A 3d content library for universities should be owned, not rented

Renting access or owning a library

When a university buys content in the usual way, it buys a subscription: a number of seats, a catalogue chosen by the supplier and a renewal date. That is easy to start. It also means the university's teaching depends on content that sits on somebody else's platform, in somebody else's format, under terms the university did not write.

The alternative is to treat 3D teaching content as the university treats a textbook its staff wrote or a laboratory it built: as an academic asset. Someone in the owning college checked it, someone is named as responsible for it, and it can be used, changed and kept without asking permission.

This is not a claim that a global XR content library sold per seat is wrong for every institution. It can be a sensible way to start. The question here is narrower: what should a university be able to say it holds once the contract is signed, and what happens to its teaching on the day that contract ends? For a 3d content library for universities, that question matters more than the size of the catalogue.

What ownership concretely means

“Ownership” is easy to claim and hard to test, so we reduce it to four checks. If one fails, the university is renting.

• A declared licence. Every model, scene and 360° capture states the licence it is held under, so staff can see for each asset what they may do with it.
• A named owner. Every asset has a person or college accountable for it. A lecturer knows who to ask about an error, and a dean knows who approved it.
• An open format in the university's storage. Models are stored as glTF, an open and widely supported format for 3D, in the university's own storage. They are readable and exportable without the supplier's software.
• Survival of the contract. The library still works the day any supplier contract ends, including a contract with us.

Why declare the licence per asset? Because a library has mixed origins. A model a student built, a scene a lecturer captured on campus, a draft generated by AI and a model commissioned from a studio do not share one set of rights. Stating the licence beside each asset, and the owner who accepted it, is what lets a university reuse the content in a new course, hand it to another college or export it, with no search through old agreements.

The fourth check decides the rest. Format and licence matter because they decide whether the library outlives the relationship that produced it.

Coverage, not asset count

Suppliers often advertise a total, sometimes in the millions. It is an impressive number and a poor measure of what a university needs, because a lecturer teaches a module, and a module needs one specific turbofan, ward or hangar.

MIHWAR's planning target is about 660 assets, intended to cover every practical module across eleven colleges. Read that as a design target, not a delivered count. We chose it because it is a number a university can reach, audit and own.

The useful measure for a 3d content library for universities is the count of assets that pass three tests:

1. The asset is mapped to one of the university's own course modules.
2. It carries a licence the university can defend.
3. An academic in the owning college has checked it.

This also changes how a dean reads progress. “How many assets do we have?” becomes “which modules are covered, and by whom were they checked?” The second question can be answered module by module, college by college, and it points directly at the gaps still worth filling.

An asset that fails any of the three is shelf space, not coverage. A library of 660 that passes all three will serve teaching better than a catalogue of millions that passes none.

Four ways to fill it, one gate to pass through

An owned library has to be built. MIHWAR uses four production routes: student coursework, campus capture, AI-assisted generation and commissioned modelling. All four end at the same academic review gate. A student uploads a model, an academic reviews it, and only then does it become an asset the whole university teaches from.

That is why the library grows by coursework rather than by purchase order. Because students and faculty build the library, the design intent is that cost per module falls each year instead of recurring.

Three further choices keep the content useful:

• Places, not just objects. A lecturer captures a ward, a hangar or a dig site on a consumer 360° camera, drops numbered teaching hotspots onto it and stands library models inside it. It is the cheapest content a university can make, in a format it already has most of. See our post on 360° captures and teaching hotspots.
• One scene, four ways out. The same experience runs in a browser with no install, in a headset over WebXR, on a phone in AR from the same URL, and inside Unity or Unreal as glTF plus a scene manifest. Nothing built is trapped. We explain the routes in one scene, four ways.
• AI as a drafting route, never a shortcut. You describe what you need to teach, a draft model comes back in under a minute, and then a human decides whether it is fit for students. The review gate is the part that matters.

MIHWAR: the prototype of this approach

MIHWAR is a working prototype built by Nplay for Interactive Solutions, Amman, and prepared for a Jordanian university. Its tagline is “The 3D and XR teaching platform the university owns.” Eleven colleges share one core, and the library is made of interactive 3D models and 360° captures that the university's own students and faculty build, review and keep.

It is not a mock-up or a video. The models are real geometry rendered in the browser, the level builder composes scenes, the review queue publishes and the export writes files. Open the turbofan, run the animation, and fan, compressor and turbine turn on one shaft. A guided tutorial of 22 stops, about six minutes, walks through three roles: student, instructor and university.

The limits are as important. The 3D models are indicative rather than teaching-grade, the 360° captures are painted rather than photographed, and four stages of the ingest pipeline are represented rather than executed. In AI Studio the geometry is synthesised in the browser from the words typed, as a stand-in for a provider call; what is real is the review gate, the licence declaration and the flow. Nothing needs an install, plugin or network connection to try.

One idea in it needs a warning label. A student marketplace, where students might earn from models they contribute, is a proposal for the university's legal office only. It touches intellectual property, student contracts and tax, and we expect it to change in review. No earnings are promised.

MIHWAR is a prototype, not a deployment, and we claim no contract, users or results. For the wider thinking behind interactive models, see why a model you can pull apart beats a diagram.