Jump to content
  • 2

Point size control for Point Clouds


JeremyLondonRMLA

Question

Does anyone know if there's a way in Vectorworks to adjust the point display size on imported point clouds, such as LAS files.

 

At present I find that point clouds display well at a distance, see attached image 1, but when I zoom in to inspect closely and snap a 3D locus/stake object to a point to confirm its coordinates, the point cloud becomes very difficult to interpret due to the sparsity of points against the surrounding background, see attached image 2.

 

I've tried adjusting the pen thickness of the point cloud object in the attributes pallette / class settings, and searched in Vectorworks Preferences for any relevant settings, but had no luck so far.

 

I've discovered a very useful open source program called CloudCompare for the inspection of point clouds. This program allows the user to adjust point display size which really helps with interpretation at close range, see attached images 3-5. Note the 'default point size' adjustment buttons in the top left. 

 

If similar control over point size display could be added to Vectorworks I believe it would greatly improve the point cloud workflow.

 

Lidar_Vectorworks.jpg

Lidar_Vectorworks_Zoom.jpg

Lidar_Cloud Compare.jpg

Lidar_Cloud Compare_Zoom.jpg

Lidar_Cloud Compare_Zoom_Increase Point Size.jpg

Edited by JeremyLondonRMLA
  • Like 1
Link to comment

Recommended Posts

  • 0

Ok I found out what the issue was: my point clouds didn't like being imported into a georeferenced file. When I import them into a new blank document with the User Origin + Internal Origin both at 0,0 in the page centre + select ‘Center Import on Internal Origin’ in the import dialog they display like they should do:

2050865726_Screenshot2021-08-16at19_51_31.thumb.png.4b5afa02a6554bffaa64bf4f5abf1925.png900949435_Screenshot2021-08-16at20_14_07.thumb.png.3ee8e9a3f070a79396a256ea09a5d47f.png

 

So in order to import point clouds into my georeferenced file + have them display without half the points missing I have to first set the user origin to the internal origin. Once the clouds are in + looking as they should I can reset the user origin to match the georeferencing coordinate system but then I need a way to give the clouds their correct X/Y/Z coordinates. To achieve this I got my very helpful surveyor to add a special datum point to each cloud (my survey consists of three) that shares a common set of X/Y/Z coordinates then placed a 3D Locus with the same coordinates in my file to allow me to move each cloud one by one to this locus point so they aligned with each other + have the correct coordinates. Then I can import my 3D topo survey + everything is aligned as it should be:

211327518_Screenshot2021-08-16at18_18_30.thumb.png.cf25726c3b2cb0fffae5af6a4e26de8e.png308458439_Screenshot2021-08-16at19_54_15.thumb.png.6d3da2ebfb937f568b1660c6af960847.png

 

So basically it was a toss-up between importing them conventionally + having them come straight into the file with the right coordinates but not display correctly, or have them display correctly but not benefit from the georeferencing + having to manually reposition/align them myself.

 

 

Not sure what other people's experiences are with this? Be nice to be able to import a point cloud straight into a georeferenced file just like I do with other surveys/maps/etc.

Link to comment
  • 0

Hi Pat my understanding is that it is the Internal Origin (the 'Vectorworks Origin' represented by the blue crosshairs) that one needs to be concerned about when it comes to rendering/performance issues not the User Origin (the user-defined 0,0 point that the rulers, O.I.P., dialog boxes etc refer to).

 

From Help:

 

2099432378_Screenshot2021-08-17at04_43_37.thumb.png.5692d55ba4c8f0fe48062e8f942b839d.png

 

1941666372_Screenshot2021-08-17at04_43_55.thumb.png.caf22c4b329c8697a661f886a1ae813f.png

 

I always ensure my Internal Origin is in proximity to my drawn geometry but in a georeferenced file the User Origin will by default be moved to match the coordinate system being used otherwise it wouldn't be georeferenced.

