Figure 5 - uploaded by Kai Riemer
Content may be subject to copyright.
The experimental layout illustrating the 'Wizard of Oz' researcher behind a wall controlling the avatar.
Source publication
The 40-year-old Uncanny Valley theory is influential in the discussion surrounding acceptance of realistic graphical agents. This theory was formulated by observing robots. While it has been shown to be valid when observing digital characters, little has been studied about acceptance when people interact with avatars, rather than simply observe a r...
Citations
... It should be noted that some studies demonstrated that people are prone to find realistic characters more attractive than their stylized versions [McD12,SRK17]. Schwind et al. ...
... Interestingly, this result highlights that a character could be considered attractive, while being perceived as more eerie at the same time. It could be argued that this difference between perceived attractiveness and eeriness scores may be due to the fact that a more human-like character could be perceived as more appealing [McD12,SRK17]. Indeed, Schwind et al. ...
The perception of virtual human faces is impacted by several interrelated factors, such as shapes, materials, textures, shading or animation. Previous research demonstrated that animation flaws and inconsistencies tend to be less accepted on realistic characters compared to stylized ones. To further investigate the relationship between stylization and facial expressions, we designed an online investigation to assess participants' perception of virtual human faces with regards to perceived realism, attractiveness and eeriness. We modeled a female and a male character with three levels of stylization (stylized, semi-realistic, realistic) and four expressiveness intensities (idle, low, moderate and exaggerated). 135 participants took part in the experiment which consisted of watching a set of 24 videos lasting 30 seconds each. Interaction effects were observed between the independent variables. Overall, results indicated that perceived realism was mainly driven by the stylization level. Visually realistic characters were perceived as more realistic even with exaggerated facial expressions. In terms of attractiveness, scores tended to decrease as the intensity of facial expressions increased. At the highest expressiveness intensity, the semi-realistic female character was preferred over the stylized and the realistic ones, while the realistic male character scored higher than the others. In line with previous studies, results indicated that perceived eeriness rises with the intensities of facial expressions. Taken as a whole, this study provides valuable guidelines for 3D artists and developers to design avatars and autonomous agents.
... Recientemente, soluciones como Metahumans de Epic Games [19], han creado nuevas oportunidades, ya que pone a disposición del público una amplia gama de humanos digitales realistas y gratuitos, listos para ser utilizados dentro de procesos profesionales. Esto convierte a los humanos digitales realistas en herramientas sencillas y asequibles, que con tecnologías como la captura de movimiento y los gráficos 3D en tiempo real, se logran resultados de interacción mejorados [15].Todos estos avances tienen la capacidad de cambiar la percepción de un ser humano digital gracias a la construcción de un vínculo emocional [25]. Este vínculo es necesario para impulsar el realismo, a fin de superar el Valle Inquietante [23], un término creado para describir el punto donde la respuesta emocional a representaciones humanas que tienen una apariencia y comportamiento similar al de un ser humano, causan una reacción negativa de extrañeza e inquietud. ...
... El análisis para evaluar la percepción de muestras emotivas visuales en la interacción humano-computadora, por lo general involucra la medición del Efecto del Valle Inquietante, que se puede realizar a través de la evaluación de las dimensiones de la personalidad con la encuesta Big-Five [8] como en Hyde et al. [18], a través de la percepción de las dimensiones afectivas definidas por Ho y MacDorma [17] o a través de ensayos interactivos como la modificación al experimento del Mago de Oz, aplicado con humanos digitales por Seymour, Riemer y Kay [25], donde un humano digital se expone a sujetos humanos, mitigando el efecto del Valle Inquietante con interactividad. ...
... El diseño del estudio propuesto expone a cada sujeto humano a una sola tecnología de captura facial y una sola exposición interactiva, evitando cualquier sesgo debido a experiencias previas o expectativas emocionales. [25] para integrar la interactividad en tiempo real como una variable clave, que se complementa con el uso de un visor de RV en el presente trabajo. ...
