pyode-user — PyODE User Discussions

[Pyode-user] Collision detection and coordinates

[Matthias T. <Sta...@gm...>]

Hi!

I am very sorry that i am posting again my test program, but i have a little 
problem which i am dialing with the whole day!
I was inspired from tutorial2 and tutorial3 and i added in tutorial2 a block 
which should be connected to the 
ode.environment and with collision detection! But there is a little problem 
with the coordinates and i don't know why!
My real life coordinates are 6 (meters) in the x and 4 in the y coordinate!

Thank you!
matthias


Here is the source:

import pygame
from pygame.locals import * 
import ode

WIDTH = 900
HEIGHT = 600

srf=None
world=None
space=None

def coord(x,y):
	"Convert world coordinates to pixel coordinates."
	xf = WIDTH / 6
	yf = HEIGHT / 4
	#return int(400+150*x), int(500-150*y)
	return int(xf * x), int(HEIGHT - (yf * y))

def near_callback(args, geom1, geom2):
	
	contacts = ode.collide(geom1, geom2)
	for c in contacts:
		c.setBounce(0.5)
		e = ode.ContactJoint(world, args, c)
		e.attach(geom1.getBody(), geom2.getBody())
		print 'Position of geom1 and block(geom2): ', geom1.getPosition(), 
geom2.getPosition()
		print 'Length of the block: ', geom2.getLengths()
	

def doSimulation(body, block):
	fps = 50
	dt = 1.0 / fps
	loopFlag = True
	
	clk = pygame.time.Clock()
	
	contgrp = ode.JointGroup()
	
	while loopFlag:
		events = pygame.event.get()
		for e in events:
			if e.type == QUIT or e.type == KEYDOWN:
				loopFlag = False
		
		srf.fill ( (255,255,255) )
		
		srf.lock()
		xv, yv = 3, 4
		
		# block zeichnen
		x,y,z = block.getPosition()
		l,h = block.laenge
		
		pygame.draw.rect(srf, (0,0,0), (coord(x,y + h), (l * (WIDTH / 6), h * 
(HEIGHT / 4) )), 0)
		
		# bodies zeichnen
		for i in range(len(body)):
			x1,y1,z1 = body[i].getPosition()
			
			pygame.draw.circle(srf, (50,0,200), coord(x1,y1), int(body[i].radius * 
(WIDTH/6)) )
			pygame.draw.line(srf, (100,0,200), coord(xv,yv), coord(x1,y1), 2)
	
			xv, yv = x1, y1
		
		srf.unlock()
		pygame.display.flip()
		
		for i in range(2):
			space.collide(contgrp,near_callback)
		
			world.step(dt / 2)
			contgrp.empty()
		
		clk.tick(fps)
		
	# end while
	
def createBody(world, space):
	
	radius = 0.2
		
	body = ode.Body(world)
	
	mass = ode.Mass()
	mass.setSphere(2000, radius)
	body.setMass(mass)
		
	geom = ode.GeomSphere(space, radius)
	geom.setBody(body)
	
	body.radius = 0.2
		
	return body

def createBlock(space, pos):
	
	geom = ode.GeomBox(space, (1.5, 2.5, 0) )
	geom.setPosition((4.5,0,0))
	geom.setBody(ode.environment)
	
	geom.laenge = (1.5, 2.5)
	
	return geom

if __name__ == '__main__':

	pygame.init()
	srf = pygame.display.set_mode( (900,600) )
	
	world = ode.World()
	world.setGravity( (0,-9.81,0) )
	world.setERP(0.5)
	space = ode.Space()
	
	bodies = []
	
	for i in range(3):
		body = createBody(world, space)
		body.setPosition((i+4,4,0))
		bodies.append(body)
		
	joint = ode.BallJoint(world)
	joint.attach(ode.environment, bodies[0])
	joint.setAnchor((3,4,0))
	
	joint2 = ode.BallJoint(world)
	joint2.attach(bodies[0], bodies[1])
	joint2.setAnchor((4,4,0))
	
	joint3 = ode.BallJoint(world)
	joint3.attach(bodies[1], bodies[2])
	joint3.setAnchor((5,4,0))
	
	block = createBlock(space, (4.5,0,0))
		
	doSimulation(bodies, block)

Re: [Pyode-user] Newb Problem with tut2!

