So, I had mentioned releasing the Liquid-Surface-System
to the WWW this weekend, but stumbled upon a little
unpleasing artifact which may occur in tight corners,
especially when the wave is created there, e.g. using
the Stomp-Macro, which places a wave around all objects
interacting with the water.

Anyways, as you can see in the edge at the red box, there
is some weird jiggling going on, and I'm working on the
dampening and internal multiplying parts (which take
care of dissipation of waves and retaining the hills) to
work slightly different dependant on amount of neighbour-
nodes.

Anyways, this animation isn't to show off any features or
such, its an excuse why its taking so long... :-)

I was thinking about adding an algorithm which checks
if there is a checker-like pattern (all horizontal and vertical
neighbours are of a much different height than the one in
question) and adjusts the height of the center node to a
certain degree. This would add a whole new loop, because
I'd have to do this after the step has been generated completely,
and needs to take dead and rim nodes into account, which
implies that I'd have to put a larger effort into that, then a
next bug-testing phase... Would result in at least another two
or three weeks of programming. Well, I'm thinking about an
easier solution right now, and hope to get something useful
till the weekend. Sorry for the delay!

Regards,
Tim

-- 
Tim Nikias v2.0
Homepage: http://www.digitaltwilight.de/no_lights
Email: Tim### [at] gmxde

Post a reply to this message

Attachments:
Download 'turb_err.mpg' (363 KB)

Um?

I'm not sure its wrong. I think you have chosen a starting point that is
very unlikely. I think the block and the side form a sort of channel. It
looks like you have started with water up the side of the block. This is a
bit strange. Channels usually have waves going up and down them and not all
the water piled up on one bank.

I think if you chose a better starting point it would be better. Perhaps
take one of the surfaces later in the render where the problem has died down
and feed it back in as a starting point multiplied up a bit.

You will probably get similar funny effects if you start with the water in a
cube shape and just "let it go".


Ta BB.
---------------------------------------------------------------------------
 Bill Bagshaw                                     Tel +44 (0) 117 929 9733
 SN Systems Limited                               Fax +44 (0) 117 929 9251
 sup### [at] snsyscom                                         sal### [at] snsyscom
---------------------------------------------------------------------------
The views expressed in this message do not necessarily constitute the views
of SN Systems Ltd and information in this message is confidential and may
be privileged.  It is intended solely for the person to whom it is
addressed.  If you are not the intended recipient, please notify the sender
and please delete the message from your system immediately.
---------------------------------------------------------------------------
"Tim Nikias v2.0" <tim### [at] gmxde> wrote in message
news:3eb78ed6@news.povray.org...
> So, I had mentioned releasing the Liquid-Surface-System
> to the WWW this weekend, but stumbled upon a little
> unpleasing artifact which may occur in tight corners,
> especially when the wave is created there, e.g. using
> the Stomp-Macro, which places a wave around all objects
> interacting with the water.
>
> Anyways, as you can see in the edge at the red box, there
> is some weird jiggling going on, and I'm working on the
> dampening and internal multiplying parts (which take
> care of dissipation of waves and retaining the hills) to
> work slightly different dependant on amount of neighbour-
> nodes.
>
> Anyways, this animation isn't to show off any features or
> such, its an excuse why its taking so long... :-)
>
> I was thinking about adding an algorithm which checks
> if there is a checker-like pattern (all horizontal and vertical
> neighbours are of a much different height than the one in
> question) and adjusts the height of the center node to a
> certain degree. This would add a whole new loop, because
> I'd have to do this after the step has been generated completely,
> and needs to take dead and rim nodes into account, which
> implies that I'd have to put a larger effort into that, then a
> next bug-testing phase... Would result in at least another two
> or three weeks of programming. Well, I'm thinking about an
> easier solution right now, and hope to get something useful
> till the weekend. Sorry for the delay!
>
> Regards,
> Tim
>
> --
> Tim Nikias v2.0
> Homepage: http://www.digitaltwilight.de/no_lights
> Email: Tim### [at] gmxde
>
>
>

