Tower Bridge AR

The brief

What if you could walk
around an AR spectacle?

VFX spectacles are one-sided by nature. A video of a character perched on a skyscraper looks dramatic from one fixed camera angle. What if the same illusion worked in real life, from wherever you were standing?

This was an R&D pitch for Oraar Studio: explore world-locked AR at the scale of a real landmark. The question was not just technical. It was about whether branded spectacle could become a genuinely shared, perspective-correct experience in public space.

Tower Bridge was the obvious choice. Iconic, globally recognisable, and meaningful to the studio. Multiple viewing angles were baked into its architecture.

Studio

Oraar Studio R&D

Platform

Snapchat

Role

Concept · Development · Technical Implementation

Type

R&D pitch project. Not a consumer campaign.

Year

2023

Deliverable

World-locked AR lens with tap-to-cycle branded bag interaction.

Process

Scale before spectacle.
Test in the real world.

Scale is a reading problem

The character had to look natural when viewed from ground level, which means it needed to be enormous in real-world units. Early iterations looked toy-like. The fix was not just increasing the scale value but understanding how size reads relative to architectural reference points. The bridge itself became the calibration tool.

World-locking as the idea

Most AR experiences lock to a face or a flat surface. Anchoring to a complex structure from multiple angles required perspective-correct rendering from wherever the user was standing. The multi-angle viewability was not a technical bonus. It was the entire point of the concept. If it only worked from one angle, it was just a video.

On-site testing changed everything

Digital rehearsal showed a clean, stable experience. On-site, real-world variables introduced tracking instability, lighting changes, and device variance that were invisible on a desktop. Added flexibility into the anchor logic after the first location test. The gap between simulated and real was larger than expected.

What this proved

Landmark-scale AR works.
Scaling it is a different problem.

PoC Multi-angle world-locked AR validated at landmark scale

Multi-angle Perspective-correct rendering from multiple viewing positions

Brands engaged Several brands engaged with the concept following the pitch

Design for mid-range from day one

The experience worked well on high-end devices. On lower-end hardware, tracking stability broke down. Mid-range optimisation cannot be a post-launch fix. It needs to be the starting constraint, not an afterthought. Next time, the reference device is not the flagship.

Real-world testing is not optional

What works in controlled testing does not always translate to a public environment. Lighting, crowd movement, device variety, and network conditions all affect the experience in ways that desktop testing cannot replicate. Must build flexibility into execution, not add it after.

Context before content

Casual passers-by do not intuitively understand AR. No context equals confusion. A technically impressive experience that no one knows how to activate has failed at the last metre. These experiences work in controlled environments or with clear audience education and a reason to engage.

Proof of concept is not the same as scalable

Validating an idea and making it ready to deploy at scale are two separate projects. The R&D proved the concept worked. Turning it into a repeatable, device-agnostic, activation-ready experience would require a second build with different priorities from the start.

What if you could walkaround an AR spectacle?

A mannequin on Tower Bridge.Visible from where you stand.

Scale before spectacle.Test in the real world.

Landmark-scale AR works.Scaling it is a different problem.

What if you could walk
around an AR spectacle?

A mannequin on Tower Bridge.
Visible from where you stand.

Scale before spectacle.
Test in the real world.

Landmark-scale AR works.
Scaling it is a different problem.