EmoBlend: Revealing the Impact of Facial Blendshape Intensity on Emotional Perception

We investigated the impact of facial blendshape intensity on users’ emotional perception experience and and the uncanny valley effect through semi-structured interviews and 7Likert-scale questionnaires, according to six types of emotions.

Furthermore, we proposed a specific implementation method to alleviate the uncanny valley effect and enhance the efficiency of emotional perception for users, allowing both conversational parties to have a better experience.

As the experiment’s platform, this repository has implemented face2face chat in VR, utilizing the official SDK for Oculus Movements.

• Face2Face Chat: Through TCP socket programming, I have developed a scene that allows two virtual avatars to engage in face-to-face communication within a virtual environment. Both users can see each other’s body movements and facial expressions in real time, providing a new form of social interaction.

• Skeletal-driven avatar: During the chat, users can enable a recording feature, allowing Unity to record the users’ skeletal motion states and audio. This recorded data can be utilized after the chat session ends to drive a new virtual avatar to replicate the previously recorded actions and speech. This enhances the playability and interactivity for users.