A collaboratively formulated protocol was established and ratified by all contributing authors. This protocol functioned as a foundational framework for the inclusion of scientific articles that align with the fundamental objectives of the research—specifically, the representation of virtual entities (avatars), the determinants influencing their design, the technologies utilized for their manipulation, and their varied applications across multiple disciplines.

A principal eligibility criterion mandated that the articles directly engage with these thematic areas. In particular, studies were anticipated to incorporate key terminologies such as “avatar” or “digital body,” as these terms are extensively acknowledged within the academic discourse. This linguistic stipulation further augmented the accuracy of indexing within academic repositories and facilitated the effective retrieval of pertinent literature.

Research that employed the term “avatar” in extraneous contexts—such as religious symbols (e.g., Hinduism), fictional portrayals (e.g., television series), or philosophical notions—was systematically excluded. This ensured that the review remained concentrated on VR applications and evaded peripheral subjects.

In instances where multiple studies addressed analogous themes, precedence was accorded to highly cited articles. This methodology guaranteed that the review encompassed influential and widely acknowledged research, thereby augmenting the credibility and impact of the findings.

For the section elucidating factors influencing the appearance of virtual characters, efforts were undertaken to incorporate studies that examined a comprehensive array of psychological, social, ethical, and legal dimensions. This expansive approach ensured a thorough comprehension of the multifaceted influences on avatar design.

In the section devoted to reviewing technologies, studies were included if they discussed tools—either in part or in full—utilized to control the appearance of virtual characters. The objective was to identify and showcase a set of tools that could be employed in forthcoming VR projects or as components of other associated endeavors.

The selection process was executed through several systematically organized stages:

Initial Screening: Titles and abstracts were initially scrutinized independently by Hamid Ouhnni and Acim Btissam predicated on their correspondence with the pre-established eligibility criteria.

Dispute Resolution: In cases of contention regarding the inclusion of a particular article, the matter was referred to an internal arbitration committee composed of Meryam Belhiah and Soumia Ziti, who evaluated the disputed articles and rendered the final decision concerning their inclusion or exclusion.

Thematic Categorization: Subsequent to the selection process, a qualitative categorization of the included studies was undertaken in two collaborative phases. The initial round was led by Hamid Ouhnni, who collaborated with Btissam Acim, Belhiah Meryam, and Benachir Rigal to allocate the studies to preliminary thematic categories.

Validation and Refinement: In the subsequent round, the categorization underwent review and refinement by a group led by Soumia Ziti, with contributions from the remaining co-authors. This phase sought to ensure thematic consistency, analytical clarity, and methodological coherence.

All procedures strictly adhered to the predetermined protocol, which was conceptualized and supervised by Hamid Ouhnni, who also coordinated the ultimate validation of the review process to ensure its quality and methodological rigor.

Ultimately, studies were included if they investigated the practical applications of VR, particularly those involving the utilization of avatars. It is noteworthy that certain papers addressed multiple dimensions of the research, providing a broader perspective on the subject matter.

The databases selected for this comprehensive review encompass Scopus, Web of Science, ACM Digital Library and IEEE Xplore, all of which are esteemed for their rigorous, peer-reviewed content and extensive interdisciplinary research coverage. Nonetheless, Scopus was designated as the principal source owing to its particular advantages in indexing literature pertinent to computer science, human-computer interaction, and virtual reality, which are integral to the focus of this research. For this source, we consistently utilized the search term queries outlined in (Table 1). The search terms were mandated to be incorporated within the title, abstract, or author keywords of the respective articles. We limited our selection to articles published between the conclusion of 2009 and January 2023.

The end of 2009 was strategically selected as the cutoff point, marking the start of a pivotal shift in the evolution of virtual reality (VR) technologies and avatar customization. This temporal milestone aligns with the advent of crucial technological innovations that propelled VR towards mainstream acceptance. For instance, in 2010, Palmer Luckey engineered the inaugural prototype of the Oculus Rift, which boasted a 90-degree field of vision and groundbreaking software solutions to mitigate lens distortion (Biju Kunnumpurath and Tribe, 2023). This prototype established the groundwork for contemporary VR headsets and rejuvenated interest in VR applications, including the aspect of avatar customization. Prior to this epoch, VR technologies were predominantly experimental and constrained in scope, with a scarcity of studies concentrating on user-centered dimensions such as avatar customization. By choosing the end of 2009 as the cutoff point, we ascertain that the literature under review accurately reflects the contemporary landscape of technological advancements and user anticipations, which are essential to the significance of this research.

