Standard comment in these threads that, while OpenSCAD is a ton of fun and absolutely a breath of fresh air for code geeks wanting to apply their skills to the 3D world without having to hand-draft an object in scary GUI software...<p>It's actually really limited, based on a needlessly simplified data model and using a kludged up DSL syntax instead of a proper software development environment.<p>Look at CadQuery and build123d, both of which are python packages build on top of OpenCASCADE[1]. They have wildly different syntax[2], but very similar capabilities. You can do the same CSG work in them that you do in OpenSCAD with similar complexity, but have access to a far more expressive underlying toolkit and a real programming language with which to manage your own parametrization needs.<p>[1] A CAD-focused boundary representation toolkit, which you can actually use directly if you want but which is aimed more at "build a CAD tool" use cases than "design a speaker cabinet".<p>[2] Truthfully I don't love either, but CadQuery is at least explicit in most cases about what is happening where build123d relies on some dynamic scoping tricks that irk me. Seems like most of the community disagrees with me though.
Fyi the latest builds of OpenSCAD have a new fast-csg library.<p>My compiling never takes more then a minute, usually it's less than a second.<p>Waiting over an hour to compile a CAD drawing seems untenable to me.<p><a href="https://ochafik.com/jekyll/update/2022/02/09/openscad-fast-csg-contibution.html" rel="nofollow">https://ochafik.com/jekyll/update/2022/02/09/openscad-fast-c...</a>
> As the design was fully parametric, I could change a single variable to move to a floorstanding design.<p>How far are we in 2025 from defining a Differentiable design, setting a target (e.g. maximally flat frequency response for a certain speaker placement in a certain room) and solving this automatically?
I am an (amateur) developer so when I needed to create some simple 3D objects to print them afterward OpenSCAD looked like a dream come true ("object as code" :)).<p>I tried three times (over two years) to use it but it was absurdly difficult to build a piece, with something that looked like "callbacks from hell" in JS (boxed in boxes in boxes of code. I had to keep track of the position all the time and there are no constraints.<p>Compared to that I tried Onshape (the free version which is almost the full one but all models are public) and it took me an afternoon to create my pieces. It was really simple, on a generic laptop without any power (a Lenovo P14s)
Showing OpenSCAD without showing any code, or even really talking about the code is a real shame - there are some cool ideas in here, but none are shown (e.g. "I even made an algorithm to calculate the screw positions evenly").
I recently build a Hexibase subwoofer and a lot of stuff you described was along the same lines. Even better, I have an UMIK-1 on order and the calibration parts will come useful too. Didn't know about CamillaDSP, gotta check it out if I don't like Equalizer APO. Thank you for posting this!
This guy might be good at designing but he is really bad at 3D-printing. All of that post-processing would've been completely unnecessary if he'd made better choices about material and print orientation.
Tangent<p>I feel like such a noob still using SketchUp but it works, Fusion360 is on my list to learn at some point<p>There are some great tools like GLB export (Khronos) for ThreeJS and STL mesh check for 3D printing.