Unreal high res shots! (56k --> NO!)

[PC-Engine]

GT3 is not photorealistic by any measure!
Even though it does look more real than any xbox racer, be it due to hardware better suited, or more skilled artists, or combination of both.

Umm..and that's why I said "Anyway let's move over to realtime apps for a minute therefore I'll loosen the definition of photoreal."

And that's why I said "GT3 is not photorealistic by any measure, as there is no such thing as "loose photorealism". It either is photorealistic or not.
How far can you "loosen the definition of photorealism" for it to still remain photorealistic

If someone were to look at a tv displaying a replay of a GT3 car racing around the track and mistakened it for a real car then that would be a loose definition of photorealism.

 

[rabidrabbit]

Ok, I agree.
But the animation and camera direction has a lot to do why GT3 looks loose-term-photorealistic. If one looks at the screenshots of GT3, they can hardly be mistaken as real world photos, but in motion, the eye is easier fooled.

Maybe for videogames, the term "photorealistic" is not suited as well as for still images. Should talk about about "tellyrealistic" indstead

 

[PC-Engine]

Maybe for videogames, the term "photorealistic" is not suited as well as for still images. Should talk about about "tellyrealistic" indstead

Photoreal means it looks like a photograph or videocamera footage or real life through ones eyes. An automotive commercial with a real car going down the street is photoreal. A videogame replay that looks like a real actual car going down the track is photoreal even thow it's low resolution. I mentioned realtime for a reason.

IMO it is still CGI technology limited, otherwise ther would be a computer generated human character, that was completetly undistinguishable from real human actor. But as it is, there is not even a short animated demonstration of an animated (especially facial) human character that looks real.
One can assume that if there were a hyphotetical super computer with unlimited power, it would be able to do a completely real human character, thus the current CGI is technology limited.

So an incompetent artist with a supercomputer will be able to render a photoreal human being?

How about even a photoreal 2D render?

Supercomputers only do what it's told to do...

 

[rabidrabbit]

............
So an incompetent artist with a supercomputer will be able to render a photoreal human being?
....

Yes, the hyphotetical supercomputer would only need minimal input from the artist, and render the photorealistic human using the data it is fed (3D scanning etc..). No artistic skills needed, at least if an existing human was modelled & rendered. An imaginative character would need some more artist input, but the hyphotetical supercomputer would still be able to do it mostly by itself, it would just maybe look a little 'odd' (maybe disfigured), but still a photorealistic, disfigured human model

 

[PC-Engine]

What you're trying to argue is the advancement in desktop PCs which is irrelevent. JP uses dedicated render farms from SGI. They don't use desktop PCs. Computing power was never an issue.

The hell it wasn't. Your home PC is a dozen times faster than an SGI machine they've had back then.

LOTR had good CGI but it's hardly more advanced than JP.

That's a joke, right? You cannot be serious... or are you having problems wiht your sight?

And a PC still can't render Toy Story with a $300 videocard, your point?

Oh and there's nothing wrong with my vision. I don't like to wear goggles.

Please tell me what's visually more advanced that can be seen between a JP Raptor or T-Rex and a troll from LOTR.

 

[cthellis42]

[quote="PC-Engine]Thanks for taking one sentence out of context to build your whole pathetic attempt at an argument.

Maybe you should go back and read every one of my posts in this thread. Selective quoting seems to be your specialty. [/quote]
Oh gee I'm sorry. You phrased it like a deliberate troll because...?

 

[Polarbear53]

Please tell me what's visually more advanced that can be seen between a JP Raptor or T-Rex and a troll from LOTR.

Although they might look close in terms of visual quality LOTR used new technology. They created everything in a 3d world for the troll scene in The fellowship of the ring, the people the troll the room. Then they hooked Peter Jackson up with a piece of wood(the "camera") and some t.v. glasses, then they used motion capture on the block to move the camera in the 3d world. So the director was walking around in the 3d world and filming what was happening from where he wanted with his virtual camera.