[Chris B. <chr...@gm...>]

On 07/11/05, Matthias <Sta...@gm...> wrote:

>                 kx, ky, kz =3D kugel.getPosition()
>                 print kx, ky, kz
>                 pygame.draw.line(srf, (0,0,0), coord(2, 3.8), coord(kx, k=
y), 3)
>
>                 pygame.draw.rect(srf, (0,100,50), (coord(kx, ky), coord(k=
x + 0.1, ky -
> 0.1)), 0)

from http://www.pygame.org/docs/ref/rect.html#pygame.Rect we find the
definition of Rect:
pygame.Rect((left, top), (width, height)): return Rect

But you used absolute coordinates instead of (width,height) - your
rectangle size is shrinking as kx and ky approach the origin!

This works better:
pygame.draw.rect(srf, (0,100,50), (coord(kx, ky), coord(0.1, 0.1)), 0)

Now your next problem is that the joint with the ode.environment
doesn't seem to be working. The body is always moving towards the
origin. Even if you set the gravity to (0,0,0) it does this. This
indicates a prolem with the joint - since the only force in your
simulation is generated by ode trying to correct the errors in this
joint. The problem is command order, you need to attach the joint
before setting the anchor:

       bj.attach(kugel, ode.environment)
       bj.setAnchor( (2, 3.8, 0) )

That should fix your problems.

Chris

[Pyode-user] Newb Problem with tut2!

[Matthias <Sta...@gm...>]

Hi!
I am very new to ODE! I had a look at tutorial 1 and 2, and i wanted to make
my own little test sample!

The program is very similar to tutorial2 but very selfmade, but it doesnt
work! Why? I dont know!! I think there is a problem with the coordinates! 
I wrote the following code:

#! /usr/bin/env python

import pygame
from pygame.locals import *
import ode

WIDTH = 800
HEIGHT = 600

def coord(x, y):
	x = x * (WIDTH / 4);
	y = HEIGHT - y * (HEIGHT / 4);
	return x, y

def initPygame():
	pygame.init()
	srf = pygame.display.set_mode( (WIDTH, HEIGHT) )
	pygame.display.set_caption('Kran Beispiel mittels ODE')

	return srf

def initOde():
	world = ode.World()
	world.setGravity( (0, -9.81, 0) )

	return world

def makeKugel(w):
	kugel = ode.Body(w)
	m = ode.Mass()
	m.setBox(7900, 0.2, 0.2, 0.3)	
	kugel.setMass(m)
	kugel.setPosition( (3, 3.8, 0) )

	return kugel

def drawKran(srf):
	
	srf.lock()
	
	pygame.draw.rect(srf, (100, 200, 40), (coord(0, 4), coord(2, 3.8)), 0)
	pygame.draw.rect(srf, (100, 200, 40), (coord(0, 4), coord(0.2, 0)), 0)

	srf.unlock()
	

if __name__ == '__main__':

	loop = True
	fps = 50
	dt = 1.0 / fps
	clk = pygame.time.Clock()

	srf = initPygame()
	world = initOde()

	kugel = makeKugel(world)
	bj = ode.BallJoint(world)
	bj.setAnchor( (2, 3.8, 0) )

	bj.attach(kugel, ode.environment)

	while (loop):

		for e in pygame.event.get():
			if (e.type == QUIT):
				loop = False


		srf.fill( (255,255,255) )

		drawKran(srf)

		kx, ky, kz = kugel.getPosition()
		print kx, ky, kz
		pygame.draw.line(srf, (0,0,0), coord(2, 3.8), coord(kx, ky), 3)

		pygame.draw.rect(srf, (0,100,50), (coord(kx, ky), coord(kx + 0.1, ky -
0.1)), 0)	
		
		pygame.display.flip()

		world.step(dt)
		clk.tick(fps)
	



if you find the bugs please talk to me like talking to a four year old child
(maybe some explanation would be nice ;) )! 

thanks,
Matthias

-- 
Highspeed-Freiheit. Bei GMX supergünstig, z.B. GMX DSL_Cityflat,
DSL-Flatrate für nur 4,99 Euro/Monat*  http://www.gmx.net/de/go/dsl

RE: [Pyode-user] General question about ode

[Paul K. <pki...@in...>]

From memory - you need to keep a global copy of your joint (since pyode
doesn't store them).  Your joints are potentially being destroyed when
drop_object() exists.
 
Paul
 
-----Original Message-----
From: pyo...@li...
[mailto:pyo...@li...] On Behalf Of Ricardo
Kirkner
Sent: Friday, 28 October 2005 8:08 AM
To: pyo...@li...
Subject: [Pyode-user] General question about ode
 
Hi:

I am really new to this ode/pyode thing. I am trying to develop a
simulation using pyode. I finally made it (using several "borrowed
example code" :-)) to have my objects get rendered on screen.

