visualpython-users

[Visualpython-users] Vpython plot to JPG

[Richard H. C. S. <rhs...@ma...>]

Hello,  I'd like to generate a plot in VPython and output it to a JPG on 
the fly so I can display the results of vector comparisons on a web 
page.  If this has been discussed before, please point me to 
it.  Otherwise, anyone have a suggestion for how to go about it?
Thanks,
Dick S.


-
   rhs...@ma...
  Anne  Arundel  Community  College
  (410)541-2424    ***    Speed the net!
  http://enterprise.aacc.cc.md.us/~rhs

RE: [Visualpython-users] Future directions

[ruth c. <rc...@an...>]

--On Wednesday, October 18, 2000 9:30 AM -0400 David Scherer 
<dsc...@cm...> wrote:

>
> (I don't think convex can be converted to blob - it has to be converted to
> triangles using a convex hull algorithm.  That could be done in Python by
> basically copying the code from convex.cpp in cvisual)
>

Duh. You're right, of course - a convex thingy has flat sides.

RE: [Visualpython-users] Future directions

[David S. <dsc...@cm...>]

> >> > rendering.  If VPython were able to take a snapshot in povray format,
> >> > that would be cool.
>
> Actually it would be pretty easy to write such a routine
> yourself. Probably
> the only tricky thing is converting -convex- to -blob-.  By writing your
> own routine you could specify textures, instead of having everything
> default to RedPlastic, etc.

Good idea!  The only caveat is that there isn't currently an easy way to
enumerate the objects in a scene, so the user program would have to keep
track of all its objects.  This could be very easily added, though -- we had
this capability in the prototype -- and then Povray output could be
implemented in Python.  I like it.

You could use optional user attributes to control the material:

for object in scene.objects:
    if hasattr(object,"material"):
        material = object.material
    else:
        # xxx I don't know Povray syntax:
        material = "material { color %s } " % str(object.color)

(I don't think convex can be converted to blob - it has to be converted to
triangles using a convex hull algorithm.  That could be done in Python by
basically copying the code from convex.cpp in cvisual)

Dave

Re: [Visualpython-users] Future directions

[ruth c. <rc...@an...>]

> On Tue, Oct 17, 2000 at 03:23:13PM -0400, Bruce Sherwood wrote:
>> --On Monday, October 16, 2000, 12:41 PM -0700 Dethe Elza
>> <de...@an...> wrote:

>> > rendering.  If VPython were able to take a snapshot in povray format,
>> > that would be cool.

Actually it would be pretty easy to write such a routine yourself. Probably 
the only tricky thing is converting -convex- to -blob-.  By writing your 
own routine you could specify textures, instead of having everything 
default to RedPlastic, etc.

Re: [Visualpython-users] Future directions

[Ari H. <ahe...@an...>]

On Tue, Oct 17, 2000 at 03:23:13PM -0400, Bruce Sherwood wrote:
> --On Monday, October 16, 2000, 12:41 PM -0700 Dethe Elza
> <de...@an...> wrote:
> 
> 
> > Is there any support for texture mapping in the works?
> 
> Ari Heitner has expressed strong interest in implementing this. However, he
> is also very busy with his computer science courses this semester, so
> there's no date attached.

soon. :)

well before heat death. :)

i have some more minor stuff to hack thru on the linux side before i play
with any of that stuff. and what i *don't* have is time to hack right now :)

at the very worst, there will be a flurry of activity as i recover from this
semester.

> 
> > rendering.  If VPython were able to take a snapshot in povray format, that
> > would be cool.
> 

mmph. but it's a real pain. now, as simple as vpython scenes are, it could
probably be done. that said ... it would also look *identical* to what
vpython shows you. i mean, short of texture support, what would you be
wanted to raytrace, exactly ? vpython doesn't give you any control over
materials (let alone GL's limited material support compared to a raytracer).




cheers,

-- 
Ari Heitner
  DC: 703/5733512  PGH: 412/4229837
Non c'è più forza nella normalità: c'è solo monotonia.

Re: [Visualpython-users] Future directions

[Bruce S. <ba...@an...>]

--On Monday, October 16, 2000, 12:41 PM -0700 Dethe Elza
<de...@an...> wrote:

> Will there be an update to Python 2.0 when that is released?

