Jump to content

P Retondo

Member
  • Posts

    1,914
  • Joined

  • Last visited

Posts posted by P Retondo

  1. Christian, I use OpenGL a lot. My system:

    VW 10.1.2

    Win2k/WinXP

    P4 2.0G

    768M RAM

    NVIDIA GeForce3 Ti200

    You may be suffering from too slow a processor, or maybe your video card / driver has some kind of problem. I would suggest there might be some kind of file-specific problem, but it seems as though you have experiemented with a variety of files.

    [ 09-03-2003, 02:13 AM: Message edited by: P Retondo ]

  2. Chris, you probably know about multiple extrusion for tapering objects.

    I would try using a nurbs / extrude along path for the curving eave. For the path, you can create a 3d polygon that basically traces the eave line.

    Robert - is there any reason not to use this approach? I've tried the polygon / "short segments" approach in the past for curving eaves, but it would be nice to get some more precise geometry and better 3d rendering using nurbs. The big question is whether this creates a huge number of calculations for rendering and other tasks. I know that when I take a nurbs extrusion and cut a 3d section from it I get an enormous number of 3D polygons.

    [ 09-01-2003, 03:49 PM: Message edited by: P Retondo ]

  3. Cools, you need to give tech support your system characteristics - VW version, OS, platform, etc.

    I have had this happen. Hope you have a recent backup, because that was the only way I could salvage the file. Lost a few hours of work.

    VW 10.1.2

    WinXP

    G4 2.2G

    1G RAM

    PS: I now save a backup version of my file every hour or so to a local (non-server) hard drive using the "save copy as" command.

    [ 09-03-2003, 02:15 AM: Message edited by: P Retondo ]

  4. Rodger, if you are saying that you want to start all over again, I have had difficulty with this, as well. I assume that when you say "deleted the existing site model" you mean that you have used the "delete" command in the sitework menu - or maybe you mean that you have deleted the symbol that was created for that model. In either case, VW still has remnants of that original model - a class instance of some kind still exists, maybe - and you will have trouble starting from scratch.

    The "proposed" site model is created to allow the program to calculate cut and fill comparisons with the "existing" model.

    I think the "delete" command does not do the job it should, and that a programmer should look into making sure that all vestiges of a terrain model are wiped out so that we can start over again. Or better yet, allow more than one or two terrain models to be created, under different names.

    The workaround is to copy your layers, except for anything having to do with DTM, into a new file. That file will then contain no reference to a DTM, and you can start fresh.

    On the other hand, if you say that you are having trouble creating the "proposed" model by making modifications to the "existing," then I suggest that you carefully follow the manual's instructions. It's a little non-intuitive, but it will work.

    [ 08-25-2003, 12:48 PM: Message edited by: P Retondo ]

  5. I just created a new class in a document, and attempted to assign it to a group of lines by entering the group, selecting all the lines, and changing the class assignement in the Object Info Palette. This causes the program to crash, invariably. I can select one line, and assign it to the new class. If I select more than one object (e.g., 2 lines), the program crashes. If I assign the group of lines to class other than the new one, no problem.

    Ungrouping the lines and attempting the assignment to the new class also causes a crash (BTW, this is a multiple window runtime error crash, in other words, I have to click on a runtime error message at least three times, sometimes many more, before the program quits).

    This looks like a bug. Is NNA aware of this, and are there any work-arounds? I am sending a file and a copy of this post to bugsubmit.

    VWA 10.1.2

    WinXP

    P4 2.2G

    1MB RAM

    [ 08-24-2003, 02:49 PM: Message edited by: P Retondo ]

  6. Archken, there are improvements, but not as dramatic as you would want! I don't think this is the decade for great leaps in digital technology. But, obviously, I've felt it was worth upgrading to version 10. I'd say that 10.1.2 is the best version since 8.5.3, and offers some significant new capabilities over version 8. The "compose" and "decompose" commands are worth checking out.

    Mac users seem to have more complaints about OS-related instability than Windows users. I hear that v 10.5 is coming out in a month or so, which may have solved some nagging performance problems with version 10.1.2.