They are currently generated a bit above the ground, so when I rendered
them, they fall a little (inspired by the pyode tutorial2). I have
created a model that consists of two boxes that should be connected
through a stick. All three objects get correctly rendered, but they
don't keep together (they just split apart as soon as they fall on the
ground)

The code I am using to created those objects is:

------------------------------------------------------------------------
-----------------------------------------------
def create_box(world, space, density, lx, ly, lz):
    """Create a box body and its corresponding geom."""

    # Create body
    body = ode.Body(world)
    M = ode.Mass()
    M.setBox(density, lx, ly, lz)
    body.setMass(M)

    # Set parameters for drawing the body
    body.shape = "box"
    body.boxsize = (lx, ly, lz)

    # Create a box geom for collision detection
    geom = ode.GeomBox(space, lengths=body.boxsize)
    geom.setBody(body)

    return body

def drop_object():
    """Drop an object into the scene."""
    
    global bodies
    
    blockA = create_box(world, space, 1000, 1.0,1.0,1.0)
    blockB = create_box(world, space, 1000, 1.0,1.0,1.0)
    link = create_box(world, space, 1000, 1.0, 0.2, 0.2)
    x, y, z = 0.0, 1.0, 0.0
    blockA.setPosition( (x,y,z ))
    link.setPosition((x+1,y,z))
    blockB.setPosition((x+2,y,z))
    m = mat4().rotation(0, (0,1,0))
    blockA.setRotation(m.toList())
    blockB.setRotation(m.toList())
    link.setRotation(m.toList())

    j1 = ode.HingeJoint(world)
    j2 = ode.HingeJoint(world)
    j1.attach(blockA, link)
    j2.attach(blockB, link)
    j1.setAnchor((0,0,0))
    j2.setAnchor((1,0,0))
    j1.setAxis((0,0,1))
    j2.setAxis((0,0,1))
    j1.setParam(ode.ParamFMax, ode.Infinity)
    j1.setParam(ode.ParamVel, 0)
    j2.setParam(ode.ParamFMax, ode.Infinity)
    j2.setParam(ode.ParamVel, 0)
        
    bodies.append(blockA)
    bodies.append(blockB)
    bodies.append(link)

------------------------------------------------------------------------
-----------------------------------------------

Maybe someone can help me to get this three objects remain connected all
time? Maybe I should have written to the ode mailing list, but since I
am writing code in python, I thought of this one.

With regards,

Ricardo Kirkner

[Pyode-user] General question about ode

[Ricardo K. <ric...@gm...>]

Hi:

I am really new to this ode/pyode thing. I am trying to develop a simulatio=
n
using pyode. I finally made it (using several "borrowed example code" :-))
to have my objects get rendered on screen.

They are currently generated a bit above the ground, so when I rendered
them, they fall a little (inspired by the pyode tutorial2). I have created =
a
model that consists of two boxes that should be connected through a stick.
All three objects get correctly rendered, but they don't keep together (the=
y
just split apart as soon as they fall on the ground)

The code I am using to created those objects is:

---------------------------------------------------------------------------=
--------------------------------------------
def create_box(world, space, density, lx, ly, lz):
"""Create a box body and its corresponding geom."""

# Create body
body =3D ode.Body(world)
M =3D ode.Mass()
M.setBox(density, lx, ly, lz)
body.setMass(M)

# Set parameters for drawing the body
body.shape =3D "box"
body.boxsize =3D (lx, ly, lz)

# Create a box geom for collision detection
geom =3D ode.GeomBox(space, lengths=3Dbody.boxsize)
geom.setBody(body)

return body

def drop_object():
"""Drop an object into the scene."""

global bodies

blockA =3D create_box(world, space, 1000, 1.0,1.0,1.0)
blockB =3D create_box(world, space, 1000, 1.0,1.0,1.0)
link =3D create_box(world, space, 1000, 1.0, 0.2, 0.2)
x, y, z =3D 0.0, 1.0, 0.0
blockA.setPosition( (x,y,z ))
link.setPosition((x+1,y,z))
blockB.setPosition((x+2,y,z))
m =3D mat4().rotation(0, (0,1,0))
blockA.setRotation(m.toList())
blockB.setRotation(m.toList())
link.setRotation(m.toList())

j1 =3D ode.HingeJoint(world)
j2 =3D ode.HingeJoint(world)
j1.attach(blockA, link)
j2.attach(blockB, link)
j1.setAnchor((0,0,0))
j2.setAnchor((1,0,0))
j1.setAxis((0,0,1))
j2.setAxis((0,0,1))
j1.setParam(ode.ParamFMax, ode.Infinity)
j1.setParam(ode.ParamVel, 0)
j2.setParam(ode.ParamFMax, ode.Infinity)
j2.setParam(ode.ParamVel, 0)

bodies.append(blockA)
bodies.append(blockB)
bodies.append(link)

---------------------------------------------------------------------------=
--------------------------------------------

Maybe someone can help me to get this three objects remain connected all
time? Maybe I should have written to the ode mailing list, but since I am
writing code in python, I thought of this one.

