I’ve been working on a WebAR engine called encantar.js. You can use it in combination with Babylon.js to create Augmented Reality experiences. It’s open source, easy to use, GPU-accelerated and it can be used anywhere (it’s not WebXR; it’s Augmented Reality made from scratch with computer vision).
At the present time the engine features image tracking. There is a Babylon.js demo that you can try with your mobile device or with your PC. I created this account to share it with you, guys!
I currently use AR.js with marker tracking in my project, I wonder how this compares to it and what differences there are? I see it’s “just” image tracking (no marker tracking), presumably that will result in slightly lower tracking quality than AR.js marker tracking.
I also see that in the demo the input video is quite low resolution (320x568), it would be interesting to see how much it affects performance to increase that, to allow the markers to track better slightly further away…
Totally rooting for you with this project! Having basically only AR.js for open source AR tracking isn’t very good, especially considering it’s not in development any more…
I keep a low resolution by default to have it run on all kinds of devices of various audiences, but you can use the API to increase the resolution of the tracking, of the camera, and of the rendered scene. Performance is affected by different factors such as upload times (GPU).
Do you plan to continue updating this project over time? (not asking for you to see the future just short term at least, do you consider it finished and done with or might you continue working on it a bit?).
Just trying to decide whether to spend some time testing porting one of my projects to it from AR.js!
@alemart I’ve purchased your system and it works perfectly in frontal angles but fails in narrow ones, I think that it is related to the angle transform since the markers are perfectly attached to the reference image.
My impression is that the transformation of the tracker to the camera orientation has something wrong and at those narrow angles the camera is less rotated than it should be. This is an effect that can be seen very well in the babylon demos you provide.
Thank you for your feedback. I know the cause of it and I’m already looking into it.
It is not that it is “wrong”, but rather, it is the case that a underlying mathematical model can be enhanced. In a nutshell, the plan is to add a more sophisticated model. The tracking was improved in the latest release (0.4.0), and I intend to do more work on it still.
No, but I’m glad to see that you’re enchanted by my magic!
I’d like to ask everyone who is interested in this project or who is already using it to support it by purchasing a copy or by becoming a sponsor.
If you’re able to tell others about the project, that also helps.
I dream of expanding this work, but having support is essential. If you’re like me and would also like to see the project expand, that’s how you can help.
Purchase your copy from the website in order to support the project, or get a source release and run the build script via npm.
and I’m new here.
Having basically only AR.js for open source AR tracking isn’t very good, especially considering it’s not in development any more…
just short term at least, do you consider it finished and done with or might you continue working on it a bit?).
