Facebook Pixel Code
Now Reading
The “Uncanny Valley” in Character Creation

The “Uncanny Valley” in Character Creation

During the character creation process for our brand, we constantly faced challenges due to design errors. These caused deformations, mutilations, or a generally unsettling appearance in our characters. Known as the Uncanny Valley, these issues became a crucial concern that we handled with care in our project. In this article, we will guide you on how to identify and eliminate these errors in your own final designs.

Encounter with the Uncanny Valley

Our quest for realism and perfection in character representation led us to face the challenge of the Uncanny Valley several times. Each advancement forced us to reflect on our design decisions, the strategies we needed to implement, and the path we should follow.

What is the «Uncanny Valley»?

More than four decades ago, Masahiro Mori, a robotics professor at the Tokyo Institute of Technology, published an essay detailing his predictions on how people would react to robots that closely resemble humans. He theorized that as a robot’s appearance became almost human, a person’s feelings toward it would suddenly shift from empathy to disgust, a phenomenon he called the Uncanny Valley.

This essay was initially published in a less-known Japanese publication, Energy, in 1970 and went largely unnoticed for many years. However, the notion of the Uncanny Valley has gained significant traction recently, capturing the interest of both the scientific community and the public. Researchers have delved into its effects on human-robot interactions and the animation industry, while others have studied its underlying biological and social causes. This growing fascination sets the stage for our own journey into the heart of the Uncanny Valley, a journey marked by the ambitious development of our character.

Identification of the Uncanny Valley

The Uncanny Valley manifests when a digital object or character eerily resembles a human being and provokes an aversion response in observers. Consider this discarded image generated by Midjourney.

The details that fall into the «Uncanny Valley» can be clearly observed when zooming on the image. The shape of the mouth lacks harmony, and even though its face is human, we can clearly identify something about it that causes unease and automatically generates rejection toward this character. On the other hand, the character’s hands merge into each other, showing us an image where it seems there are several hands, several fingers, or a single muscle mass which we immediately categorize as a fake image.

  • uv mouth
  • uv hands

The «Uncanny Valley» is characterized by a sense of discomfort or repulsion when perceiving an almost human object as «almost, but not quite» human. Midjourney provided another example of this image, in which the arms lack human coherence, quickly generating discomfort.

  • uv cinema
  • uv cinema zoom

When we are creating a character, we want people’s reaction to our character to be one of empathy or sympathy. Therefore, it is important to take care of all the details and carefully observe the final image we decide to use to show the public. In the following image, the character we obtained generated sympathy, but upon enlarging the image again, we noticed that the hands fall into the «Uncanny Valley». This is something Midjourney continues to work on to provide better results, and users must be vigilant to identify to avoid presenting a defective product.

See Also
Estudiante Foráneo Disfruta la aventura
  • uv food 1
  • uv food zoom 1

Designers can avoid the Uncanny Valley by adjusting the design to be more realistic, modifying the character so that movements are natural, and avoiding disturbing behaviors or expressions. In the following image, we asked Midjourney to give our final character with a half ponytail hairstyle. Unfortunately, none of the images were good enough, and all fell into the valley we wanted to avoid.

Design Strategies to Avoid the Uncanny Valley

  • Deep Understanding of the Uncanny Valley: It is fundamental to deeply understand what the Uncanny Valley is and why it occurs. This allows identifying specific elements that may lead to discomfort in almost all human representations and how to avoid them.
  • User-Centered Emotional Design: Emotional design involves considering how users will feel when interacting with the character. Psychological principles should be used to create a persona that feels familiar yet clearly artificial, avoiding the sensation of strangeness.
  • Visual and Behavioral Consistency: To prevent the character from falling into the Uncanny Valley, it is crucial to maintain high visual and behavioral consistency. This includes physical appearance, movements, gestures, and voice, which should be coherent and realistic within the context of artificial intelligence.
  • Constant Iteration and Experimentation: The design process should be iterative and experimental. Different approaches should be tested, and the design adjusted based on the outcomes to find the right balance between realism and familiarity.
  • Feedback from Users and Experts: It is important to obtain continuous feedback from users and design experts to identify possible areas for improvement and adjust the design accordingly. This helps ensure that the character is appealing and avoids eliciting a negative response from users.
  • Utilization of Advanced Technology: Advanced technology, such as Midjourney’s text-to-image model, can help create more realistic characters and avoid falling into the Uncanny Valley. These tools can provide suggestions and guidance to improve character design effectively.

Results and Learnings

Our exploration of the concept of the Uncanny Valley in character creation has revealed the critical importance of finding a balance between realism and familiarity. By using Midjourney, we have developed consistent and appealing characters with a wide range of options available. However, we have also been confronted with the risk of falling into unsettling results and deformations by following certain guidelines.

Through our adaptability and analysis, we have identified and understood the model’s capabilities to avoid the Uncanny Valley as much as possible. This process has taught us valuable lessons about the importance of visual consistency, users’ reactions to unreal characters, and the need for a clear story and purpose to create characters that represent us.