With regards,

Ricardo Kirkner

[Pyode-user] Compose XODE-Documents

[Diez B. R. <de...@we...>]

Hi,

I'm in the process of writing a sort of "generic" ODE application that 
reads its definition from a xode document. And I try to use two classes 
- world, and robot - to decompose the xode declarations. Basically, I 
try to extract the robot's definition from the second xode document and 
add it to the world.

Things work in so far that I can add the xode-Object, and the simulation 
  works in so far that the robots geoms get rendered. But they aren't 
subject to gravity for example, no collision detection and so on.

Has anybody managed to get a similar thing to work and give me some hints?

Regards,

Diez

Re: [Pyode-user] Why don't my joints work? (PyODE with PyOgre)

[Andrew D. <ad...@gm...>]

On 10/4/05, Matthias Baas <ba...@ir...> wrote:
>
> I couldn't run your code as I don't have PyOgre but it seems the problem
> is you store the BallJoint in a local variable which gets deleted at the
> end of the method. And as j1 is the only reference to the BallJoint
> object the BallJoint is garbage collected when j1 is deleted which will
> also delete the underlying ODE BallJoint. To fix this you simply have to
> keep a Python reference to the joint (self.j1 =3D ode.BallJoint(...)).

Yes, that fixes it -- although it seems a little unnatural. I would
have expected that the world would maintain the object, and that PyODE
would not delete the joint merely because the Python reference gets
collected...

Anyway, thanks! I've got a solution now.

[Pyode-user] PyODE and gcc 4.0

[Paul K. <pki...@in...>]

Hi
 
I'm trying to compile PyODE on gcc 4 (Darwin).  I've had a few problems.
Firstly - it appears you need to compile with g++ rather than gcc.
Secondly, when I try to "import ode", in python, I get an error:
ImportError: Loaded module does not contain symbol _initode2
 
Does anyone know how to fix this?
 
Paul

Re: [Pyode-user] Why don't my joints work? (PyODE with PyOgre)

[Matthias B. <ba...@ir...>]

Andrew Durdin wrote:
>         # Connect body1 with the static environment
>         j1 = ode.BallJoint(self.world)
>         j1.attach(body1, None)
>         j1.setAnchor((0, 200, 0))

I couldn't run your code as I don't have PyOgre but it seems the problem 
is you store the BallJoint in a local variable which gets deleted at the 
end of the method. And as j1 is the only reference to the BallJoint 
object the BallJoint is garbage collected when j1 is deleted which will 
also delete the underlying ODE BallJoint. To fix this you simply have to 
keep a Python reference to the joint (self.j1 = ode.BallJoint(...)).

- Matthias -

Re: [Pyode-user] pyODE in Blender 3D

[Matthias B. <ba...@ir...>]

Gabriel Beloin wrote:
> So I'm limited in the number of bodies i can include in the same space?

As far as I could see, the crash seems not to be related to the Python 
binding but ODE itself. The crash happens in the quickStep() method 
which does nothing but call the corresponding ODE C function.
Have you tried to visualize the simulation up to the crash to see what 
event triggers it? Is it when the boxes hit the ground? Maybe there are 
too many contacts then (which might result in a stack overflow)...?
You could try not to use all the contacts in the near callback to reduce 
the number of contact joints.

- Matthias -

[Pyode-user] Why don't my joints work? (PyODE with PyOgre)

[Andrew D. <ad...@gm...>]

Hi everyone,

I'm experimenting with PyODE and PyOgre at the moment, and have been
working through PyODE's Tutorial 2 (two bodies connected with ball
joints to form a compound pendulum).

I've tried to rewrite Tutorial 2 using PyOgre instead of PyGame. I'd
managed to get an ogre head to fall under gravity without trouble, but
attempting to attach the head to a joint (with the other end fixed),
the joint seems to be completely ignored, with the head merely falling
and not swinging as it ought.

I can't see at all what I'm doing wrong here -- can anyone help me?

