I'm not really sure whether this qualifies for erosion, but it produces
interesting terracing effects. 

As said in p.b.i. the algorithm is based on ingo's description of his
3d-erosion, but it is quite different when used on heightfields.  There is some
flattening effect on the bottom of the valleys, which is quite opposite to what
ingo's pictures show.

I did this without rescaling after every calculation, so the terrain really
lowers during the process.  

BTW, this is my first MPEG animation, i used avi2mpg1, not sure whether it got
right, so tell me please.  

Christoph

--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/

Post a reply to this message

Attachments:
Download 'i_ero1.mpg' (266 KB)

Yep, it's viewable.
I see squaring off in the lowest point near the camera and on the mountain
opposite that.  Maybe that's what you are calling terracing, makes sense if
so.  Resolution dependent or do you think it's something about the way
height fields are made to begin with?  I know they seem to be built up of
pairs of triangles arranged in a squared grid.

Bob

Post a reply to this message

>As said in p.b.i. the algorithm is based on ingo's description of his
>3d-erosion, but it is quite different when used on heightfields.  There
>is some flattening effect on the bottom of the valleys, which is quite
>opposite to what ingo's pictures show.

Except that some slopes get steeper, it doesn't look unnatural. 'Think the 
big difference is that a voxel represents the hardness/softness of the 
material, where on a HF a pixel only represents the hight. When a mountain 
is young, it's height says "nothing" about it durability.

Just an idea, how about using two different HF's, one for height and 
image, one for hardness. 
Shoot "waterdrop" at the first HF/pixel, check the value of the same pixel 
on the hardness HF, according to that value determine the amount of damage 
to be done to the first HF.

-- 
Photography: http://members.home.nl/ingoogni/
Pov-Ray    : http://members.home.nl/seed7/

Post a reply to this message

Bob Hughes wrote:
> 
> Yep, it's viewable.
> I see squaring off in the lowest point near the camera and on the mountain
> opposite that.  Maybe that's what you are calling terracing, makes sense if
> so.  Resolution dependent or do you think it's something about the way
> height fields are made to begin with?  I know they seem to be built up of
> pairs of triangles arranged in a squared grid.
> 

Not sure what you mean with "squaring off" but 3 points:

- the heightfield is generated with subdivision algorithm, that produces    
rectangular structures

- ingo's code uses horizontal/vertical movement and no diagonal.  

- HF resolution is only 256x256

maybe one of these is the reason for your observation.

Christoph

--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/

Post a reply to this message

ingo wrote:
> 
> Except that some slopes get steeper, it doesn't look unnatural. 'Think the
> big difference is that a voxel represents the hardness/softness of the
> material, where on a HF a pixel only represents the hight. When a mountain
> is young, it's height says "nothing" about it durability.
> 

That sounds reasonable.  

> Just an idea, how about using two different HF's, one for height and
> image, one for hardness.
> Shoot "waterdrop" at the first HF/pixel, check the value of the same pixel
> on the hardness HF, according to that value determine the amount of damage
> to be done to the first HF.
> 

That's worth some tries, even though, height and hardness array used at the
beginning must fit together somehow.  

Christoph

--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/

Post a reply to this message

do you deposit material as you erode?  do you do both wind and water
erosion?

Christoph Hormann wrote:
> 
> Bob Hughes wrote:
> >
> > Yep, it's viewable.
> > I see squaring off in the lowest point near the camera and on the mountain
> > opposite that.  Maybe that's what you are calling terracing, makes sense if
> > so.  Resolution dependent or do you think it's something about the way
> > height fields are made to begin with?  I know they seem to be built up of
> > pairs of triangles arranged in a squared grid.
> >
> 
> Not sure what you mean with "squaring off" but 3 points:
> 
> - the heightfield is generated with subdivision algorithm, that produces
> rectangular structures
> 
> - ingo's code uses horizontal/vertical movement and no diagonal.
> 
> - HF resolution is only 256x256
> 
> maybe one of these is the reason for your observation.
> 
> Christoph
> 
> --
> Christoph Hormann <chr### [at] gmxde>
> Homepage: http://www.schunter.etc.tu-bs.de/~chris/

Post a reply to this message

go for it!

Christoph Hormann wrote:
> 
> ingo wrote:
> >
> > Except that some slopes get steeper, it doesn't look unnatural. 'Think the
> > big difference is that a voxel represents the hardness/softness of the
> > material, where on a HF a pixel only represents the hight. When a mountain
> > is young, it's height says "nothing" about it durability.
> >
> 
> That sounds reasonable.
> 
> > Just an idea, how about using two different HF's, one for height and
> > image, one for hardness.
> > Shoot "waterdrop" at the first HF/pixel, check the value of the same pixel
> > on the hardness HF, according to that value determine the amount of damage
> > to be done to the first HF.
> >
> 
> That's worth some tries, even though, height and hardness array used at the
> beginning must fit together somehow.
> 
> Christoph
> 
> --
> Christoph Hormann <chr### [at] gmxde>
> Homepage: http://www.schunter.etc.tu-bs.de/~chris/

Post a reply to this message