n'Space Immersive Projection System

TL;DR: n'Space turns ordinary rooms into immersive 3D interactive spaces without a headset. As the team's only Technical Artist, I built the content team from the ground up to six people, addressed rendering consistency and performance across five projection surfaces, and turned repeated prototype explorations into a reusable pipeline. Over three years, we delivered 15+ commercial and exhibition projects.

Table of Contents

• At a Glance
• Building and Leading the Content Team
• Core Technical Challenge: Five-Surface Projection
• Content Pipeline: From Prototypes to Reusable Systems
• Outsourcing Standards and Art Direction
• Hardware and Software Integration
• Delivery: 15+ Projects Across Multiple Domains
• Reflections

At a Glance

My role	Content Group Lead and the team's only Technical Artist
Period	2017–2020
Team	Grew from 2 to 6 people, supported by 2 outsourcing teams
Reported to	CTO
Collaborators	Hardware, central control systems, marketing, and clients
Tools	Unreal Engine 4, Houdini
Key results	15+ delivered projects · 6-person team · 60 FPS · Standardized workflows · Shared asset library

---

1. Building and Leading the Content Team

Starting point: The team was based in Salt Lake City, and the original content team members came from my graduate game studio at the University of Utah. The product was in early R&D, with no established content workflow.

What I did:

• Hiring: We started with myself and one 3D environment designer. As the team's needs developed, I recruited programmers, environment designers, and game interaction designers from my graduate game studio. The hiring process included an HR phone screen, portfolio reviews by the CTO and me, coding interviews for programmers, and an interview with the CEO.
• Production workflow: Our graduate game studio had already used an established game production workflow, and every team member had worked within it. Bringing that shared experience into the company helped the team find its rhythm quickly. Each project moved through brainstorming → direction setting → design documentation → prototyping → testing and iteration on the hardware → feature completion → art optimization → final delivery. I refined each stage to fit our production needs. Over time, we built a standardized library of reusable assets and developed design guidelines from repeated hardware testing, substantially accelerating subsequent projects. As delivery became faster, environment art production became the bottleneck. We brought in environment art vendors to work to our standards, while the internal team focused on exploring new interaction patterns and visual approaches.
• Prioritization: Clients came first. When several projects needed attention, I balanced the urgency of client requirements, project complexity, and staffing to set priorities in line with the company's overall goals and make the most of our small team.
• A difficult decision: Because we had all come from the same graduate game studio, having to stop our first full game project after the client ended the engagement was disappointing for everyone. I explained the decision in a team meeting, then spoke with each member individually to hear their concerns. The team soon understood the situation. After discussing our options, we retained the game's content and later adapted it for delivery in another project, giving the work a useful outcome that the team was pleased with.

Result: We grew from early internal prototyping into a six-person production team supported by a repeatable outsourcing pipeline.

---

2. Core Technical Challenge: Five-Surface Projection

Test environment: All major tests took place in our physical cave lab: a cube-shaped projection room with four walls and a floor coated in specialized projection paint, and a projector mounted at the top center.

Why it was challenging: Building content for n'Space shared many challenges with VR, but required deeper adaptation to custom hardware.

• Five surfaces, multiple virtual cameras: Content was projected onto five physical surfaces from different directions. Each virtual camera's output had to remain visually consistent, without discontinuities at the seams.
• Screen-space effects did not translate: Many standard screen-space effects did not work in this format because each camera produced a different screen-space result, creating discontinuities between surfaces.
• Performance was a hard constraint: Stable, high frame rates were essential for immersion and user comfort, while multiple views had to be rendered simultaneously.

Approach:

• Camera setup: We rendered with five custom cameras. Their settings were identical except for orientation and resolution. The cameras moved with the user inside the scene's virtual cave, adjusting the viewpoint to keep scene perspective consistent with the user's viewing position.
• Effects and post-processing: All screen-space effects were disabled. Post-processing used camera-position-based calculations, and other effects were also changed from camera-angle-based to camera-position-based calculations.
• Performance: The initial target was 30+ FPS, later reaching 60 FPS, with 4K resolution per wall. Optimization covered scene logic, art assets and materials, lighting calculations, hardware upgrades, distributed computing, and more.

---

3. Content Pipeline: From Prototypes to Reusable Systems

Early Exploration

With the product still in early R&D, we began with rapid prototypes focused on abstract spatial interactions, then gradually moved into fully realized 3D environments. I was responsible for shaders, VFX, and many interaction and gameplay systems across multiple projects.

Houdini Procedural Tools

Before Unreal's PCG framework was available, we built procedural mesh generation tools in Houdini and integrated .hda assets directly into UE4.

