hopefully I'm in the right section...
I have a little problem to get the object recognition running and it would be great if anyone could help me out with it.
I have followed the instructions for unity: https://www.wikitude.com/external/doc/documentation/latest/unity/objecttrackingnative.html#object-tracking and also the tutorial on youtube: https://www.youtube.com/watch?v=mpflhDwhSwM
Sadly I couldn't get anything to work as expected.
I'm not quite sure if there is a problem with the wto file or a general one.
Currently I'm using SDK Version 8, so I had to upload 30 images to the targets manager and got a wto file which I have placed into unity as in the tutorial. My WikitudeCamera Object is also set and there is the licence key inserted. Also I have created a small cube which should be shown after the recognition. If I try to run the application on my phone there comes a message with trial all over the screen and then everything is freezing. If I run it on my desktop I see the webcam video but nothing happens if the object comes in view...
I have also tried to change the wto file in the sample with mine, but that doesn't change anything.
Would be very cool if someone has a idea what I could try next, or maybe what I have not done right...
Thanks a lot!
Do you have any other device you can try on? We weren't able to reproduce your issue here.
I've tried on Vuzix M300 but the app didn't respond to buttons and then the android launcher crashed. I'm not trying to get an app optimized for smart glasses. It should have run as a regular android app at least... This is another problem, but it would be great to get object reco for Vuzix and Moverio SDKs in the future.
I'll try another phone tomorrow, but my goal are glasses.
The sample app runs without a hitch on Xiaomi Redmi 5 Plus with API level 25... So maybe the issue is in graphics API? Tho, glasses have OpenGL ES >3.0 as well as Xiaomi phone.. But slightly lower API level.
Pardon me, I haven't noticed that Epson Moverio SDK has already provided 3d object tracking. The sample project for Epson Moverio works, though in 3d mode rendered 3d objects are seem to be squeezed along Y-axis.
Maybe I haven't callibrated a cam carefully enough, maybe issue is in interpupillary distance or something, but in 3d mode objects are highly indistinct, hurt my eyes
Please try to recalibrate the device, and also check the 3D Model On Target sample to see if Image Tracking is affected by this as well. If the issue persists, please send us some screenshots so that we can have a better understanding of what the problem is.
I've checked the sample and I've found out that the scene is clear only on a certain distance, probably focal length. But if camera a lil bit closer to or a lil farther from the target, the scene appears doubled. Is this a feature of 3D rendering and there's no workaround?
And yes, the model is slightly flattened (but not that much as in the first screen) and stretched, but not that much as in the second screen, though it's visible.
And also it's probably not taken into account that the cam on Epson is to the right, so scenes in the sample app are rendered with the offset. Especially it's visible with firetruck tracking.
P.S. the screenshots have swapped their places
By the way, maybe you have a ready to go calibration file for Moverio bt-350 for developers?
Regarding the focus issue, the device can only render the augmentations at a fixed focus distance, so when the target doesn't match this distance, you can either focus on the target, or the augmentation, but not on both at the same time.
The screenshots seem to look fine. The device stretches the rendered image to cover both eyes, so if you stretch the screenshots to double their original width, they don't appear distorted anymore. Does it look different when looking through the glasses?
Unfortunately, we don't have a calibration file for the bt-350.
Ah, that's why! In the glasses it isn't stretched too much.
Well, thank you a lot for your help and patience, Alexandru!