What’s new about Web AR in 2022
A very-personal brief technology state of the art for Web AR in 2022, mostly for Open Source technologies.
This non-exhaustive, very personal list is about the current technology and services options available to develop Web AR applications in 2022.
Before go on reading, I assume the reader knows what is Augmented Reality (AR), Web AR and have already give a brief look about the most used libraries in this topic, like AR.js or 8th Wall. If not, I suggest to go reading stories like these:
As we have reached 2022, most of latest year’s trending Web AR technologies are still here to stay. 8th Wall continues its journey on being the top choice, as for features and for performances, with its target being professional Web AR applications; still, its prices are still unaffordable for tiny companies or privates. Another continuing trend is the growing of Web AR Authoring platforms; such platforms offers — as 8th Wall does — AR tracking, SDKs and most of the time a WYSIWYG editor to develop and publish AR applications without writing code. Still, we are still talking about paid technologies and services that keep user’s data and code hosted on third-party servers.
On the Open Source side, WebXR device API is growing but in my opinion, without Apple support on iOS devices, and no possible polyfilling, its wide adoption is still blocked. Anyway, WebXR is becoming better and better but as for now, I do not see it as a real option for professional applications, but more of an experimental choice.
Google’s Model-Viewer is still my top choice for world tracking web AR (markerless) in Open Source (and in my knowledge, the only markerless option truly free of charge). I still found issues about performances on Android when “Scene viewer” options kicks in, with random crashes even on “recent” smartphones (like Huawei P20), but in general the experience is good and the support is great. Model Viewer has now a brand new editor for 3D scenes, supports annotations on 3D models and features a great documentation with lots of tutorials and examples.
As AR.js is still the only Open Source technologies with Location Based Web AR, a new, exciting Open Source library has entered the game. And my focus on this very article will be about it: it is called mindAR and it features Face Tracking Web AR but most of all, a machine learning based Image Tracking AR. The author is HiuKim Yuen (https://github.com/hiukim).
mindAR has a fully based Javascript SDK, and Three.js as rendering engine for 3D content (it is also possible to use A-FRAME). I give it a look and the Image Tracking system is very solid and reliable. It is possible to interact directly with content, to customise the User Interface, to play classic video and also transparent videos, to play audio and more.
The great thing about this new library is that it relies on well maintained and recent libraries, so every aspects of it is potentially enhanceable. So, I’d like to sponsor this library as I think it is a great alternative and competitor to other Web AR paid technologies, and it totally worths a try.
Currently, these are my top choices for free, Open Source Web AR libraries divided per “AR type” (aka tracking type):
- World Tracking (markerless): Model Viewer
- Marker Tracking (it still makes sense to have markers if you have to support lots of targets, like more than 30 or even hundreds): AR.js
- Location Based: AR.js
- Image Tracking: mindAR
- Face Tracking: mindAR
Hope this helps anyone who is struggling to find out what’s the best choice to develop a Web AR app, as of 2022, without going into paid services (well, you might have to, but I guess you better try every choice before that, don’t you?)
