Hi, I just found 2 small issues
scene.useOrderIndependentTransparency = true;
https://playground.babylonjs.com/#T8UQTA#188 
Hello @MarianG , how are you doing?
Thank you very much for letting us know! We will be updating the documentation to fix those issues!
Issue #2 is fixed and should be up soon.
For issue #1 I’m pinging @sebavan , @carolhmj , so they can take a look on the rendering side.
ccing @CraigFeldspar too for #1 as they’re the OIT author 
@CraigFeldspar do you have any ideas ?
Quickly looked into it, framebuffers destinations are wrong. Axisviewer renders in the default FB, while OIT renders in a separate framebuffer. So when compositing to the final frame, the axisviewer result is overwritten.
I think this is not a specific problem to axisViewer, we’d need a manager that knows at which point in the pipeline, what is the rendering source and destination, so we don’t let individual classes handle where they want to render.
@sebavan want me to create an issue to track this?
I think we should add it in the limitations of OIT in the doc as it is not something we intent to fix at the moment. this is an issue with rendering groups.but as it relies on post processes we would lose the previous content.
A workaround is to force the gizmo in the rendering group 0: https://playground.babylonjs.com/#T8UQTA#217
Add comment on OIT limitations by carolhmj · Pull Request #614 · BabylonJS/Documentation (github.com) added this to the docs, if anyone else knows any limitations feel free to add