So the way I see it, the technology doesn't make the CG great, but it helps the artist get what they want without having so many limitations.

 

[Laa-Yosh]

And a PC still can't render Toy Story with a $300 videocard, your point?

Don't try to change the topic - nobody was talking about realtime TS here.
Toy Story was rendered on a renderfarm with at least 500 processors, if not more, for a long, long time, and each of those machines (SUN mainframes) cost a fortune.
The power of today's PCs is at least a hundred times bigger, if not more, and it has indeed allowed many advances in rendering technology, for a few thousand dollars per dual CPU rendernode. Ask Pixar about how much the detail has increased between TS1, TS2, Monsters and Nemo.

There's no such thing in CGI as enough computing power, by the way. Artists push the detail until they reach a threshold in rendering times, and then they stop. That threshold for average render times has not changed in the past 10 years! That's why you still don't really see full blown Global Illumination and raytracing in most movie effects - there's no time to do that. Also keep in mind that there are many test renders and re-renders until a shot is considered complete.
Thus, faster computers allow more revisions and more detail, which in turn lead to better imagery.

Please tell me what's visually more advanced that can be seen between a JP Raptor or T-Rex and a troll from LOTR.

JP dinos used simple phong shaders with color, displacement and specular maps, simple matrix skinning, and a few lights per scene.

LOTR Trolls are using complex shaders (in the case of Gollum, with subsurface scattering too), full blown muscle simulation complete with secondary muscle dynamics (remember the Morannon gate in TTT?), ambient occlusion in the lighting, and they are much closer to the camera and have more screen time, too.
And if you can't see that, than you either have a problem with your eyes, or are simply trying to defend your point beyond reason.

By the way, I think you should spend some time and re-watch JP again - although the difference would be easier to see in a cinema, because DVD res is only ~720 pixels vs. 2048 in the movie theater.


I also think that you are a bit out of your expertise in movie VFX things... not that I'm an expert, but I've been living from CG for 4 years now. Even your definition of photoreal is totally wrong, because what you see in movies is absolutely nothing like real life - if it's not a documentary, then it's highly stylized, with lighting defying the real world, actors wearing makeup, colors pushed all around (there've been analog ways to do that for many years, and with CG color grading it's getting even further every day).

And to add a funny fact in the end, there are many, many effects that you just won't realize to be there, in lots of movies - effects that are totally, 100% realistic. It's just that some things are incredibly hard to do, and we don't even understand why, because we don't know how the human brain and sight works. Human beings are probably the hardest to do, whereas nonexistent prehistoric animals can't really be 'wrong'.


Edit: spelled Gollum in hungarian should go to sleep

 

[PC-Engine]

Oh gee I'm sorry. You phrased it like a deliberate troll because...?

So you've glossed over the same post you were replying to..is this becoming a habit? Since you seem lost, here it is: "There's nothing more annoying than someone who reads only the last few posts then jumps on the chance to argue a point that's already been addressed without reading the previous posts because they perceive what seems to be a flaw in a post." And for the previously addressed post wih regards to FFTSW's style, skip over to page 2.

Although they might look close in terms of visual quality LOTR used new technology. They created everything in a 3d world for the troll scene in The fellowship of the ring, the people the troll the room. Then they hooked Peter Jackson up with a piece of wood(the "camera") and some t.v. glasses, then they used motion capture on the block to move the camera in the 3d world. So the director was walking around in the 3d world and filming what was happening from where he wanted with his virtual camera.

I'm not doubting newer or more advanced technology is used in LOTR. As a matter of fact JP3's models used 3 times as much geometry as JP1. The point I was making was with respect to photorealism. The photorealism in LOTR isn't anymore advanced than that used in JP1. The technology to create photoreal images were already available back in the early 90s. It's up to the artist to make the CGI photoreal even back then.

There's no such thing in CGI as enough computing power, by the way. Artists push the detail until they reach a threshold in rendering times, and then they stop. That threshold for average render times has not changed in the past 10 years! That's why you still don't really see full blown Global Illumination and raytracing in most movie effects - there's no time to do that. Also keep in mind that there are many test renders and re-renders until a shot is considered complete. Thus, faster computers allow more revisions and more detail, which in turn lead to better imagery.

