just a quick update, it appears that the problem manifests with vray sun/sky, but not with domelight or planes with vraylightmtl.. given i was only using sun/sky for very first test, and will be using dome/vraylightmtl for actual lighting, this is now no longer super urgent. later on ill create a test example. cant share the actual renders unfortunately.
ok here is a simplified example scene. one shiny sphere inside another (fog coloured) refractive one. vray sun and sky. the outer refractive sphere has no reflections enabled.
and to pre-empt a question, the effect is the same with a pure 100% transparent refractive material, so attenuation due to fog colour is not the issue.
Also, on closer inspection, there is a difference between the specular with no caustics, but “affect shadows” turned off - the specular is massively reduced but there is still a hint of something present
and with caustics enabled -specular completely absent
thats quite an odd behaviour. i wonder if making the sun larger (much) means you are getting a reflection of a brightly emissive object rather like when you use a self illum plane. when the sun is small its treated as a point light source and uses whatever system vray uses for doing speculars from point sources.
to be honest ive never quite understood the separation of reflections and speculars and where one becomes the other… i understand the need for a “fake” specular when we deal with infinitely small lightsources, but beyond that its all a bit unclear to me.
Whilst it does solve the specific problem, it of course creates another, in that the shadows are vastly different and so is the sharpness of the caustics in a given scenario, so not ideal if you want the speculars.
I think the separation was maybe because speculars are wholly fake whilst reflections are purely that, so a physically correct approach vs the fake specular and it was simpler to separate them.
