Get position/rotation of image target to instantiate object there, but not as a child of the Image Target

Question

Hi,

I've seen numerous threads about this going back years, but not answers

With the likes of Vuforia and the NReal SDK itself, you can get the target image coordinates and instantiate objects at that exact position/rotation, but without the need to track the image, or extend tracking etc... just using the image itself as a trigger and means to get the intial position/rotation data.

Having used EasyAR for the past few days, I can easily tell when a target image has been found and trigger the string of events, and switch off the tracking once the object is instantiated, BUT the position/rotation isn't accurate at all. Not like Vuforia and the NReal SDK. I've even tried using an empty game object as a regular child of the image target to get it's position/rotation to use, but again... not at all acurate.

Is there a way to get the exact position of the image target, so you can just instantiate at that location, and not have any form of image tracking going on in the background? It's such a simple thing in other packages

Answer 1

Don't get it when you say " the position/rotation isn't accurate at all", It sounds like the SDK not working, but it seems you are asking for a simple feature, not reporting a bug through the whole thread.

So I guess it is some default value behavior if you are not telling the algorithm is totally wrong. Try change these to different values and see what happens first, you can change them in the ARSession/ImageTarget while running in editor.

https://help.easyar.com/EasyAR%20Sense%20Unity%20Plugin/latest/ApiReference/ARSession.ARCenterMode.html

https://help.easyar.com/EasyAR%20Sense%20Unity%20Plugin/latest/ApiReference/ImageTargetController.html#activecontrol

Maybe you can attach some video if they don't work, and we would know better what you are asking for.

Comments

BTW, if you simply does not want an object to be child of image target, it is really a matter of basic math to sync transform each frame, or, like above link I attached, simply make the target not move at all. And... what is the difference between an API getting the pose directly and getting empty target node transform each frame? I didn't see anything different. Not to mention you can always touch the low level Sense API whenever you want.