Post a reply to this message

"Tim Nikias v2.0" wrote:
> 
> So, I had mentioned releasing the Liquid-Surface-System
> to the WWW this weekend, but stumbled upon a little
> unpleasing artifact which may occur in tight corners,
> especially when the wave is created there, e.g. using
> the Stomp-Macro, which places a wave around all objects
> interacting with the water.
> 
> Anyways, as you can see in the edge at the red box, there
> is some weird jiggling going on, and I'm working on the
> dampening and internal multiplying parts (which take
> care of dissipation of waves and retaining the hills) to
> work slightly different dependant on amount of neighbour-
> nodes.

I don't remember how exactly the rims are handled in that method but in
real life you have the effect of stronger dissipation at the rims because
of friction between the liquid and the solid surface.  I am not sure what
would be the best method to handle this effect but you could use stronger
damping in all elements neighbored to a rim element.  

Another thing that might be worth trying: making the smoothing step
non-linear.  If the waves become very high they will probably spread out
and lower more quickly than when they are very small.  

Christoph  

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 28 Feb. 2003 _____./\/^>_*_<^\/\.______

Post a reply to this message

You're answer doesn't make much sense to me,
which I assume originates in a misunderstanding
of what the animation is supposed to show.

The System is a set of Macros which will simulate
liquid surfaces like the one presented. These are
User-Defined but refer to a rectangular shape, and
are based on nodes.
To add waves into the simulation, there are several
macros, which add either a drop, or entire wave-fronts.

"Stomp" is one of the wave-front types, it will add
a wave surrounding all objects in the water and
interacting with it.

In this case, a problem becomes apparent: in tight
corners, the step-calculations run into trouble
when there are too few neighbours for useful smoothening
of the waves. That's something I want to avoid before
releasing the system.

The scene itself is just a test-scene to check the
algorithms and if they're working properly, its not an
actual scene I'd use in an animation.


-- 
Tim Nikias v2.0
Homepage: http://www.digitaltwilight.de/no_lights
Email: Tim### [at] gmxde

> Um?
>
> I'm not sure its wrong. I think you have chosen a starting point that is
> very unlikely. I think the block and the side form a sort of channel. It
> looks like you have started with water up the side of the block. This is a
> bit strange. Channels usually have waves going up and down them and not
all
> the water piled up on one bank.
>
> I think if you chose a better starting point it would be better. Perhaps
> take one of the surfaces later in the render where the problem has died
down
> and feed it back in as a starting point multiplied up a bit.
>
> You will probably get similar funny effects if you start with the water in
a
> cube shape and just "let it go".
>
>
> Ta BB.
> --------------------------------------------------------------------------
-
>  Bill Bagshaw                                     Tel +44 (0) 117 929 9733
>  SN Systems Limited                               Fax +44 (0) 117 929 9251
>  sup### [at] snsyscom                                         sal### [at] snsyscom
> --------------------------------------------------------------------------
-
> The views expressed in this message do not necessarily constitute the
views
> of SN Systems Ltd and information in this message is confidential and may
> be privileged.  It is intended solely for the person to whom it is
> addressed.  If you are not the intended recipient, please notify the
sender
> and please delete the message from your system immediately.
> --------------------------------------------------------------------------
-
> "Tim Nikias v2.0" <tim### [at] gmxde> wrote in message
> news:3eb78ed6@news.povray.org...
> > So, I had mentioned releasing the Liquid-Surface-System
> > to the WWW this weekend, but stumbled upon a little
> > unpleasing artifact which may occur in tight corners,
> > especially when the wave is created there, e.g. using
> > the Stomp-Macro, which places a wave around all objects
> > interacting with the water.
> >
> > Anyways, as you can see in the edge at the red box, there
> > is some weird jiggling going on, and I'm working on the
> > dampening and internal multiplying parts (which take
> > care of dissipation of waves and retaining the hills) to
> > work slightly different dependant on amount of neighbour-
> > nodes.
> >
> > Anyways, this animation isn't to show off any features or
> > such, its an excuse why its taking so long... :-)
> >
> > I was thinking about adding an algorithm which checks
> > if there is a checker-like pattern (all horizontal and vertical
> > neighbours are of a much different height than the one in
> > question) and adjusts the height of the center node to a
> > certain degree. This would add a whole new loop, because
> > I'd have to do this after the step has been generated completely,
> > and needs to take dead and rim nodes into account, which
> > implies that I'd have to put a larger effort into that, then a
> > next bug-testing phase... Would result in at least another two
> > or three weeks of programming. Well, I'm thinking about an
> > easier solution right now, and hope to get something useful
> > till the weekend. Sorry for the delay!
> >
> > Regards,
> > Tim
> >
> > --
> > Tim Nikias v2.0
> > Homepage: http://www.digitaltwilight.de/no_lights
> > Email: Tim### [at] gmxde
> >
> >
> >
>
>