 

What was confusing about my point clouds was that even with this check box selected:

816231526_Screenshot2021-08-17at04_52_11.png.f28494555cd6ffde50c48070b199eead.png

 

...the cloud still didn't import properly whilst the User Origin was remote from the Internal Origin. So as long as the User Origin was set to match the georeferencing coordinate system I could import to a location 50 metres from the Internal Origin or centre the import on the Internal Origin + the cloud would be missing points. It would only import with the full compliment of points if the User Origin + Internal Origin were coincident as per a new blank file. After importing it I could then move the User Origin to its correct georeferenced location + the cloud would be unaffected display-wise + everything is fine from then on. The problem this way of course is that you have to manually give it the correct coordinates which defeats the purpose of georeferencing.

 

I have never had issues with any other geometry in my georeferenced files.

 

Link to comment
  • 0
16 hours ago, Tom W. said:

Ok I found out what the issue was: my point clouds didn't like being imported into a georeferenced file. When I import them into a new blank document with the User Origin + Internal Origin both at 0,0 in the page centre + select ‘Center Import on Internal Origin’ in the import dialog they display like they should do:

2050865726_Screenshot2021-08-16at19_51_31.thumb.png.4b5afa02a6554bffaa64bf4f5abf1925.png900949435_Screenshot2021-08-16at20_14_07.thumb.png.3ee8e9a3f070a79396a256ea09a5d47f.png

 

So in order to import point clouds into my georeferenced file + have them display without half the points missing I have to first set the user origin to the internal origin. Once the clouds are in + looking as they should I can reset the user origin to match the georeferencing coordinate system but then I need a way to give the clouds their correct X/Y/Z coordinates. To achieve this I got my very helpful surveyor to add a special datum point to each cloud (my survey consists of three) that shares a common set of X/Y/Z coordinates then placed a 3D Locus with the same coordinates in my file to allow me to move each cloud one by one to this locus point so they aligned with each other + have the correct coordinates. Then I can import my 3D topo survey + everything is aligned as it should be:

211327518_Screenshot2021-08-16at18_18_30.thumb.png.cf25726c3b2cb0fffae5af6a4e26de8e.png308458439_Screenshot2021-08-16at19_54_15.thumb.png.6d3da2ebfb937f568b1660c6af960847.png

 

So basically it was a toss-up between importing them conventionally + having them come straight into the file with the right coordinates but not display correctly, or have them display correctly but not benefit from the georeferencing + having to manually reposition/align them myself.

 

 

Not sure what other people's experiences are with this? Be nice to be able to import a point cloud straight into a georeferenced file just like I do with other surveys/maps/etc.

I might be going a bit off topic for this thread but what happens next with the point cloud info? Is it sufficient to generate meaningful "as existing" sections/elevations/plans?

 

And if so, when you come to drawing whatever the newbuild elements are, does part of the model stay as a point cloud or is there a process where you have to convert it into objects that VW understands what they are? Eg those roof beams or the walls; do you convert them into solid objects at some point?

Link to comment
  • 0

@line-weight this is all new to me too but the point cloud is just a mass of snappable points which you can then take dimensions off or trace over in order to draw your model. It is just a fancy set of measurements really. You can use it to quickly generate dimensioned floors plans, sections + elevations plus some nice visuals for reference purposes but there is no automated way to translate it into a VW model. From the surveyor's point of view it is a quick, easy + accurate way to measure a building (once you have invested in the kit) even if it is just to produce conventional 2D plans/sections/elevations in DWG format (they will trace over the point cloud sections to create their DWG sections).

 

I will have to see how I get on modelling from the clouds. It may be that I just use the clouds to generate a load of dimensioned sections/elevations (viewports) then work from these to produce the model. As you know trying to recreate the building in all its un-square, non-plumb splendour is not really feasible or desirable so I think I need to first translate it into more VW-friendly proportions then take it from there. But the main thing is I have all the dimensional information regarding the structure at my fingertips + can refer to it at any time + check how far my interpretation of the building is deviating from reality + which areas it's doing so.

 

The point clouds will be on separate layers to the modelling layers + they will also be duplicated + chopped up to allow me to concentrate on smaller isolated areas on their own. In the end the VW model layers will replace the point clouds which will just be in the background (turned off) for reference like other survey info.

 

That's the theory at least...

