Grasshopper

algorithmic modeling for Rhino

# pointcloud(polygon) reduction

Hi,

I have a long term problem...I am trying to make a terrain model which would be triangulated depending on the steepness or curvature of the landscape...for example places which are more complicated will be more populated with points or triangles and the ones closer to flat would have less of geometry...the best thing would be if the amount of the geometry could be overall also increased or decreased, depending on the needs...

thank you for help...

Views: 2044

### Replies to This Discussion

FILES? IMAGES?

I have been doing this before with ArcGis software...this is what i mean http://resources.arcgis.com/en/help/main/10.1/index.html#//00q90000...

the basic pointcloud from a grid of heights...here is with Z Tolerance:here is after VIP simplification according to significance in certain volume:

here it is after triangulation:

so you can see, that the more complex areas contain more information than the almost flat surfaces...of course this process can have many levels of precision...but the main idea remains...and my question would be...if something like this could be possible in rhinoceros...I have been trying to use random distribution or weaverbird but with not a satisfying result...here is the beginning and rough topography I have been working on...

Attachments:

What about the ReduceMesh command in Rhino?

--

David Rutten

david@mcneel.com

thank you...reducemesh was one of the possible options I have tried...however i dont really see into this command and upon what conditions it decides....could you maybe explain your second thought?

perfect grid of points is not always the case...the example model is obtained from a raster tiff data...however projection of perfect grid of points onto an arbitrary surface (eg.the example file) is not a problem...

Since the input is a perfect grid of points (is this always the case?) one simple approach would be to assign a value to each point which indicates how much it deviates from the elevation implied by its neighbours. If it's exactly at the average elevation it provides no real information and will thus get a low value. Then all you have to do is cull the points with the lowest assigned value. You can add some random noise to the values to avoid too many consecutive points from being culled, but those are already details.

--

David Rutten

david@mcneel.com

So, this is what I have produced so far...but still cannot figure out how to configure the deviation correctly...I would be really grateful for any help...thank you...

Jiri

Attachments:

I have neither Kangaroo nor Weaverbird installed here, I can't work with your file.

--

David Rutten

david@mcneel.com

I am sorry...here it is without weaverbird neither kangaroo...it might be a little messy at the end...due to my tryouts...

thank you david

Attachments:

• View All