Yes, we definitely want to track Python and stay current.

> Is there any support for texture mapping in the works?

Ari Heitner has expressed strong interest in implementing this. However, he
is also very busy with his computer science courses this semester, so
there's no date attached.

> How about support for custom lighting?

There was in the past some support, but it was not documented because Dave
Scherer wasn't quite happy with it. I'll ask him what is current view is on
this.

> Finally, not really a question but a comment.  I've also been reading
> about Kirby Urner's work with Python, using it to teach math
> (http://inetarena.com/~pdx4d/ocn/numeracy0.html).  He uses Python to write
> out povray files, but most of his examples would work great with VPython
> instead, and have much more immediacy (and I plan to tell him so on the
> EDU-SIG mailing list), but sometimes it's nice to do a static, ray-traced
> rendering.  If VPython were able to take a snapshot in povray format, that
> would be cool.

To make sure I understand the request, I think you're asking for an option
to write out a file in the form of a Povray "program" which could then be
processed by Povray to make a high-quality image. That would allow one to
use VPython as a fast prototyping environment. Makes sense. Probably not
high on our own list of priorities, but maybe someone will step forward to
implement such a thing if it is of general interest.

Bruce

Re: [Visualpython-users] Plotting

[Bruce S. <ba...@an...>]

--On Tuesday, October 17, 2000, 2:37 PM -0400 David Porter
<dp...@dr...> wrote:

> Thanks Bruce for the answer to my last question, it wasn't the answer I
> was looking for, but it was an answer. Now I have another question. If I
> create a plot window say 640*480, can I address each point individually?
> I have an application in which I need to combine several different kinds
> of data into a composite picture.

I'm not sure I understand the question. If the question is "can I address
individual pixels" the answer is no. Both regular (3D) visual windows and
the graphing windows are addressed in nonpixel coordinates. The main
difference between these two kinds of windows is that for 3D images it
almost never makes sense for x and y scale factors to be different, whereas
on a graph it is normally the case that the scale factors are quite
different (for example, 1E-10 meters horizontally by 1E-20 joules
vertically).

Bruce

[Visualpython-users] Plotting

[David P. <dp...@dr...>]

Group
	Thanks Bruce for the answer to my last question, it wasn't the answer I was looking for, but it was an answer. Now I have another question. If I create a plot window say 640*480, can I address each point individually? I have an application in which I need to combine several different kinds of data into a composite picture.

			Regards
			Dave Porter  

[Visualpython-users] Future directions

[Dethe E. <de...@an...>]

Hi folks,

I'm just discovering VPython and I think you're doing a great job.  I have a
few questions about the future, for your virtual crystal balls:

Will there be an update to Python 2.0 when that is released?

Is there any support for texture mapping in the works?

How about support for custom lighting?

Finally, not really a question but a comment.  I've also been reading about
Kirby Urner's work with Python, using it to teach math
(http://inetarena.com/~pdx4d/ocn/numeracy0.html).  He uses Python to write
out povray files, but most of his examples would work great with VPython
instead, and have much more immediacy (and I plan to tell him so on the
EDU-SIG mailing list), but sometimes it's nice to do a static, ray-traced
rendering.  If VPython were able to take a snapshot in povray format, that
would be cool.

Thanks for the great tool.

Dethe Elza

Re: [Visualpython-users] Graphing

[Bruce S. <ba...@an...>]

> I am trying to graph data that has an offset above zero (y axis). How can
> I center the data in the graph window? I tried using xmin and ymin but
> they only seems to work with negative numbers. I am new to VPython and I
> think it is great.

At present the only way to offset a graph is to offset the function:

	xxx.plot(pos=(x,y+offset))

As you have seen, the graphing machinery currently uses xmin and ymin as
negative numbers in deciding how to handle the various quadrants. It is
reasonable to ask that this be made more general to handle offsets. 

To be honest, I found it difficult to get the plotting to work properly
just in the current restricted sense and gave up at least temporarily on
generalizing that aspect. For example, how would I choose sensible tick
marks and labels if the offset is 1.73?

Glad you like VPython!