Link to comment
  • 0

@Tom W.thanks. I normally tend to do my own surveys and one reason for this is that it seems to take as much time translating a surveyor's (conventional) 2d plans and sections into a useful 3d model, as it does for me to draw up a 3d model directly from my own survey info. Also, when measuring-up, I know what's actually important to measure.

 

The laser survey/point cloud method is obviously very attractive if it cuts out the tedious process of physically taking the measurements... the question is whether the process of making the point cloud into a usable model then becomes an equally tedious process in itself. I'd be interested to hear how you get on with it in practice.

Link to comment
  • 0

Hmmh,

Bricscad has a Tool that detects what is likely a Surface.

So you can walk around and create lots of surfaces of Walls, Floors, ...

You can later stitch them and or create Rooms or Objects from it.

 

But basically you will get an average position, leveled from all inaccuracies.

And I don't want a Wall at 89,99993567273° on 12,2434678396 m in X and

17,9238648649 m in Y position.

I would always try to calculate in my head what was the Wall or Column

position meant, when planning and would move it back into correct location.

 

This is where I am doing easier in VW than in any other CAD.

Being able to recognize which Objects are off and readjusting them.

So I would rather go with a tight Grid Snap and heavy use of SHIFT and T-Key.

 

I like how it should be, not how it got built and finished in the end.

But that is my personal problem 🙂

Link to comment
  • 0
On 8/17/2021 at 8:40 PM, zoomer said:

Bricscad has a Tool that detects what is likely a Surface.

So you can walk around and create lots of surfaces of Walls, Floors, ...

You can later stitch them and or create Rooms or Objects from it.

 

Zoomer if I’ve understood correctly that sounds exactly what I needed for what I’ve just done! The buildings I’m working with on this particular project are really all over the place + although I do ultimately want to translate them into a nice neat BIM model I felt like I first needed to model them in a more authentic way where much of the lean, twist + sag of the walls was represented in the model. So the way I did this was to cut a horizontal section through my point clouds at the base of the walls, trace over this to generate 2D polygons then extrude these a nominal distance to create 3D solids. Once I had 3D objects I could switch to working in an isometric view + use the Push/Pull + Taper Edge tools to ‘sculpt’ the walls in situ to match the surfaces of the point clouds + in this way build a 3D model from the inside out.

 

image.png.e6754b0f6f18dcfbb658359d4dc246fa.pngimage.png.5a2521ce7f6d8886273f14e6c7558f73.png

image.png.a66b3645bcf1925cfb16674904697626.pngimage.png.e1f23493e4ffd599b2a81025f2ffdea2.png

On 8/17/2021 at 3:29 PM, line-weight said:

@Tom W.thanks. I normally tend to do my own surveys and one reason for this is that it seems to take as much time translating a surveyor's (conventional) 2d plans and sections into a useful 3d model, as it does for me to draw up a 3d model directly from my own survey info. Also, when measuring-up, I know what's actually important to measure.

 

The laser survey/point cloud method is obviously very attractive if it cuts out the tedious process of physically taking the measurements... the question is whether the process of making the point cloud into a usable model then becomes an equally tedious process in itself. I'd be interested to hear how you get on with it in practice.

 

I normally measure stuff myself but generally I'm ironing out any out-of-squareness as I take the measurements + looking to draw it up in VW in a rationalised uniform manner. But even then it's often several hours on site + several return visits to check dims or take ones I forgot the first time. On this project the building is so all over the place I really wanted to model it fairly authentically initially at least + it would have been a nightmare trying to capture all that info myself with pad + laser measure (+ laser level for out of plumb walls). The point cloud was definitely the way to go. 

 

So the idea when I've completed this initial 'true' model is to use it as the basis for my proper BIM model which will be drawn using the parametric tools. I can keep the former in the file + use it for checking dimensions as I go + even incorporate it into certain viewports. It's not the easiest thing navigating the point cloud + understanding what you're seeing so translating it into a fairly realistic model initially seemed like a good step rather than going straight into the BIM model. Anyway we'll see what happens...

  • Like 1
Link to comment
  • 0

Well done !

