 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
hobBIT wrote:
> last week I started to write a little ivy generator
Looks great!
> Please leave comments on how to increase realism
In the tower image, it looks like a large curved section
near the bottom is too distant from the wall to be attached?
Also, there is a lot of ivy on top of the objects while the
walls are more thinly covered. I'm no expert on ivy but this
seems a bit odd (but probably it just needs more starting
points to cover a wall).
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Christian Froeschlin <chr### [at] chrfr de> wrote:
> hobBIT wrote:
>
> > last week I started to write a little ivy generator
>
> Looks great!
>
> > Please leave comments on how to increase realism
>
> In the tower image, it looks like a large curved section
> near the bottom is too distant from the wall to be attached?
>
> Also, there is a lot of ivy on top of the objects while the
> walls are more thinly covered. I'm no expert on ivy but this
> seems a bit odd (but probably it just needs more starting
> points to cover a wall).
Thanx !
The biggest problem with my generator is the fact it is nearly uncontrollable, a
can weight the growing direction using random, last direction up and down
vectors, twig len and thickness, chance to split and last but not least the
seed base. Every little change results in a totally different "grow". I think,
some adjustments to these parameters will bring a better ivy :) ... Distant
wall<->ivy, you are right, I will make this controllable by 2 new parameters :)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Please leave comments on how to increase realism
Real ivy gets little hair roots where it attaches to things,
they're a little yellower than the main vine.
It looks perty good already though.
Here's a link with a good photo...
http://www.moplants.com/blog/?p=1081
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"hobBIT" <bla### [at] gmxde> schreef in bericht
news:web.4a4e17009a7f7b0e28c001c10@news.povray.org...
> last week I started to write a little ivy generator which can interact
> with
> every povray object. I was a bit inspired by the generator by Thomas Luft
> (http://graphics.uni-konstanz.de/~luft/ivy_generator/) which only work
> with
> mesh objects. Please leave comments on how to increase realism,
>
This is already looking really good! When you say 'every povray object' you
mean 'including mesh2'? That is what I understand.
Thomas
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Christian Froeschlin" <chr### [at] chrfrde> schreef in bericht
news:4a4e1f30$1@news.povray.org...
> Also, there is a lot of ivy on top of the objects while the
> walls are more thinly covered. I'm no expert on ivy but this
> seems a bit odd (but probably it just needs more starting
> points to cover a wall).
In some way, real ivy seems to concentrate on the top of walls/objects,
maybe because sunlight is stronger there. At least, that is what I observe
in my garden, so the ivy generator is doing a good job :-) Thomas Luft's
generator shows exactly the same phenomenon.
Thomas
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Plants will grow towards a light source, it's called phototropism (which I
learned about a long time ago and never thought to find a use for). you'll find
that in a dark room with light coming through a hole in the ceiling, vines will
grow along the floor toward the light, and will generally try to find the hole
in the ceiling. Perhaps your model could account for the direction of the light
source as well. It would be a matter of weighting the probability of a split in
a given direction by the amount of light in that direction or the probability of
reaching light by growing in that direction.
Otherwise, I think it's excellent work, and I wonder if it could be applied to
other plant life, like trees and shrubs.
Are you predetermining the starting location, or is it random?
A.D.B.
"Thomas de Groot" <tDOTdegroot@interDOTnlANOTHERDOTnet> wrote:
> "Christian Froeschlin" <chr### [at] chrfrde> schreef in bericht
> news:4a4e1f30$1@news.povray.org...
> > Also, there is a lot of ivy on top of the objects while the
> > walls are more thinly covered. I'm no expert on ivy but this
> > seems a bit odd (but probably it just needs more starting
> > points to cover a wall).
>
> In some way, real ivy seems to concentrate on the top of walls/objects,
> maybe because sunlight is stronger there. At least, that is what I observe
> in my garden, so the ivy generator is doing a good job :-) Thomas Luft's
> generator shows exactly the same phenomenon.
>
> Thomas
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Tim Attwood schrieb:
>> Please leave comments on how to increase realism
>
> Real ivy gets little hair roots where it attaches to things,
> they're a little yellower than the main vine.
