RonR
-
Posts
21 -
Joined
-
Last visited
Never
Content Type
Profiles
Forums
Events
Articles
Marionette
Store
Posts posted by RonR
-
-
I'm trying to loft (with no rail mode) a series of complex closed NURBS curves that contain both sharp angles and smooth sections, making a container with flanges on the edge. The curves are all related shapes, but vary in size- a boat hull would be a rough analogy. The part is destined for injection molding.
It seems that to get smoothly contoured results I need to have the same number of vertexes functioning as alignment points on each curve in the series- otherwise the shape twists wildly. Am I correct in assuming that the vertexes work as alignment points only if they are interpolation points, rather than control points? When I select the curves with the loft tool and step through the series of vertexes on each curve in sequence, the alignment guide "skips over" some points that seem to me to be interpolation points. I have tried turning them into points that would be used as alignment points by using the "convert points" option on the 3D reshape tool, but the results are unpredictable. As close as I can look at them the points appear to be on the NURBS curve, but the alignment guide (the little red line) in the loft tool won't snap to them. This seems as if it might be related to an earlier problem I had (I believe it was confirmed as a bug ) in which vertexes would "drift off" the NURBS curve and could not be brought back into alignment.
Am I using the right approach? Is there a way to know the type of point I am dealing with, or is that not what determines the alignment of the lofted shape? Do the degree settings have an influence I should be exploiting? These shapes started as 2D polylines, and were converted to NURBS- would I possibly do better drawing them as NURBS from the start?
Any suggestions for using the lofting tool would be appreciated.
Thanks, Ron R
-
Interesting message from Thomas register. I would guess that since the site does not work with Mac, they see VERY few Mac users.
It's not clear that even their feedback system works with Mac. This would be a great service- I emailed them directly letting them know I would like to use it. More feedback might help.
TRSupport@thomastechsolutions.com
RonR
-
I'm using Mac OSX 10.3.3 on a G5. With Netscape (current version) I came closest to being able to download- I could see the part, but pressing the Download button did nothing.
We do programming here as well, including web development, and the few places I have seen this type of behavior before the culprit is often ActiveX or other Microsoft developed, Windows specific software solutions. These technologies seriously threaten the independence of the Web environment, because they "embrace and extend" (read as ignore and subvert) web standards. Many web developers are trained only to use Windows specific methods, even though standards compliant solutions exist that are just as effective and easy to do. It's the kind of thing that gets my programmer's blood boiling, because there is NO technical or practical reason that the kind of content on this site cannot be delivered in a platform independent form. It's not as if they are writing an application for the Mac- or Linux, Unix, or whatever. It's delivering files over the web.
I can't see any reason why downloading the parts on a Windows machine and copying them to the Mac wouldn't work- though I have not yet tried it. For IGES files coming from Windows you may need to open the files in a text editor (I use BBEdit, which I believe has a free version) and change the line break from DOS to Macintosh and do a "save as" in order to be able to open the files.
RonR
-
Has anyone been able to use this site successfully with a Mac? I have tried with several browsers and can't seem to get it to work.
RonR
-
This is probably very basic, but I'll ask anyway. When drawing solid shapes- mechanical parts, in my case, I often want to create a cutaway view. But my parts, when split, often appear to be shells rather than solids. That is to say the sectioning surface doesn't create a flat surface on the part, and if I split an object in half, what I see is the interior of the skin, rather than the sectioned surface.
What am I doing to cause this to happen?
Thanks, RonR
-
When I have had this problem it has been because of a problem with the underlying object- edges that are not joined is the usual culprit. Try the compose command, or redraw the object. Some modifications to objects seem to "break" them such that the fillet command will not work on all edges. Drawing the same shape via a different method can sometimes allow the filleting to proceed. Perhaps someone with more knowledge can offer insight?
RonR
-
I've had some IGES imports from Solid Works that took a very long time to process (complex mechanical assemblies imported as a single file), and when imported the line and object relationships made them for all intents and purposes impossible to edit. Single lines or arcs came in as many segments, NURBS surfaces as many individual surfaces, etc.
My solution was to import the various parts as individual files, compose them in VW, and reassemble the object in VW. You might try importing a small part of the file, maybe a single object, to prove the process. It would also show you the kind of results you will get.
RonR
-
I don't know if it's related, but check out my thread in 3D solids re. the problems I have been having with snapping NURBS curves consistently- I am also on a G5. A summary:
I have snap to edge points turned on from the constraints palette. When I draw the curve, the "POINT" cue comes on, and the curve I'm drawing appears to snap to the other curve at a vertex. However when I try to create a surface a good portion of the time there will be an error: "surface could not be created from curve network". Usually when I look up good and close, the curves are not intersecting- though the cues were there and made a nice "snap" sound.