    [ 08-21-2003, 01:21 AM: Message edited by: P Retondo ]

  7. The workaround suggestions here are sound, but, all the same, propstuff's request is a great one. We shouldn't have to go to the trouble of creating a view or go through extra steps to be able to look at a given object from all sides quickly and simply. I'll bet implementing this function would be pretty easy - VW already does this kind of thing with the zoom commands, where if an object is selected it is centered in the zoomed view. In fact, this is what I use to get my object back in the center of the picture again after changing point of view.

  8. propstuff, I tried the circle and line experiment, in a slightly different way. I used the "split by point" tool to make the circle into an arc at the intersection with a line, and found what I think is a bug. When I split a circle at such an intersection, it became an arc of less than 360 degrees. One side of the arc lay on the intersection, the other side some distance away. This appears to be a replicable problem. When I split that arc again at another line intersection, the arc end point did not lie on the intersection. I could actually draw a line between the split point and the actual intersection that had a fairly large magnitude (L = .08" on an arc of 3' radius).

    I will submit a file illustrating this problem to bugsubmit. It is likely that the same chunk of code involved in this bug is responsible for the problem you are observing.

    On further experimentation, it appears that a 360 degree arc is an illegal object - try to make one and see what happens. I also found in a couple of test files that the "split" tool failed to operate on an arc at the intersection point, and I had to create a fresh file, in which the exact same operation worked.

    [ 08-20-2003, 01:54 AM: Message edited by: P Retondo ]

  9. If the property line doesn't close, there are two possible explanations: 1) you made an error, or 2) the description of the property is defective. I usually find that there is a small gap to close at the end - less than an inch. You can just reshape the polyline to close the gap, or let it be.

    In addition to being aware of the 180 degree thing, you have to be up on your degrees-minutes-seconds. Take a look at the manual or the help topic - it's easy once you get the hang of it.

  10. Jan15 is absolutely right. The problem using the polyline tool is that when you offset the centerline, you will get thousands of little facets as the road edges instead of a set of arcs and curves. One solution is to "decompose" the polyline into arcs and straight lines before you offset. If you have your straight lines and arcs meeting in a tangential relationship, by offsetting them their intersection will still be in a tangential relationship. This is the correct way to draw roads, even if they are laid out in the field less precisely than they are drawn.

  11. broc, I use the Property Line Tool, which is part of VW Architect (that is, when I don't have a CAD file from a surveyor or civil engineer). Enter the bearings and distances between points in sequence to create the boundaries of the property. I can then see if the legal description of the property closes or not, and be aware of any discrepancy in that regard. I sometimes have to adjust the bearing by 180 degrees, because often it is hard to tell whether the bearing on the map is taken from point A to B, or from point B to A. That's easy to do - if your new line goes off in the wrong direction, back up a step and enter the bearing as before, but add "-180" to the angle.

  12. I think it would be a groundbreaking advance to be able to set the appearance of windows and doors (line weights and fills of various parts) using a graphic interface. I.e., the user would be presented with a representation of the window, could then edit that representation using standard VW tools until it looks the way he/she wants it, then have the option to save this appearance as a default preference or to apply it only to the given instance of the object.

    This kind of interface seems to me to be the best way to deal with the often complex settings for parametric object tools. I can forsee the time when we will have more 2D/3D interaction, and we will want to represent entities differently from different viewpoints (plan, section, elevation). To see how NOT to manage the graphics, look at ACAD's ADT "styles" matrices.

  13. Extrusion along two rails, or some equivalent, is really important. Here's an example: try to create a curving stairs rail that comes down a stairs, levels out, and ends in a volute. You would want the base of this rail to always be level in cross section. If you attempt to extrude the shape along a nurbs curve, the shape and hence the base twists unpredictably. If you could set two rails that will define the lower corners of this cross section, the desired effect could be achieved. Alternatively, the extrude along path tool could be provided with some other user control over twist.

    limbrost: setting the smoothing angle in the preferences dialog box, as Alexandre suggests, really helps with the extraneous line clean up. There are occasional glitches, but usually it works.

    I personally prefer the way VW marquee selection works, and as iboymatt points out, you can have the ACAD method with very little extra trouble.

    [ 08-06-2003, 12:56 AM: Message edited by: P Retondo ]

  14. Fred, is this utility a lot simpler than entering the commands "ctrl + d" to duplicate the wall (assuming you have "duplicate in place" selected), them "ctrl + m" to move it? I admit that I use the offset command quite often with lines, as opposed to the above procedure.

