| XObjView: Select Object and Opacity Problem [message #41600] |
Tue, 09 November 2004 02:26  |
robertschaefer
Messages: 10
Registered: August 2004
|
Junior Member |
|
|
Hello,
I want to select an object in XObjView and rotate it,
but other objects should be fixed.
I read in the refguide, that it is possible
to add Arrays to XObjView. So I think it should work,
but how?
Is there a posibility to realize this idea?
My first try was with the SELECT_TARGET property
but it seems to me that nothing changed in View.
An other way I thought about is to create to Models
theModel_1 = Obj_New('IDLgrModel') and theModel_2 = Obj_New('IDLgrModel')
give them rotate,scale and Translate Options, because I think
every Model would need its own zeropint. But when I zoom, in
XObjView it always depends on one Point.
Any ideas?
My second problem deals with the alphachannel.
It works, and I can change the opacity. the front
view in XObjView (x-y level) is good, but when
i rotate my data and have a look at y-z level i have
stripes. Could this depend on the alphachannel?
Or maybe the slices, because i scaled the stac?
Thanks for any help and ideas.
Regards Robert
|
|
|
|
| Re: XObjView: Select Object and Opacity Problem [message #41690 is a reply to message #41600] |
Tue, 09 November 2004 10:53  |
Rick Towler
Messages: 821
Registered: August 1998
|
Senior Member |
|
|
Robert Schaefer wrote:
> I want to select an object in XObjView and rotate it,
> but other objects should be fixed.
Aside from David's suggestion, I think that you have moved beyond the
realm of XOBJVIEW. XOBJVIEW throws everything in a top level model,
gets the extents of that model, sets the viewing parameters and draws
the scene. There is no way to select and manipulate individual
objects/models.
You'll have to build your own viewing system.
You may want to look at my camera object and some of the demos. I don't
have anything that demonstrates picking objects, but maybe you'll get
some ideas/inspiration from them.
http://www.acoustics.washington.edu/~towler/
The CSS is broken in Firefox and has been for months but you still
should be able to navigate and d/l files. I apologize. It will be
fixed some day...
> My second problem deals with the alphachannel.
> It works, and I can change the opacity. the front
> view in XObjView (x-y level) is good, but when
> i rotate my data and have a look at y-z level i have
> stripes. Could this depend on the alphachannel?
> Or maybe the slices, because i scaled the stac?
This is a problem with rendering order. A weak explanation follows:
Assume you have 2 planes orthogonal to the vector [0,0,1]. A
semi-transparent one containing the point [0,0,0.5] and the other opaque
containing the point [0,0,-0.5]. Also remember that your viewing vector
is *always* fixed at [0,0,-1].
When you rotate these two objects 90 deg. it gets difficult to separate
the objects properly in terms of rendering order. At 85 deg. the front,
semi-transparent object, is still in front of the back object in regards
to Z. But at 95 the "front" object's vertices are now more negative in
terms of Z than the back object. In between you have a mix, some of the
verts of the front object are +z some are -z. It is in this case where
you see the banding as IDL renders some of the triangles for the back
plane *after* rendering some of the triangles of the front plane.
You can verify that this is your issue by continuing to rotate the
objects. If, as you rotate, the banding fills in and you lose part or
all of your semi-transparent object then you are in fact seeing this.
For the gory details of rendering order and semi-transparent objects,
google the group for "Pimento Problems". In short, there aren't any
easy clean fixes. If you are using the alpha channel you have to
manually sort all of your objects in regards to Z. With simple scenes
this isn't that difficult. With complicated scenes/geometry it is very
difficult.
If the banding only occurs where your objects are very close to each
other then it is probably a traditional z-buffer precision issue. In
this case you'll need to adjust the eye property of your view, moving it
further away from the near clipping plane to increase z-buffer precision
at distance. But I doubt you would have this problem with XOBJVIEW.
-Rick
|
|
|
|
| Re: XObjView: Select Object and Opacity Problem [message #41697 is a reply to message #41600] |
Tue, 09 November 2004 05:35 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Robert Schaefer writes:
> I want to select an object in XObjView and rotate it,
> but other objects should be fixed.
> I read in the refguide, that it is possible
> to add Arrays to XObjView. So I think it should work,
> but how?
Well, here is an example of how you can get this
to work, sorta:
;********************************************************
PRO Example
; Create contour object:
oCont = OBJ_NEW('IDLgrContour', DIST(20), N_LEVELS=10)
; Create surface object:
oSurf = OBJ_NEW('IDLgrSurface', $
DIST(20),INDGEN(20)+20, INDGEN(20)+20)
; View model:
XOBJVIEW, [osurf], Stationary=[ocont]
end
;********************************************************
I think you are probably going to have to eventually
give up on XObjView though. You are pushing it now
almost beyond the breaking point of where it was
meant to go. :-)
>
> An other way I thought about is to create to Models
> theModel_1 = Obj_New('IDLgrModel') and theModel_2 = Obj_New('IDLgrModel')
> give them rotate,scale and Translate Options, because I think
> every Model would need its own zeropint. But when I zoom, in
> XObjView it always depends on one Point.
I think this is a result of what XObjView was designed
to do originally. As I say, I think you are pushing it
well beyond this point.
> My second problem deals with the alphachannel.
> It works, and I can change the opacity. the front
> view in XObjView (x-y level) is good, but when
> i rotate my data and have a look at y-z level i have
> stripes. Could this depend on the alphachannel?
> Or maybe the slices, because i scaled the stac?
It could depend on a lot of things. One of the problems
with object graphics, of course, is that *everything*
is possible. It makes the Universe of potential problems
rather large as well. :-(
A better description of the code might help. And, of course,
it always pays to read about the "pimento problem" in the
IDL newsgroup archives. :-)
Cheers,
David
--
David W. Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http:/www.dfanning.com/
Phone: 970-221-0438, IDL Book Orders: 1-888-461-0155
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
