Jump to content

line-weight

Member
  • Posts

    2,409
  • Joined

  • Last visited

Everything posted by line-weight

  1. A cavity closer is a different thing to a cavity barrier of course.
  2. Any reason not to use simple vertical extrudes? Or is the problem then how they interact with wall objects when seen in section? How do you use slabs for the horizontal ones - do you somehow wrap certain wall components around them? If a tool were to be developed for this I'd suggest considering a general-purpose "elevation embelishment" tool that could deal with different kinds of linear elements within/on an elevation, so that it could also be used for things like edging trims, or maybe even rainwater goods. (Really I'd prefer all the other tools to be fixed before developing an inevitably half-completed and ten abandoned new one though)
  3. I think I'd definitely spend the money on extra RAM (and perhaps also HD size) rather than the better processor, as far as VW is concerned. My impression is that VW currently barely takes any advantage of the extra processing power except for RW renders. It's the memory that slows things down. On big files my M1 mini is pretty much constantly using as much of the 16GB RAM as it can get its hands on, and then frequently using 20 or 30GB of swap memory.
  4. By the way something I don't understand about the memory figures in Activity Monitor... If I add up the memory usage of all the apps it comes to >48GB But if I add "memory used" + "cached files" + "swap used" I only get about 32GB. Why don't they match up?
  5. Well I just spent a bit of time doing my favourite procastination/avoiding real work activity: unpaid beta testing for VW. I've compared VW2021 SP5 with VW2022 SP4 (as part of my decision making process about whether I dare move to using VW2022). This is using quite a large file and it's on my mac mini M1 16GB running monterey 12.2. The numbers are gb of memory use by VW according to Activity Monitor. Note that 2022 seems to behave better, memory wise, up until step 7. However, once I untick "save VP cache" in document preferences everything starts to go wrong. I have several sheet layers in this file which contain a large number (eg 44) of renderworks viewports. Where these have been rendered out in a previous session, and the completed render images are retained in the file thanks to "save VP cache" being ticked, then VW2022 can deal with that sheet layer. It doesn't use loads of memory, and I can pan and zoom about it reasonably well. But once I've unticked that box, and reopen the file, all of those VPs are shown with a wireframe view of the model, because the rendered images haven't been cached. This causes issues in VW2021 and VW2022 - a much larger amount of memory is used, and panning/zooming around that sheet layer becomes very slow indeed. But it's worse in VW2022 than in VW2021. And when I want to switch to another, similar sheet layer, then VW2021 can cope with this but VW2022 can't. What this means for me, at least with this file, is that I potentially can't operate using VW2022 because as far as I can see, there may be certain sheet layers that I simply can't access. I potentially can't even get at those sheet layers to reduce the number of viewports to a level that VW2022 can deal with. Maybe by keeping "save VP cache" ticked, I can render a sheet layer, save and close the file, reopen it, move to another sheet layer, but that's very laborious especially if it's necessary each time I change something. Or maybe if I left things running with memory at 56gb and beachballs spinning for long enough it would eventually settle, but I suspect not. I'd be interested if anyone else can replicate the same thing. My guess is that you can, if you just increase the number of VPs on a sheet layer to some threshold amount.
  6. So frustrating that it's not possible to add memory to the M1 minis.
  7. You're more likely to get help if you do as much work as you can yourself. So, provide a vectorworks file for us to look at, with the DWG already imported. Anyway, I imported the DWG into a VWX file. Looks to me like the spot heights for the general groud level are labelled "NS" in pink along with a height value (eg 13.15). There is a crosshair symbol superimposed on a 3d locus for each. Examining the 3d locus suggests its X, Y and Z values are correct (but you will need to tweak import settings to make sure the units match what you want). So it looks like it's these 3d loci you want to extract to make your site model. Your surveyor seems to have provided a reasonably well organised drawing (not always the case!) and those 3d loci seem to be in the class "NS". So it's quite easy to extract them all using the "custom selection" tool using these criteria: Use this to select those 33 3d loci, and follow the good advice from @jeff princeabove - get them into a new layer. And ignore the contour lines drawn on the survey. Then you can use them to create the basis of your site model. It'll be up to you to interpret the other height data on the survey, some of which relate to building heights. Some of it might be useful measurements of local ground level adjacent to buildings ... quite how you integrate building models with a site model is less straightforward (and in my opinion not well catered for in Vectorworks). You'll need to do a fair bit of homework to understand what the best approach is for you. It'll depend what you need to use the models/drawings for.
  8. Does SP4 continue to have the problem of memory usage gradually creeping up and up and up the longer the file is open? (I've only done a bit of testing so far, haven't tried to spend a day working in it)
  9. Again no improvement, as far as I can see, in SP4. @jblock or @Dave Donley are you willing to give us any update on this?
  10. I've also repeated this test in VW2022 SP4. Things are perhaps a little better - such that VW2022 behaves not significantly worse than VW2021. Navigating a large sheet layer with multiple viewports remains painfully slow.
  11. I've just repeated these tests on VW2022 SP4 (again on my 16GB M1 mini) and am sad to say that I see no improvement.
  12. I'd be interested to know if anyone has seen any improvement with SP4 now released. My initial testing suggests: not really.
  13. Yes please! Worth putting the explanatory video in this thread too:
  14. That detailer tool looks incredibly useful. Why don't we have it already!
  15. I don't understand what the second image is showing - what are the blue boxes? As for the first one, the "successful" viewport on the left, I'd just duplicate that, then adjust the crop on the duplicate so it only takes in the bottom part of your worksheet, this will give you a viewport that's not in the location you want but you can then just move this alongside the other one. I use this approach quite often - it's a pain to have to do it, and there ought to be a better way, and people have asked for this for some time, but with no success.
