Thursday, 16 March, 2023 UTC


Summary

MUSE Museum AR Experience on a mobile application

Finding the best platform for your WebAR project

Introduction

I have been building Augmented reality apps since 2006, and I have worked with various systems: PC, Mobile, Adobe Flash FlarToolkit, Depth sensors, RGB Sensors, Magicleap, Microsoft Hololens, Google Tango andARCore, and Apple AR Kit. But I have skipped the hype around WebAR in the past few years.
Finally, I had the time and opportunity to get my hands dirty with WebAR creation. These last few weekends were jam-packed with Javascript and various online frameworks: old and new.
I needed to develop a 3D experience coming out of a poster for a huge conference, so there was an image marker, but it had to be stable and high-quality tracking.

Evaluation Factors

My report compares the various available services, their quality, and what they offer. I hope you will find this useful, as I would be happy to read such a report 3 weeks ago…
I decided to evaluate them by the following factors:
  1. Quality of tracking. How actually good the tracking quality is and how efficient it can become.
  2. Quality of documentation. How easy it was to get to a first working prototype, and how well and accurate the setup guide really was.
  3. Quality of support. How easy was it to find answers to my questions, and how fast I received answers by email.
  4. Pricing of system. How much it costs to publish a project, and is it worth it.

Brief of Results

If you wanna know my final answer before reading the rest, then my vote goes to Zappar, while I would recommend thinking twice if you want to use 8th Wall framework. But as with any other such report, there are other reasons that might change your might, so read through.

Tested Frameworks

8th Wall

A well-known library that was lately purchased by Google Niantic. Seen as one of the leaders in web AR solutions and my first go-to solution. Offers an online editing studio (which has quite an extended feature list, requires writing code but provides extensive documentation. This is quite a powerful option.
https://medium.com/media/d7058b73340079b900dc4f0972c37b62/href

MindAR

An open-source javascript library, completely free to use. Nicely documented, although the final working example took me some extra effort. But I managed to create a full working project in less than 3 hours. This is a code-only framework that requires you to host the code.
Image tracking by MindAR

Blippar

The company offers the most affordable and fair solution, with a pay-per-view mechanism. They have an online creator studio, or an SDK which has several plugins. I did not manage to use their Unity package yet.
https://medium.com/media/c859b5a12387d1a83358e5ae118a6329/href

Zappar

A veteran in the market and a player since probably 10 years ago. They started with a mobile app that ran all content, but today they also offer full web-powered solutions. Documentation is super accurate, and their Unity component is by far the best of these options. Zappar still offers to publish through their mobile app, but also as a mobile app of your own (which I do not recommend since you can just use AR Foundation with Unity), or using their web creator studio, and last, using their Unity plugin.
https://medium.com/media/1563135037c195416318012e533b88e2/href

Comparison and Scoring

Compare quality of tracking
Quality of tracking.
8th Wall: 8 of 10. Capable of combining marker tracking with world tracking, which is completely innovative. But still not perfect by the distance of detecting marker and loses marker from time to time.
MindAR: 5 of 10. Quite a nice solution, offering a lot of options to improve tracking that I never saw in other engines.
Blippar: 6 of 10. The camera was working quite slowly compared to other solutions.
Zappar: 7 of 10. Tracking quality was quite reasonable, and after little changes to the sample code, I managed to fix most of the issues in marker tracking stability. One project can contain more than one tracking method, but incorporating both to increase detection and tracking quality is not out of the box.
Compare quality of documentation
Quality of documentation.
8th Wall: 8 of 10.
MindAR: 8 of 10.
Blippar: 6 of 10.
Zappar: 10 of 10.
Compare quality of support
Quality of support.
8th Wall: 4 of 10. Took me some time to get answers from the Slack team and the email team. It took me almost 2 weeks to realize that I am making a mistake using this engine (for the pricing error).
MindAR: X of 10. I didn't really try to get support from an open-source code project.
Blippar: 9 of 10. Super speed responses and very short and effective.
Zappar: 7 of 10. Their documentation was so high quality that I did not really need support to get started. They have more posts than Blippar around the internet, and this helped, but I am still working on finding solutions.
Pricing of system.
8th Wall: 0 of 10. Not only do they have the highest monthly fee among all options, but you also need to pay a monthly fee of 3000$ per project you publish. This is by far the most expensive offer out there, and a huge point against this product. It took me 2 weeks to realize that pricing mechanism, and I would never recommend anyone to use it
MindAR: 10 of 10. Completely free. The best option possible. Only requires that you have your own hosting service, but I assume that you already have it.
Blippar: 9 of 10. I would give a 10, but I never knew it can be free. Blippar has the cheapest offer you can ever ask for. around 0.1$ each time someone watches your app. You pay in packages as you go.
Zappar: 8 of 10. Zappar offers a monthly fee to use, but it doesn't have any other cost to publish a commercial project.

Final Verdict

Even though 8th Wall does have a great product system, I would never recommend using it for its overpriced system.
MindAR would be my recommendation for code-oriented students or educators who would like to do something for free. But the quality is not yet ready for commercial production quality.
If Blipper had better tracking and their Unity samples were more organized, I probably would have recommended them since they have great pricing and a friendly support team.
Therefore, Zappar is the winning framework overall. With their awesome set of tools, terrific samples, getting starter guides, and a great Unity plugin which many SDKs can learn from — they can take a spot on the top of my list. I would be happier if their tracking quality would improve if they could combine world tracking to stabilize the image tracking and increase the distance of image detection.
Leave me a comment if you find other useful frameworks.

Building Augmented Reality for the Web: Which Platform is the Best? was originally published in AR/VR Journey: Augmented & Virtual Reality Magazine on Medium, where people are continuing the conversation by highlighting and responding to this story.