 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Rune <run### [at] mobilixnet dk>
> How was this done? One mesh per frame? How long was the animation? And how
> much disc space did it take?
The horse movement was created in Poser. It's cyclical and uses 24 frames.
Each mesh was exported and converted (using the batch procedure of 3DWin).
There was 50 Mb of meshes. The POV procedure loads each mesh and puts it
where it should be. I never finished it for time reasons and the best I did
was a 400 frame animation of the whole herd which is far from perfect. I
plan to redo it with a mocap file anyway.
> Would it have been possible to make the horses run over a height field and
> other such demanding interactions with the surroundings?
Running on a height field can be possible using the trace function. Other
interactions could be simulated within Poser using dummy props and then
exported.
> And yes, I *do* believe quality character animation can be done writing
> numbers in a text editor. With the proper macros of course.
My own *** limited *** experience with animation indicates that it's already
very difficult to do with a GUI, as you need precise visual references to
adjust the speed and movements, and for the most complicated stuff a mo-cap
file becomes necessary. A Povray-only boning system could certainly be
implemented and fun to use but only for a limited number of users in a
limited number of situations (movements that can be described
algorithmically). For starters, I suggest that you try animating a
walking/running dog-like quadruped (not a biped !) using simple cylinders to
see what I mean. And make it a system flexible enough to be adaptable and
usable by other people...
Now, if Povray was able to communicate smoothly with apps providing this
sort of service, that would open very different perspectives.
G.
--
**********************
http://www.oyonale.com
**********************
Graphic experiments
Pov-ray gallery
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Nekar Xenos" wrote:
> I think it's a good idea, considering there are so few
> Pov animation entries on IRTC. It might just help to get
> more animations on there.
Exactly, that was one of my considerations too.
Rune
--
3D images and anims, include files, tutorials and more:
Rune's World: http://rsj.mobilixnet.dk (updated June 26)
POV-Ray Users: http://rsj.mobilixnet.dk/povrayusers/
POV-Ray Webring: http://webring.povray.co.uk
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Christoph Hormann" wrote:
> Just like to point out that deforming meshes
> (meaning translating the vertices in a certain
> defined way) is really not much problem in current
> pov/megapov. Just define a vector field in a user
> defined function or pigment and use it to translate
> the vertices. It will work like displace warps for
> patterns.
While that kind of mesh deformation can be use for very simple deformations,
it definitely can't be used for things like character animation. I'd like to
see you define a pigment or function that deforms a character mesh so that
the right arm bends in this way, the left arm in this, the legs in that way,
and the head looks to the right!
Besides, it doesn't work for smooth meshes.
> just imagine blobs could be nicely 'boned' too
No they couldn't. But bezier pathes could perhaps.
Rune
--
3D images and anims, include files, tutorials and more:
Rune's World: http://rsj.mobilixnet.dk (updated June 26)
POV-Ray Users: http://rsj.mobilixnet.dk/povrayusers/
POV-Ray Webring: http://webring.povray.co.uk
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Gilles Tran" wrote:
> The horse movement was created in Poser. It's cyclical and
> uses 24 frames. There was 50 Mb of meshes.
50 MB for just one second of animation seems like a pretty inefficient
solution to me!
> Running on a height field can be possible using the trace function.
Well, didn't mean a smooth flat height field but one with a rather bumpy
ground so each foot have to adapt to the ground...
> Other interactions could be simulated within Poser using
> dummy props and then exported.
Again, I think this works only in a limited extent...
> My own *** limited *** experience with animation indicates
> that it's already very difficult to do with a GUI, as you
> need precise visual references to adjust the speed and
> movements, and for the most complicated stuff a mo-cap
> file becomes necessary. A Povray-only boning system could
> certainly be implemented and fun to use but only for a
> limited number of users
Well, users could choose to animate the bones in a different program and
then export the animation of the bones to an include file. The mesh
deformation in POV-Ray would then be controlled by the bones in that include
file. The advantage here is that it would only require about 10 to 50 lines
of code per character per frame, much much less than exporting a whole mesh
per frame.
I'm sure the proper converter programs would pop up over time.
> in a limited number of situations (movements that can be
> described algorithmically).
Splines are available in POV-Ray.
> For starters, I suggest that you try animating a walking/running
> dog-like quadruped (not a biped !) using simple cylinders to see
> what I mean. And make it a system flexible enough to be adaptable
> and usable by other people...
I've already made a walk and a running cycle for a biped, but not yet for a
quadruped. To make it usable by other people is something I'm working hard
on.
Rune
--
3D images and anims, include files, tutorials and more:
Rune's World: http://rsj.mobilixnet.dk (updated June 26)
POV-Ray Users: http://rsj.mobilixnet.dk/povrayusers/
POV-Ray Webring: http://webring.povray.co.uk
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Rune wrote:
>
> Besides, it doesn't work for smooth meshes.
>
No problem with some extra calculations.
