First, Tracking on the video lags — this means that you need to change or replace the Reference Image to improve the quality score.
Next, learn these features of tracking for each platform (iOS and Android) from my tutorial.
ARCore (Android) Behavior
ARCore Requirements: Supported Devices & Augmented Images.
ARCore API assumes that AR Images are static in the environment (2), so once they are recognized successfully, they will always appear in the list of anchors (updated list) (1) until the session is reset.
This means 2 things:
-
AR Foundation will never mark the Tracked Image as removed, and it will never mark the Tracking State for the Tracked Image as None.
-
If you try to test 2 Different Images via your monitor (desktop/laptop) and you will switch images with CMD+Tab (macOS) or Alt+Tab (Windows), so Images will be in the same position on the monitor, then ARCore will not understand that.
ARKit (iOS) Behavior
You can learn more about ARKit Requirements for Image Tracking here.
-
Tracking State for the Tracked Image can be set as None when the camera stops seeing it.
-
2 images can be successfully tested on the monitor in the same position, as it described above for ARCore.
So, you have an ARCore example on the video as I understand, and this is the case of item 2 in the list from the tutorial above. As an alternative way of implementing, you can:
- Use another AR engine that works differently.
- Using AR Foundation, scan only 1 image (book cover for example) and implement a virtual book as I show in the next example (related links are in the YouTube description).