Resumen. El creciente universo de creadores de contenido virtuales, avatares del metaverso, y humanos digitales en general, ha creado una oportunidad para integrar soluciones de captura facial en un amplio panorama de nuevas aplicaciones para la industria de creación de contenido. Junto con este crecimiento se ha incrementado la demanda por humanos digitales que generen empatía y cuenten con un mejor desempeño en sus expresiones faciales. Es por esto que en el presente artículo se exploran las principales codificaciones faciales empleadas para la captura de movi-miento facial y las diversas soluciones existentes para dar vida a humanos digitales. Adicionalmente, se presenta un experimento realizado con un humano digital dentro de un ambiente de realidad virtual para medir el vínculo empático creado a partir de algunas tecnologías recientes de captura facial, Faceware, Live Link UE, y Avatary. Los resultados explo-ran la percepción de determinadas expresiones emocionales, la respuesta empática, y la semblanza de familiaridad que reflejan las soluciones disponibles. Finalmente, se discute la necesidad de alternativas nuevas y accesibles con una codificación más expresiva, como medio para abrir el panorama a un amplio campo de investigación que busca mejorar la captura facial. Palabras clave: Realidad virtual, captura de movimiento facial, interfaces humano-computadora, humanos digitales, codificación facial. Abstract. The growing universe of virtual content creators, metaverse avatars, and digital humans in general has created an opportunity to
... The uncanny valley effect suggests that as the realism of a human representation increases, affinity increases, but only to a certain extent. Once realism exceeds a certain point where it becomes too human-like, there is a sharp drop in acceptance or affinity (Rokeman et al., 2020;Seymour et al., 2017). ...
Background
Zoomorphism, an obverse of conventional anthropomorphism, has been shown in many cases to be a powerful tool with respect to the metaphorical use of animals.
Objectives
This study introduced zoomorphism into the instructional video design of pedagogical agents (PAs).
Method
We first developed three image sets, each of which corresponded to a type of zoomorphic PA image construct (full panda, a human face with a panda body, and a human body with panda head imagery). Then, we conducted an image‐screening experiment in which 177 second‐grade students from a primary school were asked to select their favourite PA image for each image construct set. We used a realistic human PA as the baseline. Next, 114 second‐grade students from another school were randomly assigned to learn instructional videos on the topic of pandas using a PA with one of four imageries.
Results and Conclusions
ANOVA and Kruskal–Wallis tests indicated that students presented with PAs with full‐panda imagery achieved significantly higher learning performance, learning experience, social presence, and learning interest than those presented with other imageries of PAs. The regression results showed that the effect of zoomorphic PAs with full‐panda imagery on learning performance was exerted in a direct way and that its effect on learning experience was mediated by social presence and learning interest.
Implications
Our findings suggest that zoomorphic appearance is an essential quality for video learning that merits further investigation for the efficient imagery construction of PAs.
... robots) and stillness (e.g. images) in eliciting eeriness, the UVH has more recently been applied to computer generated and moving images (Seymour et al., 2017). ...
Virtual influencers engage in emotional sharing to gain and keep followers. However, given that many people use social media for diversion purposes, this emotional sharing may hinder users‘ ability to escape from everyday emotional experiences, particularly for highly empathetic individuals. Using a between subjects, randomised experimental design, we explore how empathy affects reactions to virtual vs. human influencers, showing that those
highest on empathy are more likely to follow a virtual influencer, and rate her as more socially attractive, than a comparable human influencer; these results disappear when the influencers‘ true nature is unknown to participants. We postulate that these results represent an ―escapism effectǁ , where the virtual influencer is expected to provide greater diversionary benefits from everyday human emotional experiences and require fewer cognitive resources in the form of emotional sense making. We present practical implications and future research opportunities arising from this effect.
... Based on the interaction with a Digital Human, the "Wizard of Oz" experiment [9] suggests using a digital marionette exposed to humans. The experiment presented in this work adds a VR headset as a new layer to such experiments, as observed in Figure 2. The experiment can be operated locally or remotely (with video conference tools). ...
... The "Wizard of Oz" Experiment with Digital Humans used by Mike Seymour [9] offered a new perspective to these experiments, integrating real-time interactivity as a key variable, which is completed perfectly by using a VR headset. ...