Bruce Sherwood

[Visualpython-users] Graphing

[David P. <dp...@dr...>]

Group
	I am trying to graph data that has an offset above zero (y axis). How can I center the data in the graph window? I tried using xmin and ymin but they only seems to work with negative numbers. I am new to VPython and I think it is great.
		Regards
		Dave Porter

[Visualpython-users] New graph.py

[Bruce S. <ba...@an...>]

I draw your attention to a newer update zip file at

	http://cil.andrew.cmu.edu/projects/visual

which contains an improved graph.py (which goes in the Python\visual folder
and is invoked by from visual.graph import *; see the Visual reference
manual). The improvement is to manage better the margin around the graph,
so that no data is plotted outside the window.

Bruce Sherwood

[Visualpython-users] is there a better way to build animated GIFs using Python than what I'm doing?

[Stephen H. <sh...@ua...>]

Hi.  I'm one of the new folks who first heard of VPython at the Canada AAPT
meeting last month.  After a couple of weeks of learning, I think I'm making
pretty good progress.

I found a brute force way to capture images of the visual window (I dug up a
shareware program called ScreenTaker that works OK) and then paste I them
together on my Mac with GifBuilder.  It works, but eventually it would be
nice to have this happen more automatically.

Does anyone already have a better way?

[If you want to see a few things in progress, go to
http://comp.uark.edu/~shighla     ]

I know I don't need to tell you folks this, but VPython is a wonderful
thing!  The idea of having all the graphics happen without me having to "do"
anything about it was truly brilliant.

Thanks!
Steve Highland 

[Visualpython-users] new demos

[Bruce S. <ba...@an...>]

On the "Course Documents and Software" page at

	http://cil.andrew.cmu.edu/mi.html

you will find some lecture-demonstration programs we have used in our own
physics teaching. Currently available are a demonstration of the nature of
rotating vectors (used when discussing the time derivatives of rotating
vectors), three-body orbits, and a model for the propagation of sound in a
solid.

Bruce Sherwood

[Visualpython-users] Label bug

[Bruce S. <ba...@an...>]

There was a serious memory leak associated with the new label object. If
you installed Visual-2000-09-12.zip, you should fix the bug by installing
Visual-2000-09-13.zip, now available at

	http://cil.andrew.cmu.edu/projects/visual

Bruce Sherwood

[Visualpython-users] New: label and graphing

[Bruce S. <ba...@an...>]

I draw your attention to the fact that at 

	http://cil.andrew.cmu.edu/projects/visual 

there is a new update zip file available (Visual-2000-09-12.zip) which
contains the new label object for displaying text, the new graphing
machinery using the label object, related minor updates to demo programs,
and corrected/updated documentation, including how to use the label and
graphing machinery.

Currently these new features are available only for Windows, not Linux.

Bruce Sherwood

[Visualpython-users] Re: Visualpython-users digest, Vol 1 #8 - 2 msgs

[Bruce S. <ba...@an...>]

Another useful thing to try is to run the demo program glinfo.py and tell
us what you see printed in the output window (hoping the program gets that
far). This program is a diagnostic for OpenGL.

Bruce Sherwood

RE: [Visualpython-users] Running VPython on Win 98

[David S. <dsc...@cm...>]

> Having recently downloaded and installed VPython under Win 98 I am
> unable to progress very far with any of the demo programs that come with
> the product. What happens is that the computer hangs, no mouse cursor or
> ctrl+break action works. Has any body else had problems like this. My
> hardware is running on an AMD chip set rather than Intel but I would not
> expect this to have caused any problems.

Nor would I.  Can you tell me:

  - The version of VPython you downloaded?
  - The resolution and color depth of your display?
  - The make and model of your video card?
  - Have you tried downloading the latest OpenGL drivers for your video
card?
  - Can you run a simple Python program that doesn't use the Visual library,
like the following:

for i in range(100):
  print i, i**2



Thanks,

Dave

[Visualpython-users] Running VPython on Win 98

[John P. <jo...@pi...>]

Having recently downloaded and installed VPython under Win 98 I am
unable to progress very far with any of the demo programs that come with
the product. What happens is that the computer hangs, no mouse cursor or
ctrl+break action works. Has any body else had problems like this. My
hardware is running on an AMD chip set rather than Intel but I would not
expect this to have caused any problems.
-- 
John Pickard

Re: [Visualpython-users] surfaces and gradients

[David S. <dsc...@cm...>]

> To kick off, below is my amateur attempt to create a potential surface and
> its gradient (field). I was wondering whether there's a better way of
> handling this. This isn't very smooth but it's fast and gives the
> flavor of
> what I want. Would there be a better way to handle exceptions, like the
> singularites that can result at the origin?

