Chris B wrote:
> 
> "aelgoa" <wra### [at] gmailcom> wrote in message 
> news:web.498811b6f0bf7a9c626c89e40@news.povray.org...
> 
>> i'm trying to wrap several dozens of height-fields around a curved shape
>> (similar to an S-shape)
>>
>> how would i go about this?
>> do i use a spline or an iso-surface? and how do i correct the 
>> orientation?
>>
>> the result should look as if the shape has an irregular "skin" or "hair"
>> covering the entire (cylindrical) surface
>>
>> thank you in advance
>>
> 
> Nobody seems to have taken up on this question so I'll give it a shot.
> 
> You can use a heightfield as a function in an isosurface so, if you can 
> define the cylindrical S-shape you want as a function, you may be able 
> to suitably combine a single heightfield with  the S-shape. Mike 
> Williams' isosurface tutorial may give you some inspiration for this 
> sort of thing.
> 
> I can't think how you'd go about wrapping several dozen height fields 
> around such a shape because it would be difficult to imagine how you 
> could get them to mesh together in any sort of a decent way.
> 


If the curved surface is "cylindrical" meaning it snakes through two 
dimensions, but is constant in the third then I can picture it but it 
would be quite invovled.

I would picture the surface first as a flat 'ribbon' that is then 
adapted to the curve by slicing it into arbitrary tiles that would 
correspond to segments along the actual curved surface. Each tile would 
correspond to a separate hf used.  The tiles could be laid along the 
surface in arbitrary segments for which the anchor points and normals 
would be known.  Hf's could be oriented with the normals and translated 
to fit the anchor points of each segment.

Each hf could be 'elevated' with a pattern before being translated to 
its place on the surface.  The pattern could be made continuous and 
coherent at the edges by 'applying' the pattern to the ribbon of hf's 
while the 'ribbon' is flat, so to speak, in a primary plane.  To do this 
you'd have to translate the pattern, by sucessive tile lengths, in the 
plane, before creating the hf with it, since you can only apply the 
pattern in the 1 unit box at the origin.  This is doable though a bit 
mind bending because if the way the hf function reverses the axis.  Once 
the hf is created and scaled if necessary it is oriented and translated 
to the receiving curved surface.

The hf would not of course reflect the pattern as it intersects  the 
curved surface in space, but rather how it would intersect the primary 
plane used along the 'ribbon'.

Post a reply to this message