In article <Xns### [at] netplexaussieorg>,
 ingo <ing### [at] tagpovrayorg> wrote:

> Is the pattern infinite?

It's infinite, but not very interesting far away from the points. As you 
can see in the examples I posted, it turns into uniform rays.


> A thought, triggerd by your pattern: In the crackle pattern, would it be 
> possible to make the position of the centeroids controlable by some 
> function? 

It certainly would be, though slower.


> I'm thinking of something like, put all the centeroids in the centre and 
> than give them a random deviation from there with a certain (Gauss) 
> distribution. 

I'm not sure what exactly you're thinking, but you pretty closely 
describe the first 2 images...just random points, more dense in the 
center.


> Or, make the position of the centeroid in the box depend on another 
> pattern. Imagine a wood pattern, in the light parts of it the centeroids 
> are "more random" where as in the dark parts they are very close to the 
> centre of the box. That would give bands of more and less uniform 
> crackles.
> Or, with the same wood pattern, in the lighter area's the centeroids are 
> closer to the sides of the box and in the darker area they ar closer to 
> the center.

I don't think this would have the effect you expect...you would just get 
a strong grid pattern where the points were less random. Varying point 
density would be more interesting. Or point importance...maybe if offset 
were controlled by a pattern...

Post a reply to this message

Christopher James Huff wrote:
> Here's a little patch I've been working on. It is very similar to the 
> crackle pattern, but rather than using a 3D grid of points that is 
> randomly jittered, it uses a user-specified set of points. It is 
> actually very fast, usually faster than the crackle pattern. It stores 
> the points in a binary space partition tree, which it can then search 
> very quickly, while the crackle pattern must always check a large number 
> of points, and sometimes recompute all of them. The options available 
> are the same as for the crackle pattern, with the addition of the 
> points. You can specify the points alone, or with a blend map index for 
> each point, giving you an effect similar to the "solid" feature, but 
> with much more control.

That looks very nice and offers a lot of interesting possibilities (for 
both textures and isosurfaces).  Could the speed improvements also be 
used for the regular crackle pattern?

Christoph

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 25 Oct. 2003 _____./\/^>_*_<^\/\.______

Post a reply to this message

In article <f0d### [at] tritonimagicode>,
 Christoph Hormann <chr### [at] gmxde> wrote:

> That looks very nice and offers a lot of interesting possibilities (for 
> both textures and isosurfaces).  Could the speed improvements also be 
> used for the regular crackle pattern?

Very unlikely. The crackle pattern stores a cache of 125 points in an 
array. Running through a BSP tree for those points would be faster than 
testing each of them for every evaluation of the pattern, but building 
that tree would be much slower than building the array. However, I have 
found several other, completely unrelated ways to speed up the crackle 
pattern...my updated version is a few percent faster with the default 
settings, and up to 6 times as fast with non-0 offset and a metric of 3. 
A pure 2D version would also be much faster...it would only need to 
handle points in 21 squares, rather than 125 cubes.

My voronoi patch is getting closer to finished...it now detects the case 
where all points are in an axis-aligned plane, and switches to a 2D 
partition tree. It will still break when there's a mixture of the 
cases...not all points coplanar, but a large group of coplanar points. 
I'm still thinking about the best way to handle that case...

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: <chr### [at] tagpovrayorg>
http://tag.povray.org/

Post a reply to this message

Christopher James Huff wrote:
> 
>>That looks very nice and offers a lot of interesting possibilities (for 
>>both textures and isosurfaces).  Could the speed improvements also be 
>>used for the regular crackle pattern?
> 
> 
> Very unlikely. The crackle pattern stores a cache of 125 points in an 
> array. Running through a BSP tree for those points would be faster than 
> testing each of them for every evaluation of the pattern, but building 
> that tree would be much slower than building the array.

I feared something like that.  But in many cases where the crackle 
pattern is nice to use only a relatively small area of the pattern space 
is actually evaluated.  In these cases i could imagine a system like this:

- initially build a BSP tree for all points within a certain area 
(possibly customizable with a parameter).
- when a point outside this area is evaluated use the old technique.
- when a certain number of points outside needed to be evaluated rebuild 
the BSP tree for a different/larger area.

I don't know if this idea is feasible or would be efficient of course.

Christoph

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 25 Oct. 2003 _____./\/^>_*_<^\/\.______

Post a reply to this message

In article <8ei### [at] tritonimagicode>,
 Christoph Hormann <chr### [at] gmxde> wrote:

> I feared something like that.  But in many cases where the crackle 
> pattern is nice to use only a relatively small area of the pattern space 
> is actually evaluated.  In these cases i could imagine a system like this:
> 
> - initially build a BSP tree for all points within a certain area 
> (possibly customizable with a parameter).

It would have to be customizeable with a parameter...POV has no way of 
determining where you are going to use the pattern.


> - when a point outside this area is evaluated use the old technique.

This much could be done with the Voronoi patch. Just use an object 
pattern with a box, the voronoi pattern with appropriately placed points 
inside the box, and the crackle pattern outside it. Computing the right 
points to get a smooth transition would be tricky though...a couple 
extra functions such as hash3d() would make it easier.


> - when a certain number of points outside needed to be evaluated rebuild 
> the BSP tree for a different/larger area.

This would require actual patch work. Figuring out how big the cache has 
to be to make it worth having, when to rebuild it, etc...an interesting 
problem.

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: <chr### [at] tagpovrayorg>
http://tag.povray.org/

Post a reply to this message

Christopher James Huff wrote:
>In article <3febe5bb$1@news.povray.org>, "Slime" <fak### [at] emailaddress>wrote:
>> [...]
>> Can they be 4D? =)
>
> I'd also like to implement a 4D noise pattern, which is something I've 
> seen used for things like flames...

I've made a cute swirling smoke using 4D noise from this guy... :-)

http://www.slimeland.com/slimepov/

Post a reply to this message