Thomas de Groot wrote:
> "stbenge" <THI### [at] hotmailcom> schreef in bericht 
>> All I'm really doing is using trace() on a height_field and updating the
>> vectors to correspond with the surface normal. The objects are cylinders
>> which grow bigger the longer they are "alive."
> 
> Very neat! Shouldn't cones not be more appropriate than cylinders in this 
> context?

Yes, of course, but I wanted to test the proof-of-concept first, and 
worry about cone-sphere connection issues later. As it is, this project 
isn't working out too well...

Sam

Post a reply to this message

Tek wrote:
> Ah of course! That's a really nice effect already but I look forward to 
> seeing the "Alaska style" results :)

Well, you might be looking forward for a long time, because this 
technique isn't working out! All I've managed to do is scratch the 
surface down the sides of the height_field. Maybe I should post the 
source so other people can look at it. A free, fully-automated erosion 
setup producing 16-bit gray scale height maps... that would be nice.

Sam

Post a reply to this message

Christian Froeschlin wrote:
> stbenge wrote:
> 
>> There seems to be a lack of good, free implementations of it
>> anywhere...
> 
> The basic edition of World Machine is free, although the
> output resolution is then restricted to 512x512. Still, could
> be enough to get the basic erosion done and then add some
> finer details in POV using f_ridge.

That's a good idea, especially since it seems POV 3.7b handles 16-bit 
gray scale images better. I discovered yesterday that you can 
encapsulate them inside pigment statements without them turning back 
into black-green green-red ramp wave nightmares. It makes the 
possibilities open up ;)

Sam

Post a reply to this message

stbenge <THI### [at] hotmailcom> wrote:
> Tek wrote:
> > Ah of course! That's a really nice effect already but I look forward to
> > seeing the "Alaska style" results :)
>
> Well, you might be looking forward for a long time, because this
> technique isn't working out! All I've managed to do is scratch the
> surface down the sides of the height_field. Maybe I should post the
> source so other people can look at it. A free, fully-automated erosion
> setup producing 16-bit gray scale height maps... that would be nice.
>
> Sam



Suddenly I am reminded of playing with the 'flow' command in hf_lab in an
attempt to build eroded height fields... I had limited success.  I am very
interested in your technique(s), though.  I've wondered whether a decent
erosion routine could be made into functions for use with isosurface based
landscapes.

-Reactor

Post a reply to this message

Reactor wrote:
> Suddenly I am reminded of playing with the 'flow' command in hf_lab in an
> attempt to build eroded height fields... I had limited success.

That's been my experience with everything I've tried. World Machine has 
good erosion, but the free version limits the size of the map :( 
Christian Froeschlin has a good idea for making use of the limited 
files. The way POV now handles 16-bit grayscale images makes his idea 
even easier to implement, I think.

> I am very
> interested in your technique(s), though.  I've wondered whether a decent
> erosion routine could be made into functions for use with isosurface based
> landscapes.

Anything can be done, but at what cost? Isosurfaces are already slow, 
and to incorporate erosion... well, I hope you won't be using your 
computer for a while :)

I'd rather make it work with height_fields, but in a more dynamic way 
than it is now. I don't know how to do this yet, though.

Sam

Post a reply to this message

stbenge <THI### [at] hotmailcom> wrote:
..
..
..
> Anything can be done, but at what cost? Isosurfaces are already slow,
> and to incorporate erosion... well, I hope you won't be using your
> computer for a while :)
>
> I'd rather make it work with height_fields, but in a more dynamic way
> than it is now. I don't know how to do this yet, though.
>
> Sam


The reason why I am interested in isosurfaces as landscapes is their resolution
independence and the fact that they permit overhangs.  As fast and simple as
the heightfield is, certain landscapes cannot be done well with them.  I've
developed a fascination with canyon landscapes recently, and I'm always on the
lookout for alternate techniques.

Attached is a picture created using meshes (which does ok except getting the
texture to match the rock layers is very difficult.)

-Reactor

Post a reply to this message

Attachments:
Download 'cliffscene6ft.jpg' (81 KB)

Preview of image 'cliffscene6ft.jpg'

"Reactor" <rea### [at] hotmailcom> schreef in bericht 
news:web.48a09d3e4025b6ec9bfe5600@news.povray.org...
>
> The reason why I am interested in isosurfaces as landscapes is their 
> resolution
> independence and the fact that they permit overhangs.  As fast and simple 
> as
> the heightfield is, certain landscapes cannot be done well with them. 
> I've
> developed a fascination with canyon landscapes recently, and I'm always on 
> the
> lookout for alternate techniques.
>
> Attached is a picture created using meshes (which does ok except getting 
> the
> texture to match the rock layers is very difficult.)
>

Nice one indeed! Good work.
The texture is indeed not truly right here. UV mapping?

Thomas

Post a reply to this message

"Thomas de Groot" <t.d### [at] internlDOTnet> wrote:

> Nice one indeed! Good work.
> The texture is indeed not truly right here. UV mapping?
>
> Thomas

No, unfortunately.  I was having trouble with the Wings uv mapping export at the
time.  I might revisit it at another point, but it was really an experiment into
finding a good way of making overhangs in landscapes.  The next thing I want to
try is something my brother described: a six-sided heightfield in which 6
images determine the shape of the landscape.  I am not sure how to implement
that yet, but that and another idea of blending heightfields and meshes are
things we want to test out and see if they would be feasible.

-Reactor

Post a reply to this message

"Reactor" <rea### [at] hotmailcom> schreef in bericht 
news:web.48a1e2994025b6ec993f4ab30@news.povray.org...
>
> No, unfortunately.  I was having trouble with the Wings uv mapping export 
> at the
> time.  I might revisit it at another point, but it was really an 
> experiment into
> finding a good way of making overhangs in landscapes.  The next thing I 
> want to
> try is something my brother described: a six-sided heightfield in which 6
> images determine the shape of the landscape.  I am not sure how to 
> implement
> that yet, but that and another idea of blending heightfields and meshes 
> are
> things we want to test out and see if they would be feasible.
>

I see. I have not been overly enthousiastic about Wings uv mapping. Nowadays 
I prefer Silo to it (not free, however).

In the past, I have made objects composed of different heightfields, unioned 
and differenced together. In particular, a six-sided tower! That works very 
well indeed. Overhangs can also be created by heightfields placed 
vertically. Differenced with a horizontal heightfield on top, and you have a 
(ragged)cliff.

I think that blending of heightfields together and/or with meshes is not 
difficult in itself, the problem is the seam which will often show up as a 
fairly straight line under some viewing angles. That can be hidden to some 
degree by an appropriate global texture or by carefully placed objects in 
the line of sight, but it is tricky.

Thomas

Post a reply to this message

Some sort of trees seen from below, against the sunny sky..?

Post a reply to this message