Seemingly buggy interaction between real-time reflection probe and smoothness in PBS

Hey everyone, I'm experiencing a buggy interaction between my materials and a real-time reflection probe, wherein using a smoothness value less than 1 causes the reflection to quickly degenerate into a mass of giant blurred pixels. To replicate the effect, create a new material using the standard shader, set the metallic and smoothness both to 1, and set it up to receive reflections from a real-time reflection probe. You should notice that everything reflects like it should; however, when you lower the smoothness down to around 0.9, you will see that the reflection, rather than blurring normally, looks like it drops its sampling resolution enormously and generally looks terrible.

As far as I can tell, increasing the resolution of the probe doesn't help (it looks bad even at 2048), nor does switching the shader from standard to specular. Shaders created through ShaderForge also experience the same problem. What *does* help, however, is to use a baked reflection probe instead; in this case, the reflection on the material behaves exactly like you would expect when altering the smoothness, but unfortunately I need real-time for my current project.

Presumably there's some sort of bug related to the interaction between real-time reflection probe data and smoothness (since again, it looks perfectly fine if smoothness is set to 1), unless I'm just totally missing something. Has anyone else encountered this problem before, and if so, were you able to come up with a solution?

 

One more thing I just tested; disabling HDR on the reflection probe and then adjusting the smoothness down from 1 causes the reflection to rapidly disappear altogether rather than pixelate. Again, this only applies to real-time reflection probes, baked ones seem to do just fine in this scenario.