Creating immersive content using generative AI is a likely future. In this paper, we take a sneak peek into that future. We collaborated with sixteen users to create their memories with them. Different methods, including three generative AI pipelines and Gaussian Splatting, were used to recreate twelve personal objects. Twelve participants were instructed to select one version and subjectively assess each 3D reconstruction in four dimensions: realism, aesthetics, resemblance, and reminiscence. Our results show that Gaussian Splatting is generally preferred for its fidelity, while image-to-3D pipelines are competitive on aesthetic appeal. Crucially, reminiscence, how well a reconstruction evokes the original memory, emerged as the strongest driver of preference, suggesting that for this group emotional resonance matters more than visual realism.

Improving user experience during the delivery of immersive content is crucial for its success for both the content creators and audience. Creators can express themselves better with multisensory stimulation, while the audience can experience a higher level of involvement. The rapid development of mulsemedia devices provides better access for stimuli such as olfaction and haptics. Nevertheless, due to the required manual annotation process of adding mulsemedia effects, the amount of content available with sensorial effects is still limited. This work introduces an innovative mulsemedia-enhancement solution capable of automatically generating olfactory and haptic content based on 360 video content, with the use of neural networks. Two parallel neural networks are responsible for automatically adding scents to 360 videos: a scene detection network (responsible for static, global content) and an action detection network (responsible for dynamic, local content). A 360 video dataset with scent labels is also created and used for evaluating the robustness of the proposed solution. The solution achieves a 69.19% olfactory accuracy and 72.26% haptics accuracy during evaluation using two different datasets.

HiruXR is a Javascript library aimed at creating collaborative and interactive data visualizations in eXtended Reality (XR). Developers can use it to create environments where multiple users can communicate and collaborate around one or many visualizations. We want to open these environments to users with 2D displaying and interaction capabilities, e.g. allowing them to collaborate from their laptops and phones with others using Head Mounted Displays (HMD). But displaying the same interface is not optimal, users should see an interface according to their device capabilities. Therefore, we are steering the design of HiruXR towards supporting responsiveness without putting all the burden in application developers. The library defines visualization, interaction and collaboration components that adapt to the user device capabilities, i.e. 2D or XR. In this paper, we share the design philosophy, initial implementation examples and lessons learned so far building some of these components for collaboration between VR and 2D devices.

In recent years, artists and communities have expressed the desire to work with tools that facilitate co-creation and allow distributed community performances. These performances can be spread over several physical stages, connecting them on real-time towards a single experience with the audience distributed along them. This enables a wider remote audience consuming the performance through their own devices, and even grants the participation of remote users in the show. In this paper we introduce the Co-creation Stage, a web-based tool that allows managing heterogeneous content sources, with a particular focus on live and on-demand media, across several distributed devices. The Co-creation Stage is part of the toolset developed in the Traction H2020 project which enables community performing art shows, where professional artists and non-professional participants perform together from different stages and locations. Here we present the design process, the architecture and the main functionalities of the tool as well as the results of the first user evaluation with Opera houses and artists.

The burgeoning capabilities of Web browsers to exploit full-featured devices can turn the huge pool of social connected users into a powerful network of processing assets. HTML5 and JavaScript stacks support the deployment of social client-side processing infrastructure, while WebGL and WebCL fill the gap to gain full GPU and multi-CPU performance. Mobile Grid and Mobile Cloud Computing solutions leverage smart devices to relieve the processing tasks to be performed by the service infrastructure. Motivated to gain cost-efficiency, a social network service provider can outsource the video analysis to elements of a mobile grid as an infrastructure to complement an elastic cloud service. As long as users access to videos, batch image analysis tasks are dispatched from the server, executed in the background of the client-side hardware, and finally, results are consolidated by the server. This paper presents SaW (Social at Work) to provide a pure Web-based solution as a mobile grid to complement a cloud media service for image analysis on videos.