Microsoft ends support for Internet Explorer on June 16, 2022.
We recommend using one of the browsers listed below.

  • Microsoft Edge(Latest version) 
  • Mozilla Firefox(Latest version) 
  • Google Chrome(Latest version) 
  • Apple Safari(Latest version) 

Please contact your browser provider for download and installation instructions.

Open search panel Close search panel Open menu Close menu

October 10, 2025

Escape from the Uncanny Valley

Sometimes you can get the ick when you're watching a superhero movie. A character leaps from a rooftop or gets thrown across a battlefield, and there's just something about the way they move through the air that feels... off. It's too floaty, the landing is too soft. Like gravity isn't working properly. And it takes you right out of the movie until all you can think about is how much you wish they'd spent a bit more on the special effects.

We Can Tell...

There's actually a name for this sensation: the "uncanny valley." It was originally a description of the queasy feeling people get when a humanoid robot or digital face is almost, but not quite, lifelike, but is now more generally used for anything that feels close to real, but slightly wrong. When it comes to full-body character animation in movies and games, you can tell when a jump doesn't land right.

We tend to notice when something is a bit off, but not always. It turns out our brains can be surprisingly forgiving when it comes to watching a ball bounce, for instance. A new study by NTT Communication Science Laboratories has found that our sense of naturalness changes depending on whether we're observing a simple object or a complex, coordinated human motion.

Humans are Complicated

In brief: if a rubber ball flies through the air and bounces more gently than it would under Earth's regular gravity, most people will still perceive the motion as natural. But when it comes to watching a human jump, we're a lot more discerning.

For decades, researchers in perceptual psychology have known that humans use internal, approximate models of physics to judge motion. These intuitive, rule-of-thumb models work well enough, but are not completely accurate. Research from NTT Communication Science Laboratories shows that we sometimes see object motion as natural, even when it's inconsistent with real-world gravity. For example, when a ball appears to bounce under gravity slightly weaker than Earth's—say 60% of normal gravity—people still tend to view it as natural. This leads us to believe that our brains aren't especially strict about physics, at least when it comes to simple movements.

But human jumping is something else entirely. Unlike when a ball bounces, a jump involves the whole body, with legs, torso and arms moving in coordination. Complex acceleration and deceleration patterns come into play.

Prisoners of Gravity

NTT researchers tested whether our brains treat these more complicated motions differently. Using human motion capture data, they created dozens of video clips where the height and duration of jumps were adjusted, making some jumps higher or longer, others lower or faster. Participants then watched these clips and rated how "natural" each one appeared.

And the results were clear: participants consistently saw jumps as natural only when they followed trajectories that aligned closely with Earth's gravitational laws. Even if the jumps didn't exactly mirror real human capabilities—say, being a bit taller or longer than normal—as long as the motion matched Earth gravity, it was usually thought to be realistic.

By contrast, jumps that were similar to the kinds of arcs that feel natural for a bouncing ball under weak gravity were rated as unnatural. How come?

To explain the difference, NTT researchers developed a computational model of how people might process jump motion. They programmed the model to assume that observers look at the initial velocity of the jump from the video, use their internal understanding of Earth's gravity to predict how that jump should progress, and then compare the predicted motion to what they see. The closer the match, the more natural the motion appears. The model was very accurate: the correlation between predicted naturalness and participants' ratings was -0.89. It's a high score. If the correlation were 1, it would mean the model's guesses and people's ratings always matched exactly.

The model's predictive power dropped when gravity was tweaked; only when it used natural Earth gravity did results align well with human perception. This seems to suggest that our brains use different internal models for different types of motion: an approximate one for bouncing balls, and a much more precise, gravity-bound one for human movement.

Better, More Real Characters

All very interesting, you might think, but what does this mean in the real world? Well, for one thing, it's extremely useful information for designers of virtual worlds, video games, and films. Making virtual, game and movie characters "real" is more complicated than you might imagine. Along with creating anatomically accurate characters, designers also have to try to match the invisible physics we subconsciously expect, especially for movements like jumping, which are parts of the lives we have all lived under Earth's gravity.

... along with Better Rehab, More Agile Robots and Superior Athletes

Outside of entertainment, NTT's research has genuine value. In physical rehabilitation, for example, where recovering patients relearn everyday movements, systems that detect whether motion appears "natural" could help monitor progress more reliably. In robotics, models that understand how to move properly could guide the development of humanoid machines or prosthetics that move in ways people find easier to trust and interact with. And in sports science, comparing an athlete's movements to what the brain perceives as physically correct might help flag subtle inefficiencies or the early signs of strain or injury. The possibilities are wide and exciting.

NTT Communication Science Laboratories aren't finished yet. They now plan to look into whether the same principles apply to animals or other entities whose motion might fall somewhere between a ball and a person. Once we can understand the brain's internal physics engine properly, we can make digital motion feel both far more believable and immersive. Watch this space.

NTT—Innovating a Sustainable Future for People and Planet

For further information, please see this link:
https://group.ntt/en/newsrelease/2025/02/17/250217b.html

If you have any questions on the content of this article, please contact:
Public Relations
NTT Science and Core Technology Laboratory Group
https://tools.group.ntt/en/rd/contact/index.phpOpen other window

Picture: Daniel O'Connor

Daniel O'Connor joined the NTT Group in 1999 when he began work as the Public Relations Manager of NTT Europe. While in London, he liaised with the local press, created the company's intranet site, wrote technical copy for industry magazines and managed exhibition stands from initial design to finished displays.

Later seconded to the headquarters of NTT Communications in Tokyo, he contributed to the company's first-ever winning of global telecoms awards and the digitalisation of internal company information exchange.

Since 2015 Daniel has created content for the Group's Global Leadership Institute, the One NTT Network and is currently working with NTT R&D teams to grow public understanding of the cutting-edge research undertaken by the NTT Group.