It would be nice to have a "surface" primitive in Visual, but I think your
basic approach is the best way to approximate one given what's in there now.

I would suggest moving the arrows completely above the surface:
   arrow(pos=(i,j,z+0.2), axis=gradf((i,j,z)), ...

As for singularities: One approach is to change the gradf() function to
return the zero vector at the origin.  A terse way of writing this is:

def gradf(v):
    return vector(v) and norm(v)

Alternatively, you could allow norm(v) to raise an exception, and catch it:

def gradf(v):
    try:
        return norm(v)
    except ZeroDivisionError:
        return vector(0,0,0)

> As a suggestion, when I want to clarify for my students the order of some
> vectors (like the unit vectors forming the coordinate basis), I always use
> the order RGB (red green blue) since that is the canonical ordering of the
> colors. Further ordering uses CMY (cyan magenta yellow) since that's the
> common abbreviation for printer inks. This way I don't have to
> use labels to
> slow things down (but the real reason is that I haven't figured those out
> yet -- help?).

The .label attribute does nothing for most primitives and will eventually
vanish.

If you have a very recent build of visual, you can use the new
(experimental) label primitive.  For example:

xhat = vector(1,0,0)
xaxis=arrow(pos=origin,axis=xhat*10,shaftwidth=vsize,color=color.red)
label( pos = xaxis.pos + xhat*8, text = 'X' )

> Finally, I can't seem to define "up" for the screen or view or camera
> objects. What is the format?

The direction the camera points is controlled by the "forward" attribute,
not by "up".  Up only controls the camera's rotation around that axis.

Try this:

scene.forward = (-1,-1,-1)
scene.up = (0,0,1)

Dave

RE: [Visualpython-users] VPython from wxPython?

[David S. <dsc...@cm...>]

> I wonder if there is the possibility to use VPython from a wxPython
> framework, i.e. embedding the VPython scene window into an wxPython
> application. wxPython has OpenGL capabilities but all these lower level
> things are in cpp which don't know (yet?).

Some C++ work would definitely be needed to embed a VPython window into a
wxPython window, but as far as I know you can open a (separate) VPython
window from inside a wxPython application without difficulty.

Dave

[Visualpython-users] Re: Visualpython-users digest, Vol 1 #6 - 2 msgs

[Bruce S. <ba...@an...>]

To Mike Mueller: I'm not competent to comment on the wxPython question you
ask; perhaps someone else will comment.

Concerning mouse control of the scene, there are two things to say. I just
learned that the display object (which governs overall behavior of the
window, such as its height and title) has attributes userzoom and userspin.
These are normally set to 1 (true), but you can turn off the capability:

	somedisplay.userzoom = 0  # user cannot zoom in or out of the scene
	somedisplay.userspin = 0     # user cannot rotate the scene

You can write a program which detects mouse clicks and toggles these
features.

Dave Scherer has proposed a change to the default behavior: Left button
would not zoom unless you hold down some key (such as control), so that
left button down would be used for dragging. This would be more intuitive.
I have seen new users be confused about dragging with the mouse buttons up.
We don't expect to make this change immediately, but it is on the list.

Bruce Sherwood

[Visualpython-users] Re: Visualpython-users digest, Vol 1 #5 - 2 msgs

[Bruce S. <ba...@an...>]

To Robert Irie: Fairly high on our priority list is the ability to import
3D models. Ari Heitner in particular has expressed a strong interest in
this. But it won't happen for many months. There is the problem that there
doesn't seem to be a generally accepted data format for 3D models. DXF is
fairly widely used, but it is very defective in what it can represent.