Post a reply to this message

That is an idea which I've already had and am currently
experimenting with. The problem is I'd like to avoid to
check even more neighbours than the original 8, and
the problem only enters in tight corners, so there's
actually no need to widen the smoothening radius. But
using different values dependant on neighbours is definitely
what I'm trying to do.

The problem with adjusting the smoothing in relation to
the heights is that waves overlap all the time, if I'd dampen
them more, that would be counter-realistic, as waves just
travel through each other (unless they crash together,
which isn't, as you probably know, possible with the algorithm).

Thanks for the suggestion anyways!

-- 
Tim Nikias v2.0
Homepage: http://www.digitaltwilight.de/no_lights
Email: Tim### [at] gmxde

>
>
> "Tim Nikias v2.0" wrote:
> >
> > So, I had mentioned releasing the Liquid-Surface-System
> > to the WWW this weekend, but stumbled upon a little
> > unpleasing artifact which may occur in tight corners,
> > especially when the wave is created there, e.g. using
> > the Stomp-Macro, which places a wave around all objects
> > interacting with the water.
> >
> > Anyways, as you can see in the edge at the red box, there
> > is some weird jiggling going on, and I'm working on the
> > dampening and internal multiplying parts (which take
> > care of dissipation of waves and retaining the hills) to
> > work slightly different dependant on amount of neighbour-
> > nodes.
>
> I don't remember how exactly the rims are handled in that method but in
> real life you have the effect of stronger dissipation at the rims because
> of friction between the liquid and the solid surface.  I am not sure what
> would be the best method to handle this effect but you could use stronger
> damping in all elements neighbored to a rim element.
>
> Another thing that might be worth trying: making the smoothing step
> non-linear.  If the waves become very high they will probably spread out
> and lower more quickly than when they are very small.
>
> Christoph
>
> -- 
> POV-Ray tutorials, include files, Sim-POV,
> HCR-Edit and more: http://www.tu-bs.de/~y0013390/
> Last updated 28 Feb. 2003 _____./\/^>_*_<^\/\.______

Post a reply to this message

Looks like unstable behaviour to me, but I might be wrong. Maybe with higher
order integration methods or with noise filters you could do something.

Post a reply to this message

It is a stable behaviour, but an unexpected side-effect
in tight corners. I've tried to get rid of this, but didn't want
to implement a parsing-intensive check for the
checker-pattern. Might do that as an addition to clean
the buffer, with options to modify the buffer on the
go. I did find a solution which will at least eliminate
this problem in VERY tight corners, using a higher
dampening. The effect only gets introduced in
corners anyways, and seems to get erased with a
wave. Shouldn't appear to often, I think.

-- 
Tim Nikias v2.0
Homepage: http://www.digitaltwilight.de/no_lights
Email: Tim### [at] gmxde

> Looks like unstable behaviour to me, but I might be wrong. Maybe with
higher
> order integration methods or with noise filters you could do something.
>
>

Post a reply to this message