Well, you should follow more or less the same way. Blend mask would be obtained by taking a dot product between vertex normal and world up direction in a classic landscape case. For a spherical world, everything remains the same, with exception that world up direction varies with location.
Quite simply… I could have looked for weeks! Thank you very much Deathrey!
EDIT: Oops! This is half-resolved because my entire world is in rotation on itself! So the material changes with this rotation
I’m trying to find a solution…
Sorry to bump and old issue, but I am looking for some clarification of this issue as well. When Deathrey said that the up vector varies by location on the sphere, how can that be expressed in a material graph?
Ok I got it. I’m using the local position node normalized , and the pixel normal ws node normalized, and getting the dot product from that. add a multiply, power and clamp to control the falloff