Live data from Hacker News

CadQuery –- A Python parametric CAD scripting framework based on OCCT

github.com

11–20 of 45 posts

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#11
post #4

Slight aside, does anyone know of a more modern open source parametric CAD kernel than OpenCascade (which this is built on)? When I have looked into it before it seems to be criticised for being somewhat dated and not have the power and developer ease as Parasolid (which costs big bucks). Having both a software and mechanical engineering background (I spent a lot of time in SolidWorks) I regularly get the itch to try…

The only other open source full function CAD kernel that I know of is BRL-CAD, have only poked around a bit in it and OCCT.

It Coin3D not full function?

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#12
post #4

Earlier quoted context omitted.

The only other open source full function CAD kernel that I know of is BRL-CAD, have only poked around a bit in it and OCCT.

It Coin3D not full function?

Coin3D is a 3d graphics scene graph toolkit, it has no parametric solid models tools.

I believe most tools built with OpenCascade use Coin3D for rendering.

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#14
post #13

For anyone interested in this, I would also recommend checking out the [antimony project]( https://github.com/mkeeter/antimony ). It's a parametric modeling program, but you can also pretty easily produce new scripts for its visual nodes.

The developer has since moved on to libfive and libfiv-studio.

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#15
post #13

For anyone interested in this, I would also recommend checking out the [antimony project]( https://github.com/mkeeter/antimony ). It's a parametric modeling program, but you can also pretty easily produce new scripts for its visual nodes.

The developer has since moved on to libfive and libfiv-studio.

For the curious: https://libfive.com/ and https://github.com/libfive/libfive

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#16

I made this keyboard case with CadQuery. In the long run I'm working towards generating cases (and PCBs) for keyboards with many parameters available for configuration (e.g. number of columns, number of rows, number and position of thumb keys, case style, rotary encoders, etc). https://imgur.com/a/VfOyAjP https://github.com/possibilities/unkeyboard Note I chose a somewhat non-idiomatic approach (in terms of cadquery)…

I've been actually considering multiple CAD options to design a keyboard... pretty cool to see that you did so!

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#17
When I read parametric CAD I was hoping to see a tool that can automatically assemble full floor plans from just XYZ dimensions. Wiring, plumbing, structure, lighting, insulation, etc. Everything maybe except paint color. It's just another constraint solver, not sure why that doesn't exist yet and why we need to cough up billions in engineering and architecture costs every year.

With a tool that can provide normalized, structured data corresponding to buildings, then for sure we could save on manufacturing too, every wall panel +- the same you could mass produce them. With a clever concept with datacenter-like floating floor you could even change and adapt the rooms (including plumbing) to suit needs without having to rebuild.

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#18

Slight aside, does anyone know of a more modern open source parametric CAD kernel than OpenCascade (which this is built on)? When I have looked into it before it seems to be criticised for being somewhat dated and not have the power and developer ease as Parasolid (which costs big bucks). Having both a software and mechanical engineering background (I spent a lot of time in SolidWorks) I regularly get the itch to try…

same itch here.

Here are two quite remarkable statements from a MIT lecture on CAD (https://stellar.mit.edu/S/course/6/fa19/6.807/courseMaterial... ):

"[there are] ASIC and Parasolid, others are not really as good"

and "[CAD] Research disconnected from reality [compared to CG research]"

One might expect a whole range of CAD kernels floating around, since these kernels are such a fundamental part of civilization in some way. Doesn't seem to be the case, tho..

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#19

Slight aside, does anyone know of a more modern open source parametric CAD kernel than OpenCascade (which this is built on)? When I have looked into it before it seems to be criticised for being somewhat dated and not have the power and developer ease as Parasolid (which costs big bucks). Having both a software and mechanical engineering background (I spent a lot of time in SolidWorks) I regularly get the itch to try…

I've never tried BRL-CAD but it looks really big and complex. The only other one I'm aware of is Solvespace: https://solvespace.com/index.pl

It uses it's own B-Rep for NURBS and can export STEP files. The basics are there for extrudes, revolves and helixes. Booleans are still a bit buggy, but we occasionally fix a bug or two.

It's definitely the smallest NURBS kernel around at about 8000 LoC and the full program comming in at ~5MB. There is no real API for geometry though, just the constraint solver.

BTW the constraint solver has also been used in one of the FreeCAD assembly workbenches, and also in this sketcher add-on for blender: https://blenderartists.org/t/geometry-sketcher-constraint-so...

Re: CadQuery –- A Python parametric CAD scripting framework based on OCCT

#20
post #17

When I read parametric CAD I was hoping to see a tool that can automatically assemble full floor plans from just XYZ dimensions. Wiring, plumbing, structure, lighting, insulation, etc. Everything maybe except paint color. It's just another constraint solver, not sure why that doesn't exist yet and why we need to cough up billions in engineering and architecture costs every year. With a tool that can provide normalize…

The "design space" of architectural CAD models is mostly illegal (e.g. does your AI know how many parking spaces you must provide?) and/or lethally dangerous (e.g. is that beam strong enough?) so in the real world a constraint solver that saves time by automatically drafting, according to detailed specifications, detailed plans that certified architects and engineers can then analyze and sign off is the most useful possible tool.
Post reply on HN