I have been playing around with the text recognition example that Vuforia supplies, I am trying to read the points of the corners that are used to render the mesh that overlays the text. What I get though is a very wierd printout that jumps between what looks like the correct points and some extremely out of place ones.
The printout onto the console is here:
https://www.dropbox.com/s/7ykcfexnbvb26bg/VuforiaDebug.JPG?dl=0
APRIL is the word name, followed by the 4 Vector2 corner printouts, and the 0 was just the amount of words being tracked.
Seems wierd that it jumps from normal to extremely high and I don't think I can debug further to see why as I dont have access to the source code.
This was taken directly from the text recognition example.
var normals = new Vector2[4];
for (var i = 0; i < 4; i++) {
var p0 = corners[i];
var p1 = corners[(i + 1)%4];
normals[i] = (p1 - p0).normalized;
normals[i] = new Vector2(normals[i].y, -normals[i].x);
}
//add padding to inner corners
corners = ExtendCorners(corners, normals, mBBoxPadding);
//computer outer corners
var outerCorners = ExtendCorners(corners, normals, mBBoxLineWidth);
print (outerCorners[0] + " " + outerCorners[1] + " " + outerCorners[2] + " " + outerCorners[3]);
- Development OS (Mac OS X, Windows, Linux): Windows - Mobile OS and Version: Android 5.0 - Mobile Device Manufacturer and Model name: LG G3 - Do the Vuforia Sample Applications show the same behavior?: Yes, this was taken from the sample