Nobody said there's enough computing power. I said there's enough computing power to render photoreal CGI as seen in JP1&2 way back in 93. Also render times are relative. For movies that are 100% CGI like Shrek, Toy Story, FFTSW, it would obviously take a long time to render. For movies like JP1&2 which are comprised of ~ 50% CGI and the reast live action, it wouldn't take that long to render. Some scenes in JP1&2 look photoreal and can be mistakened for animatronics. If you take a photograph of a real animatronic dinasour on the set of a movie and then compared it with the CGI images from the movie, they would be almost identical. Keep in mind I'm only talking about specific scenes like where the two Raptors are walking around in the lab where you seen them close up.

JP dinos used simple phong shaders with color, displacement and specular maps, simple matrix skinning, and a few lights per scene.

Actually in JP1 or 2 or both, used muscle simulation. They also use physics as seen in the Brontosaurus's jiggling flesh as they move. That's pretty much where those effects used today come from. Regardless how technical or simple it was, the end result is still photorealism in specific scenes.

Edit: Also keep in mind that SGI has been in the scientific computing industry for decades. In other words they make supercomputers as well as the simulation software that drive them. They have a leg up on algorithms using real physics modeling for various industries which would also hugely benefit the CGI movie studios for accurate physics from accurate muscle simulation used in medical imaging to barometric pressure simulation for everyday containers like coffee cans.

LOTR Trolls are using complex shaders (in the case of Gollum, with subsurface scattering too), full blown muscle simulation complete with secondary muscle dynamics (remember the Morannon gate in TTT?), ambient occlusion in the lighting, and they are much closer to the camera and have more screen time, too. And if you can't see that, than you either have a problem with your eyes, or are simply trying to defend your point beyond reason.

Regardless how complex the CGI used for Gollum, it still does not look like an animatronic. It looks like CGI.

By the way, I think you should spend some time and re-watch JP again - although the difference would be easier to see in a cinema, because DVD res is only ~720 pixels vs. 2048 in the movie theater.

I thought movie theaters used analog film reels? Regardless, I'm only talking about specific scenes in JP1&2. Yes there are scenes in JP that look fake, I never said all the scenes look real.

I also think that you are a bit out of your expertise in movie VFX things... not that I'm an expert, but I've been living from CG for 4 years now. Even your definition of photoreal is totally wrong, because what you see in movies is absolutely nothing like real life - if it's not a documentary, then it's highly stylized, with lighting defying the real world, actors wearing makeup, colors pushed all around (there've been analog ways to do that for many years, and with CG color grading it's getting even further every day).

If you can't tell if it's a real actor or a CGI actor then it's photoreal. If you can't tell if it's an animatronic puppet or a CGI model, it's photoreal. It's very simple. Yes movies rely heavily on artificial lighting, but that's irrelevent with regards to my example above because the lighting still has to follow the laws of physics.

And to add a funny fact in the end, there are many, many effects that you just won't realize to be there, in lots of movies - effects that are totally, 100% realistic. It's just that some things are incredibly hard to do, and we don't even understand why, because we don't know how the human brain and sight works. Human beings are probably the hardest to do, whereas nonexistent prehistoric animals can't really be 'wrong'.

Well like I already stated earlier in my previous post, if you substituted the dinousours with say a normal sized CGI alligators, the end result would still be the same...photorealism.

I think the majority of artists will agree that human beings are the hardest to do, that's why the human beings in FFTSW don't look 100% real. You hear that cthellis42?

 

[archie4oz]

Photorealism is based entirely on lighting. For example you can take a VERY simple 3D model and put it in a simple environment and have it look photoreal. Here's are some examples:

Actually photorealism has nothing to do with 3D rendering at all, it's an early 1900's style of painting...

