So Ingo's link to that historical Voro++ image has piqued my interest in an SDL
version.

Here's my first iteration of a basic Voronoi cell generator in 3D, completely in
SDL. This is 100 seed points in a cubic volume, and was pretty slow to parse (19
minutes total render time, almost 18 of which was parsing). The output from the
algorithm is an array of cells, each cell an array of faces, each face an array
of lines, and each line an array of 2 points. The cells were then rendered as
simple meshes, each shrunk slightly towards its seed point.

The method is probably painfully naive and could benefit from all sorts of
optimization, which will probably be my next focus. However, this image is a
good proof of concept!

Bill

Post a reply to this message

Attachments:
Download 'cracklebobs3_02.png' (117 KB)

Preview of image 'cracklebobs3_02.png'

"Bill Pragnell" <bil### [at] hotmailcom> wrote:
> The method is probably painfully naive and could benefit from all sorts of
> optimization

Found an easy way to cut out most of the edge trimming loop. This image is 200
seeds, parsed in 4m30s, rendered in 90s.

I reckon there's more gainz to be had, onwards I go

Bill

Post a reply to this message

Attachments:
Download 'cracklebobs3_03.png' (149 KB)

Preview of image 'cracklebobs3_03.png'

"Bill Pragnell" <bil### [at] hotmailcom> wrote:
> "Bill Pragnell" <bil### [at] hotmailcom> wrote:
> > The method is probably painfully naive and could benefit from all sorts of
> > optimization
>
> Found an easy way to cut out most of the edge trimming loop. This image is 200
> seeds, parsed in 4m30s, rendered in 90s.
>
> I reckon there's more gainz to be had, onwards I go
>
> Bill

Hi Bill,

These are great and very interesting images, reminds me of the method I used for
my giants causeway image which I only managed with some good advice from these
forums. I think if I tried to create your fully 3D example my brain would break
;-)

Sean

Post a reply to this message

Le 2022-12-03 à 16:46, s.day a écrit :
> "Bill Pragnell" <bil### [at] hotmailcom> wrote:
>> "Bill Pragnell" <bil### [at] hotmailcom> wrote:
>>> The method is probably painfully naive and could benefit from all sorts of
>>> optimization
>>
>> Found an easy way to cut out most of the edge trimming loop. This image is 200
>> seeds, parsed in 4m30s, rendered in 90s.
>>
>> I reckon there's more gainz to be had, onwards I go
>>
>> Bill
> 
> Hi Bill,
> 
> These are great and very interesting images, reminds me of the method I used for
> my giants causeway image which I only managed with some good advice from these
> forums. I think if I tried to create your fully 3D example my brain would break
> ;-)
> 
> Sean
> 
> 
Personally, I'd rather use an isosurface for that task...

Post a reply to this message