The Houdini tools primarily supported the generation of various real-world piping systems, cloth animation, and complex VFX before Niagara was available.

Asset Library and Base Plugin

Problem: As the number of projects grew, many assets, interaction patterns, and technical components were repeated across projects.

Approach:

• Generalized frequently reused content into documented, data-driven assets, including mesh props, master materials, Niagara systems, control Blueprints, Houdini tools, and reusable interaction components.
• Extracted core n'Space functionality into a base plugin. Adding the plugin to a new project provided n'Space integration, including central control communication, heartbeats, and input handling.
• Organized standardized art assets in a separate library project with a consistent folder structure, so the relevant assets could be migrated into new projects as needed.

Result: New-project setup time dropped from one day to half an hour.

---

4. Outsourcing Standards and Art Direction

We introduced outsourcing after receiving our first commercial order. Client projects required a higher and more consistent quality bar than internal R&D, making vendor production an important part of the pipeline.

Through vendor collaboration, we gradually established formal art and interaction guidelines. Alongside the technical constraints described above, each project's level design had to account for the physical structure of the cave showroom. For example:

• Sense of space: Although content was projected onto fixed room surfaces, the environment needed to feel spatially extended rather than flat.
• Height limits: With projection coming from above, interactive objects had to stay below a certain height so they would not extend above the viewer and outside the rendered area.
• Points of interest: Points of interest had to be arranged around the central cave area.

Documenting these and other constraints clearly was essential to making vendor output predictable and ready for delivery.

---

5. Hardware and Software Integration

n'Space was a complete immersive system spanning real-time content, the cave structure, projection hardware, and central control software. As Content Group Lead, I worked closely with the hardware and systems teams to ensure that content ran reliably within the complete product.

• Delivery format: Final projects were delivered as packaged executables. We invested substantial effort in the runtime stability and performance of packaged UE4 builds.
• Monitoring: The central control system launched the content applications and exchanged heartbeat signals with the main control unit to monitor application status.
• Input expansion: Gesture and mobile control modules were added later. As system complexity increased, reliability across team boundaries became an important production goal.

Result: By the time I left Archifiction, the product had a more stable monitoring system and an extensible foundation for new types of immersive projects.

---

6. Delivery: 15+ Projects Across Multiple Domains

Domain	Representative projects	Technical focus
Architectural visualization / Real estate	Virtual apartments and showrooms	Day/night cycles, lighting scenarios, material swaps, furniture set replacement, and skybox selection; mobile control panels added in response to client feedback
Nature and extraordinary environments	A seasonal garden, a volcano with flowing lava, a tropical lake, a lost desert city, a glowing underwater cave surrounded by fish and sharks, and a space station orbiting Earth	Environmental realism, specialized shaders such as underwater and lava effects, a sense of expansive distance, and appropriate lighting intensity
Entertainment	Virtual anime concerts, a 360 video player, virtual KTV rooms, and dance-party experiences	Continuity in virtual character animation and smooth interactions
Games	A shopping-mall interactive game installation enjoyed by guests across age groups	Game mechanics with immediate feedback and engaging interactions designed for the custom hardware

---

Reflections

• Start with how the client will use the product. Content has to prove its value in the client's actual use case. Adding a mobile control panel in response to feedback reinforced this for me: technical choices should help clients present and interact with the experience successfully, and feature decisions should follow that goal.
• Let test results guide the solution. Content for custom hardware has to account for projection geometry, viewing position, interaction range, and runtime performance together. I put more weight on testing in the physical cave and used those results to adjust visuals and interaction so the content fit the product in practice.
• Give new ideas a chance to be tested. Projects such as the virtual concert began as experiments with new use cases. Their results revealed broader possibilities for the product. That experience made me more willing to test unfamiliar directions with small prototypes, then use actual feedback to decide whether to invest further.
• Establish standards at the right time. Defining a complete rule set too early can turn untested assumptions into unnecessary process; waiting too long allows repeated work and communication costs to accumulate. I look at whether requirements are stabilizing, problems are recurring, and the team needs a shared way of working.
• Look for optimization opportunities across the whole system. Optimization should extend beyond the models, materials, and effects in a scene to include the rendering approach, coordination between content applications and central control, and how the hardware operates. Locating the bottleneck before choosing a fix helps avoid repeatedly investing in familiar areas with limited returns.
• Move from doing the work to enabling the team. As the team's only TA, I increasingly focused on making experience reusable. Asset libraries, the base plugin, and outsourcing guidelines turned individual knowledge into methods the team could use. This became my most important lesson about leading: solve the immediate problem while helping the team deliver more independently in the future.