How to write an augmented reality application?

Question

Android

He came up with the idea of ​​making photos come to life, like in Harry Potter newspapers: you point your phone at a printed picture and instead of a static image, a video is played.
Let, for cars, I leave a qr-code in the corner of the image, then the process of the application is something like this:
- find a qr-code on the image from the camera
- read the code
- if a potentially correct link is sewn in the code, determine the rectangle framing this code (for humans these are the borders of the printed image)
- in place of the rectangle, start showing the video via the link
As I understand it, the task is completely solvable in the forehead on OpenCV, if it weren’t for such a possible case that part of the image rectangle could go beyond the area captured by the camera and the program would stop perceiving the remaining area as a rectangle for video and stop showing it.
This is user-agly, if the fact that the hand twitches, the entire playback will be destroyed.
It turns out that some tracking of objects is needed, and here specialized AR libraries enter the arena.
There is a problem that they are paid mostly, but I don’t want to see a watermark on the image at all. Yes, and a lot of it is sharpened for 3d animation and games, I don’t want to pull a lot of extra stuff into a nonsense application.
Kick, please, in the right direction, which tool is best to choose for solving my problem?

Answer 1

Unity + ARKit

Answer 2

Either Vuforia or ARCore/ARKit. The first option works on all devices, the second only on the newest ones, and you need to write separate code for each platform. I think that the cost of the Vuforia license will cover the cost of self-written code. You can also take Unity, there is a wrapper over arcor and arcite, but the number of target devices will still be small.