Hello, i been trying to do an object recognition test with an specific object, i have tried taking photos directly of it but wikitude studio doesnt generate many points, so tried a 3d version of it with screenshots and it generated a decent amount, but when i try it in unity, the webcam only recognizes different screenshots on my phone (like the sample of the 360°firetruck that comes with the pack) but it does not recognize the actual real size object. The scale of the points are accurate (height:10cm)
I wonder if its a webcam issue or if i need more points directly from the object photos.
My unity version is 2021.3.22f and wikitude professional edition 9.13.
Heres the photo of the object, the screenshot of the 3d model that i uploaded and the generated points. Thanks in advance.
despite having a nice looking object it's really hard to be tracked. First of all the shiny surface is a big challenge because reflected light can impact the detection. Despite that since the object's texture is almost the same color it's really hard to find keypoints in order to track and also distinguish between front/back/side. Try adding some textures / paint your object with some kind of pattern of high contrast that makes detecting and distinguishing certain surfaces easier (non-repeating tiger stripes, QR codeish patterns, ...). Changing the shape of your object to something that's less symetrical could also help, but changing the texture will probably do the trick. By the way: you did a great job having a plane white background for taking the pictures!
Here are some tips on how to create good Image Trackable (this also applies to surfaces for Object Tracking) : https://www.wikitude.com/blog-image-recognition-tracking-best-practices-and-target-guidelines/
Also here are some tips regarding Object Trackable: https://www.wikitude.com/external/doc/documentation/latest/android/objecttargetguide.html#image-based-conversion-best-practice
I hope that this helps making your tracking work.
With best regards,