There are many cases BTW, where a field based deformation is much more
convenient than something based on transforms.
> > just imagine blobs could be nicely 'boned' too
>
> No they couldn't. But bezier pathes could perhaps.
>
What's so different about blobs (the blob component coordinates can be
compared to the mesh vertices)
Christoph
--
Christoph Hormann <chr### [at] gmxde>
IsoWood include, radiosity tutorial, TransSkin and other
things on: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> 50 MB for just one second of animation seems like a pretty inefficient
> solution to me!
Absolutely. But nice enough in the present case (and I hope impressive if I
manage to finish it one day).
> I've already made a walk and a running cycle for a biped, but not yet for
a
> quadruped. To make it usable by other people is something I'm working hard
> on.
My point exactly. What I'm suggesting is that this sort of large-scale
proposal would be much more convincing and "saleable" to other people if
you'd do first some sort of feasability study of what you propose and
demonstrate it on a smaller scale. After all, there's an infinity of
features that one would like to find into POV-Ray and I could post one of
them every day, with every detail of how I would like to see them
implemented.
So, making a working, usable include file of a generic (bones only)
quadruped (not biped) walking/running animation would be a first step in
that direction. It will show you the various problems involved, and, if
successful and appreciated by other Povers (not just the "wow cool!"
reaction, I mean something actually used like Chris Colefax' macros are),
will constitute a basis for a larger project that would get developers
interested.
G.
--
**********************
http://www.oyonale.com
**********************
Graphic experiments
Pov-ray gallery
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Gilles Tran" wrote:
> What I'm suggesting is that this sort of large-scale
> proposal would be much more convincing and "saleable"
> to other people if you'd do first some sort of
> feasability study of what you propose and demonstrate
> it on a smaller scale.
Quite understandable. That's what I'm working on, I just wanted to seed a
discussion and awareness of the possible potential at an early stage. Then
when I post my animated demonstrations people will already have this in
mind.
> So, making a working, usable include file of a generic
> (bones only) quadruped (not biped) walking/running
> animation would be a first step in that direction.
While I can understand that a quadruped will be a greater challenge, I can't
see why a biped would be about equally useful. But in the long run I'll have
a try with both.
Rune
--
3D images and anims, include files, tutorials and more:
Rune's World: http://rsj.mobilixnet.dk (updated June 26)
POV-Ray Users: http://rsj.mobilixnet.dk/povrayusers/
POV-Ray Webring: http://webring.povray.co.uk
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Christoph Hormann" wrote:
> Rune wrote:
> > Besides, it doesn't work for smooth meshes.
>
> No problem with some extra calculations.
Deforming the normals too involves knowing the positions of the accompanying
vertices, after which for each normal at least 4 samples must be taken to
estimate the normal, and then various vcross operations are performed to get
the final result.
While it is possible if the number of normals specified is equal to the
number of vertices, it sure is not a simple operation, and it's extremely
computationally expensive for large meshes. I estimate a deformation of a
smooth mesh would take about six times as long as a deformation of a
non-smooth mesh.
> There are many cases BTW, where a field based deformation
> is much more convenient than something based on transforms.
Sure, but not for character animation which is the primary goal here.
> What's so different about blobs (the blob component
> coordinates can be compared to the mesh vertices)
The surface of a mesh is entirely dependent on the vertices. The surface of
a blob is dependent on many other things than just the center coordinates.
This is also why you can't UV-map a blob.
Rune
--
3D images and anims, include files, tutorials and more:
Rune's World: http://rsj.mobilixnet.dk (updated June 26)
POV-Ray Users: http://rsj.mobilixnet.dk/povrayusers/
POV-Ray Webring: http://webring.povray.co.uk
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> While I can understand that a quadruped will be a greater challenge, I
can't
> see why a biped would be about equally useful. But in the long run I'll
have
> a try with both.
There have been many bipeds animated successfully in Povray. I even did a
basic walking stick figure in 1993 with Povray 1. So it's difficult, but
it's been done before, at least partially (see the anim "For a claw full of
gold" http://www.irtc.org/anims/2001-01-15.html).
However, I've never seen a Povray-made quadruped trotting or galloping
(which was why I used Poser to do it, BTW). So I guess that solving such a
problem in a way that's appealing to users would be a much better proof of
Povray's ability to support a native, flexible character animation system.
G.
--
**********************
http://www.oyonale.com
**********************
Graphic experiments
Pov-ray gallery
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Rune wrote:
>
> The surface of a mesh is entirely dependent on the vertices. The surface of
> a blob is dependent on many other things than just the center coordinates.
> This is also why you can't UV-map a blob.
>
Sorry, but i don't see a reason for limiting new, more sophisticated
transformation capabilities (which is what this is all about as far as i
understand from your proposal) to mesh2 objects.
Christoph
--
Christoph Hormann <chr### [at] gmxde>
IsoWood include, radiosity tutorial, TransSkin and other
things on: http://www.schunter.etc.tu-bs.de/~chris/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