>
> It looks perty good already though.
>
> Here's a link with a good photo...
> http://www.moplants.com/blog/?p=1081
Thanx, I didn't know this before, but I think this would increase the
scene complexity too much, but I will think about this :-)
hobBIT
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Anthony D. Baye schrieb:
> Plants will grow towards a light source, it's called phototropism (which I
> learned about a long time ago and never thought to find a use for). you'll find
> that in a dark room with light coming through a hole in the ceiling, vines will
> grow along the floor toward the light, and will generally try to find the hole
> in the ceiling. Perhaps your model could account for the direction of the light
> source as well. It would be a matter of weighting the probability of a split in
> a given direction by the amount of light in that direction or the probability of
> reaching light by growing in that direction.
>
> Otherwise, I think it's excellent work, and I wonder if it could be applied to
> other plant life, like trees and shrubs.
>
> Are you predetermining the starting location, or is it random?
>
> A.D.B.
>
> "Thomas de Groot" <tDOTdegroot@interDOTnlANOTHERDOTnet> wrote:
>> "Christian Froeschlin" <chr### [at] chrfrde> schreef in bericht
>> news:4a4e1f30$1@news.povray.org...
>>> Also, there is a lot of ivy on top of the objects while the
>>> walls are more thinly covered. I'm no expert on ivy but this
>>> seems a bit odd (but probably it just needs more starting
>>> points to cover a wall).
>> In some way, real ivy seems to concentrate on the top of walls/objects,
>> maybe because sunlight is stronger there. At least, that is what I observe
>> in my garden, so the ivy generator is doing a good job :-) Thomas Luft's
>> generator shows exactly the same phenomenon.
>>
>> Thomas
>
>
>
>
At the moment, growing is weighted by 4 factors:
1. follow direction of last segment
2. random
3. gravity
4. "up",
while the last can be seen as growing towards light. It's easy to add
another weight which grows to a given vector, but growing to "light" can
be very complex, depending on scene complexity, maybe a path can help
out to accomplish this.
The starting point and primary growing direction are input parameters.
I think about a total replace of the current growing algo, as it always
clamps to any object, so hanging ivy is not really possible atm.
hobBIT
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thomas de Groot schrieb:
> "hobBIT" <bla### [at] gmxde> schreef in bericht
> news:web.4a4e17009a7f7b0e28c001c10@news.povray.org...
>> last week I started to write a little ivy generator which can interact
>> with
>> every povray object. I was a bit inspired by the generator by Thomas Luft
>> (http://graphics.uni-konstanz.de/~luft/ivy_generator/) which only work
>> with
>> mesh objects. Please leave comments on how to increase realism,
>>
> This is already looking really good! When you say 'every povray object' you
> mean 'including mesh2'? That is what I understand.
>
> Thomas
>
>
Thanx :-),
Imho mesh2 should work, I'm not that pov expert. The algorithm is quite
easy:
- calculate direction from current position, this can be manipulated by
different weight factors
- find nearest intersection to the given object instance, this is done
using 'trace', so it should work with mesh2
(but this can heavily increase parsing time, I've implemented a
little macro which searches in all directions, so 96 traces for every
twig step can result in horrible parsing times if objects are too
complex, I thought about saving this to a file, so re-reandering with
the same params will go much faster)
- the resulting end position is than moved to this intersection
position, leaving a little distance into normal direction of the
intersection
hobBIT
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
hobBIT schrieb:
> last week I started to write a little ivy generator which can interact with
> every povray object. I was a bit inspired by the generator by Thomas Luft
> (http://graphics.uni-konstanz.de/~luft/ivy_generator/) which only work with
> mesh objects. Please leave comments on how to increase realism,
>
> many thanks,
> hobBIT
>
>
> ------------------------------------------------------------------------
>
If someone is willing to try it, here's the current state, info on how
to use it can be found in ivytest.pov, near eof around the macro call
itself, good luck and show us your results :-)
hobBIT
Post a reply to this message
Attachments:
Download 'ivygen.zip' (57 KB)
|
|
| |
| |
|
|
|
|
| |
