World-in-Miniatures (WIMs) are an established technique in virtual reality (VR) that provide an overview of the surrounding virtual environment. While traditionally used to support navigation over larger distances and distant manipulation in large-scale spaces, their potential for comparing alternative versions of a virtual environment has not been exploited. In this paper, we introduce a Multi-WIM paradigm that facilitates the detailed exploration and comparison of urban planning alternatives. Each alternative is visualized in a separate WIM. The WIMs can be arranged side by side (juxtaposed) or stacked. Stacking allows toggling between alternatives or using a swipe tool where on each side one alternative is shown. These three modes (juxtaposed, toggle, swipe) provide flexible ways of exploring differences and similarities between alternatives, with highlighting functionalities that reveal differences of functional units across designs. We evaluated the Multi-WIM paradigm in a single-factor VR user study (N=24), combining a single-user experiment with a multi-user exploration. Participants had to identify differences and estimate distances, areas and heights across urban planning alternatives. Results show that the stacked toggle mode was significantly faster for finding differences while the juxtaposed view better supported a landmark recall task. Furthermore, toggle was the overall preferred technique in the single-user mode. In contrast, in multi-user settings participants favored juxtaposed, as it fostered collaborative discussion and shared sensemaking. The identified design trade-offs between efficiency and collaboration indicate the need to incorporate all three WIM modes, together with seamless transitions among them, in urban planning applications so as to accommodate both loosely and tightly coupled phases of collaboration.

Embodied conversational agents are an established approach for immersively conveying the narratives and information of virtual museums. Recent advances in large language models and text-to-speech systems enable the active guidance of users by agents and raise the question of how such assistants should be visually embodied to engage visitors and promote knowledge transfer. This demo showcases a stylized humanoid guide, a novel animism-based approach with speaking objects, and a combination of both concepts to interactively guide users through a digital replica of the famous Gropius Room at the Bauhaus University in Weimar.