We formulated the search term queries in two distinct sets of combinations: In the initial set of combinations, we utilized “avatar” as the primary keyword “AND” (“customization” OR “personalization” OR “embodiment” OR “creation” OR “generation”) as the secondary keyword. Moreover, the second set of combinations was structured as follows: “virtual body” as the primary keyword AND (“customization” OR “personalization” OR “embodiment” OR “creation” OR “generation”) as the secondary keyword. For those databases that allow for the specification of the language of the publications, we opted for “English.” Additional articles identified through supplementary sources and citation indexing were also incorporated to guarantee a thorough review.

The Mendeley Reference Manager and its associated browser extension were employed to collect and organize the data. Information from each entry, such as titles, authors, abstracts, and the source of information, was compiled. This was done either by importing RIS files generated from the databases or by using the Mendeley Web Importer extension. In some cases, original work was manually searched through Scopus or captured using the Mendeley Web Importer. The authors express their gratitude to the library of their host institution for providing access to all the necessary data.

The selection methodology is outlined in the flow diagram presented in (Figure 1) above. Initially, superfluous items were expunged. Subsequently, we undertook a screening process predicated on the title and abstract, executing level 1 evaluations, and perused a select few through comprehensive text analysis. During this phase, we eliminated manuscripts that were not pertinent to the avatar-related themes of “Customization, Personalization, Embodiment, Creation, Generation,” as well as those lacking relevance to our research inquiries, non-English publications, and those with inaccessible abstracts and full texts. Subsequently, we scrutinized the remaining items to ascertain eligibility by performing meticulous readings as deemed necessary. Articles that were not accessible in full text were excluded, alongside those deemed irrelevant or diverging in focus, in addition to those lacking a robust foundation and empirical evidence to substantiate their conclusions. Ultimately, the items retained from the preceding processes were subjected to analysis within this systematic review. In summary, 45 scholarly articles were selected, as illustrated in Table 2, which we considered appropriate for subsequent examination.

Customization of avatars is therefore an essential feature of user experience and interaction in VR, as the study results show. Avatar customization allows users to create digital representations of themselves or how they would like to be, either in real life or in their dreams (Lin and Latoschik, 2022; Wu et al., 2023; Ho et al., 2023). Such an approach not only helps to develop a better feeling of ownership of avatars in virtual reality but also contributes to the improvement of the quality of interactions in VR environments, as well as trust between the users (Lin and Latoschik, 2022; Wu et al., 2023). In addition, the link between online avatar representation and real-life identity also highlights the users’ desire for consistency between the two worlds, the importance of avatar customization in the formation of online identities (Ho et al., 2023).

Situational factors that determine the appearance of avatars in the virtual world are as follows: Firstly, the type of customization that is possible has a large part to play in the whole process. The studies show that people are more inclined to identify with their avatars when the options available for customization are realistic and can be changed in real life, for instance, hairdo or clothing. This identification is further supported by the use of real life factors such as friends or events in the customization process which makes users more attached to their avatars and encourages them to interact with others on the site (Wu et al., 2023).

Secondly, the type of virtual environment has a large influence on the level of customization of avatars. Since the level of public self-consciousness is low in casual and comfortable virtual social environments, users create a persona that is an imaginary one. On the other hand, in the more formal and serious virtual environments, where users feel higher public self-consciousness, they are willing to represent avatars that are closer to their real life. This implies that the level of self-similarity of avatars differs with the virtual environment and the level of public self-consciousness of the users (Alldieck et al., 2018a).

The need to construct a new self and to come out in cyberspace is another reason for this. People tend to design their avatars to have a new body which replaces the real one and enables them to express themselves more openly in the virtual environment (Roth et al., 2019). In addition, avatars in virtual spaces enable users to conceal their real-life identities and thus free themselves from social expectations, which results in more permissive behaviors and higher levels of self-actualization (Shang et al., 2012).

Also, the social context and the goals of interactions in virtual environments influence avatars’ look. For instance, in social VR environments, the users can decide to have avatars that resemble their real-life physical appearance or have avatars that are enhanced in some way. This flexibility of customization improves the level of interest, identification with avatars, and the realism of interactions, which leads to trust and stronger social relations within the virtual environment (Lin and Latoschik, 2022).

The innovations elucidated in this research were systematically derived from an exhaustive review of scholarly articles centered on the customization of avatars within virtual reality environments. The principal objective was to discern and classify technologies predicated on their methodologies for avatar construction, thereby ensuring a lucid and methodical presentation of the most consequential advancements within this domain.