 

It depends for what you need the 3D Model.

I personally would usually prefer an idealized Model.

 

But in this case, with conical Walls and such things,

it is very interesting to model the reality.

Maybe it can even help a carpenter to prepare furniture

that fits the inaccuracies of Wall Surfaces.

 

 

  • Like 1
Link to comment
  • 0

Yes, well done! This is definitely the approach I would prefer for really old buildings - rather than creating the "idealised" model, because if walls are leaning/twisting significantly enough, then it does actually affect the geometry of what you can subsequently fit into it.

 

Your method makes sense to me as a pragmatic way of extracting some of the detail into a usable VW model. It would of course be nice if it could somehow be automated, and what @zoomer describes Bricscad being able to sounds really useful. Probably we need to develop some pretty sophisticated AI before such a process could be fully automated - until then I guess there will always need to be a fair bit of human intervention in the process.

 

Did you consider taking a horizontal slice at vertical intervals going up the building, and then trying to use some kind of loft command to generate the solids for the walls?

  • Like 1
Link to comment
  • 0
5 minutes ago, line-weight said:

Did you consider taking a horizontal slice at vertical intervals going up the building, and then trying to use some kind of loft command to generate the solids for the walls?

 

That's a good point. My first stab at modelling the walls involved taking a slice at the bottom + one at the top + drawing 3D Polygons to represent the vertical (or not so vertical) faces of the wall in between then using Add Solids to convert all the faces into a 3D object. But I found the resultant solid still needed tweaking with Push/Pull + Taper Face to align it with the clouds so in the end it was more pragmatic to just create a quick Extrude from the profile of the wall at floor level then immediately manipulate it using those tools to bring it to the right height + align its surfaces to the clouds. With Taper Face you can tilt the wall + as you do it immediately see where it needs to be in order to best represent the real surface. Obviously the real walls are not perfectly flat planes like in my model so you need to judge where best to position the VW wall to achieve a good likeness. It sounds very imprecise + fiddly but in reality was really quick + effective!

 

I think with Loft Surface you are restricted to two profiles so would only have achieved the same results as the hand-drawn 3D Polygons (albeit more efficiently!). However your comment has reminded me of the Multiple Extrude command which would in theory have allowed me to draw multiple slices rising through the building vertically then join them up into a single solid...? I would need to see whether the work of setting up clip cubes + working planes + tracing multiple polygons over the walls would have been worth the extra effort. I suppose as always it's a case of determining how accurate a likeness you're trying to achieve. Although the walls are massively out of plumb they don't actually twist that much within each separate section so it was quite easy to create good representations of them using single flat planes.

 

The roof however I did want to model fairly authentically because single flat planes for each pitch would have meant ironing out the way it sags/rolls. I ended up making it using a Site Model!

 

45 minutes ago, line-weight said:

This is definitely the approach I would prefer for really old buildings - rather than creating the "idealised" model, because if walls are leaning/twisting significantly enough, then it does actually affect the geometry of what you can subsequently fit into it.

 

The thing is I do need a BIM model for the project as well which involves using the parametric tools which discounts representing the building as it actually is as far as the leaning walls, etc are concerned. So I'll have to see how working with two representations of the building simultaneously works. But the more I think about it the more I think it could work quite well: the BIM model for all the reporting, construction drawings, etc then the 'true' model for rendered 3D views + other things. Presumably when working on the BIM model + placing objects in the file I can bring up the 'true' model to ensure like you say that things are going to fit in reality + not just in the idealised BIM model. I can imaging section VPs where I'm showing the slab + objects (kitchen, windows, doors, light fittings, etc) from the BIM model with the walls from the 'true' model behind...?

