4SweetB
12-13-2004, 10:12 PM
I have gotten myself into an interesting predicament while playing around with NURBS in XSI (which I'm very much enjoying, coming from LW.. that, and having an operator stack!)
I'm trying to make a lamp cord by extruding along a guide that I can later move around (or hopefully even animate). I made a very nice profile curve of the insulation. With but a few clicks of the mouse, I had an excellent cord. Of course, I had to get ambitious. :)
I decided to make it one of those speaker cord type deals, where you can see the two conductors through the (semi-)transparent plastic insulation. I pulled the curve out of a disc to get a couple nice clean round profiles for the two conductors, and proceeded to extrude them along the same guide as the insulation.
Doing this of course meant that I had to offset the profiles and use "snap to profile", so that each conductor is on its own side of the cord. That looked good, so I went ahead and surfaced the conductors to look like two different metals, as they often do in speaker cords.
From my vantage point, it looked great, but when I checked other angles, I found that as the cord twisted around, the conductors crossed each other!
I immediately realized that this was because snapping to the profiles was causing the paths of the conductors to fail to account for their offsets from the center of the cord (and therefore the positional offsets caused by its rotation).
If I turn "snap to profile" off, then the conductors will correctly follow the rotations of the cord, but will also happen to both run down its center, which is, of course, wrong.
Mathematically speaking, what I should do is leave "snap to profile" off, but have the object centers of both profile curves at the same center as the insulation's center. In theory, this would make XSI have the conductor run alongside the guide curve instead of right through it, while still correctly accounting for the cord's rotations.
Unfortunately, it seems I've found an oversight in XSI (surprising as that is), because it doesn't actually do that; if I offset the centers, it still pulls the geometric centers of the profile curves through the guide curve, NOT the *official* centers of the profile curves like I need it to in this case. I even tried deleting the surfaces and extruding again in case it had compensated for the change in the centers, to no avail.
Can this be for real? Have I really found a limitation? Can anyone think of any magic that might solve this? I suppose I could use three separate guide curves, but that sounds overly complex and harder to animate. There must be an easier way...
Cheers,
Kevin
I'm trying to make a lamp cord by extruding along a guide that I can later move around (or hopefully even animate). I made a very nice profile curve of the insulation. With but a few clicks of the mouse, I had an excellent cord. Of course, I had to get ambitious. :)
I decided to make it one of those speaker cord type deals, where you can see the two conductors through the (semi-)transparent plastic insulation. I pulled the curve out of a disc to get a couple nice clean round profiles for the two conductors, and proceeded to extrude them along the same guide as the insulation.
Doing this of course meant that I had to offset the profiles and use "snap to profile", so that each conductor is on its own side of the cord. That looked good, so I went ahead and surfaced the conductors to look like two different metals, as they often do in speaker cords.
From my vantage point, it looked great, but when I checked other angles, I found that as the cord twisted around, the conductors crossed each other!
I immediately realized that this was because snapping to the profiles was causing the paths of the conductors to fail to account for their offsets from the center of the cord (and therefore the positional offsets caused by its rotation).
If I turn "snap to profile" off, then the conductors will correctly follow the rotations of the cord, but will also happen to both run down its center, which is, of course, wrong.
Mathematically speaking, what I should do is leave "snap to profile" off, but have the object centers of both profile curves at the same center as the insulation's center. In theory, this would make XSI have the conductor run alongside the guide curve instead of right through it, while still correctly accounting for the cord's rotations.
Unfortunately, it seems I've found an oversight in XSI (surprising as that is), because it doesn't actually do that; if I offset the centers, it still pulls the geometric centers of the profile curves through the guide curve, NOT the *official* centers of the profile curves like I need it to in this case. I even tried deleting the surfaces and extruding again in case it had compensated for the change in the centers, to no avail.
Can this be for real? Have I really found a limitation? Can anyone think of any magic that might solve this? I suppose I could use three separate guide curves, but that sounds overly complex and harder to animate. There must be an easier way...
Cheers,
Kevin