The procedure commenced with a thorough examination of the selected studies, wherein pivotal technologies or methodologies employed in the creation and customization of avatars were identified. These innovations were subsequently categorized into six principal segments predicated on their construction techniques:

Images (Section 3.3.1): Technologies that utilize static imagery for the creation or modification of avatars. Videos (Section 3.3.2): Methods that exploit video data for the generation or enhancement of avatars. Scans (Section 3.3.3): Techniques that incorporate 3D scanning to produce detailed and lifelike avatars. Real-Time Video (Section 3.3.4): Innovations that facilitate avatar customization or interaction in real-time through the utilization of video input. Additional Effects (Section 3.3.5): Instruments or features that augment avatars with special effects, animations, or enhancements. Software Libraries (Section 3.3.6): Frameworks or libraries that furnish developers with the requisite tools for avatar creation and customization.

This classification was meticulously designed to encapsulate the heterogeneity of methodologies in avatar customization while accentuating the most significant advancements in each specific domain. By concentrating on innovations that exhibited discernible technical contributions and practical applications, we ensured that the selected technologies were both pertinent and impactful.

The final curation of 20 scholarly articles (Table 3) was predicated on their capacity to address critical challenges in avatar customization, such as enhancing realism, increasing user engagement, and facilitating greater personalization. Each innovation underwent thorough analysis, with its distinctive contributions and potential applications articulated in the pertinent sections of the research paper.

This methodical approach enabled us to furnish a comprehensive overview of the current state-of-the-art in avatar customization technologies, while simultaneously preserving a clear and logical arrangement of the findings.

As illustrated in (Figure 2), our review and analysis of the collected scientific research revealed that the majority of applications for avatar customization are concentrated in three primary social contexts: education (9 works), mental health (7 works), and communication (6 works).

In the course of the first analysis of the research, these three spheres were distinguished as the most extensive areas where avatar customization is employed. Although, we did not set out to specifically target these contexts in our analysis, the findings from the reviewed studies pointed towards their relevance. Those that did not help in the advancement of virtual reality or were not insightful about these key social contexts were omitted; those with little room for future investment. This process helped in the elimination of any study that was not considered to be very relevant and important. After this, a full-text screening was done in order to make a more selective choice of the articles. This led to the elimination of articles that were related to non-central or non-social uses of avatar customization to minimize the study on areas that are less relevant. In this way, 23 works were selected for the final analysis, which were grouped according to their relevance to education, mental health, and communication.

This work establishes that the determinants of avatars’ appearance in virtual worlds are customizability, types of virtual environments, social relations and user requirements for representation and anonymity. All these factors define the level of realism of avatars and real or imaginary selves of the users, interaction and reality of virtual environments. The analysis in this paper reveals that the options that are present in virtual environments play a role in the manner in which users perceive and interact with avatars. For instance, the look of avatars influences physical outcome and perceived effort, muscular avatars reduce perceived effort and improve strength in tasks (Kocur et al., 2020). The personalized avatars are also better in increasing the body ownership, presence, and emotions than the general ones, and hence, the need to establish high fidelity and personalized avatar creation processes (Waltemate et al., 2018). Another critical aspect is the kind of virtual environment that is being used. In social VR environments, motion-controlled avatars with full body representation improves presence and co-presence, it shows that avatar completeness is significant for social presence (Heidicker et al., 2017). One of the emerging trends was the role of social relationships and the nature of the interaction in the appearance of avatars. The avatars chosen by users in social VR environments are similar to the real life appearance of the users or an even better version of the users and thus, the user is more immersed, trusting, and building better relationships with other users in the virtual environment (Radiah et al., 2023).

However, the study has the following limitations: The degree of customization offered by the various virtual platforms may also be a limitation because users’ experiences may be influenced by the platform being used. Moreover, the research is mainly conducted in social contexts of VR; more investigations are required to examine avatar customization in professional or educational environments of VR (Buck et al., 2023). Future research should focus on the effect of various virtual environments on avatar customization in professional and educational contexts. Longitudinal studies should also be conducted to examine the impact of avatar customization on the usage and relationships within virtual environments. One of the other research directions is the examination of cultural differences in preferences regarding avatars and their influence on virtual communication (Thomas et al., 2017). In the research, the authors demonstrate the significance of avatar customization on the level of interest and actuality of users and interactions. Choices and preferences, virtual environment, social interactions, and the need for uniqueness affect avatar look. These considerations indicate the necessity of providing a wide range of diverse and flexible options for customization of virtual platforms to support better user experiences and interactions in virtual worlds (Reinhard et al., 2019). Therefore, avatars in virtual worlds are determined by the choice of the appearance, characteristics of the virtual environment, interaction with other avatars, and the need for self-identification and anonymity. Knowledge of these factors is highly essential in developing virtual environments that will increase user participation and promote real-life interactions.

