Posted by philjord on Feb 17, 2016; 2:54pm
URL: https://forum.jogamp.org/Testing-text-antialiasing-Java3D-1-6-0-pre8-vs-1-6-0-pre12-tp4036282p4036291.html

Alexei,

A quick look at the Java3d source only shows one possible commit that might cause this.

pre8 vs pre12
https://github.com/hharrison/java3d-core/compare/1.6.0-pre8...1.6.0-pre12

Has 2 changes that look like they might impact anything in this area

initial fix for stereoscopic and double buffering graphics configurations
https://github.com/hharrison/java3d-core/commit/8593fa0991feb9c9baa85a83269fc50c755dff80
Doesn't seem to do much that would cause trouble

prepare to refactor the Font3D tesselation code
https://github.com/hharrison/java3d-core/commit/7a6102dbbb3660c6ce15aa4a64f3d289cbf09b59
This one add normals to the Font3D, which given you have lighting enabled does definitely affect the edge color.

On my screen I can't see the same difference you are seeing, but I can definitely see a difference.

You could run your tests again with pre8 and pre12 but replace the line
ap.setMaterial(new Material);

With these 3
Material mat = new Material();
mat.setLightingEnable(false);
ap.setMaterial(mat);

In your sample code and see it that removes the issue?

If it's not that then possibly you are using different Jogl version with a different default AA pass number? Can you confirm it's 2.3.2 with each?

You did already spot that it claims to be using multi-sampling in both cases.


Thanks,
Phil.