And "photorealism" is a pretty crappy term anyways since photo's aren't very real. Exposure, emulsion and chemistry can easily alter the scene (and usually they are) from a "realistic" scene. And of course CCD's and CMOS sensors exhibit pretty different light gathering qualities as well. Hell I did more convincing representations with paints and pastels in school than I could do with a photo...

Now that to me looks real. It looks like a real object that I can pick up.

No, that doesn't look real, it looks convincing (to you), and there *is* a difference between the two. (for one thing, "real" scenes don't have that much aliasing... )

Same here.

Not really... Photon dispersion isn't all the great under the fenders in that scene, and there's visible faces in the tires....

That looks like a real room.

No it doesn't, real chairs don't exhibit that sort of aliasing, no can I believe that there are arm rests that are that sharp that they do not emit a specular...

FFTSW doesn't use any global ilumination

Actually it did for some scenes....

Now the reason why FFTSW looks fake is because the artists do not know how to create a photoreal human being.

No the reason they don't look "convincing" is because the artist didn't have the time nor computational resources (to generate the scenes within the alloted time)...

The technology to create photoreal images were already available back in the early 90s. It's up to the artist to make the CGI photoreal even back then.

Aside from you're annoying over-abuse of the term "photo-real", I didn't ANY decent works using BSSRDF until after 2000, so I think you're quite wrong in that regard...

Also render times are relative. For movies that are 100% CGI like Shrek, Toy Story, FFTSW, it would obviously take a long time to render. For movies like JP1&2 which are comprised of ~ 50% CGI and the reast live action, it wouldn't take that long to render.

I think Laa-Yosh was referring to a per-frame basis not overall... Although less CG doesn't make things any quicker... You just shove more time and effort away from rendering to you composting teams...

Regardless how complex the CGI used for Gollum, it still does not look like an animatronic. It looks like CGI.

That's has little to do with rendering... Has more to do with the fact that your brain is registering that there is no animatronic that moves that fluid, so you make the assumption that it is CG...

I thought movie theaters used analog film reels

Most do (assuming your local theatre isn't DLP). However film emulsions have a known LPI resolving capability and you're going to see more details than you would via DVD...

If you can't tell if it's a real actor or a CGI actor then it's photoreal.

No, that's called "convincing"...

I think the majority of artists will agree that human beings are the hardest to do

Depends on the context... The humans in Titanic are a lot more convincing than in Final Fantasy... Granted it's because they're just extras in the distance used at filler rather than key actors...

 

[PC-Engine]

No, that doesn't look real, it looks convincing (to you), and there *is* a difference between the two. (for one thing, "real" scenes don't have that much aliasing... )

Actually it does. If they were real objects photographed with a digital camera in average resolution they would look identical. When I take a JPEG picture from my digital camera and downsample it, there will be aliasing in the final picture. Does the content of the picture look any less photoreal? Do the people in the picture suddenly look like Aki or a Ben Affleck manequin? No.

Not really... Photon dispersion isn't all the great under the fenders in that scene, and there's visible faces in the tires....

And if you ask any normal person, they would say it looks real. It looks like a snapshot of a really good clay sculpture actually.

No it doesn't, real chairs don't exhibit that sort of aliasing, no can I believe that there are arm rests that are that sharp that they do not emit a specular...

To a normal person without a magnifying glass it does. You keep talking about aliasing, but it's irrelevent see above for digital camera example.

No the reason they don't look "convincing" is because the artist didn't have the time nor computational resources (to generate the scenes within the alloted time)...

If artists in general can create a photoreal human being then why isn't there 2D renders of photoreal human beings? Are you saying all the render farms in the world can't render a single frame of a photoreal human being?

Aside from you're annoying over-abuse of the term "photo-real", I didn't ANY decent works using BSSRDF until after 2000, so I think you're quite wrong in that regard...

Doesn't really matter what YOU did, all that matters is that ILM did it back in 93...

That's has little to do with rendering... Has more to do with the fact that your brain is registering that there is no animatronic that moves that fluid, so you make the assumption that it is CG...

