Friday 20 February 2015

What trackable type should I use in Vuforia project?

ROAR! The Augmented Reality family board game.

There are many different types of targets and markers you can use in your Vuforia project. Image targets or frame markers are a couple of the best examples of the possibilities of the Vuforia plugin for Unity. If you are developing an ambitious or commercial project you want to know what is your best shot in terms of performance, cost and reliability.

Making Augmented Reality apps is getting easier and easier. The future of the free version Vuforia 4.0 is not as clear as it should be and people is concerned about the price of it. Anyhow is a bright future for AR in general! Meta glasses and Microsoft Hololens prove that there will be hardware that will live up to the expectations of AR potential consumers and that in a matter of few years this gadgets will be in everyone life just like the smart phones today.


I this post I will go though the different types of markers I have used in my real work experience and I will explain which one works better in any situation. Deciding what is the best option is up to you :)



Frame Marker

Frame marker is the simplest way of making AR work.
PROS:
  • Faster performance: is the only way of tracking 100 markers at the same time.
  • Resource saving: up to 512 different markers without any real picture.
  • Reliable when small: the marker that gives the best results when the real life token is small or far away from the camera. You can have a 3cm marker recognised more that one meter away.
  • Easy to set up: you don't need to go to the web page and create the target database as they are built-in in the plugin.
  • Flexible: you can put whatever you want inside the marker as long as you respect the black border and the white line around it. You don't have to care if it has enough features.

CONS:
  • The frame: you need to have this black a white frame that ruins the magic of the AR a little bit.
  • Not reliable when big: in other to work properly the frame has to be totally visible in the camera, that means that it will not work if you are too close or something is on top of it.
  • Not so flexible: you need to respect the frame and you cannot change it in any way (shape or colour).

Image Target

Image Target is the most common target in AR.
PROS:
  • No frame: people is still amazed today to see that any picture can be used as an AR target. This is a great plus to impress users.
  • Reliable when big: it will work even if you just see a part of when for example you are very close.
  • Virtual buttons: some of the best Vuforia features like Virtual Buttons are only available for Image Targets and therefore making it the only option in many cases.

CONS:
  • Pictures need features: not any picture performs well in AR, it need to follow a certain set of rules that sometimes required tweaking the original picture to make it work better.
  • Not reliable when small: if the image target is too small or you are very far away it will most likely not work very good. Even if you have an amazing printer.
  • Not so easy to set up:to create image targets databases you need to go to the web page and do it there. Additionally there are few more steps to follow in order to settle it up inside your project that are not super intuitive.

Other Targets

That Object Recognition looks AWESOME!

OTHER OPTIONS:

  • Multi Targets: this targets have 6 sides, like a box. Very good if you have a box to begin with. It's important to note that if the box is poorly crafter the recognition will suffer greatly.
  • Cilinder targets: very good to put on cans or bottles. The same as the multi target, it need to be prepared carefully.
  • User-Defined targets: a great target in the situations when there is not any target! The user creates it on the fly making a picture. The recognition could be better, but is ideal for free apps on the store since the user doesn't even have to print anything to use the AR.
  • Cloud-Recognition targets: do you have thousands of targets and thousands of users? This is how you do it. However using the Vuforia servers in this way is not free.
  • Text-Word targets: nice make AR translators. I believe is quite limited at the moment and not very used. I can still find interesting uses for in anyway
  • Object-Recognition targets: I cannot say much since I haven't tried it yet, but the possibility of tracking the shape of any object imaginable opens a new world of oportunities. Most of all in the toy sector.

CONCLUSION

My general recomendation is: if you have a big board to track with things on top of it use Image Targets. If you have an small tokens without anything on top of them go for Frame Markers. For the special situations you can use the rest of them, but it will not be usually the case.

Do you have any experience with AR trackers? If so, please share it in the coments! :)
PROS:
  • A veteran: NGUI out stand in every possible feature to uGUI and the plus is that plenty of examples will help you to implement faster.
  • Unity old standard: the king of the UI in Unity so far, with the greatest community you will not have trouble finding help.
  • Stable and up to date: NGUI is constantly fixing bugs and what is more, integrating new Unity functionality with every patch. That means that NGUI will continue to grow thanks to uGUI.
  • Many complex component that works out of the box: that's it, the most common UI setups are already implemented in a flexible way so in most of the times you just need to combine them.

CONS:
  • Has expire date: one day it will stop to receive updates and most of the people will be using uGUI at that time.
  • Performance capped: it doesn't matter how optimised NGUI will be, uGUI will be always an step ahead.
  • Paid: money is money. It can be a high investment if you are an individual or you just want to compare to other options
- See more at: http://unitydojo.blogspot.com/2015/01/unity-46-14-tips-to-choose-your-ui.html#sthash.SfNfl1Z0.dpuf

5 comments:

  1. Do you know how to get the cloud recognition set up the tutorials online are out of date.

    Also for text recognition is there a way to make where it reads in the text larger when I run it on unity there is only a small bar at the top where the word has to fit I'm hoping that I can use the full camera view to read in the text.

    Also is there a way to make text recognition read the text and display the image faster.I feel like it takes a long time for unity to read the text opposed to it reading a qr code.

    Thank you for the help!

    ReplyDelete
    Replies
    1. Hello Kionte!

      I never used the cloud recognition... if the tutorials are outdated I recommend you to ask in the forums. There always answered me there :)
      https://developer.vuforia.com/forum

      About text recognition, the only thing I can imagine is using Image Target instead of text reco itself. Create a picture that contains the text and upload it to the web to create a target database and see how good it could work! :)

      Good luck!

      Delete
  2. Do you know how to get the cloud recognition set up the tutorials online are out of date.

    Also for text recognition is there a way to make where it reads in the text larger when I run it on unity there is only a small bar at the top where the word has to fit I'm hoping that I can use the full camera view to read in the text.

    Also is there a way to make text recognition read the text and display the image faster.I feel like it takes a long time for unity to read the text opposed to it reading a qr code.

    Thank you for the help!

    ReplyDelete
  3. Nowadays, such situations occur, you have to get used to it.

    ReplyDelete