 can scale to 1024 CPUs in a single system image - it's one OS, no partitions. Sure, that's for a niche market, but it doesn't change the fact that there is a serious IO contention issue that needs to be solved in that scenario. SGI have then taken that solution and thrown it into their graphics workstations.
CPU speed and gfx speed will not help here - neither will the OS. Linux or Windows, ATI or Nvidia, PC solutions will not cut it in this area - the underlying architecture is poorly suited to those sorts of tasks.
Niche markets for people like SGI will always remain small, but despite the ill-informed nay sayers, they will never die - because there is a need there, and there is no commodity kit that can do the job.
I actually work with the guy who did those! They were done by Touch Animation AFAIK - now defunct. Apart from the fact that The Day Today's gfx didn't have a huge budget, they were also done quite a long time ago now - still excellent and groundbreakingly satirical (can gfx be satirical? you bet).
On the Partridge point, you're forgetting my combat-based gameshow that runs on digital TV.
That was classic intercourse!
It depends on which part of the production world you're talking about...
For rendering, you need raw CPU power and middle of the road networking. A rack of dual proc PCs and a 100BaseT switch is plenty for most 3D people.
For 3D modeling, a good graphics card and a strong PC behind it is what's needed. You want a card that can handle the polygons, can handle the textures, and has enough cache for all of the display lists. A 3DLabs Wildcat-series card on a modern PC is good enough for almost any 3D animator. And will be faster than all but the fastest SGI.
For compositing, you really do need the big iron. There is no PC out there that can handle 3+ streams of HD at the same time, let alone layer them, adjust the colors, add effects, etc. in real time. This can be done without much sweat on an big modern SGI running IFX Piranha or Discreet Inferno. The big catch is RAM and the RAID performance. 8 GB of ram is a good start and a sustained 500+ MB/sec coming off your multiple channels of fibrechannel is a must.
Don't get compositing confused with "painting". Painting is where you take one frame of a movie and touch it up with paint tools (Amazon Paint, FilmGIMP, etc) then flip thru the other frames noting any needed changes. This can be done on Mom's Pentium 3.
There are, of course, cheaper solutions that don't work as well... but for some folks, that may be good enough. Shake on a Mac, Linux box, or low end SGI does a good job. So does Combustion on a modern wintel PC. It's *can* work with HD, but you need patience and realistic expectations. It's not realtime in any way, by any stretch of the imagination.
So yeah... there are lots of machines to choose from, and many tools for many jobs. If the Shuttle could hold a real gfx card (Wildcat 7210, for example) it would be a 3D modeler's dream. With the ATI 9700, it's only for games or light use of Maya and friends.
You also have to keep in mind what gets bought with the money. A $2M SGI usually includes lots of RAID hardware... and a machine that can handle 100s (often 750+) sustained MB/sec without problems. It's required to do the heavy lifting jobs.
Also note that SGI doesn't sell "Render Farms". Big SGI gear is usually used to support one task... generally compositing, somtimes multiple projector realtime 3D work (such as a six-sided reality CAVE). The spare CPU cycles are sometimes used to render in the background.... but rarely is SGI hardware used to render when a cheap PC cluster can do it faster anyway. It'd be like trying to enter a F1 race with your Kenworth 18 wheeler.
There are also SGI workstations like the O2 and Octane... they used to be used for 3D modeling and low-res realtime work. Still used by many studios, and even broadcasters. Still very popular for the weather maps on the local news for for drawing the yellow "virtual first down line" on live football games. Not heavy work, but there are some gotchas involved. The stability is handy there.
As much as I love ILM and what they are able to do with technology and movies, this doesn't seem like that great of a thing. If all they are writing about in the article is being able to see how the film they just shot will line up with a rough animatic, then thats not that great. I'm guessing what they have is much like what Weta Digital had to make the cave troll and other stuff in Balin's Tomb. Now, I would have been shocked and surprised if they said they could render a CGI scene with full effects, shaders, and the like in real time. That will be an accomplishment. What they have now (if its really like what Weta has) is no more than a video game with input based on the positions of sensors rather than a controller.
SIGFAULT
ILM CTO Cliff Plumer attributes this amazing leap to the increase in processing power and a migration from using Silicon Graphics RISC-Unix workstations to Intel-based Dell systems running Linux.
Well, I hope all you open-source advocates are happy now. You worked to develop Linux and other open source software because it was "cool," and I'm sure you all had a great time making it more and more powerful. I'll bet you never gave one minute of thought to the fact that the software you were producing might make it easier to make those awful, awful movies, did you?
Well, it's too late now. I just hope you're satisfied!
When all you have is an axe, everything looks like a grindstone.
Apparently there is not much an SGI machine can do that a PC cannot do (or other unix machine) since SGI has not posted a yearly profit since 1997!
It's amazing they still have a few dollars in the bank. They've sold some patents and sold off Cray (albeit for pennies on the dollar from what they originally paid), but to last that long is impressive in itself.
Any any rate, SGI does offer unique and unmatched products *in certain areas*...
Do you need a shared-memory supercomputer that can scale to 512 processors with the same exact kernel that runs on a desktop model? Or how about 1024 processors with a simple kernel patch? Very few people need that much IO across that many processors, but for those that do, there is no better choice.
Do you need a machine that can handle dozens of channels of 2gbit fibrechannel without breaking a sweat?
Do you have an insanely complicated set of HD video files and other material that need to be layered/composited? Does this job need to be done yesterday? Is full-resolution/full-quality realtime effects work needed? Piranha or Inferno running on an Onyx 3000 (plus gobs of ram and disk arrays on several channels) can do this for you.
Are you interested in seeing the true potential of Linux? Do you want to work with a true Itanium2/Linux supercomputer... one that is way more than a cluster? Want to see a single machine (again, not a cluster) with 64 processors and 512 GB of RAM? Yes, Linux can handle it too, because of SGI's kernel patches and hw/sw architecture.
Not many people need or can afford SGI big iron... but for those that do, nothing beats the SGI Origin and it's baby cousin, the Altix.