The image recognition and tracking algorithms do have certain limitations regarding the target's minimum size on screen or the maximium distance from it respectively. This is due to the limited processing power of current mobile devices and may improve as smartphone CPUs get faster / more cores.
Right now the best advice is to choose target images that show larger image features as opposed to very small and detailed features. This could also explain why a 3-star image with less but large features is able to track better than a 5 star image with many small features.
I hope that helps to acheive your goal.
over 10 years ago
I need to track a small image say about 10 X 10 cm. This works fine when tracking it up close but it will loose it's target very fast. This ofcourse also result into not being aible to track the image from a greater distance(2 meters).
Is there anyway to boost the tracking distance? I tested this with a 5 star image and the tracking was actually very bad. I have a 3 star image that works way better then the 5 star image. Maybe i am still doing something wrong when uploading the image. The only thing i might be doing wrong is entering the size in mm when uploading the image into the database.