To Larry Martin: The only object with labels at the moment is sphere, and
its label attribute is considered experimental and likely to go away in
favor of a much more generally useful label object which Dave Scherer is
currently debugging. It has the property that you specify a point in world
coordinates, then specify a line in pixel coordinates from there to a box
containing a label, with text whose height is in pixel coordinates. You can
optionally not show the line or the box.

If you say "scene.up = (1,0,0)" a vertical line on the screen now
represents the x axis. There does seem to be something problematic about
making the z axis the up vector. Note too that individual objects have up
attributes.

Bruce Sherwood

[Visualpython-users] VPython from wxPython?

[Mike M. <mmu...@dg...>]

Hi everybody,

last night I discovered VPython. Playing around with demos and writing some 
of my own small scripts I found it very useful for some real world 
scientific visualization. I like it. With some rather simple commands I can 
get impressive output.

I wonder if there is the possibility to use VPython from a wxPython 
framework, i.e. embedding the VPython scene window into an wxPython 
application. wxPython has OpenGL capabilities but all these lower level 
things are in cpp which don't know (yet?).

I also would like to change the default behavior of the mouse so that if 
anybody wants to zoom or rotate he/she has to turn on a tool (e.g. wxPython 
icon in a tool palette) rather than have the left and right mouse button 
move/clicks doing it as default. This way accidental zooming and rotating 
can be avoided. Can I do this in VPython?


Thanks.

Mike

[Visualpython-users] surfaces and gradients

[Martin, L. <LM...@No...>]

Thanks for signing me up for this list. I'd just started wondering how to
best find answers without pestering any one person. A users group is a good
way of posting ideas and asking questions with possibility of a fast
response.

To kick off, below is my amateur attempt to create a potential surface and
its gradient (field). I was wondering whether there's a better way of
handling this. This isn't very smooth but it's fast and gives the flavor of
what I want. Would there be a better way to handle exceptions, like the
singularites that can result at the origin?

As a suggestion, when I want to clarify for my students the order of some
vectors (like the unit vectors forming the coordinate basis), I always use
the order RGB (red green blue) since that is the canonical ordering of the
colors. Further ordering uses CMY (cyan magenta yellow) since that's the
common abbreviation for printer inks. This way I don't have to use labels to
slow things down (but the real reason is that I haven't figured those out
yet -- help?).

############################################# f-gradf.py
from visual import *
import floatdivision

origin=(0,0,0)
vsize=0.1
xhat=(1,0,0);
xaxis=arrow(pos=origin,axis=xhat,shaftwidth=vsize,color=color.red,
label='x')
yhat=(0,1,0);
yaxis=arrow(pos=origin,axis=yhat,shaftwidth=vsize,color=color.green)
zhat=(0,0,1);
zaxis=arrow(pos=origin,axis=zhat,shaftwidth=vsize,color=color.blue)

def f(r): #the scalar field
    return mag(r)

def gradf(r): #the gradient of the scalar field
    return norm(r)

for i in arange(-10,10,1):
    for j in arange(-10,10,1):
        v=(i,j,0)
        z= f(v)
        a = convex(color=(1,z/10,0))
        v=(i+.5,j+.5,0)
        z= f(v)
        a.append(pos = (i+.5,j+.5,z))
        v=(i-.5,j+.5,0)
        z= f(v)
        a.append(pos = (i-.5,j+.5,z))
        v=(i-.5,j-.5,0)
        z= f(v)
        a.append(pos = (i-.5,j-.5,z))
        v=(i+.5,j-.5,0)
        z= f(v)
        a.append(pos = (i+.5,j-.5,z))
        v=(i+.5,j+.5,0)
        z= f(v)
        a.append(pos = (i+.5,j+.5,z))
        if not (i==0 and j==0):
            v=(i,j,0)
            z= f(v)
 
arrow(pos=(i,j,z),axis=gradf((i,j,z)),shaftwidth=vsize,color=color.red)

#############################################


Finally, I can't seem to define "up" for the screen or view or camera
objects. What is the format?
Any responses, critiques, etc. are welcomed.

Larry Martin, Ph.D.
Professor of Physics
North Park University
Dept. of Physics, box 30
3225 West Foster Ave.
Chicago, IL  60625-4895
(773) 244-5668  FAX (773) 279-7952
http://www.admin.northpark.edu/LMartin
LM...@no...