Here's the code:


import pyogre.ogre as ogre
import SampleFramework as sf
import ode

class TutorialApplication(sf.Application):
    def __init__(self):
        sf.Application.__init__(self)

        # Set up the physics world
        self.world =3D ode.World()
        self.world.setGravity((0, -9.81, 0))
        self.bodyList =3D []


    def _createScene(self):
        sm =3D self.sceneManager

        # Set up the lighting conditions
        sm.ambientLight =3D (0.7, 0.7, 0.7)


        # Create a disembodied ogre head
        e =3D sm.createEntity("OgreHead", "ogrehead.mesh")
        n =3D sm.rootSceneNode.createChildSceneNode()
        n.attachObject(e)
        pos =3D (100, 200, 0)
        n.position =3D ogre.Vector3(*pos)

        body1 =3D ode.Body(self.world)
        M =3D ode.Mass()
        M.setSphere(2500, 0.05)
        body1.setMass(M)
        body1.setPosition((100, 200, 0))

        # Add it to the body list
        self.bodyList.append((body1, n))

        # Connect body1 with the static environment
        j1 =3D ode.BallJoint(self.world)
        j1.attach(body1, None)
        j1.setAnchor((0, 200, 0))


    def _createFrameListener(self):
        """Creates the FrameListener."""
        self.frameListener =3D PhysicsFrameListener(self.renderWindow,
self.camera, self.world, self.bodyList)
        self.frameListener.showDebugOverlay(True)
        self.root.addFrameListener(self.frameListener)


class PhysicsFrameListener(sf.FrameListener):
    def __init__(self, renderWindow, camera, world, bodyList):
        sf.FrameListener.__init__(self, renderWindow, camera)

        self.world =3D world
        self.bodyList =3D bodyList
        self.time =3D 0.0
        self.dt =3D 0.08


    def frameStarted(self, frameEvent):
        # Step the physics
        self.world.step(self.dt)
        self.time +=3D self.dt

        if False:
            body =3D self.bodyList[0][0]
            x,y,z =3D body.getPosition()
            u,v,w =3D body.getLinearVel()
            print "%1.2fsec: pos=3D(%6.3f, %6.3f, %6.3f)  vel=3D(%6.3f,
%6.3f, %6.3f)" % \
                (self.time, x, y, z, u,v,w)


        # Update the positions of objects
        for body, node in self.bodyList:
            pos =3D body.getPosition()
            node.position =3D ogre.Vector3(*pos)

        return sf.FrameListener.frameStarted(self, frameEvent)



if __name__ =3D=3D "__main__":
    app =3D TutorialApplication()
    app.go()

[Pyode-user] Compiling PyODE for the MacIntosh

[Paul K. <pki...@in...>]

Hi
 
I was wondering if anyone had been able to compile PyODE for the Mac.
I've compiled ODE with gcc (3.3 I think), however the Distutils that the
PyODE build uses wants to compile the ode files from Pyrex using
Metrowerks Codewarrior 7 (which I don't have, and which appears to not
be supported any more - judging by the Metrowerks website).
 
Paul

Re: [Pyode-user] Trimesh collision problem

[<th...@po...>]

> Are you sure that vert and ind do carry=20
> correct values?=20

Thanks for your answer, I just figured out that I got a stupid error in cre=
ating my set of indices. Now it works good, but slow... :( But I can surely=
 decrease my vertices set and I hope it'll be ok then.=20
Przemek

----------------------------------------------------------------------
Jestem niesamowita... ;-) >>> http://link.interia.pl/f18b8

[Pyode-user] pyODE in Blender 3D

[Gabriel B. <gb...@ne...>]

Hi everybody,

I'm Gabio from the Blender community. Artist and python coder. Blender 
is an opensource 3D package using python as main scripting language. The 
API is accessible 
here:http://www.blender.org/modules/documentation/237PythonDoc/frames.html .
I recently tried to simulate some tests scenes in Blender using the 
pyODE module. And it worked relatively well. I got a problem though. A 
crash to be more correct, when having a lot of bodies in the same space. 
The fact is I want all of them to be able to collide. I'm not aware ODE 
have limit in the number of collisions it can manage.
To test it out. you can get Blender 3D @ www.blender.org. have python 
2.3 installed and pyODE.
Then download this file: 
http://web.netrevolution.com/gbeloin/ODE_bug_gabio.blend
You can open it in Blender by pressing F1 and browsing to the .blend 
file or double click on it (Windows)
You will then face a screen with the scene (plane and ball) and a script 
windows on the right. to start the test move your mouse cursor in the 
text windows and press ALT-P. for me it crash. in the near_callback. and 
the debug give me a stack overflow. If you go at line 95 and change     
"for x in range(20):" to "for x in range(5):" the script doesn't crash 
anymore. you then can see the anim by moving your mouse in the 3Dspace 
on the left and pressing ALT-A.
or you can see the result of the best i could get out of pyODE before 
crash: http://media.putfile.com/Render_legowall0001_0200
So I'm limited in the number of bodies i can include in the same space?

