The Great Library of Tomorrow

Developed for "The Great Library of Tomorrow" exhibition in Barcelona, this interactive installation acted as a narrative bridge for visitors entering a large-scale VR experience. The objective was to invite the audience into a participatory ritual, utilizing AI to merge the faces of attendees into a single "Collective Portrait." This living character acted as a digital manifestation of the exhibition's community, born from the real-time presence of the public.

The system employed Generative Adversarial Networks (GAN) to move beyond simple image overlays and into true digital synthesis. To achieve a seamless blend, the system computed the vector representation of each visitor's face within the GAN’s latent space. By calculating the mean of these vectors while specifically maintaining a dynamic buffer of the last participants the installation ensured that every new visitor’s contribution was immediately visible, causing the collective visage to subtly "evolve" with every interaction.

The installation required a robust, multi-platform infrastructure to handle continuous public use over a three-month period. A fleet of iPads served as the capture frontend, feeding data to a centralized in-house server. A custom Python-based pipeline handled the latent space computations and generated real-time transition animations. These were then streamed via NDI protocol to the exhibition's primary media server, while OSC messaging was utilized to synchronize the AI output with localized particle effects and scenographic animations. The result was a high-performance, low-latency loop that integrated individual identity into a massive, collective digital narrative.