… it still seems to be broken with more complex setups with custom datasources and dynamic callback texts. It draws the “unintended gradient”.
I was not able to create a simple sandcastle to reproduce the problem yet but this far I know that issue goes away if I set whitePixelSize to be 64x64. Even still we can see a very thin black border around label backgrounds but the issue is way smaller.
Thanks for the report! Would you be able to provide a Sandcastle code example which reproduces the issue? The previous issue with the label backgrounds arose due to race conditions, so an example would help us track that down.
I got something done now; Its not yet the exact issue but something a bit similar. This sandcastle draws label without background at all. It really seems like a race condition of some sorts, because I had to tweak with requestRenderMode and timer ti get this to fail…
It gets fixed also by forcing all textures to use nearest filtering… My current undestanding is that timing might cause background image to go to another place in texture atlas and that makes some other texture atlas tex-coord bug surface. But I’m just guessing really…
Other thing that I was puzzled with is that whitePixelBoundingRegion does not seem to affect anything.
Hi,
I am ran into the same issue.
Any updates on a solution?
About setting “Sampler.NEAREST in TextureAtlas”, I couldn’t find how/where to do it, or what are the cons of using this method compared to the default method.