Of course if you don't feel like downloading Blender and python 2.3, 
I've made an example you can run on any system with pyODE. The crash is 
instant:

--------------------------------------
import ode

def draw_body(body):
    global frame
    x,y,z = body.getPosition()
    print x,y,z

# Collision callback
def near_callback(args, geom1, geom2):
    """Callback function for the collide() method.
    This function checks if the given geoms do collide and
    creates contact joints if they do.
    """
    # Check if the objects do collide
    contacts = ode.collide(geom1, geom2)
    # Create contact joints
    world,contactgroup = args
    for c in contacts:
        c.setBounce(.5)
        c.setMu(7000)
        j = ode.ContactJoint(world, contactgroup, c)
        j.attach(geom1.getBody(), geom2.getBody())

# Create a world object
world = ode.World()
world.setGravity((0,0,-9.81))
world.setERP(0.2) #error correction
world.setCFM(1E-10) #softness of collision

lobj = []
lbody = []
contactgroup = ode.JointGroup()

print "Create a space object"
space = ode.Space()

print "Create a plane geom which prevent the objects from falling forever"
floor = ode.GeomPlane(space, (0,0,1), 0)

print "create array"
a = 1.5
for z in range(20):
    for x in range(20):
        lobj.append((x*3+a,1,z+.5))
for obj in lobj:       
    body = ode.Body(world)
    M = ode.Mass()
    M.setBox(5000,3,1,1)
    body.setMass(M)
    body.setPosition(obj)
    geom = ode.GeomBox(space,(3,1,1))
    geom.setBody(body)
    lbody.append(body)
    body.disable() #wait for the cue
   
bball = ode.Body(world)
M = ode.Mass()
M.setSphere(700,5)
bball.setMass(M)
bball.setPosition((9,-19,5))
geom2 = ode.GeomSphere(space,5)
geom2.setBody(bball)
bball.setLinearVel((0,20,0)) #give some speed.
lobj.append((9,-19,5))
lbody.append(bball)

print "Simulation loop..."

fps = 24
dt = 1.0/fps
frame = 1 #first frame
totframe = 300
while frame < totframe: #tot of frame to do
    for i in range(2):
        space.collide((world,contactgroup), near_callback) #colision
        for b in range(len(lbody)):
            draw_body(lbody[b])
        world.quickStep(dt) #shift time
        contactgroup.empty() #empty mem
    frame += 1
ode.CloseODE()
print "done!"

-------------------------
Thanks for looking.

Re: [Pyode-user] Trimesh collision problem

[Matthias B. <ba...@ir...>]

th...@po... wrote:
> trimesh=ode.TriMeshData()
> trimesh.build(vert,ind)
> trimeshgeom=ode.GeomTriMesh(trimesh,space)

So far, this looks fine to me. Are you sure that vert and ind do carry 
correct values? (each item in vert should be a sequence of 3 floats and 
each item in ind should be a sequence of 3 ints)
Have you tried reverting the order of the vertices per face to flip 
their direction? (but I'm not sure if this makes a difference in 
ODE/Opcode anyway)

- Matthias -

Re: [Pyode-user] Tut3 without cgkit

[Matthias B. <ba...@ir...>]

Jean de Largentaye wrote:
> I think this would be quite useful for people who can't / don't want
> to install cgkit, because cgkit1 doesn't exist in a packaged form for
> python2.4, or they're afraid to try to replace it with cgkit2 (oooh,
> 'alpha'! fear!). After all, reducing dependancies is a Good Thing, no?

Even though I'm biased I do agree with you. :)
I'm sorry that you had to spend your time on that, it actually wouldn't 
have been necessary because this work has already been done before. 
Pierre Gay has already sent me a version where he removed the dependency 
on cgkit and pygame (using GLUT instead which is part of PyOpenGL). We 
just haven't had the time yet to update the web site and cvs repository. 
My apologies for that.

(By the way, your version doesn't run over here (Python 2.3, WinXP). I 
get a segmentation fault by pygame right after the window pops up... 
maybe it has something to do with the PyOpenGL binary using numarray 
instead of Numeric...?)

Thanks for your efforts anyway, and sorry again that it has already be 
done but not posted yet...

- Matthias -

Re: [Pyode-user] amotor with ball joint?

[Matthias B. <ba...@ir...>]

Chris Bainbridge wrote:
>>The tutorial is *not* referring to AMotors but to the internal motors of
>>the joints (which a ball joint doesn't have).
> 
> Do you mean you can't just set ParamVel and ParamFMax on a ball joint?
> You have to use an AMotor? I didn't know that.. don't remember seeing
> it in the manual.