State-of-the-art Digital Humans are now easy and affordable tools
because of the availability of realistic models [5], Motion Capture
(Mocap), and Real-Time 3D graphics that archive improved interaction
results. The use of performance capture tools increased in
the middle of the 2010s decade [1], along with the improvement
of video processing in real-time. The first options were based on
Head-Mounted Cameras (HMC) with video capture processed in
postproduction. Then, live tracking tools based on facial recognition
became popular, along with the introduction of the TrueDepth
camera on iPhones. These facial tracking tools are mostly based on
standards known as Facial Codifications. The two most common are
the Face Animation Parameters (FAPs) and the Facial Action Coding
System FACS. FAPs are the most common standard in traditional
animation scenarios; however, as realism became a priority, the
need to overcome the uncanny valley [8] pushed its limits. Followed
by this, the FACS methodology, proposed by Paul Ekman in the 70’s
[2], started to be used in Digital Humans a few years ago [7] since it
considers muscle movements. Additionally, Ekman’s Emotion FACS
(emFACS) model [3] defines movement combinations resembling
specific emotions, which makes it possible to evaluate the empathic
response to Digital Humans.
Based on this, and to determine how state-of-the-art solutions
contribute to these realistic and emphatic interfaces, this document
presents an experiment designed and tested with Virtual
Reality (VR). The experiment focuses on the interaction with an
Epic MetaHuman [4], driven by a contemporary set of solutions:
Live Link Face UE by Epic Games, Faceware Studio, and Facegood
by Avatary. All of them use the 51 blendshapes method and are
options available for research and content creation.
... In addition to this article, based on the author's study, it was noticed that avatars cannot be very similar to the individual, therefore, as they lead to the Uncanny Valley (Seymour et al, 2017) and also not very distant from the user's appearance because they enable the Proteus Effect (Yee et al, 2009), thus not causing psychological implications that will affect the usability of the metaverse. ...
The metaverse has garnered global interest due to the advancement of virtual reality (VR). Avatars play a pivotal role in the metaverse, acting as a 'second self' for users. This ongoing research aims to identify the optimal approach to avatar creation, considering effectiveness, accessibility, and acceptance. Consequently, this study fills a gap in scientific literature by providing a comparative analysis for avatar creation in the metaverse, utilizing the Design Science Research (DSR) methodology. The research will provide valuable insights for the development of more effective platforms, maximizing user identification and interaction in the metaverse.
... The authors performed perceptual experiments, which explored how different anomalies have relative importance. Regarding the interactivity of experiments with VR [Schwind et al. 2018, Pelechano and Allbecky 2016, Musse et al. 2021, in the work of [Seymour et al. 2017] the authors observed the effect of UV theory on user interaction with the photo-realistic human avatar without just looking at its appearance. The area of group perception has grown in recent years. ...
Virtual characters can elicit an uncomfortable sensation usually known as Uncanny Valley (UV). Evidence suggests that animation exacerbates the UV effect. This paper revisits the UV hypothesis to assess its effects on people’s perception of virtual characters, to try to answer the questions “What happens to the perceptual comfort if, in addition to images and videos of characters, we include those characters in an interactive Virtual Reality (VR) environment?”, and “Considering the VR environment, do people feel equally comfortable if we increase the number of characters?”. The results indicate that there are differences in the perception of scenarios with low and high densities.
... These limitations may result in avatars that cause a feeling of uncanniness or eeriness in end-users Seymour, Riemer [8], and may be associated with Mori and MacDorman [9] Uncanny valley theory. This theory was originally associated with robotics but has transitioned into HAI, where human-like avatars that have been created using computer graphic capabilities with a focus on boundary-pushing realism [10,11]. ...
Despite the variety of applications that use avatars (virtual humans), how end-users perceive avatars are not fully understood, and accurately measuring these perceptions remains a challenge. To measure end-user responses more accurately to avatars, this pilot study uses a novel methodology which aims to examine and categorize end-user facial electromyography (f-EMG) responses. These responses (n = 92) can be categorized as pleasant, unpleasant, and neutral using control images sourced from the International Affective Picture System (IAPS). This methodology can also account for variability between participant responses to avatars. The novel methodology taken here can assist in the comparisons of avatars, such as gender(sex)-based differences. To examine these gender(sex) differences, participant responses to an avatar can be categorized as either pleasant, unpleasant, neutral or a combination. Although other factors such as age may unconsciously affect the participant responses, age was not directly considered in this work. This method may allow avatar developers to better understand how end-users objectively perceive an avatar. The recommendation of this methodology is to aim for an avatar that returns a pleasant, neutral, or pleasant-neutral response, unless an unpleasant response is the intended. This methodology demonstrates a novel and useful way forward to address some of the known variability issues found in f-EMG responses, and responses to avatar realism and uncanniness that can be used to examine gender(sex) perceptions.
