I hope I did not waste your time with this but there is no "Real Working" problem behind this.
Simply because the renders looked like this I am trying to figure out and understand what is going on.
pokoy: yes, upping the GI sample quality helps to some degree
dune: see above. It`s only playing around with the settings and coming across stuff I can`t make sense of.
The examples are completely overdone, of course. Reflection softness was set to 0.9, 2, 4, and even higher numbers.
Surely I did not know what suitable numbers to put in, at first.
But your solution would be a good workaround for sure.
kadri: also a good workflow to get this out of the way.
And your observations regarding the cores and threads seem to come close to mine.
It is simply puzzling me, why that happens. Take a look at the new attachments.
I think the softness in this picture might be one might want or use in a project.
Maybe like a blurry smeary magic crystal ball reflection or something.
I set the cores to 16 and to maximum 16 threads as well.
The softness in 01...jpg does not look too exaggerated, I think. This was set to 0.2.
Reflection softness set to 0.1 and the edges are visible still.
Picture 02...jpg has softness set to 0.05 and it is ok regarding the edges but already too "sharp".
It`s about finding the right balance between these values, as it always is (yawn
) and there are workarounds.
That is good to know. Because upping all the samples settings available in the involved nodes makes this almost unusable.
The render times simply go through the roof.
I really wonder why it occurs at all.
ps: fleetwood - that`s very helpful. Thanks. I did not search long enough...