geo edge displacement

[Follyx]

Hi,

well I'm frozen. See the file for my problems.

I cannot get the chunks from the quader correctly displaced. 2 sides are always missing in the geodisplace process.

Use the blue main switch to check a smaller object (fragment of an extern wal)l.

Thanks in advance.

Cheers
Follyx

 

 

displacement_detai_basic_01.1.hip

Edited June 23, 2016 by Follyx

[f1480187]

 

You need to transfer normals on further distance, probably. IIRC, this setup was intended for thin walls, to displace in 2D plane. Another promising implementations was in edge displacement thread. My addition is /BUILD_DISPLACEMENT_DIRECTION1/attribtransfer1 node.

 

edge_displacement_fix.hipnc

Edited June 24, 2016 by f1480187

[Follyx]

Wow, thats all? Instead of transfer the normals, just build an inheritance for all available attributes?

Works perfect, thanks a lot. Thats so frustrating. Its such a heavy way to understand and think of every chance... I'm not sure to understand the logic behind your change. And is a wired pointcloud filter really slower than the attributetransfer as written here (last post)?

What would you suggest to flatten the chunks on its broken sides near to the middle? More noise at the edges, more planar in the middle of it or kinky in certain areas (think this would be the next step) Simple more naturally. Any Idea?

The distance threshold and blendwidth parms in the attribtransfer node are to limited.

I think I have one but dont know now how to realise it. A gradient wich control the amount of noise, yes. But how to define the area. Would it be easier to arrange that with VDBs. Questions, questions, questions.... lol.

How do you get the idea? Did you already used such an approach or just thought about it? Well as more as I know as more I see how less it is...

Would be great if you or anyone else could assist me in here.

 

Cheers

Follyx

Edited June 24, 2016 by Follyx