When reshaping a vertex, likewise, the points appear to snap together. They show as being in the same position in the object info palette. The points appear exactly coincident, even when I zoom in 50,000% or more.
If I use the NURBS analysis tool to check the intersection, only one 3D locus is placed at the intersection, not 2 as would be if they did not intersect.
But when I select "create surface from curves" from the model menu, I get the message "surface could not be created from curve network".
By only adding one curve at a time, getting a failure, then systematically eliminating intersections with other curves until the surface creation succeeds I can slowly isolate the problem intersections- but snapping the vertexes again does not always succeed. Sometimes it does, and I can add another piece to the surface. But if it doesn't snap right away it does not seem to matter how many times I reposition the vertex with either the "POINT" or "OBJECT" cue showing- the create surface command will fail if that intersection is included. I don't see any difference between the successful and the failed intersections. With a minimum of 170 intersections it's pretty tedious- one failed intersection means redrawing a line with many intersections, any one of which could fail as well.
I'm assuming curves not intersecting is why the Create Surface from Curves fails, but maybe I'm missing something basic here?
RonR
-
Well, now this is getting frustrating- I am still having difficulty getting NURBS curves to make a surface. Here's what I'm doing:
I have snap to edge points turned on from the constraints palette. When I draw the curve, the "POINT" cue comes on, and the curve I'm drawing appears to snap to the other curve at a vertex. However when I try to create a surface a good portion of the time there will be an error: "surface could not be created from curve network". Usually when I look up good and close, the curves are not intersecting- though the cues were there and made a nice "snap" sound.
When reshaping a vertex, likewise, the points appear to snap together. They show as being in the same position in the object info palette. The points appear exactly coincident, even when I zoom in 50,000% or more.
If I use the NURBS analysis tool to check the intersection, only one 3D locus is placed at the intersection, not 2 as would be if they did not intersect.
But when I select "create surface from curves" from the model menu, I get the message "surface could not be created from curve network".
By only adding one curve at a time, getting a failure, then systematically eliminating intersections with other curves until the surface creation succeeds I can slowly isolate the problem intersections- but snapping the vertexes again does not always succeed. Sometimes it does, and I can add another piece to the surface. But if it doesn't snap right away it does not seem to matter how many times I reposition the vertex with either the "POINT" or "OBJECT" cue showing- the create surface command will fail if that intersection is included. I don't see any difference between the successful and the failed intersections. With a minimum of 170 intersections it's pretty tedious- one failed intersection means redrawing a line with many intersections, any one of which could fail as well.
I'm assuming curves not intersecting is why the Create Surface from Curves fails, but maybe I'm missing something basic here?
VW- it would be a lot easier to troubleshoot this stuff if the curves missing an intersection were indicated.
RonR
-
Thanks for the suggestions- I've tried to use the snap to edge constraint, and it does work to create intersections when drawing or dragging vertices, but does not appear to work when entering values from the keyboard.
What I would like to be able to do is to draw a NURBS curve constrained in one dimension that will snap to intersect with my other NURBS curves - like a topographical contour line, or the waterlines on a boat hull intersecting with cross sections through the hull.
I may be missing an obvious way to do this, as my experience is limited, but if I have
1) the constrain to working plane constraint turned on, and
2) the snap to edge points checked, and try to draw a NURBS curve, the snap to edge points does not work to place the vertex at the intersection of the working plane and the curve I'm trying to intersect with.
If I don't have constrain to working planes switched on, the vertex is not placed at the working plane either. Is there a simple way to draw something like a contour line that would intersect other curves and constrain it in one dimension?
My workaround has been to create a NURBS surface, position it sectioning through the curves, use the NURBS analysis tool to place a 3D locus at the intersections, then draw a NURBS curve through the loci. However when I do this many of the curves don't actually intersect, although they appear to do so in the drawing. I'm going to try this again, though, I'm not too sure that I have had the snap to edge constraint turned on when I was doing this.
RonR
-
I'm using 10.5. I may not totally understand the snap features in the new release, though. Will they work to snap a vertex to another object when you are only moving a single vertex, not the whole object?
I'm fairing complex curved shapes by adjusting the NURBS vertex of the defining curves through keyboard entry, and I'd like to be able to make sure the two curves intersect when I move the two vertices to the same location.
Thanks,
RonR
-
What would be incredibly useful to me- maybe there is a way to do this, but I haven't seen it- is the ability to snap a given NURBS vertex precisely to a vertex on another NURBS object.
I'm creating very complex curves and would like to use the create surface from curves tool for some objects, but find that I can't reliably edit the vertexes to create intersections. Snap to objects works if things are drawn exactly right, but the intersection is lost if you do any keyboard editing of the vertex positions- even keying them into the same X/Y/Z positions does not create an intersection.
Does anyone have a suggestion for ensuring that given curves intersect at a particular location in space?
