An overlay platform for virtual tabletop play

Virtual tabletop platforms surged during the pandemic, yet none of the major platforms added voice customisation or physical component integration. As a fifteen-year player, I saw the gap firsthand and designed 4DnD as my Pratt Institute capstone: an overlay platform paired with a physical dice set, built to restore what remote play was losing without replacing the platforms players already use.

Challenge

The problem was not that VTT platforms were bad. Roll20, Foundry and Fantasy Grounds had well-established ecosystems with years of accumulated user assets and session data. Building a competing platform would mean asking users to abandon all of it.

The real challenge was narrower: identify the specific experiential gaps that made remote play feel hollow, and solve only those, on top of whatever platform a group already used.

Competitive audit comparing Roll20, Foundry and Fantasy Grounds across feature categories, with the 3D audio and physical integration columns empty for every platform. — The competitive audit. Two columns are empty across every platform on the market: spatial audio and physical component integration. That gap is the entire product thesis, and the reason an overlay beats a competitor.

What I did

I chose the overlay model deliberately. Auditing the existing platforms confirmed none offered voice or video customisation or physical integration. An overlay adds what is missing without disrupting what works: users rebuild nothing, migrate nothing, and lose no session history.

I ran 10+ user interviews across player types, in-person and virtual players, across multiple game systems. Three themes came back consistently: the loss of physical dice tension, flat audio killing immersion, and no tools for character embodiment. The three personas in the full case study are composites drawn from real interviewees.

I prioritised features from that research rather than from my own assumptions. 3D spatial audio to simulate sitting around a table. AI voice and video filters to support character embodiment, which is a core tabletop behaviour rather than a novelty. Bluetooth physical dice to restore the tactile ritual of rolling. All three map directly to the most-cited pain points.

A hand rolling a die from the physical dice box on a desk, with the result appearing moments later in the virtual tabletop session on the monitor above. — The core idea, demonstrated: roll the physical die, the result lands in the session on screen. This is a visualisation against an older build, not a working prototype. Bluetooth integration was outside the project scope, and I am not going to imply otherwise.

I delivered end to end. A full hi-fi Figma prototype across all flows, plus the physical dice box modelled and built. Bluetooth integration was outside the project scope, and I scoped it out explicitly rather than pretending otherwise.

Exploded view of the 4DnD dice box, showing the lid, dice tray, electronics housing and base as separate components. — The dice box in exploded view. The electronics housing sits under the tray so the rolling surface stays a rolling surface. The tactile ritual was the point, and burying hardware in the lid would have compromised it.

The dice box open, its lid lifted clear, with three translucent red polyhedral dice suspended between lid and base. — The box open. The dice sit in the lid rather than loose in the base, so opening it presents them rather than revealing them, the ritual the whole product is trying to give back.

Close detail of the closed dice box lid, showing the illuminated red ring inset into the surface and a small port on the chamfered edge. — Lid detail. The ring is the only status indicator on the object. The design brief was a piece of tabletop equipment that happens to be connected, not a gadget that happens to hold dice.

Outcome

This was a school project: no real users, no adoption metrics, and I will not dress it up as more. The deliverables were a complete hi-fi prototype, a physical dice model, and a competitive audit grounding every design decision. Faculty and peer feedback was strong. There was no formal usability testing beyond the interview phase.

10+ interviewsEvery prioritised feature traced to a named pain point

Learning and reflection

Insider bias needs active management. Fifteen years of domain fluency helped, but I had to work against assuming my own experience was representative. Interviews surfaced perspectives I would not have predicted, and would not have looked for.

Wireframes deserved more time. Moving to hi-fi too quickly meant catching structural problems late, when they were expensive to fix.

Accessibility was an afterthought, and that is a real failure. Customisation is not accessibility. I did not follow WCAG and did not test with users with specific needs. It is the first gap I would close if this continued.