I am facing big trouble, my Object Recognition app only work on the first object i tracked but the second. If I restart my app the track the other, it work but not with the remain object. Any one know this please help me!

I am also having the exact same issue, even tested it with different light in the Scanner application and it works perfect, however using the same phone with the application I made in Unity, It could still recognize all sides but constantly gets lost by the camera.

For anyone that is having trouble with Light:

If you are using the Mobile Scanner app, before you start scanning your target, grab another smartphone and place it underneath yours with the flashlight turned on! That way when you scan your object, you will allways get perfect lighting conditions.

Kind Regards,

Hristakis.

mfontoura wrote:

Having the same problem. Test is better than app deployed by Unity in the same light conditions.

I don't know why this thread was left unanswered.

i think it's a very interesting feature and I don't know why my cheap webcam in unity preview works better than a Galaxy Tab S!

let me know if you will make some other test.

best.

Having the same problem. Test is better than app deployed by Unity in the same light conditions.

hi,

i did some test with other complex objects and I have the same problem.. i can track them very well with scanni App but they work bad in a compiled Application, on the same table, same position and same lights.

any suggestion?

please help us to solve this problem

thanks

I don't think so..

the object is placed in the same position, without the paper marker.

same behaviour with different objects.

why the scanner app works? What's the difference?

thanks

Thank you for the screenshots. In the capture app screenshots the back is lit better than in the tracking screenshots. Also, the back has less feature point areas. The pants seem to be a solid color without feature points and the hair is similar. You may want to compare number of feature points (front vs. back) during capture and try to increase lighting on back of character during object tracking.

hi,

these are some screenshots.

using scanner and test App, I can see the augmented solid from every position.

using the same OD file in a productio App, I can see my cube (and bounding box volume) only if I'm in front of the object, or a little shifted.

you can see the max distance I can reach from the very frontal position.

i've talked about this problem with your rapresentative at unite2015 in amsterdam..

in front of the object, with correct lights and environment, the recgnition is very fast and stable

let me know, please..

thanks

Emiliano

Tracking depends on feature points and can be influenced by background environment. It sounds like your objects don't have enough feature points in some areas to maintain the tracking. If you can attach screenshots, it could help explain.

hi,

sorry for late answer.

I have enough points from every position.

I've just scanned other 2 objects with about 200 pints.

in test mode averything works very well. no lag, no jerks.

using data files in unity play mode, it doesn't works.

If I build a test App, it works only if i'm in front of the object.

if I rotate my object, traking is lost.

why there is so many difference in production environment? 

performance in test mode using the scanner are what I want and I think it should be possible to have it.. or not?

thanks for your support

Emiliano

Were enough feature points captured on all the sides of the object? When you turn the object around, are there other objects behind it?