RonR
-
I appreciate VW's timely and frequent updates and upgrades. But the link doesn't give any information as to the nature of the incompatibility or the suitability of the Beta update for the current OSX versions. I'd like to know more so as to make a decision to update/ upgrade, Specifically; does the beta update run under OSX 10.2.xx, or is it ONLY for Panther?
RonR
-
will confirm what Mike B says, with one small caveat. While VW is not aware of dual processors, the OS is, and some other apps, notably Photoshop, are so there is still a processing speed advantage with the dual system.
It does somewhat depend on what you are doing. I'm doing complex 3D solids modeling, all curved shapes, no straight lines, and lots of rendering. This is the kind of math the G5 processor shines at. I moved from a G4- 500/1 gig to a G5 1.8 single processor/1.5 gig (only because the dual wasn't shipping- today I'd get the dual). A render that took my old system 4.5 minutes completed in about 12 seconds on the G5. Solids operations are incredibly faster. It's a really big leap forward.
I'd guess RAM is very important if you are doing large (lots of pixels) renderings. At one point I was running VW and several other apps while working on two 400 MB photoshop files, and with 1.5 gig of RAM I could see some delays.
I'm not sure about VW, but most apps can't at his point use nearly the amount of RAM the G5 can install, and won't be able to until they are rewritten to take advantage of the 64 bit processor.
It's not just processor speed, either. The data pathways in the G5 system- at least the 1.8 and dual- are a lot faster, and the systems should get more responsive as the OS takes advantage of the processor architecture.
Video cards make a big difference with animations, movies, and games. As fast as the G5 is I doubt you would detect a difference among them in using VW.
Actually, one of the parts of the new system that speeds me up the most is the 23" cinema display...
RonR
-
How are you generating the curves for the lofted surface?
I've noticed that the way the original curves (that the the loft is generated from) are created can make quite a difference in the number of vertexes, and thus calculation times.
Draw a NURBS curve with 7 or 8 control points that sweeps back and forth. Check the number of vertexes. Then use the Offset by Distance tool to create another curve from the first a short distance away, and check the number of vertexes. It will increase dramatically.
I needed to create a shell object by first drawing the inside contours so I could clear components, then offsetting to get the outside contour so I could shell to the inside- shelling to the outside left gaps. On lofting I ended up with over 23,000 vertexes.
I went back and redrew the outside contours with the minimum number of vertexes to maintain the shape, using the offset curves as a guide. I lofted a shape with only a few hundred vertexes.
RonR
-
We have a mixed PC and Mac environment here, I got a G5 1.8 mz single processor about two weeks ago specifically because of the lengthy calculation times in VW as my models increased in complexity. I would have gone for the dual processor, but I was shooting for a deadline and they weren't shipping yet.
The performance reports for this system on the NNA site are, if anything, understated. At least they seem so to one who waited a long time to upgrade and thus was using a slower system than the ones they are comparing to. A complex render I did for comparison took 4.5 minutes on my old G4-500 mz, on the G5 it was done in just over ten seconds. Other processes are comparably improved. And it's quiet, too. I can't decide which I'm happier with- the G5 or the 23" LCD display- both provide tremendous improvements in productivity. Now I'm glad waited so long to upgrade.
RonR
-
re. Turboman, "I need to loft a precise airfoil surface using 2 or more section profiles originally entered as 2D loci. While the 2D "compose curve" command does an acceptable job converting the loci to smooth curves, the subsequent "convert to NURBS" (required for lofting the 3D surface) introduces too much distortion.'
I have been generating complex curved 3D shapes by drawing 2D cross sections, converrting them to NURBS, moving them in 3D to the appropriate seperation, and using the lofting tool to generate the shape. The lofting tool does not require the intersecting curves. Compose curves has worked well in some situations, but for the airfoil shape you are describing I suspect the lofting might be easier.
For some irregular shapes I needed to fair I placed a NURBS planar surface at various points intersecting the curves, used the NURBS analisis tool to place a locus at the intersections, then drew a NURBS curve that snapped to the loci. This seemed to show up any distortions in the shape before using the lofting tool, so I could adjust the curves precisely using keyboard commands to move individual vertex points.
As has been mentioned, the irregularity you see in the display is the 3D conversion resolution, not the actual data. I put my conversion settings up high for smoother display since I got the new G5 Mac- the performance reports for this system on the NNA site are, if anything, understated. A render I did for comparison took 4.5 minutes on my old G4-500 mz, on the G5 1.8 single processor it was done in just over ten seconds. And it's quiet, too. Now I'm glad waited so long to upgrade.
