anisotropy on separate UV islands

Sempoo

I am also wrapping my head around anisotropy on the dome - and while radial scratches set to spherical projection work fine, anisotropy angle fill does not. It breaks between UV islands.
Any hints?

Eric Chadwick

Anisotropy typically uses the tangents of the model to influence the stretch direction. Tangents are influenced heavily by the uv, so I would recommend redoing your uv to be one contiguous shell.

Neox

Eric Chadwick said:

Anisotropy typically uses the tangents of the model to influence the stretch direction. Tangents are influenced heavily by the uv, so I would recommend redoing your uv to be one contiguous shell.

flowmaps can help with this, but yeah UV seams tend to be an issue and are best hidden where they are not as visible

Sempoo

I have set it up in Blender, and it works fine:

Eric Chadwick

All that tells us is that Blender uses a different way to generate the tangents.

Eric Chadwick

As Neox suggested, you could try overriding the tangents in Substance, using a spiral gradient. As an example, here's one I made for glTF's anisotropy:
https://github.com/KhronosGroup/glTF-Sample-Assets/blob/main/Models/AnisotropyBarnLamp/README.md#anisotropy-directions