 

404998344_Screenshot2021-08-27at11_16_04.thumb.png.a8d4bd614d20abb218abe7b2de542f8b.png

 

The external walls will have hempcrete sprayed on them internally then lime plaster so the lean will be present in the final product. Other parts of the project there is just one existing wall + the remaining walls + roof will all be newly built. So a mixture of new build + renovation. So do need to be using Wall, Slab, etc tools for most of it

 

Link to comment
  • 0

I'm bandying 'BIM' around but all I mean by that really is that it's an 'intelligent' model that I can extract data from + which uses wall, slab, roof (maybe not...), window, door, etc styles so as to utilise their functionality. There won't be any sharing with other consultants or anything like that. I am managing this project from start to finish so want to know material quantities/costs, thermal performance, etc

Link to comment
  • 0
5 minutes ago, Tom W. said:

I'm bandying 'BIM' around but all I mean by that really is that it's an 'intelligent' model that I can extract data from + which uses wall, slab, roof (maybe not...), window, door, etc styles so as to utilise their functionality. There won't be any sharing with other consultants or anything like that. I am managing this project from start to finish so want to know material quantities/costs, thermal performance, etc

 

There's obviously a judgement to be made, whether it's more work to (for example) model walls as solids and then extract areas by measuring manually off elevations, and so on, or two build two parallel models and extract that information more automatically from parametric objects.

 

For me I think I'd always tend towards the former, seeing as I am quite hardline about getting away the drawing-things-twice workflows that I used to have when I drew construction drawings in VW in 2d and had a parallel sketchup model for presentations, trying out design options and so on.

 

Obviously it totally depends on the project, what makes sense.

Link to comment
  • 0
1 hour ago, Tom W. said:

but all I mean by that really is that it's an 'intelligent' model that I can extract data from

 

 

You can Tag all your custom Solid Elements with IFC Tags and apply all kind of information.

For Bim you don't need any VW PIOs. You could also import Blender Meshes and tag them in VW.

 

PIOs are just more comfortable for standard Arch Geometry and already auto-tagged.

(That is only true as long as the geometry fits into PIO's capabilities)

AFAIK,

you can assign even IFC Tags via Classes by VW BIM Manager, so no need to do it manually.

Just assign Objects to proper Classes.

Link to comment
  • 0

I definitely don’t want to model everything as 3D solids. Even if I can add data to the objects automatically using the Data Manager I wouldn’t be able to report on component quantities, edit/add/remove components, assess R-values, etc in the way I can with Wall/Slab/Roof styles. And utilize component offsets to generate construction details. Like I say a good proportion of the project is new-build anyway (my screenshots show less than half of the footprint) so will be drawn by default with the parametric tools. Perhaps the 3D solids version of the existing structure I’m drawing at the moment is an unnecessary step in achieving this + I should have just jumped straight in with the parametric tools from the off. But like I’ve said the point clouds are quite big + unwieldy + not the easiest to apprehend in 3D space: a VW ‘sketched’ version of the point clouds seemed a useful first step in building the ‘BIM’ model (I will trace over it with the parametric tools) + gives me something to refer back to in future to check ‘real’ dimensions against the rationalized ones in the ‘BIM’ model. As a bonus it may also be feasible to incorporate it in certain presentation drawings later on but this certainly wasn’t the primary reason for producing it.  

 

My file including all the point clouds plus the topo survey is 2.3GB. If I trim the clouds down to just the parts I need to see to model the existing structure I can get it down to 800MB. But as soon as I’ve finished ‘tracing’ over the scans those layers will be deleted to get me down to a reasonable size.

 

I’m not sure if I mentioned earlier but because the only way I found to correctly align the three point clouds in a georeferenced file was to have the surveyor include a special ‘datum’ point in each file they sent me. This point shared common x/y/z coordinates + allowed me to manually align each scan correctly in georeferenced space. But it also meant I was forced to import 100% of the points otherwise I’d lose the special datum point in the process! Hence the enormous file size… VW are looking into why you can’t import a point cloud directly into a georeferenced file without suffering the loss of graphics I described in my initial posts.

 

Anyway the main point here is that if it were possible to apply an angle to a Wall I would be able to model the existing structure far more accurately + satisfactorily using the parametric tools. You can slope + taper a Slab so not sure how hard it would be to provide the same control over Walls? Then I could have the best of both worlds: an authentically modelled building using the parametric tools.

Link to comment
  • 0

Got it.

