Corona renders different shading with anisotropy

[MetricaVisuals]

We have a scene where corona will produce a shading bug when using anisotropy, all objects have the same material but render way different.
We spend a LOT of time debugging this until we found the anisotropy was to blame..

[maru]

Interesting. Could you share the scene with us? Even a stripped version like the one from your screenshot would work.

[pokoy]

Try this - (in a temp scene or copied objects of ocurse):
- in 'advanced options' tab of the material, change 'anisotropy orientation' from local axis to UVW
- now apply a UVW mod in box mode to *all* affected objects so they share the same UV space
- make sure to use the same UV channel for both the anisotropy orientation and UVW modifier.

If the problem goes away, it means the objects have been modeled in different local axes. This happens when you start modeling one wall in the left view, another one in front/top etc. The local axis of these objects will be different and since anisotropy treats each one differently each object will reflect differently.
By using UVs as anisotropy orientation *and* making sure all objects have their UVs defined the same way by using a UVW modifier, it should render the same for all objects.

A good addition for the physical material would be to have a user 'reference node' for anisotropy orientation, that way users wouldn't have to worry, the specified reference node would then be used as axis reference for all objects using that material.

[arqrenderz]

Thx Pokoy !
When i read this part "objects have been modeled in different local axes" it clicked my mind, i was trying to solve it reconstructing the meshes with boxes, planes,etc.. with various results because of this!
I can now rest in peace :)
+1 for the reference node and this knowledge
 

[pokoy]

Please don't rest in peace yet! :D

Glad it was helpful.

[maru]

I have logged the "anisotropy by reference node" option as a feature request here: https://chaoscorona.ideas.aha.io/ideas/CMAX-I-155
Feel free to vote for it and/or comment. Note that it will be invisible for some time, until approved by Tom.

[pokoy]

I have logged the "anisotropy by reference node" option as a feature request here: https://chaoscorona.ideas.aha.io/ideas/CMAX-I-155
Feel free to vote for it and/or comment. Note that it will be invisible for some time, until approved by Tom.

Great, thanks! However the link doesn't work for me - it says 'record not found'.

Maybe even it would make sense to have an 'object space reference map' node where any nested map/maptree would inherit the picked object's local space?
A more generalized solution wouldn't be bad since this is useful for more properties than in this case with anisotropy.

[maru]

Great, thanks! However the link doesn't work for me - it says 'record not found'.

Tom needs to approve it first. :)

Maybe even it would make sense to have an 'object space reference map' node where any nested map/maptree would inherit the picked object's local space?
A more generalized solution wouldn't be bad since this is useful for more properties than in this case with anisotropy.

Could you please log this one at https://chaoscorona.ideas.aha.io/

[pokoy]

Oh sorry - missed that it needs to be approved first.

Will post the node map idea once it's approved and reference it so it's clear it's a follow up and a generalization of that reqeust.

Thanks Maru, much appreciated!