See section 7.5.1 "Parameter Functions" in the ODE manual 
(http://www.ode.org/ode-latest-userguide.html#sec_7_5_1).
This sections lists the functions to get/set the limit/motor parameters 
for each joint type and there is no get/set function for a ball joint.

- Matthias -

[Pyode-user] Trimesh collision problem

[<th...@po...>]

Hi,
I have problems with setting up collision between GeomTriMesh and GeomSpher=
e. I have a list called "verts" with my vertices and a list called "ind" wi=
th indices. It look that they're ok, because it is drawn as I want it to be=
 drawn with glVertexPointer/glDrawElements.=20
That's all about trimesh geom in my code:
trimesh=3Dode.TriMeshData()
trimesh.build(vert,ind)
trimeshgeom=3Dode.GeomTriMesh(trimesh,space)

And then I place a sphere over it and it just falls through the mesh, witho=
ut any collision (of course sphere has it's body and geom). What should i d=
o to make this work? Should the indices or vertices be in some specific ord=
er? I took the collision callback function from Tutorial 3.
Greets,
Przemek Muller

----------------------------------------------------------------------
Sa niesamowite, zobaczysz... ;-) >>> link http://link.interia.pl/f18b9

[Pyode-user] Tut3 without cgkit

[Jean de L. <jla...@gm...>]

Hi folks!

I rewrote the tut3 for myself to work without cgkit. I work with
python 2.4, and there's no package of cgkit1 for this, at least on
Debian... Furthermore, I couldn't get cgkit2 to work as a drop-in
replacement (some C++ ABI errors I forgot), so I decided to see if I
could remove it completely, and succeeded.

I think this would be quite useful for people who can't / don't want
to install cgkit, because cgkit1 doesn't exist in a packaged form for
python2.4, or they're afraid to try to replace it with cgkit2 (oooh,
'alpha'! fear!). After all, reducing dependancies is a Good Thing, no?
OK, maybe not for tutorials where you want things to be as clear as
possible. But I think it would be interesting to give it as an
alternate download for those who are frustrated by the dependancies
(as I was).

The result is slightly less tidy than the original. An additional
requirement is to import Numerical. But since that package is a
requirement for PyOpenGL which is already used, I don't think that's a
problem. I had to add a small function that flattens a Numerical.array
object, since the included array.tolist() function doesn't do it
correctly, at least for our use.

If you run down the sources, I have added comments "#use this..." and
"#instead of this" to show what I have replaced (hint: "use.."
precedes the new part, and "instead" precedes the old part).

in Prepare_GL(), it's actually shorter to use the libraries included
functions gluPerspective() and gluLookAt(). As we're alrady importing
GLU, why not use them?

in draw_body(), I directly worked with the flattened matrix returned
by getRotation(). Nothing to see here...

In drop_object(), I used GL's matrix operations to make and retrieve
the rotation matrix, then my flatten function I mentioned above to put
that with setRotation()

In explosion, I basically extracted the length() and normalize()
functions from vec3 :)
Same for inside pull()


Oh, and in the meantime, the Debian package isn't getting anywhere.
All of a sudden, having pyode already installed removes most of the
incentive :(

And a last thing: it's late, it's possible (probable) I'm blathering
nonsense somewhere (everywhere). Be kind ;)

John

Re: [Pyode-user] PyODE and gcc 4.0

[Timothy S. <str...@ma...>]

On Tue, 2005-09-20 at 13:21 +0200, Jean de Largentaye wrote:
> Furthermore, I'm a bit motivated to make a debian package of PyODE. I
> wonder if there have been previous efforts in that direction?

I don't believe anyone has attempted to build a Debian package so far
although there have certainly been people interested in having one. I
use Gentoo so I won't be of much help. An Autopackage[1] package might
be a good idea for the next release; that should suit most people.

[1] http://autopackage.org/

-- 
Timothy Stranex <str...@ma...>

Re: [Pyode-user] amotor with ball joint?

[Chris B. <chr...@gm...>]

On 20/09/05, Matthias Baas <ba...@ir...> wrote:
> The tutorial is *not* referring to AMotors but to the internal motors of
> the joints (which a ball joint doesn't have).

Do you mean you can't just set ParamVel and ParamFMax on a ball joint?
You have to use an AMotor? I didn't know that.. don't remember seeing
it in the manual.


On 20/09/05, Paul Kinnane <pki...@in...> wrote:
>=20
> Converting this to UserMode resulted in the limits not working at all.
> I was unable to find the problem.  My guess is that there is an issue
> with PyODE's AMotor.setNumAxes or setAngle, but I couldn't track it
> down.

