"stm31415" <sam### [at] cscom> wrote:
> I'm being told to get offline, so I haven't gotten more than a page into the
> PDF, but I assume there is some way of determining how a plant is replaced?
> Can it depend on elevation, or water concentration? That is, do I get to
> make an image (or something) that would allow variations in the ecosystem,
> have a higher plant density at a creek and lower on rocky ground, with
> varaiation in the type of population in each area? It seems like it would
> be too hard to give the replacement a type given a probability map for the
> set of possible plants. I just going off the top of my head here, but this
> is really, really cool - it's got me excited.



Thank you!
In fact such a tool is always a fake. I try not to confuse it with reality -
it only has to simulate it. This is because of the control paramenters which
I cannot adjust in a correct way simply because I don't know them. By
example, what is the optimum soil humidity for nettles? At least nettles
seem to grow everywhere, so they can tolerate many conditions from sunny to
pure shady locations. Perhaps they tolerate conditions other plants simply
don't like, but aren't good in competing with other species. In this
respect my model won't be very different from a pure guess.

For water availability I got a tip of Prof. Deussen, one of the Xfrog
authors.
The function, which defines the ground can also serve as approximation to
water content.

Rocks are easy - they cast shadows. A simple image which defines shadowing
from rocks, big plants, houses and so on is sufficient to define smaller
and less dense vegetation near rocks. Near trees you have to consider the
roots too, which lower water availability at those points.

Probability maps are nice. In fact I used simple examples for some of my
images. But with a little practice you can see the difference to an
iterative approach.


Norbert Kern

Post a reply to this message

Norbert Kern wrote:
> 
> What is a spatial data structure?

A tree stucture like an octree/quadtree or BSP tree that makes it very 
efficient to access an element at a certain location.

Note in your case (and esp. for implementation in SDL) it might be the 
best approach to use a regular grid to store your locations in (or 
alternatively using spatial hashing if a grid would be too large).  This 
technique is widely used in molecular dynamics simulation.  What would 
be necessary for this to work is a limited density or the plants.  If a 
large number of plants can occur very close to each other that would be 
quite difficult to handle in SDL with a grid.

The idea is that with a grid cell size >= the maximum 'influence 
distance' of the plants only plants in neighboring grid cells can 
influence each other.  With a limited plant density that would only take 
constant time.

Christoph

-- 
POV-Ray tutorials, include files, Landscape of the week:
http://www.imagico.de/ (Last updated 01 Jun. 2006)
MegaPOV with mechanics simulation: http://megapov.inetart.net/

Post a reply to this message

>
> Thank you!
> In fact such a tool is always a fake. I try not to confuse it with reality -
> it only has to simulate it. This is because of the control paramenters which
> I cannot adjust in a correct way simply because I don't know them. By
> example, what is the optimum soil humidity for nettles? At least nettles
> seem to grow everywhere, so they can tolerate many conditions from sunny to
> pure shady locations. Perhaps they tolerate conditions other plants simply
> don't like, but aren't good in competing with other species. In this
> respect my model won't be very different from a pure guess.
>
> For water availability I got a tip of Prof. Deussen, one of the Xfrog
> authors.
> The function, which defines the ground can also serve as approximation to
> water content.
>
> Rocks are easy - they cast shadows. A simple image which defines shadowing
> from rocks, big plants, houses and so on is sufficient to define smaller
> and less dense vegetation near rocks. Near trees you have to consider the
> roots too, which lower water availability at those points.
>
> Probability maps are nice. In fact I used simple examples for some of my
> images. But with a little practice you can see the difference to an
> iterative approach.
>
>
> Norbert Kern

Yes, I understand/see the iterative process makes a major difference - what
I am wondering is if the method you are using takes into account both the
terrain and the neigboring plants, or just the latter. Anyway, now I have
time so I'll just hunker down and read the PDF.

As far as the reality of the control paramenters -- It doesn't really amtter
if they are the real numbers, does it? The look of the thing is in the
variation itself. You could use entirely fictional plants, and the process
would still give you a result that is pleasing to the eye, because it
defies the regularity of a less intensive approach. If you wanted it to be
*just* like the real world, well, I have a very nice camera you could
borrow ;) Seriously, though, it looks fantastic.

-s
5TF!

Post a reply to this message

"stm31415" <sam### [at] cscom> wrote:

> Yes, I understand/see the iterative process makes a major difference - what
> I am wondering is if the method you are using takes into account both the
> terrain and the neigboring plants, or just the latter.


Only plant competition was considered in WIP 1.
In the mean time I integrated water availability.


> As far as the reality of the control paramenters -- It doesn't really amtter
> if they are the real numbers, does it? The look of the thing is in the
> variation itself. You could use entirely fictional plants, and the process
> would still give you a result that is pleasing to the eye, because it
> defies the regularity of a less intensive approach. If you wanted it to be
> *just* like the real world, well, I have a very nice camera you could
> borrow ;) Seriously, though, it looks fantastic.


Thank you again. I fully agree with you.


Norbert Kern

Post a reply to this message

Not sure I understand exactly how it works but the outcome looks great..

Sean

Post a reply to this message

"Norbert Kern" <nor### [at] t-onlinede> wrote:
> Hi,
>
> I want to share the first results of my planned plant distribution tool.
>
> In principle it's an solvable problem, see
> http://graphics.stanford.edu/papers/ecosys/ecosys.pdf for a basic procedure
> example.
> First you have to check distances to all neighbors of each plant. Dependent
> of the distance and whether the neighbor is another species or not,
> increase or decrease the size of the plants. A plant will be replaced after
> several iterations or if the size becomes too small.
> The problem is to avoid checking each plant against each other on each
> iteration, because this leads to n*(n-1) checks, if n is the number of
> plants.

Wow, sounds like a neat project.  I have tp try to avoid things like that
these days, otherwise everything else I have to do suffers.  If you want to
do some more technical work on the n*(n-1) problem, have a look at this
recent thread on the random placment of non-intersecting spheres
(http://news.povray.org/povray.advanced-users/thread/%3C446b33a3@news.povray.org%3E/).
 Especially the code I put together which sets up a proximity search and
speeds things up enormously.  Good luck, and I can't wait to see more of
this in action.
-tgq

Post a reply to this message

"Trevor G Quayle" <Tin### [at] hotmailcom> wrote:

> Wow, sounds like a neat project.  I have to try to avoid things like that
> these days, otherwise everything else I have to do suffers.  If you want to
> do some more technical work on the n*(n-1) problem, have a look at this
> recent thread on the random placment of non-intersecting spheres
>
(http://news.povray.org/povray.advanced-users/thread/%3C446b33a3@news.povray.org%3E/).
>  Especially the code I put together which sets up a proximity search and
> speeds things up enormously.  Good luck, and I can't wait to see more of
> this in action.


Thank you for the tip - it will take time to understand the details.

Norbert Kern

Post a reply to this message

"s.day" <s.d### [at] uelacuk> wrote:
> Not sure I understand exactly how it works but the outcome looks great..
>
> Sean

Thank you, Sean.
It's a pity not to see an IRTC entry from you last round.


Norbert Kern

Post a reply to this message

that's fantastic. by far one of the most interesting posts here in a while.

Post a reply to this message

"Ross" <rli### [at] everestkcnet> wrote:
> that's fantastic. by far one of the most interesting posts here in a while.



Thank you very much! That's encouraging.


Norbert Kern

Post a reply to this message