    To answer Robert's question, I use the above procedure very often with walls because it is the easiest way to lay out spaces with known dimensional relationships. My current method to lay out a building is to create short lengths of walls with the desired heights/thicknesses/cavities/etc., then duplicate, rotate, and move them to the correct locations, then join them all.

    [ 08-03-2003, 01:24 AM: Message edited by: P Retondo ]

  15. There is an interesting relationship between the "object oriented" concept behind BIM, interactive sections and elevations which might be the full equivalent of VW's "hybrid" 2D plan view of a 3D model, and the notion of interoperability. I think the first two ideas are pretty useful, but have doubts about the third.

    Interoperability and the ISO/STEP attempts to come up with standards are absorbing the time of a great number of talented programmers. The problem, fundamentally, is that they are dealing with a moving target. I have the feeling that by the time a "standardized" solution has been created, the real world will long have moved beyond the parameters with which they have started.

    Technology does not develop according to a grand unified plan. It just happens. That may be messy, but that's just the way it is. If NNA focuses on support for file formats of proven utility, such as .dwg for working with ACAD and maybe .stl for freeform solids (I'm sure there are others), that would be good enough for me.

    In the meantime, I find that the comments which led this topic (about idea two) are crucial to closing the gap between the object-oriented model and the ultimate work product of an architect's office.

    [ 08-03-2003, 01:06 AM: Message edited by: P Retondo ]

  16. NNA may not be willing to endorse a particular brand. I can recommend the system I've been using:

    Dell 8200, P4 2.2G processor, 1 MB RDRAM, GeForce 4 Titanium graphics card w/ 128MB memory, WinXP.

    No problems to date with VW 10.1.2, aside from a couple of bugs in the software. In fulltime daily use, I'd expect a crash once a week, maybe - unless you're doing 3D->2D conversions, due to a problem I think is a bug in VW. In a year, I've lost work only once, which took an hour and a half to reconstruct, and that was probably due to shutting down the system improperly.

  17. Wish list item: a command to turn a schedule into a regular worksheet with each cell containing an ordinary value, so that we can edit and format the worksheet at will after generating the information. Not a perfect solution, you would still have to do the conversion again after making changes, but better than the current situation.

  18. I want to tranfer this discussion from the "Tech How To" board to get some attention to what has become a real difficulty for me. I find that it has become very inefficient to translate a 3D model into 2D objects once a certain level of complexity has been reached. My model is of a large single-family residence, with windows, doors and a very moderate level of trim detail.

    VW 10.1.2

    Win2000/XP

    P4 2.2 G

    1 G RAM

    Every time I take a 3D section and render it as lines (to get interior elevations), I am required to quit out of VectorWorks and restart. Otherwise, the program slows to a crawl. In fact, it usually refuses to quit in any reasonable (minutes) period of time, and I have to end the program using task manager. Any process of selection or de-selection can take seconds, escalating to minutes, after making the 2D rendering.

    I see two problems here. One must be a bug - the program shouldn't behave abnormally after doing some processing. Others have complained about slow operation, and this bug may be related.

    The other has to do with the way 3D sections are taken. There must be a better way than converting everything to a million 3D polygons.

    Also of note: I'm still getting multiple overlapping lines when generating a hidden line rendering from a 3D section.

    Is NNA aware of these problems? Is there a chance of getting fixes on a high-priority basis?

    I would suggest that solving the structural problem with 3D sections may start with a command to convert layer links to an actual copy of the objects from the linked layers. This may or may not be useful to many people in and of itself, but from there it might be possible to sort those objects into those to be thrown out, those in the background, and those actually being cut and which will require some kind of trimming.

    Even better: when using the 3D section tool, we are often not actually after a cut up part of the model - we are after a view or representation that we can't get any other way. If we could get that without actually copying and manipulating 3D entities, things might be done much more efficiently. For example, how about creating a plane that the rendering engines interpret as boundary between completely transparent stuff and visible stuff? If that view could be then converted to 2D lines with the hidden line rendering tool, bingo!

    [ 07-21-2003, 02:45 PM: Message edited by: P Retondo ]

×
×
  • Create New...