 

 

47 minutes ago, Tom W. said:

if it were possible to apply an angle to a Wall I would be able to model the existing structure far more accurately + satisfactorily using the parametric tools. You can slope + taper a Slab so not sure how hard it would be to provide the same control over Walls?

 

 

What about AEC Wall Projection / Wall Recess / Symbol in Wall ?

 

To either add or cut off a slope into Walls.

Never used those before and not sure which to use to still allow

wall connections and calculations.

Link to comment
  • 0
25 minutes ago, zoomer said:

What about AEC Wall Projection / Wall Recess / Symbol in Wall ?

 

These won't work unfortunately.

 

This is my hempcrete + lime plaster wall style that I would be drawing in front of my sloping as-existing walls:

2111554012_Screenshot2021-08-28at10_01_24.thumb.png.5743684d1972b566ea2c6c09342d2461.png

 

I'd need the whole wall to slope (both components) whereas the way all of those methods work is to carve a shape out of the face of the wall - through the layers of components - or by planting a shape on the face of it (+ you can choose which component you want to constitute the shape but only one component).

38586657_Screenshot2021-08-28at10_06_26.thumb.png.b40a9b4e1979d2c86af0f99e29199754.png

 

The hempcrete is sprayed on to a uniform thickness across the surface of the wall then the lime plaster is applied to a uniform thickness to the hempcrete so the finished wall surface will have the same slope as the wall behind.

 

Plus Wall Projection + Recess often behave quite unpredictably + can't be used across the full length of a wall. It would get really tricky at junctions.

 

Like I say, with a Slab you can slope the whole object (all the components) or limit the slope to specified components (tapered slab). Be good to be able to do the same with walls.

 

At the end of the day I think it will be fine that the 'BIM' model has straight walls, it will just be good to have the 'true' model in the background to refer to to check that things will fit in reality: kitchens, bathrooms, etc plus spans for beams. The hempcrete/lime plaster quantities will be pretty much the same whether the walls are plumb or off-plumb.

Edited by Tom W.
  • Like 1
Link to comment
  • 0

Walls would certainly be a lot more useful if you were able to taper components or have them non-vertical. Also if your core component at least could be non uniform thickness along the length of the wall. Seeing as we don't even have curved walls that work properly, these things seem a bit of a distant hope.

Link to comment
  • 0
On 8/27/2021 at 10:17 AM, line-weight said:

Did you consider taking a horizontal slice at vertical intervals going up the building, and then trying to use some kind of loft command to generate the solids for the walls?

Just to say that I tried using Multiple Extrude to do this + it didn't work. I'd forgotten how the command worked. It requires you to have all the profiles on the same plane + specify an overall height for the Extrude. The intermediate profiles are spaced equidistantly in between

Link to comment
  • 0
2 minutes ago, zoomer said:

 

Ah, " Sauerkrautplatten" here 🙂

(or officially "Holzwolle Leichtbauplatten" ?)

 

Ha ha sauerkraut yum! But I think this is woodwool board rather than hempcrete. Hempcrete is hemp fibres + a lime slurry sprayed onto the walls:

 

2060823155_Screenshot2021-08-28at11_05_08.thumb.png.878bbfa8096f2b918f7b462df225ed12.png

 

Or you can cast it in situ into formers

Link to comment
  • 0
1 hour ago, Tom W. said:

Just to say that I tried using Multiple Extrude to do this + it didn't work. I'd forgotten how the command worked. It requires you to have all the profiles on the same plane + specify an overall height for the Extrude. The intermediate profiles are spaced equidistantly in between

I'd go for loft surface, no rail mode, ruled, create solid.

 

713033436_Screenshot2021-08-28at11_53_03.jpg.c8acce9cb7a71a78bd52c4e10b43d6b3.jpg763307380_Screenshot2021-08-28at11_54_08.jpg.a9c4ba82d856f0de1b6dfa46a9d8f6a1.jpg2014673851_Screenshot2021-08-28at11_55_37.jpg.4c121d75abe4fa23265cb9ab72219fb6.jpg1527278669_Screenshot2021-08-28at11_55_53.jpg.dc183845b42504a145c93161589c71e9.jpg

  • Like 4
Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Answer this question...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...