Umm..no, but thanks for trying to explain what my brain is/isn't registering. The lighting just looks "off" sorry.

No, that's called "convincing"...

Yes go off on your semantic crusade.

Depends on the context... The humans in Titanic are a lot more convincing than in Final Fantasy... Granted it's because they're just extras in the distance used at filler rather than key actors...

I think when the camera is so far from the characters where you can't even see any detail, a human and a stick figure would probably both look the same.

 

[Jabjabs]

There's no such thing in CGI as enough computing power, by the way. Artists push the detail until they reach a threshold in rendering times, and then they stop. That threshold for average render times has not changed in the past 10 years! That's why you still don't really see full blown Global Illumination and raytracing in most movie effects - there's no time to do that.

Couldn't of said it any better. Pixar temselves are only now starting to experiment with Global illumination for use in movies, but even this is only experimentation for the time being. Take a look at Toy story 1 and look at look at their new movie The Incredibles and it's very easy to see that the artist is not the one that has full reign when it comes to developing the movies.

Speaking of TS1, I saw it again recently and personally that when compared with the Toy story skateboarding, the only reall major diffrence is in the lighting, other than that I'm sure that current machine would be more than capable of the geometry and texture work. As soon as the light systems are available, which could be very soon, I think that the thing most graphics developers (and marketers) strive for will finally come true. A software/hardware that would be capable of TS1 like visuals with no diffrence.

If development continues as it is, I would say that by 2006/7 it will be more than viable with no short cuts.

 

[PC-Engine]

Animatronic or CGI?

http://images.movie-gazette.com/thumbnails.php?album=search&type=full&search=Jurassic Park

Here's another:

 

[Laa-Yosh]

The technology to create photoreal images were already available back in the early 90s.

Bullshit, ILM had dozens of coders developing software that became the groundwork for today's appliacations at that time. There was no f***in technology - they were creating it on the fly.

Some scenes in JP1&2 look photoreal and can be mistakened for animatronics.

That's called "bad animation" and "bad shaders" for me.

Actually in JP1 or 2 or both, used muscle simulation. They also use physics as seen in the Brontosaurus's jiggling flesh as they move. That's pretty much where those effects used today come from. Regardless how technical or simple it was, the end result is still photorealism in specific scenes.

The hell they didn't, they were happy to get their characters moving att all. There was no technology for it, all they probably had was some extra bones. It is nowhere near the stuff used in LOTR and other movies today, where skin slides over muscles, muscle volume is maintained, as well as fat under the skin, etc. It's a fake effect, which can look good but absolutely not as good as a full simulation. So, once again you're convincing yourself that there's no difference, so that you can keep your stupid point that there was no advancement in 10 years...

Edit: Also keep in mind that SGI has been in the scientific computing industry for decades. In other words they make supercomputers as well as the simulation software that drive them. They have a leg up on algorithms using real physics modeling for various industries which would also hugely benefit the CGI movie studios for accurate physics from accurate muscle simulation used in medical imaging to barometric pressure simulation for everyday containers like coffee cans.

What's this about??? SGI has never developed animation software, they bought Alias and Wavefront at a time which were indipendent companies before... and they developed visualization software, not scientific stuff (that was left to their clients). Once again you're pulling out whatever crazy stuff you can imagine to defend your arguments...

Regardless how complex the CGI used for Gollum, it still does not look like an animatronic. It looks like CGI.

That's what I can only laugh about... You know what? I think Gollum is real, and the JP dinos look like lame drawings of my sister; now argue with this...

Well like I already stated earlier in my previous post, if you substituted the dinousours with say a normal sized CGI alligators, the end result would still be the same...photorealism.

CG alligators in Arnie's Eraser were looking very fake.


It's more about you being more impressed with JP at that time, compared to what you can see today... It was a revolutionary step forward, but it's like the stone age now, and everyone else can see that...

 

[Laa-Yosh]

Those dinos are all animatronics on the picture.

