Just saw King Kong. Really liked it and Naomi Watts is great. The special effects were really good. Kong looked very realistic and showed real emotion. One major nit however; when Kong jumped it looked very fake. Not sure if it was the physics, the posture, or what, but it looked bad. Anyone else have the same take?
50 views, no replies, must be massively unclear. I was thinking in particu;lar about the scene in which Kong jumps over the brdge towards the gate. There are several other scenes of him leaping as well.
…while not quite addressing your OP, here is a wonderful article with comments from the Oregon Zoo Director on the movements of Kong…
http://www.oregonlive.com/living/oregonian/index.ssf?/base/living/113573850597490.xml&coll=7
Maybe it was just me, but I noticed that, in Kong’s earlier appearances in the film, there was a sort of stuttery stop-motion look to the animation, which seemed to gradually disappear as the film went on. I figured it was a deliberate homage to the original. Was that anything like what you noticed when he was jumping around?
That’s not what I had in mind, but I do vaugely recall what you are saying.
They still can’t get the physics right in CGI motion. I’ve complained about it for years, and I keep waiting for them to catch up. Just a few days ago I said, in reference to the werewolves in Van Helsing–which are very closely related, in the CGI universe, to Kong–that I was still looking forward to the day when CGI motion was as convincing as CGI fur.
There must be a million little variables that dictate the fluidity of animal movement, and just getting the physics of momentum and trajectory obviously isn’t good enough. There’s something off about the way the individual components of a body–arms, legs, center of gravity–move through CGI space. Some factor of air resistance, inertia, muscle-mass elasticity, or something–I have no idea–that they still haven’t gotten right. I suspect that due to what I presume is the vast number of variables involved in organic movement, chaos theory, or its practical applications, will have to be involved at some deep level.
I think with Kong, it seemed almost like during the arc of his jump, he accelerated too quickly to account for his mass and bulk. Not sure that’s it; but something like that. (And yes, I know that someone involved in the process probably tried to account for that mass, but it doesn’t work.)
Well, for one thing, I don’t think big old silverbacks do much jumping. AFAIK, they tend to stay grounded. Perhaps they had to use films of jumping chimps or monkeys as their model, and the motion just didn’t translate that well to a massive gorilla.
King Kong thread at CGTalk. Some insightful comments, if you can look past the “Can’t wait to see it!!” and “Looks cool” posts.
The problem that Dan and I are referring to is not unique to King Kong. It’s the last hurdle that CGI needs to get right. It’s how you can tell the werewolves in Van Helsing are CGI, even when they’re just shadows within shadows; because the movement is somehow still unnatural; not fully subject to physical laws.
My guess, based on what I’ve gleaned from the articles about the movie, is that the jumps might be fakier looking because most of Kong’s other movements were based on Andy Serkis’ performance, via motion capture technology. Supposedly the process is so precise now that it can even capture the motion and direction of the actor’s eyeballs, which is why the closeups of Kong appear so expressive. However, pictures of Serkis in the full-body mocap rig reveal how bulky the apparatus is; its limbs are padded and extended to give him the general proportions of an actual gorilla, and I believe that much of the more physical and gestural acting was done using special platforms to adequately support the actor. I suspect that the huge jumps were one of the actions that Serkis was simply incapable of acting out, and so the filmmakers had to program wholly artificial movements for Kong, which might be why it clashes with the rest of the CGI. If Kong had been entirely animated in this manner, without an actor’s performance to lend a more natural quality to the other scenes, the jumping might not have stood out so much.
Jumps have never been realistic in any CGI. It’s the one thing that draws me out of believing a CGI creature or human is actually real. The Spider-Man movies are full of it, it’s very distracting, but even films as recent as Revenge of the Sith and Kong have it, still, and I can’t believe that they haven’t had some people concentrate some time on figuring this out. Forget real hair, or water (they still haven’t got convincing ocean waves, river water, and wakes yet, either, by the way), or cloth dynamics. Just get the damn physics of jumping right.
The ironic thing is, in the 50s Disney animated movies, they seem to have figured out jumping really well, and there’s some amazing convincing stuff from back then. Though probably they’re based around rotoscoping, that’s fine by me as long as it works - and digital animation should use the same techniques to get their stuff right too. I’m amazed that they haven’t.
The thing about jumping is not only does your centre of gravity shift in a distinct way, but your mass follows through, all unique to each individual. It’s a tough thing to grasp, and then tougher to find a way to track it and simulate it.
It’s funny this thread is up now because I was just thinking the same thing.
lissener and GuanoLad have nailed it; it’s not King Kong, specifically, it’s EVERY SINGLE MOVIE where a CGI animation jumps or flies through the air. Absolutely every single one. It just never looks lifelike. It’s not just Kong, or lissener’s example of “Van Helsing,” it’s all of them. When Legolas hops off the Oliphant he kills in “Return of the King,” it looks phony as well. All the Spider-Man animations of Spidey jumping and landing on something seemed wrong, somehow. “The Brothers Grimm” had several examples.
Yes, I forgot to mention Legolas. And Hulk.
Another thing: they can’t get the light in people’s mouths right. People’s teeth always look funny, like they’re in some kind of exaggerated darkness. You’d think they’d have all that figured out–light on surfaces–but they haven’t yet.
There was a moment in Chronicles of Narnia where I think Peter was thrown off his horse or something, and I suddenly thought, “he’s turned into Spider-man!” just because of the way his cgi body double moved.
After seeing the movie a few nights back, the ladyfriend commented that she thought the CGI on Kong looked exceptional but the animation of Naomi Watts in Kong’s hand pulled her out of it. Specifically, the times when Kong is running and holding her and she’s flopping around like a rag doll.
I think a reason it may not “look exactly right” is because there is nothing to compare it to in real life.
What are you going to model it after? Nothing of huge mass makes giant leaps.
You can look at insects jumping, squirrels leaping, chimps leaping to branches, but what are you going to do for BIG things.
We already know that if you knock down a tower of blocks it falls pretty quickly. But when a massive building falls it looks like its falling in slo-mo. So when a huge ape like Kong makes a leap it should probably look somewhat slo-mo. But since no-one has ever seen that before in real life, it’s probably going to look strange or ‘fake’ to people.
I think the over-quickness of his falling is part of it. If Kong fell his body height (20-25 feet) as slowly as objects do in the real world it would probaby look too much like the movements in Japanese monster movies.
No, and no. Again, this problem is not unique to Kong. No CGI has *EVER *gotten it right; it’s the last frontier of CGI verisimilitude.