The advancement in technology has seen many improvements in the area of research and development of avatars and their customization. All these innovations have contributed significantly to improving the precision, realism, and individuality of 3D avatars. Some of them are single image 3D reconstruction, hybrid representation techniques, face swapping techniques, personalized 3D head model, high fidelity facial reconstruction and building avatars from videos and scans. Furthermore, the current trend in creating avatars is the AI-generated avatars and not the standard options where one is limited to choices or has to select the features manually. Some of the recent developments that have been made in single-image 3D reconstruction include the models that were made by Zollhöfer et al. (2011). Hu et al. (2017) and from the work of other authors as well. emphasize the combination of the dense semantic representation and volumetric feature transformation to increase the surface details and geometry from a small set of input data. Also, there are more complex representation methods that are half way between the previous types, such as those described by Ichim et al. (2015) and Pan et al. (2023) which combine neural fields with skinned meshes, thus enabling fine-grained local editing and cartoon-like avatars’ customization. One of the emergent positive effects is the improvement of the techniques of high-fidelity facial reconstruction from minimal data inputs, for instance, the one-shot 3D facial avatar reconstruction by Yu et al. (2023). This innovation gives practical use in the fields of VR and AR, which greatly expands the use of realistic avatars in different fields. These are some of the most important technologies, and as such, they have their drawbacks as well. For instance, the use of input data of high quality and computational power can be a limitation as highlighted by Garrido et al. (Yu et al., 2023) and Liao et al. (2023).

These studies stress the importance of developing strong data acquisition and processing methodologies to achieve realistic and accurate avatars in the reconstructed scenes. Using AI for avatars’ generation has become more frequent than the methods that involve using presets or certain settings for customization. This change can be attributed to the use of AI and speed that is offered by the same as well as the shift towards more realistic avatars. GANs stand out as the most used technology in avatar customization due to their versatility, ability to generate realistic data, and the continuous improvements in their architecture, all of which contribute to creating more dynamic and lifelike avatars.

The future work should be dedicated to the improvement of the algorithms that would allow creating high-quality avatars based on low-quality or even partial input data. Also, the further development of more convenient and affordable scanning technologies may open the possibility to create personalized avatars for everyone, as Bai et al. pointed out Bai et al. (2022) and Zhang L. et al. (2022) pointed out. These innovations in avatar customization have greatly enhanced the general possibilities of constructing avatars in 3D in terms of realism and individuality. The use of AI instead of conventional techniques is a sign of the growing demand for more effectiveness and realism of avatars. These technologies are versatile and can be applied in many fields including but not limited to gaming, entertainment, business meetings, and virtual shopping. Future studies should build upon current gaps and extend the research findings to new directions to improve avatar customization to the user convenience.

Avatar customization plays a critical role in improving the communication, education and healthcare delivery. In communication, avatars facilitate better interaction since the user is able to show nonverbal signs and has a higher social presence. Research indicates that avatars with actual human features and realistic emotions enhance interaction and communication quality in virtual environments (Hart et al., 2021). In education, avatars enhance interaction and learning in virtual classrooms and language learning platforms where students are more active and engaged (Woodworth et al., 2019). In the context of healthcare, avatars enhance patient involvement and change in behavior through the use of personalized and interactive health management (Kim, 2014). In communication, avatars are not only the visual appearance but also self-sufficient communicative actions that improve the interaction experience. In education, avatars are important enablers of learning personalization, which has been discussed in the context of fashion education where avatars improve creative self-confidence and interest (Jang and Kim, 2023). In the healthcare context, avatars assist in the delivery of behaviour change interventions; for instance, the medical avatars for lifestyle changes and management of chronic illnesses (Anam et al., 2016).

A surprising result is the large effect of avatar customization on user attachment and communication in telehealth. The more avatars physically resemble users, the more attached they become, and the better they communicate with friends and family, which is the psychological value of custom avatars, (Rice et al., 2018). Research on avatar customization to date tends to be limited in scope, with few longitudinal studies of psychological impact, or consideration of the effects across diverse populations. Further, there is a dearth of research on the ethics of avatar use in particularly vulnerable settings like healthcare and education (Hart et al., 2021). Further work needs to address the long-term psychology of avatar customization and its role in increasing user engagement and behavior change for health and education. Further research is required to understand the ethical concerns and possible biases that may arise from avatar customization in different social settings (Kang and Kim, 2020) Avatar customization is an effective strategy in communication, education, and healthcare, to increase user motivation, interaction quality, and behavioral change. Further work may wish to pursue these applications to refine avatar utilization and mitigate possible ethical issues.