And the problem here is that you can not argue with PCE - he will always just say, it doesn't look photoreal, and he thinks there's no need for any other arguments. I'm fed up with that, he can go on here with his stupid agenda from now on as far as I'm concerned.

 

[archie4oz]

Actually it does. If they were real objects photographed with a digital camera in average resolution they would look identical.

What kind've digital camera? Actually it's very rare that a digital camera will capture a scene as identical. Color is usually off (and I'm not even talking about color temp/white balance), shadow detail is reduced, and often there's jitter in high-frequency reflections...

When I take a JPEG picture from my digital camera and downsample it, there will be aliasing in the final picture.

Well for starters, a JPEG's chroma sampling is going to give blocking artifacts, but not uniform aliasing... Secondly downsampling is almost always bicubic (at least in most decent applications) which is going to reduce and smooth out aliasing...

Does the content of the picture look any less photoreal?

Yes... In fact I left out the most obvious problem (the lack of any light corona, which would have cut down on light/roof aliasing)...

And if you ask any normal person, they would say it looks real. It looks like a snapshot of a really good clay sculpture actually.

So I'm abnormal? And what scale would this model need to be to have clay spokes that thin? Injection-molded polystyrene (with no flashing) sprayed with dull-coat would be more convincing than a clay model argument...

To a normal person without a magnifying glass it does. You keep talking about aliasing, but it's irrelevent see above for digital camera example.

Didn't need a magnifying glass... Took all of 3 seconds of a quick glance... And the digital camera argument doesn't cut mustard...

If artists in general can create a photoreal human being then why isn't there 2D renders of photoreal human beings? Are you saying all the render farms in the world can't render a single frame of a photoreal human being?

2D renders? You need to be more clear on what your talking about... And no, I'm saying a convincing human actor is easy to render in a still. However a moving, breathing digital human is orders of the magnitude more difficult, and is simply too impractical to create at this time...

Umm..no, but thanks for trying to explain what my brain is/isn't registering. The lighting just looks "off" sorry.

Then perhaps you could elaborate in more specific detail what is wrong with Gollum's lighting?

Yes go off on your semantic crusade.

I don't need a crusade... But things need to be clarified if you're going to try to convince others to accept your opinion of what the definition of a word who's only definition is in reference to a 90 year-old art style...

I think when the camera is so far from the characters where you can't even see any detail, a human and a stick figure would probably both look the same.

Well that's what happens when you don't qualify your statement..

Animatronic or CGI?

Animatronic, all of them...

 

[PC-Engine]

Bullshit, ILM had dozens of coders developing software that became the groundwork for today's appliacations at that time. There was no f***in technology - they were creating it on the fly.

Actually SGI themselves did a lot of the software development. This is fact.

The hell they didn't, they were happy to get their characters moving att all. There was no technology for it, all they probably had was some extra bones. It is nowhere near the stuff used in LOTR and other movies today, where skin slides over muscles, muscle volume is maintained, as well as fat under the skin, etc. It's a fake effect, which can look good but absolutely not as good as a full simulation. So, once again you're convincing yourself that there's no difference, so that you can keep your stupid point that there was no advancement in 10 years...

LOTR probably took it to a higher level, but the ground work was laid down by ILM and SGI back in 93. I'm sorry it's not fake, it was simulated. Maybe you should do some more research on SGI's role in the JP films.

What's this about??? SGI has never developed animation software, they bought Alias and Wavefront at a time which were indipendent companies before... and they developed visualization software, not scientific stuff (that was left to their clients). Once again you're pulling out whatever crazy stuff you can imagine to defend your arguments...

SGI did software development in-house probably with help from Alias/Wavefront. This is fact. If you dont' believe me write a letter to SGI regarding the effects used in JP and who created the algorithms.

CG alligators in Arnie's Eraser were looking very fake.

Yes they were. It didn't win any awards either. JP did. The movie came out later too. Just proves that the artist makes the difference. Your mileage will vary...

Those dinos are all animatronics on the picture.

Are you sure? Also did you check the two pictures in the link?

