Refraction + low GI/AA balance

[maru]

I suppose it's not a bug but I'd like to know the explanation of this behaviour. This happens when you create a glass material with caustics enabled and render it with low pt samples value (1-5). After 6 it becomes refractive but has some strange dark outline. Also last two images show another setup with a glass material with caustics disabled, rendered with gi/aa balance @ 1. This also happens with different geometry and without hdr background. Is this expected?

[Ondra]

which corona version? can you post the scene? I am unable to reproduce

[maru]

Ahhh crap I investigated it and it turned out I had secondary solver set to none. I'm still interested to hear why this happens but I guess this is the explanation. Sorry!

[Ondra]

it is because with such low samples the refracted ray is not considered important enough to use primary GI solver, so secondary (none = black) is used

[GRouslan]

Wow. I'm stunned.