I’m working on a project with uses several image targets to maintain tracking of a single object. I was wondering if there were guidelines or tricks to use to maximize the distance I can travel away from a single image target before the xrimagelost event is triggered. I’m currently using some random images as targets that meet the recommendations of this page but I’m still only able to get a few feet away from the printed images.
Hello @Matthew_Gallagher the effectiveness of tracking an image target in augmented reality largely depends on its relative size and visibility to the camera, rather than the physical distance alone. For instance, an image target as large as a building can be tracked even from a mile away, while a target the size of a thumbtack might only be trackable from a few inches away.
The key factor here is the proportion of the device’s screen that the image target occupies, coupled with the presence of distinctive tracking features within the image. For optimal AR tracking, our technology must clearly recognize and interpret the image target. Therefore, maintaining a significant portion of the target—preferably centered in the camera’s view—enhances tracking accuracy.
Here is a photo for reference (essentially the same ratio of image target taking up the screen and tracking but the image target sizes are very different):
Then along with best feature points on an image target video here:
Hi Ian , I have a similar issue with tracking a 2 m statue, that we’d like to overlay with ar content. Making the " image target " larger seems to be the solution. Though wondering if you can advise on the best strategy re : feature points for an incomplete " image target ". Since the base of the statue will occlude parts of the target placed underneath …
-
Can " 8th wall " resolve incomplete targets …
Also roughly how much does it need to see for a 2d marker ? -
The other idea was to position the " image target " in front …
Then offset the ar content back a bit , though we will hit the scale issue …
— I will try with absolute scale …
Thanks !
Hello @Kean-Chuan_Lim if you are using an image target there is no need to use absolute scale as you can use the image target to scale the experience.
Can you share a photo of what the enviornment looks like? Is this an image of a statue?
The amount of the 2D marker our engine needs to see depends on a lot of attributes as I stated above, such as feature points, size, centering the image in the screen etc.
Ok thanks for the heads up !
I can see in your example above that " image marker " and relative position of the viewer will affect the 3d model scale. I’ll do some practical tests with this in the studio tomorrow. If I get what you’re inferring : its best to figure out the users position. Then scale up the physical " image marker " so that the 3d augmentation appears at the correct size …