green trees with fog

What Would Make An Animated Character Appear “Creepy”?



Realistic animated character that looks creepy.

Have you ever stopped to think about animated characters? With the capabilities of graphics these days it’s possible for an animated character to look just like a real person. And then there are still cartoon characters created that look nothing like real people. Have you ever experienced an animated character that “creeped” you out?

Animators have to make constant decisions about how realistic a character should be, and what that even means. Research shows that there is a point where animated characters are not “cute” anymore, and actually can become “creepy”. This point is called the “uncanny valley”.

This semester at the University of Wisconsin, Stevens Point, I worked with a student on an independent study project about the uncanny valley. This blog post is a guest post by the student, Kierstan Leaf, who describes the research study she did this semester:

The Uncanny Valley is the idea that as things, particularly robots and animated characters, become more realistic they eventually hit a point where we determine them to be creepy and nonhuman. This is due to the small inconsistencies that we see within the characters, for example, the skin texture or reflection in the eyes may seem a bit off.  We unconsciously notice these things because these are attributes that we observe daily in our interactions with people.

The Uncanny Valley theory originated from Masahiro Mori, while working with robotics in 1970. He compared the relationship between robots and their “degree of human likeness” (see the references below). Mori noticed that when robots become more lifelike they began to be viewed as being creepy. On the other hand when the robots did not have much human likeness, such as a robot in a factory, the creepy level was very low, if non-existent. 

For this study on the uncanny valley I took images from movies, cartoons, and television shows. I used images that ranged from “less realistic” (in other words, not human-like) to “more realistic”. These images were shown to 58 people to rate on a scale from 1 to 10 where “normal” was at one end and “creepy” at the other. I hypothesized that as the images become more realistic they would be considered creepier. Here’s a short video that summarizes the research shows the images I used, and the results of the study.

The hypothesis was correct. The more realistic the images were, the more creepy people rated them.

So what does this mean for decisions about animations in design? If  you would like your viewer to fall in love with your character nearly instantaneously, then perhaps you should stick with more cartoonish designs.  If you want your user to be scared of a monster or evil villain, you can push the line of realism and tip your viewer over to the creepy side. Knowing these unconscious reactions exist, you can apply them to your projects.


Mori, Masahiro. The Uncanny Valley. Trans. Karl F. MacDorman and Norri Kageki. IEEE Spectrum, 2012. Web. 11 Dec. 2014. <>. 

Karl F. MacDorman. Exploring the Uncanny Valley. Web. 11 Dec. 2014. <>.

What do you think? What do you think makes animated characters cross into the “creepy” realm?

If you have questions for Kierstan you can reach her at


5 responses to “What Would Make An Animated Character Appear “Creepy”?”

  1. Adam Ruf Avatar

    Interesting stuff, Susan & Kierstan, thanks!

    I couldn’t help but notice that the three animated characters with the lowest creepiness rating were either characters that have been around a long time (i.e., Snow White, The Flintstones) or have recently experienced high popularity/visibility (i.e., Frozen (I think)).

    Do you know if participants were familiar with any of the characters they rated?

    Also, would you expect to find a similar Uncanny Valley for nonhuman species, like animated cats and dogs? Why or why not?

    1. Susan Weinschenk Avatar
      Susan Weinschenk


      Here’s Kierstan’s reply:

      We do not know how familiar any of the participants were with the characters.

      I would expect a similar Uncanny Valley for nonhuman species, but not to the same extent. We interact with other humans on a daily basis, and have come to unconsciously notice their skin texture and light reflections. Therefore, we can then consider images to be creepy when they do not line up with our expectations.

      On the other hand, not all people interact with a cat or dog on a daily basis. Therefore, what is real/creepy may be harder to differentiate since a person may not have spent as much time with the animal to know what the norm is.

      Overall, these are my initial thoughts; it would definately make for an interesting research project.

  2. Karine Thériault Avatar
    Karine Thériault

    Interesting subject and study. But my first impression is to wonder if there wouldn’t be any other variables besides the realism of images that could have influenced the ratings. The pictures differs more than in terms of the realism level of the characters, they also present different emotions, atmospheres, and normal vs distorted body parts.

    For instance, most of the pictures at the bottom of the creepy scale show happy characters with light surroundings, not in a context of interaction (they mostly look in front of them as if they were posing for the picture. As the ‘creepy’ rating increases, characters seem worried or angry, in darker surroundings and they seem more in interactions with either another character or something we don’t see.

    As for the 2 creepiest characters, they both have something odd. One has really disproportionately big eyes and looks like a villain, the other has really small pupils giving an impression of snake eyes.

    Where these variables considered in the analysis of the results?

    1. Susan Weinschenk Avatar
      Susan Weinschenk


      Yes there are likely many variables. Since Kierstan was using existing images it wasn’t possible to control for many of the variables. It is possible that the “odd” factor is connected to the “creepy” factor. Time for more experiments!

  3. Mark Wangwiwattana Avatar
    Mark Wangwiwattana

    This research is interesting. I think it should be more carefully tested. There are many variables involved. I agree with Karine. To add more, the participants may have the impression toward on of those characters from past experience. It can influent the creepiness. For example, Courage The Cowardly Dog got very high creepiness, although it is cute cartoon, because its story is a horror.

Leave a Reply

Your email address will not be published. Required fields are marked *