6
$\begingroup$

I want to bake a curvature map in Cycles but not by using the pointiness node which is dependent of the vertex density and not with a normal map to cavity ( How to convert a normal map into a curvature map ) because it creates artefacts.

I want to create a Concavity map that generates dark areas in concave shapes. Suzanne should bake like this (baked in Knald):

enter image description here

And i need to combine it with a Convexity map that highlights the peaks like this (baked in Knald):

enter image description here

I want the closest possible bake to the images i have posted above. I have used the Suzanne's default scale and a plane to bake it.

$\endgroup$
  • $\begingroup$ did you try "dirty vertex colors"? adaptivesamples.com/2013/08/07/… $\endgroup$ – lbalazscs Jan 13 '17 at 18:26
  • 1
    $\begingroup$ It's like the pointiness node, the quality depend of the vertex density. A curvature map works with every topology. $\endgroup$ – Danyl Bekhoucha Jan 13 '17 at 18:28
  • $\begingroup$ I know it's late, and I don't have a proven, tested answer right now, but the way to handle this would be to bake world space normals and then check the dot products of each sample with its neighbors in the compositor. Where the net dot products are negative, you have convexity; where they're positive, you have concavity; where they're zero, you're either flat or have negative (saddle) curvature. That'd be in UV space, but you could resample based on world space lookups from baked world space and correct (except if you crossed a seam.) $\endgroup$ – Nathan Sep 9 at 14:48
3
$\begingroup$

Maybe this helps. Baking an Edge map Using Ambient Occlusion https://youtu.be/RhMEBKuPYXY

$\endgroup$
  • 1
    $\begingroup$ It's a solution, but having to invert the normales of an high poly with millions of triangles takes a lot of computing time/power. $\endgroup$ – Danyl Bekhoucha Jan 13 '17 at 21:07
  • 1
    $\begingroup$ Also a thin cylinder will create errors, his planar faces will generate occlusion because they will be close to each other and the thinkness low. The concavity and convexity generates informations based on the angles only. $\endgroup$ – Danyl Bekhoucha Jan 19 '17 at 15:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.