Babylon.js Template Repositories

NOTE: This documentation and the features it describes are under active development.

Quick Start Guide

  1. Follow the Quick Start Guide for the NPM Package Template repository to create your new code repository your-new-repo.
  2. Follow the Quick Start Guide for the Asset Host Template repository to create your new asset host repository your-asset-repo.
  3. Develop your experience in your two repos, adding code and business logic to your-new-repo and assets (images, 3D models, sounds, etc.) to your-asset-host. To maximize flexibility, limit your development in these repositories to only that which is required to create and test your 3D experience itself.
  4. Export your 3D experience from your-new-repo's app_package as an NPM package in order to integrate it with the experience in which it will ship.

For a detailed example showing how to use the template repositories to create a real experience, see this tutorial.

The Template Repository Workflow

The primary purpose of the template repositories is to suggest and bootstrap a workflow for taking Babylon.js 3D experiences from Exploration to Circulation. Concretely, these repositories provide an answer to the question, "How do I take my idea that I explored in the Playground, or somewhere similar, and mature it into production software that I can ship in my product?" There are, of course, innumerable viable answers to this question. Our Template Repository Workflow simply represents a "golden path" option that we believe will be well-suited to many common scenarios.

The primary goal of the Template Repository Workflow is to achieve flexibility by promoting the separation of concerns. The Workflow recognizes three primary components to a Babylon-powered experience: (1) the Babylon experience itself, (2) the assets used in the Babylon experience, and (3) the encompassing "shipping vehicle" experience of which the Babylon piece is one part.

  • An ecommerce site might have (1) a 3D product viewing experience, (2) a library of 3D models of products, and (3) the ecommerce Website itself.
  • An indie game studio might have (1) an individual game, (2) the assets for that game, and (3) the Website, PWAs, and native applications in which that game will be shipped.
  • An in-house visualizations team might have (1) a specific 3D visualization, (2) proprietary datasets to be visualized, and (3) in-house dashboards and/or distribution mechanisms for sharing the visualization throughout the company.

The Workflow provides a pattern by which all three of these components can be developed more-or-less independently -- and even interchangeably -- without sacrificing the efficiency of the development experience. For an illustration of this, let's consider a canonical example of a hypothetical independent developer using the Babylon.js Template Repository Workflow.

Canonical Studios: the Workflow in (Hypothetical) Action

Consider a hypothetical mid-sized 3D development house called Canonical Studios. This company's business is small-scale 3D experiences, including contract work (marketing, etc.) for other companies as well as their own independent games. Their contract work must ship in a wide variety of scenarios including standalone apps (native and Web) as well as pre-existing Websites operated by major brands. Their indie games also ship in a variety of circumstances, and often the same game will ship in more than one shipping vehicle.

Canonical Studios employs about 20 developers (spread out across disciplines including code, art, design, and so on) organized into four teams: three "3D experiences" teams and one "frontend/integrations" team. At any given time, the "3D experiences" teams may each be working on separate 3D experiences, largely independent of each other and independent of the frontend team does to deliver those experiences to customers. Thus, at any given time, Canonical Studios may have up to four active workstreams in flight: three under-development 3D experiences plus the integration work to deliver the most recently finished 3D experience to customers.

Because Canonical Studios leverages the Babylon.js Template Repository Workflow as part of its operations, all four of these teams are able to pursue these simultaneous workstreams highly independently thanks to the separation of concerns. Each new 3D experience begins from the first two steps of the Quick Start guide at the top of this page; thus, every individual 3D experience is its own Git repository and, eventually, NPM package ready to be integrated in any shipping vehicle capable of consuming NPM packages. Asset hosting, kept isolated from experience development from the start, will not add bloat to either the Git repository or the NPM package and can be solved in various ways as appropriate to the scenario (or even in combinations of different ways at different times, such as simple individual hosting during development adapting to asset bundles for shipping). Finally, once a 3D experience is ready to be integrated into its shipping vehicle, the frontend team merely has to integrate the experience's NPM package and inform that package of how to get the assets based on the hosting solution chosen for shipping.

So, for a contract product -- for example, let's say Canonical has been contracted to add a 3D bee character that animates around the homepage of an apiary -- the first task belongs to a 3D experiences team, which creates new repositories in which to develop the bee experience. They populate the NPM Package Template's test_package with a dummy version of the customer's homepage, then create their experience to add the animated bee to that, exposing a minimal contract from app_package to test_package. When the experience is satisfactory, it is exported as an NPM package and handed off, along with bee asset, to the integration/frontend team. The 3D experiences team can then move on to their next product (though presumably remaining available to help if needed) as the only thing the frontend/integration team should have to do is decide where on the customer's system to host the bee asset, expose the NPM package, then integrate it into the customer's site as they would any other NPM package.

The same general workflow applies to shipping an indie game, but the variety of shipping vehicles further exercises the flexibility of the approach. Again, the 3D experiences team's final product is an experience NPM package and a library of assets handed off to the frontend/integration team. However, because this experience has multiple shipping vehicles, the integration/frontend work is a bit more varied. For each shipping vehicle (Web, native hybrid app, PWA, etc.), the details of where/how assets can be hosted/accessed may vary significantly; however, since the concerns of asset hosting and shipping vehicle have been kept separate from the 3D experience development from the beginning, it is possible for all shipping vehicles to ship the exact same 3D experience NPM package, with only the infrastructure surrounding the 3D experience differing from platform to platform.


Though the Template Repository Workflow is, we believe, well-suited to many common scenarios, there are certain scenarios where this pattern will do more harm than good. Because of the Workflow's emphasis on separation of concerns, it is unlikely to be a good alternative when the concerns are not actually separate. In particular, the Workflow's primary conceit is that the Babylon experience being developed is largely isolated from the logic of the shipping vehicle that contains it; this is viable as long as the logic boundary can be condensed to a "contract" through which the shipping vehicle and the 3D experience can communicate. In some cases, however, it may be desirable for the 3D experience to be intentionally tightly coupled to the experience in which it resides, which may eventually broaden the "contract" to the point where concerns aren't really separate, at which point the appearance of separating concerns may add nothing more than inconvenience.

Another potential disadvantage of the Template Repository Workflow is that, out of the gate, it may not facilitate the use of certain tools. The Babylon.js Community-driven Editor, for example, has its own conventions and practices which may not allow both workflows to be used effeciently on the same project. Similarly, the Workflow's unrestrained embrace of NPM may make it more difficult to adapt to situations where using NPM is not an option, though certain overarching themes and patterns may still be transferrable.