... However, imperfect human-likeness can provoke dislike or strangely familiar feelings of eeriness and revulsion in observers, a phenomenon known as the uncanny valley [47]. Studies suggested that interaction and animation can overcome the valley in affinity due to matching and common human non-verbal cues [58]. In our platform, we use human-like VHs for pursuing a realistic representation of tool and machine usage in the context of TC processes simulated in a VE (e.g., most cartoon-style avatars do not have five fingers, resulting in poor and unrealistic handling capabilities with tools). ...
Visualizing human motion is a topic that has gained increasing attention in the domain of cultural heritage, due to the need for capturing intangible dimensions, existing for example in theatrical performances, dances, and crafts. In this respect, virtual humans are typically employed to re-enact human motion, executing movements reproduced through predefined animations, or physics simulation engines. In the case of traditional crafts, a defining point is how to model the interaction of virtual humans with craft-related objects and how to transfer it from the physical to the digital world. Toward a more effective and generic modeling and visualization of the interaction of humans with tools and machines utilized in crafts, this article proposes a novel methodology for the modeling and visualization of crafts and presents a platform enabling the authoring and visualization of craft processes. We contribute a way of visualizing craft processes within virtual environments, aiming to increase the usability of craft representation. As an example, we present and analyze the case of the craft of weaving with the use of a loom.
... One-fourth of these studies focused on virtual interactive representations (i.e., with a virtual human (e.g., Cafaro et al., 2016;Krämer et al., 2013)). Further, 17 percent of the studies address CAs with virtual static avatars (e.g., images (Seeger et al., 2018)), 8 percent explore agents with physical embodiment (e.g., service robots (Stock & Merkle, 2018a, 2018b), and a handful investigate and compare multiple embodiments (e.g., Gong, 2008;Seymour et al., 2017). ...
... The Uncanny Valley, in short, postulates that there is no monotonically increasing relationship between the human-likeness of a technological artifact and a human affinity toward it, but that there is a point (i.e., the Uncanny Valley) where "a person's response to a human-like robot would abruptly shift from empathy to revulsion as it approached, but failed to attain, a lifelike appearance"(Mori et al. 2012, p. 98). On this theme, for example,Seeger et al. (2018) found empirical evidence for an adverse emotional reaction to a CA that exhibits only partial humanlike characteristics, andSeymour et al. (2017) suggested that interactivity, realized through an interactive 3D avatar that matches common human non-verbal cues, can contribute to overcoming the Uncanny Valley sensation. In addition to the CASA paradigm and the Uncanny ...
Conversational agents (CAs), described as software with which humans interact through natural language, have increasingly attracted interest in both academia and practice, due to improved capabilities driven by advances in artificial intelligence and, specifically, natural language processing. CAs are used in contexts like people's private life, education, and healthcare, as well as in organizations, to innovate and automate tasks, for example in marketing and sales or customer service. In addition to these application contexts, such agents take on different forms concerning their embodiment, the communication mode, and their (often human-like) design. Despite their popularity, many CAs are not able to fulfill expectations and to foster a positive user experience is a challenging endeavor. To better understand how CAs can be designed to fulfill their intended purpose, and how humans interact with them, a multitude of studies focusing on human-computer interaction have been carried out. These have contributed to our understanding of this technology. However, currently a structured overview of this research is missing, which impedes the systematic identification of research gaps and knowledge on which to build on in future studies. To address this issue, we have conducted an organizing and assessing review of 262 studies, applying a socio-technical lens to analyze CA research regarding the user interaction, context, agent design, as well as perception and outcome. We contribute an overview of the status quo of CA research, identify four research streams through a cluster analysis, and propose a research agenda comprising six avenues and sixteen directions to move the field forward.