This post (1) on the ode list has code for user mode ball joints with
working limits (though I haven't tested it). (2) says there are some
problems though - it sounds like ODE only considers the limit in the
direction the joint is moving towards, instead of trying to satisfy
both limits at the same time.

(1) http://q12.org/pipermail/ode/2005-August/016510.html
(2) http://q12.org/pipermail/ode/2005-July/016474.html

Re: [Pyode-user] PyODE and gcc 4.0

[Matthias B. <ba...@ir...>]

Jean de Largentaye wrote:
> A little note though: on deb, the compiler is called
> 'python2.x-pyrexc' (where x={3,4}), not 'pyrexc', so I had to make
> that trivial change to the setup script.

Would a variable carrying the name of the pyrex executable at the top of 
the setup script already be acceptable? (you'd still have to modify that 
  manually then)

> Furthermore, I'm a bit motivated to make a debian package of PyODE. I
> wonder if there have been previous efforts in that direction? 

Not that I knew of, but I'm mainly using Windows anyway, so maybe 
Timothy knows something...?

> Anyhow,
> I might need some help, as I'm a bit new to PyODE and this would be
> the first time I make a deb package ;)

Thanks for your effort, I'd suggest to post any problems right here...

- Matthias -

Re: [Pyode-user] amotor with ball joint?

[Matthias B. <ba...@ir...>]

Chris Bainbridge wrote:
> Has anyone successfully got AMotors to work with universal or ball
> joints? According to the 2nd pyode tutorial, they won't work. 

The tutorial is *not* referring to AMotors but to the internal motors of 
the joints (which a ball joint doesn't have).

> mailing list have claimed to have it working. I notice that the
> ParamX3 definitions are missing from the pyode wrappers - is this
> deliberate?

No. They're in there now, I've just checked it in.
(but as was already mentioned, you can just add 256 to the ParamX2 
values to get the ParamX3 values (or add 512 to ParamX))

>>From my own tests, in User Mode I can't get the AMotor to do anything.
> In Euler mode only the 2nd axis control (ParamVel, ParamFMax2) seems
> to have any effect, the other axis controls don't do anything.
> 
> Any ideas?

I have never used AMotors so far, so I'm afraid I can't help you there 
right now... but it looks as if this is actually an ODE issue and not 
really limited to PyODE (so asking in the ODE mailing list might a good 
idea, too).

Paul Kinnane wrote:
> Converting this to UserMode resulted in the limits not working at all.
> I was unable to find the problem.  My guess is that there is an issue
> with PyODE's AMotor.setNumAxes or setAngle, but I couldn't track it
> down.

These methods do nothing but call the appropriate ODE C function, so I'd 
say the problem is already located in ODE itself...

> can't use setAngle, and 2) you can't create a limit between a body and
> the environment (well I couldn't work out how to do it anyway) since
> am.setAxis(2, 2, [0, 0, 1]) fails if the second body is the
> ode.environment.

ode.environment is not an actual body but is just an alias for None. 
Connecting to the environment is just like calling the ODE C function 
with a second body argument of 0. Obviously the AMotor requires a real 
second body...

- Matthias -

[Pyode-user] PyODE and gcc 4.0

[Jean de L. <jla...@gm...>]

Hi folks,

I just wanted to point out that the PyODE 1.1.0 distribution won't
compile on gcc-4 based systems, ie those where 'gcc' is gcc 4 (like
debian unstable, for example).
This comes from the ode_[no]trimesh.c files, generated by pyrex 0.9.3,
that are included. This is a known problem with pyrex 0.9.3, it seems.
There is a new upstream release of pyrex that solves this problem.
Maybe it'd be a good idea to make a minor PyODE release with the
corrected ode_[no]trimesh.c files?
Debian uses a patched version of 0.9.3 that already fixes the gcc 4
problem, so it should be enough to remove those 2 C files and let the
system's pyrex rebuild them. This is what worked for me.
A little note though: on deb, the compiler is called
'python2.x-pyrexc' (where x=3D{3,4}), not 'pyrexc', so I had to make
that trivial change to the setup script.

Furthermore, I'm a bit motivated to make a debian package of PyODE. I
wonder if there have been previous efforts in that direction? Anyhow,
I might need some help, as I'm a bit new to PyODE and this would be
the first time I make a deb package ;)

cheers,
John

Re: [Pyode-user] amotor with ball joint?

[Chris B. <chr...@gm...>]

On 20/09/05, Paul Kinnane <pki...@in...> wrote:
> I got AMotor limits working in Euler Mode.

Do you know if ParamVel / ParamFMax work on all 3 axes in your setup?