  16. Select the viewport In the viewport OIP press the "classes" button Brings up the "viewport class properties" dialogue Select all the classes in the list in that dialogue, and press "edit" button Then under graphic attributes>pen>Thickness, set the line weight you want (but don't touch anything else) That'll overide the pen thickness for all classes viewed in that viewport Does that work?
  17. I wonder what it does when the walls aren't flat, or vertical, as is often the case in old buildings? My human-generated surveys will often record these things but my human brain will then decide whether it matters that the model includes them. Sometimes there's a good reason to pretend a wall is straight ... sometimes it's important that the model shows it's not.
  18. @MHBrown I was curious about this because of other problems I've experienced which I feel are related to memory problems. I located some "sample" large PDF files by googling, which others could try with if they wanted to replicate. An approx 105MB file from here: https://cartographicperspectives.org/index.php/journal/article/view/cp43-complete-issue/pdf An approx 26MB file from here: https://research.nhm.org/pdfs/10840/10840.pdf I am on Monterey 12.1 and tried on VW2022 SP3.1 as follows: 1. Open blank file in vectorworks 2. Import the 105MB pdf (selecting "all pages") directly into the design layer 3. The PDF imports quite slowly, with some beachball moments, but does eventually all appear. and VW appears basically functional 4. I note that this brings VW's memory usage (in apple activity monitor) up to 6.8GB which seems a little disproportionate to the size of the PDF? 5. Import the 26MB pdf in the same way 6. Also imports slowly, with some beachballs and hangs, but eventually all appears. VW basically functional although zooming/panning might now be a little sluggish 7. Memory usage has now gone up to 9.9GB Then I tried importing the same PDfs into some "real files" with mixed results, and some investigation led me to suspect the scale of the design layer might be significant. When I open a blank file in VW, the default design layer's scale is at 1:50. But if I open a new blank file and change the scale to 1:1, then repeat steps 2 to 7 above... the imports are faster, and the respective memory usages are 2.5GB and 2.9GB (ie significantly less). If I do the same but with the design layer scale set to 1:5000, the experience is similar to the first experiment but memeory usages come in at about 6.9GB and 10.2GB. VW seems to remain functional, but I haven't tried to do anything further with the file. Don't know if that offers any clues.
  19. So it looks like the apple API provides 3d geometry in USD format I don't know anything about USD format (had to google it) but it seems like it allows objects (and surfaces?) to be defined as certain object types. So it can create a file that knows that this is a wall, this is a door, this is a table, etc. Presumably to get it into VW in a "parametric" form, VW needs to take eg each wall object, and convert it to a VW parametric wall object with accurate/useful attributes. In theory that would be very useful for architectural surveys as long as (a) the scanning bit is clever enough to accurately scan and understand multiple rooms, with walls (and floors) in between them (b) what it outputs is reasonably accurate (c) it can actually get transformed into sensible VW parametric objects where appropriate (c) presumably would rely on some work at the VW end. My experience of VW importing other file formats, so far, is pretty bad, usually ending up with a giant mess of non parametric meshes. Has, for example, REVIT import got any better recently? My experience trying it some time ago was that stuff didn't generally get converted to useful parametric objects. If that's still the case, then I suspect the ability to do a lidar scan and have computers do all the tedious work of converting the survey info into a useful model is still some way off.
  20. I'm sure that may be the reality. However, that's not how I recall 2022 being marketed to us. VW2022 was supposed to bring improved performance, thanks to running natively on M1 (as well as some other stuff such as the transition to Metal). However, my real world experience is not that VW2022 brings significant performance improvements over VW2021. In fact it brings additional problems, which means I am not yet using it, although I remain hopeful that they might be resolved in future SPs. Actually I'm not really sure we've had anyone on this thread say that they have found significant performance benefits in VW2022 compared to VW2021 - or have I missed something?
  21. This article seems to be an unquestioning parroting of what VW tells the writer, so effectively just a kind of press release. What it mainly seems to be saying though, is that we should expect various improvements in VW2023, rather than that they already exist. They talk about an upgrade of "shaded" (ex-OpenGL) view, and speeding up things like section viewports. It seems that although VW2022 shaded view already runs on Metal rather than OpenGL, they aren't actually claiming any improvements yet. Which matches my experience, where in fact "shaded" view is actually worse on VW2022 than previous versions. Likewise that we don't yet see much of a speed-up in section viewports (which quite a few people seem to be having problems with in VW2022). Am I reading that right? I find it quite frustrating because VW2022 was marketed as the first release "optimised" for M1 and running natively, but I don't really feel that's what we've got in reality. And this article seems to suggest we actually have to wait for VW2023 for that.
  22. I use a regular laser measure. Aiming at the target can be a pain, especially when there are things like furniture or foliage in the way. And it's not much good for outdoors measurements on a sunny day. And while it can give me very accurate straight-line measurements it won't give you relative elevation heights. Those always have to worked out after the event. I can see the attraction of the device in the OP. Presumably these types of things are going to become gradually cheaper and more commonplace.
  23. I also think they are quite decent so far. Unfussy and have enough to convey a sense of what the space would be like. The only obvious problem I see is that the multiple downlights are casting sharp-edged shadows, which is quite distracting. If you could adjust those so that they are more diffused, I think that would help a lot.
  24. If you leave it to render multiple viewports, does it gradually eat up all of the 32GB until everything grinds to a halt? I think this question is quite important for anyone considering a new machine - it's clear that VW struggles somewhat with 16GB but the question is whether shelling out for 32 or 64GB solve this memory problem, or just mean that you have a little bit more time before things clog up.
×
×
  • Create New...