POV-Ray : Newsgroups : povray.general : 3D convex hull ... of sorts Server Time
3 Oct 2025 01:33:14 EDT (-0400)
  3D convex hull ... of sorts (Message 1 to 2 of 2)  
From: Bald Eagle
Subject: 3D convex hull ... of sorts
Date: 18 Sep 2025 20:05:00
Message: <web.68cc9d25a1d8890f1f9dae3025979125@news.povray.org>
After Kenneth's recent post, I was contemplating the structure of meshes, and
how to mesh a raw point cloud, and my brain started connecting Voronoi with the
gift-wrapping algorithm, and somewhere in there I thought it would be
interesting to simply see the intersection of the prisms generated by projecting
the points onto the 3 cardinal planes.

An even, regular distribution of a lot of random points basically gave me a cube
with uneven edges and corners, so I cut down the number of points, and then
anisotropically scaled the random coordinates.

I think that a lot of points randomly generated inside a more complex object
would yield more interesting results.

As would intersecting more than 3 prisms.

But it was just playing with the proof-of-concept of the idea.

Kind of an axis-aligned bounding prism, if you will.

Enjoy.

- BE


Post a reply to this message


Attachments:
Download 'projectedpointcloud.png' (69 KB)

Preview of image 'projectedpointcloud.png'
projectedpointcloud.png


 

From: kurtz le pirate
Subject: Re: 3D convex hull ... of sorts
Date: 19 Sep 2025 09:55:24
Message: <68cd60cc$1@news.povray.org>
On 19/09/2025 02:00, Bald Eagle wrote:
> After Kenneth's recent post, I was contemplating the structure of meshes, and
> how to mesh a raw point cloud, and my brain started connecting Voronoi with the
> gift-wrapping algorithm, and somewhere in there I thought it would be
> interesting to simply see the intersection of the prisms generated by projecting
> the points onto the 3 cardinal planes.
> 
> An even, regular distribution of a lot of random points basically gave me a cube
> with uneven edges and corners, so I cut down the number of points, and then
> anisotropically scaled the random coordinates.
> 
> I think that a lot of points randomly generated inside a more complex object
> would yield more interesting results.
> 
> As would intersecting more than 3 prisms.
> 
> But it was just playing with the proof-of-concept of the idea.
> 
> Kind of an axis-aligned bounding prism, if you will.
> 


I'm surprised you brought that up !

I've been studying this same topic since this morning, and I'm currently 
studying Leo McCormack's code on GitHub.
<https://github.com/leomccormack/convhull_3d>


LOL


-- 
kurtz le pirate
compagnie de la banquise


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.