RonR
-
I have been trying to do precise 3D work in VW, in my case with little CAD experience, also on a Mac (under OSX 10.2.7). I've found that it is generally possible (and I have been dealing with extremely complex and irregular shapes). I've had to learn the limitations of each object type and plan ahead in the process of doing the drawing. In the process I have been down quite a few blind alleys. The examples and tutorials are a bit basic and don't address the overall strategy for creating complex 3D designs. Vector Works is easy to start using, and easy in 2D design, but there are a lot of capabilities that have not been so easy to figure out- my assumption is that someone with more 3D experience would have an easier time.
Things that have helped me:
Extensive use of the "Edit Group" command allows you to regress to an earlier phase and change objects precisely, while maintaining their position in the drawing.
Snapping the working plane to an object surface and aligning other objects with reference to the working plane.
Placing a 3D locus and aligning the objects to the locus.
In some situations precise alignment is possible using the align 3D command- it depends on what part of the object you are trying to align.
The very recent update to VW (10.5) has been incredibly useful, particularly the improved snap 3D and most particularly the NURBS analysis tool. This latter allows you to query the relative position of NURBS lines and surfaces and place lines or loci at intersections or at the nearest points. I've used this to precisely fair complex curved shapes.
I don't completely remember details of version 9, but the latest versions have much greater capability in 3D mechanical design. I feel the upgrades have been well worth it.
I still am a little unclear about how to get two NURBS curves to intersect reliably (as in using them to Create Surface from Curves). I can position the vertex of both lines with the keyboard at the same position to the level of precision specified for the drawing, but still not have them intersect. If anyone can clue me in I'd appreciate it.
Good luck.
RonR
-
I am attempting to model an injection molded instrument case that has a quite complex shape- essentially two cylindrical tubes (for batteries) blended into an upper section that is an elongated curved (submarine hull) shape. It needs to hold an LCD, circuit board, and batteries and be as small as possible given the components. There is a cap on one end and a threaded fitting on the other.
I'm using VW 10.1 Mechanical, Mac OSX 10.2.6, G4/ 500/ 1G ram.
My strategy was to
1) Place the components that I need to house in a 3D drawing
2) Draw smooth polyline contours in 2D to determine clearances and general shape
3) Use this shape to determine the boundaries of my cross sections
4) Draw a polyline cross section at the largest diameter
5) Duplicate this polyline and edit the (bezier) vertexes to describe several other cross sections
6) Convert the polyline cross sections to NURBS, and place the NURBS cross sections in 3D space where they belong.
7) Use the loft tool to create the shape.
8) Shell the solid
9) Section the solid to create two halves and an opening for a clear screen.
10) Add separately drawn interior details, supports, ends, etc with the add solids command.
Right now I am breaking down with the Loft tool. If I loft only the three cross sections that are very close in size and shape, a smooth shape is created. If I add another cross section- one where the shape changes, the loft yields a twisted or creased and unfair shape, and I can't seem to untwist it. The object is symmetrical, like a boat hull. The NURBS curves have exactly the same number of vertexes and are all variants of the same basic shape. I try to specify the alignment in the loft creation dialog so the points on the guide line line up- say along the top centerline, where they should be exactly in a straight line. But adjusting in 1 degree increments does not give me the ability to line them up exactly, and the lofted shape appears to have some "wrinkles" in it. I've tried to attach an image that shows how the lines don't quite line up.
I feel there must be something I am not understanding about how to use the Loft tool- or perhaps my strategy is not the best? Would another tool or approach such as creating surface from curves be more appropriate, or would it be better to loft the shape in sections and then combine them? I would appreciate any comments about the overall strategy, as well as any insight into the loft tool and aligning points.
Thanks,
RonR
-
I need to create a recessed rim, beaded gasket, and mating surface inside a small instrument case. I have tried using the extrude along path command, following a NURBS curve for the path and selecting all three objects to extrude so I maintain a precise relationship between them. Unfortunately I seem to have trouble separating the objects after the extrude- I need to combine each with a different solid, and when I ungroup, they separate into (many) individual NURBS surfaces, not into the shapes defied by the original extrude objects.
My questions are:
1) Is it possible to define the precise point on an object that follows the path? I don't see any reference in the manual, but the path appears to track the object center. If I could define a point that that followed the extrude path I could treat the objects separately and still keep a consistent fit after the extrude.
2) Can I extrude multiple objects along one path and separate them after the extrude?
Thanks,
RonR
Help with Lofting tool?
in Solids Modeling
Posted
Kaare,
Thanks,
Breaking the curves up into smaller segments was going to be something to try. I have tried without the flanges, and they actually seem to aid in the alignment, as the guide always finds the vertex if it's a corner. If I can get the same number of vertexes in he right positions to be found by the point alignment guide in the lofting tool, the loft works beautifully.
Can any VW gurus tell me how the tool choses the vertexes it uses for point alignment? I feel if I can understand this I will be able to solve the problem.
RonR