Hi there,
I have been working on making downloadable models looks the same regardless of the render engine they are used in.
I've been finding that in Arnold, and a lot of other engines, that they appear way too shiny compared to Blender using just the default PBR maps.
My solution has been to multiply the refl/spec map by a facing falloff and as you can see from the results above, it's worked well.
However, it occurs to me if I'm having to do this step in a lot of the major render engines, including Arnold, that there may be an issue with the workflow used to create the maps.
Has anyone got any thoughts on this? Would multiplying the refl/spec map be a fairly typical thing to have to do? Below is an example of the node setup I've used in Arnold.
Any help anyone can provide would be greatly appreciated.
Bill @ Poliigon
Hi Andrew, are you able to provide a simplified scene and/or screengrab of your shader network? An example map would be useful too.
Hi Lee,
I've updated the main post with a snippet of the nodes I've used.
Here is an example of the reflection map, in this case for the cherries
Thanks
By the way is the Arnold render in the above example the shiny example (before or after you have created the above shader network)?
"I've been finding that in Arnold, and a lot of other engines, that they appear way too shiny compared to Blender using just the default PBR maps."
Sry but if ' a lot of other engines' have a consistent look and only cycles is different wouldn't some basic logic dictate that you better change cycles setup or that is cycles doing something different ? 🙂 Also because it is never a good practice to multiply an albedo by an incident falloff.. that kind of stuff should be taken care by the material brdf.
Hi Lee,
Thanks for getting back to me. You were exactly correct in fact, we assumed Blender was 'correct' because it looked very close to our Marmoset result, which we're using as are target but it turns out that was a fluke, after performing a load of tests we realised we had an incorrect setup for a Specular workflow in Blender/Cycles.
Our current solution to match the Marmoset reference is to simply multiply the reflection map by about .5/.6 or so.
In your experience is it fairly common for realtime renderers and offline renderers to give slightly different results in terms of reflection strength?
As now, Blender, Arnold, Redshift etc all look a little too shiny whereas Unity/Unreal/Substance/Marmoset all look the same. (This is under identical lighting etc)
Thanks again for any help you can provide.
Bill @ Poliigon