Geometry Nodes - Position/Placement of CurveToMesh from extracted Curves

I want to replace faces of an existing object to gain control over resolution (subdivisions). I extract edges and convert them into curves and create a mesh from these. Unfortunately placement/position of the resulting mesh is of. (All transformations are applied!) How can I fix that? Here is the model:

Here is the nodetree:

The profile-curve used for curve-to-mesh is defined in the XY plane… so you’re getting the expected result.


Why not use Mesh subdivision? What is your goal?

Here is a method to subdivide a single face by swapping it out with a grid, using UV sampling to set the position of the grid:

… the grid vertices then control the subdivisions of the single face. :man_shrugging:
Good luck.


Thx zeroskilz,

Why not use Mesh subdivision? What is your goal?

Your second post gives the answer: I want insert edge loops freely.
Subdivide is no option, since X and Y divisions are always equal!
I already thought about your 2nd solution (replacing faces with grids) but I wasn’t able to solve this either. This helps me a lot!

However, I am still interested for answers to my initial question.
Your first solution doesn’t look right. It doesn’t replace the face. The resulting face is stretched and only this face might live in world XY. What of the other faces?
In my mind curve2mesh should work like an extrusion. Why doesn’t it follow the initial curves? I could subdivide them easely and be done!

No, it doesn’t. The Profile-Curve lives in local-space. My 1st example was not a solution. It was an example for you to study to help you understand how curve-to-mesh works.

Good luck!

1 Like

Thanks for clarification.
I start to grasp the concept.
Could you help me once more regarding your second example?
This is what I get:

Looks like that only worked for square faces. :thinking:

Seems like you need normalized coordinates that match the indices and that when doing a set-position that the Sample UVs need to be in Point domain: :man_shrugging:

…even so you get UV triangulation warping for non-rectangular fadces.

Good luck.