Scalability? with regard to lots of faces/lines/shapes

I do a lot of scale models, and lots of them have either brick siding or shingled roofing - or both. I know how to draw them, in either FreeCAD, OpenSCAD, or some other software. (XCS - I have an xtool.) It usually involves drawing a few bricks or shingles outright, then arraying them. Doing test runs is simple and very fast in all of them. But doing things at scale is a whole 'nother story.

As near as I can tell, all three of the above have some kind of algorithm that is O(n**2) or worse, where n is the number of faces. Today I was working on a building wall 140 x 21 feet, which has about 7000 bricks in it. XCS was very slow, OpenSCAD agonizingly so, and FreeCAD utterly ridiculous. Is Lightburn any better? Even n(log n) would be a lot better.

I know that I can break things up into smaller pieces, but aligning the engravings is a tedious and error-prone step that I’d like to avoid.

Use the ARRAY tool to rapidly fill in a pattern, the use Boolean tools to cookie cutter the shape you need from an oversized field of bricks /shingles

1 Like

Previous discussions may have useful suggestions:

https://forum.lightburnsoftware.com/search?q=brick%20%22ho%20scale%22

Plenty of brickmakers out there! :grin:

How to do it is not my question. As noted in my original post, I know how to do the job. What I’m asking about is scalability. All of the solutions that I know about (plus Fusion 360, as received wisdom) get slower and slower as you add faces or shapes. The algorithms appear to be O(n**2), although I have not investigated the actual source code.

My question about Lightburn is whether or not it is similar in scalability, or ideally better.

I made:

  • A single 1.5Ă—0.75 mm brick
  • A 100Ă—100 array in a few seconds
  • A 10Ă—10 array in a few tens of seconds

Which looks like this, with the front edge of the machine’s 700×500 mm workspace peeking out of the bottom:

A 1 megabrick wall definitely makes the UI laggy, but it’s running on a tiny PC through an RDP network connection.

If you can make do with only 10 kilobricks, it’s reasonably performant:

Your mileage will certainly vary. :grin:

3 Likes

That sounds considerably better than anything else I’ve seen. It took OpenSCAD more than 10 minutes to render 7000 bricks. I don’t know how long FreeCAD took, but it would have been over an hour.

10k bricks is sufficient, there are very few single items that need to be bigger than that.

I have an M1 Max, 10 cores and 64GB, so I presume I’ll have at least as good performance as your “tiny PC”.

Keep in mind that Lightburn is designed to control the laser, rather than to be a full house CAD program. The design features are a happy second order result of the software.

Do you plan to use your new “brick and tile making super powers” to actually run a laser… or something else entirely??

If not, you’ve done been robbed! :grin:

As I said in my original post, I have an xtool S1. And after I said that I’d already drawn this at least three different sets of software, and that I’m working on a wall that’s 140 x 21’ (scale), doesn’t it seem likely that I’m serious about this?

If you must know, I have about 75 buildings to construct, of which about two thirds need brick walls and probably more than half need some form of shingled roofs. Most of the non-brick need clapboard siding, which I can do either with the laser cutter or on my 3D printer. But the clapboard is many, many fewer shapes than brick or shingle, and is handled efficiently in 3D CAD such as FreeCAD and OpenSCAD. Given the information from Ed, it sounds like brick and shingle are a LOT more practical than other things, so the laser will likely be used for several hundred building sides or roofs.

I’ve asked here before trying it because I used a trial version of Lightburn a year+ ago before the S1 was actually working with it. I started with FreeCAD because that’s how I drive the 3D printer - 2D doesn’t do for that, although I can project 3D models into 2D for the benefit of a laser cutter. That was a disaster, so I tried OpenSCAD, which is considerably less bad. In the process of driving the S1 with the DXF I produced out of OpenSCAD, I discovered that XCS doesn’t scale well either. Hence my question.

and thanks to Ed for actually answering the question!

No need to get hurt. My crystal ball is out of batteries, and we ONLY know what you tell us. Often times it is very clear in your head, but nebulous at best when posted to a forum.

We, as a community, would prefer to teach you HOW to do something, rather than do it for you. Give a fish vs. teach how to fish parable applies.

The forum has help people who ultimately wanted to use LB as a design studio, rather than a laser engraver suit.

I hope you achieve your goals, and it would be awesome to see what your results are. Share your success with the forum that helped you achieve them. If you can pass on what you know to other members, maybe you can pay-it forward?

Cheers

1 Like

If you like programmatic CAD along the lines of OpenSCAD, you’ll love GCMC. It started as a way to write G-Code using a C-oid language, then added SVG and DXF output because it already had the geometry. Import the SVG into LightBurn and away you go.

I’ve done simpleminded coasters:

Guilloche patterns suitable for destroying DVDs:

Those have color-coded layers to automagically select the proper cut parameters for lines & lettering.

A reproduction of an old circular slide rule:

It’s pen-plotted, but has laser-cut outlines & cursor, so it counts.

Bricks? Shingles? Siding? No problem. :grin: