algorithmic modeling for Rhino
Built a little GH definition to build a 3d mesh up from sequential image layers from a CT scan. relies on millipede, anemone, and human.
I am interested in being able to display the images on screen and then obtain images of sections through the volumetric space defined by the images. Many thanks
Thanks for posting this a ways back. Been working on something similar. Thanks for sharing!
Would you be opposed to sharing this definition? Thanks, John
Ah ok i think i follow. Thanks for explanation. I wasn't sure why you were using Anemone, but that makes sense and i suppose even if this wouldn't crash it, you can now apply it to more and more complex files. How many vertices are in final mesh?
Just to make sure i'm not doing it wrong, do you find its the generation of the field values to be the computationally hungry element? not the isosurface itself.
For my mesh of 100k vertices the isosurface=500ms but field generation is 1.5minutes.
Hi Nick -
The beauty of marching cubes (the algorithm at the heart of millipede's isosurface) is that the mesh geometry in any given cubic cell is dictated ONLY by the field values at the vertices immediately around it. This means you can solve it layer after layer, and every subsequent layer will be perfectly continuous with the one adjacent to it. For each layer, I'm sampling two CT scan images at a time, so I have two layers of points to generate one mesh "slice" - and the top points for one layer naturally become the bottom points for the next layer.
I could have obtained the same geometric result by simply feeding the entire 3d grid of points into millipede all at once - but my suspicion was that this would crash my machine given the density at which I was sampling the source images.
Andrew i have recently starting using Millipede, specifically the isosurface component. Im enjoying using it, but cant understand how you use it this way. Could you please explain the relationship between Anemone and Millipede.
Is the mesh 'closed' after each additional layer? If so is the whole mesh recalculated with each iteration?
Or do you keep the top of the mesh 'open' with maybe the box input, then some how generate next marching cubes layer off that open mesh edge?
They're both bone scans, taken from this website: http://isbweb.org/data/vsj/
what is the object?
Sign Upor Sign In
by Naoto Tamura
by Laurent Delrieu
by Parametric House
Added by Naoto Tamura
Added by Naoto Tamura
Added by Parametric House
Added by Andres Gonzalez
Added by Andrea Di Rosa
Added by Alva Sondakh
© 2018 Created by Scott Davidson.
Report an Issue |
Terms of Service
Please check your browser settings or contact your system administrator.