Anyway I don't have an agenda. Don't take it personal. I don't work for ILM or SGI either.

 

[archie4oz]

Actually SGI themselves did a lot of the software development. This is fact.

Fiction...

SGI did software development in-house probably with help from Alias/Wavefront. This is fact. If you dont' believe me write a letter to SGI regarding the effects used in JP and who created the algorithms.

Then you can name the tools and the engineers behind them...? Especially since Alias/Wavefront didn't exist... And what about SoftImage|3D? Isculpt? Viewpaint? Cari?

Are you sure? Also did you check the two pictures in the link?

Yes... It's pretty easy to tell... Anyways, IMO scale and extensive complexity of Phil Tippett's animatronics and animation were more impressive than the CG dinos themselves...

 

[_phil_]

Meh .. moot subject. Even reality seen directly trough our eyes is a "rasterised" one.

 

[PC-Engine]

Actually I went back and checked my sources and SGI wasn't involved in the software side of the JP movie so I stand corrected. It was ILM's inhouse stuff

That still doesn't change my original argument about photorealism.

Here's some good info on the VFX used in the first JP film.

http://silicon-valley.siggraph.org/text/MeetingNotes/ILM.html

Oh and the full blown simulation stuff was used in JP3 not JP1 still real nonetheless.


JP1:

Surface modeling is accomplished by many "potato chip" patches that are sewn together to form a skin. Underneath the skin they use muscle masses that the skin is stretched over. Spheres were used as "mass objects" to control flexing and appearance of muscles. Soft tissues were held tight at the joints, and then allowed to run free in between. Special software fixed surface patches that folded over themselves in the original modeling process. (Some of this is done by the SoftImage software, other parts of it are custom).

JP3:

The film, of course, had a host of other prehistoric digital animals ranging in size from two-feet to 51-feet tall, all modeled in Alias|Wavefront's Power Animator and ILM's ISculpt. In addition to velociraptors, spinosauruses, T rexes and pteranodons, the film also starred CG brachiosauruses, ankylosauruses, compsognathuses, parasaurolophuses, corythosauruses, stegosauruses, triceratops, and ceratsauruses.

Making all these dinosaurs move realistically took the efforts of the animation team, led by Dan Taylor, but the skin simulations created by the technical teams helped. Sebastian Marino, CG software developer, based the new skin simulation techniques on volumetric simulations developed by John Anderson, CG scientist, that were first used for The Mummy.

"Basically what we simulate is the volume of the skin and essentially the fat layer of the creature," says Marino. A three-dimensional point mesh, in which any vertex could connect to any number of other vertices, created the simulation.

This "flesh mesh" has the characteristics of foam and Jell-O. Like foam, when one end of the mesh is squeezed, it compresses but the other end doesn't get larger, and when the pressure is released, the mesh returns to its original shape. And like Jell-O, it jiggles.

To create the simulation, a dinosaur's volumes-its torso, arms, legs, and tail-were filled with the mesh, and the mesh was linked to the creature's primary motion. "It's as if the mesh is attached to a centerline down the volumes of the creatures," says Tim McLaughlin, creature supervisor. "As the critter moves, the mesh goes along for the ride."

The amount the mesh moved and jiggled during that ride depended on various stiffness and damping parameters and on whether there were obstacles such as muscles or bones in its path. The mesh contoured over muscles and bones that pushed it from one place to another as they moved-imagine punching a foam pillow. Also, waves of motion propagated through the mesh-imagine shaking a bowl of Jell-O.

Thus, the simulator caused the belly of the T rex to shake like Jell-O; and, at the other end of the spectrum, caused muscles to ripple under the skin of the raptors' athletic bodies. "As the mesh gets pushed from one place to another, you can see what's driving a creature, which is something you never get with enveloping," Marino says. "First, you see muscles flexing, and then the leg goes forward and extends. As the creature steps down, you see the muscles tighten up and see that movement affecting the fat layer."

Can you tell the difference from watching the movie? No. I rest my case.