I digged today in Amazon Sumerian. Nice platform, similar to PlayCanvas. They allow AR output to 8thWall (interesting solution but not acceptable for their pricing policy and poor WebXR tracking that is not comparable to native ArKit and ArCore anchors).
But they have 2 ready to developer applications for ArKit and ArCore that fill the gap from VR PlayCanvas world and AR world.
I think you should fill this gap too if you want to do a giant leap towards AR. It’s hard for us, not for you.
We’ve just finished a Shopify site with a nice configurator app developed in NextJS + PlayCanvas for VR experience, but for AR we started with ViroMedia and then switched to native ArKit and ArCore apps. Very expensive!
For next jobs we are thinking to start with Amazon Sumerian for that reason: 1 code for VR and AR. It should be nice if we have the opportunity to develop in PC too, not with another platform.
Another issue we found in Sumerian is that they only run on Chrome, not Safari and not mobile too. This is another opportunity for you to match the requests of school that want a javascript universal plaform for their iPad.
Last: neither you and Sumerian have an app for prototype AR apps on mobile like Expo.io or ViroMedia. If you do that, I will suggest PC for lot of Italian schools (and you for the rest of the World).
Nice could be more starter projects to avoid boilerplate, anche nicer to have a basic assets store.
An update: I just finished trying their app and looking at the code I found that they put a WebView above the AR scene: the result is bad. The WebView receives the transformations of the native anchor and passes it to that in the webview, but the delay due to Javascript and the event communication system highlights a significant discrepancy between the 2 anchors. A serious solution remains that of ViroMedia because it does not use WebView (at least I think but I’m practically sure). Sumerian cannot be used for serious AR applications.
As already mentioned, the 8th wall team $ 1000 a month (at a minimum) is not within the reach of all our customers. In addition, the pay per view is very risky if the app is successful. And in any case, the tracking speed of ArKit and ArCore is much higher and more faithful than web-based solutions, so 8th wall is a very expensive compromise and we absolutely exclude it. Maybe in the future if their economic policies adapt to our needs and if WebXR technology should improve, we do not rule out taking it into consideration. Unfortunately after a later think we don’t believe that none of these solutions (Sumerian, Playcanvas) will be comparable to a native solution. The only one that we have tested and responds very well to our needs is ViroMedia because it maps each UI element with the native part. Unity and Unreal are also better. If there is a WebView between the AR and the JS, it will never achieve the performance of the native version. You need a Javascript / native conversion library.
There has been some work on Unity by Mozilla to create a WebXR application via Unity which may be more suitable for your use case of one codebase for both a native and web app.