CADmium: A local-first CAD program built for the browser
mattferraro.dev
mattferraro.dev
The biggest reason this hasn't happened so far is the lack of a truly capable parametric kernel, Truck, the kernel that Matt is using looks like an incredible project and exactly whats needed. The only other kennel till now that been close to being whats needed is OpenCascade, but its lacks important features, is buggy and at times quite unstable.
Once Truck (and CADmium) lands stable fillets (surprisingly one of the hardest features to make stable) it will prove itself as the perfect successor to OpenCascade and and the perfect platform to build the future of open source parametric CAD upon.
I've recently moved over to Alibre Atom3D, which, while not open-source, costs $200. Once. Then (as long as your host OS doesn't change too much) you can access your designs forever. It runs on the ACIS kernel.
"Browser first" and (from the readme) "Beyond that, I will try to monetize by offering a hosted version of the software as a paid product." reads to me like this project is doomed to either fizzle or to grow and transform into an open-core, subscription-requiring product, accumulating more complicated dependencies to install and "security features" that make it harder and harder to run in truly local fashion.
Your designs are wide open for anyone to see, but I can live with that for hobby stuff.
I one time had an issue where I messed up and had to post on the FreeCAD forum and got two of the best responses I've gotten anywhere on the internet ever. Two people opened up my model and took screenshots highlighting where my issue was and gave me several suggestions on how to improve my model in other ways. The initial time investment was substantial and there are still a few workflows that could be smoother for me (attaching a sketch to a different face after it's been created), but that's fairly minor.
And I asked countless questions on the forum as well. People were always very helpful.
But there’s a point where you just give up.
Having never used OnShape before, it literally took me half a day to get something done and submitted for external production. I tried that design first in FreeCAD, after those 100 hours, and got nowhere.
But there’s no denying that FreeCAD is still a bit of a lifestyle choice :-) I love it, but I don’t try to sell everyone on it yet.
It might be possible to predict which edges would have problems, but I think that would involve repeating geometry calculations outside the geometry kernel?
Specifically the main problem with fillets/chamfers in OCC is that the new face cannot completely consume/replace an existing edge.
That is to say -- I think!? -- that one of the edges of the new fillet/chamfer is always an existing edge just displaced, and no edges are destroyed. If an edge would be destroyed, then it fails?
But this is just my colloquial/folklore understanding.
There have been newer versions of OCC that FreeCAD has not yet switched to, I think. (And there is a new (Chinese) fork, called OpenGeometry.)
So maybe there are some fixes already in the pipeline or in the future. But they are somewhat out of the FreeCAD project's control, is my understanding.
I've invested several hours already in trying to learn FreeCAD and achieved nothing. Its a hot mess that desperately needs a new UI.
I don't know if you can really get away without learning general things about CAD systems if you want to make complex things, because those general things are so influential on the existence and design of real world objects.
It's like saying "but I don't really want to learn about different programming languages or software design, because I only want to write apps for iPhone."
If you want to make very simple objects and not learn deeper concepts, use TinkerCAD. It's fine. Weirdly competent in places; people pull off crazy stuff. It uses a simple physical world model.
Beyond that, you need more abstract tools. Ultimately you have to decide whether you want to exist within the (shifting) free plan limitations of a commercial cloud package, spend money on something like Alibre Atom3D, pay the cloud firms, or use something open source.
I'm not sure if it should be surprising that the open source systems are more of a commitment.
FreeCAD (main branch) has one fundamental limitation which is being mitigated -- the topological naming issues -- and some smaller issues to do with fillets/chamfers, but apart from that is a disorganised-but-competent, open source workbench-based system.
And the tradeoffs are not straightforward. This is less like Photoshop vs GIMP and more like ArcGIS vs QGIS. Is ArcGIS initially more coherent? Yes, allegedly. Is it universally more powerful? No.
FreeCAD is really a lot like QGIS, IMO.
You can learn QGIS, and when you learn it, you might decide that the freedoms and flexibility it offers are worth the pain. Or you might front up the cash for ArcGIS.
The problem was not about getting away with it without learning general principles.
Of course, those 100 hours taught me about principles that were applicable to OnShape as well, so it’s not entirely apples to apples.
BTW, the issue was never TNP. It was everything else. The absurd separation between Parts and bodies and whatnot. The 5 different assembly benches, none of them good. The crashes. The way you need pixel precision to grab a vertex or an edge. And if some operation cuts a part in multiple disjoint pieces, only one survives.
My wife told me to stop doing what I was doing because I ended the weekend with way more frustration than the work week.
> The absurd separation between Parts and bodies and whatnot.
Again, Parts/Bodies, it's confusing (and Part Design should be called "Single Body Design" IMO) but there are videos that explain the difference.
(I gather it comes from CATIA, and CATIA users are comfortable. Either way you don't have to use both; it's possible to never touch Part Design and do absolutely everything.)
I don't think I got it initially but now the difference is useful. There could be more work to allow non-Part-Design Parts to be wrapped in Part Design bodies. There's a macro (pdwrapper) for this, and Realthunder's branch built it in.
There could be better explanations on how Part Design works, how you build a Part Design Body from existing Parts and how you later combine them.
> The 5 different assembly benches, none of them good.
I feel this pain and I mostly avoided it by just using static placement. There's a new built-in Assembly workbench with a sophisticated solver that is being added in 1.0 which will land in the next two to three months (and is already in 0.22-dev and Ondsel 2024.2). I am going to dive in with that.
> The crashes.
Not a problem I have on the Mac at all now; 0.21 is really stable. Though startup crashes are common on Linux due to W****nd and the complexities of compiling support for it. The Flatpak appears to solve those. Ish.
> The way you need pixel precision to grab a vertex or an edge.
There is a preference setting (a newish one) that makes edge selection rougher/easier.
> And if some operation cuts a part in multiple disjoint pieces, only one survives.
Only in Part Design, which (in mainstream FreeCAD) is concerned with only one contiguous object. In fact, that operation actually fails. (A Body can be started with a compound of multiple pieces, but the first feature has to join them).
RealThunder's branch can support Bodies made of separate solids, and maybe that will come to mainstream once the TNP fix, which is a prerequisite, is in.
Outside of Part Design, objects split as you'd expect.
I don't think FreeCAD is perfect. That would be silly. But none of the "free" plans of the commercial cloud packages are right for me.
Alibre Atom3D is the entry level package, but the whole thing seems very much like the Affinity suite model. It looks very good.
Nice files you got there. It would be a shame if they suddenly become unreadable, eh?
Even the so called "open" core is a joke. Just try to add anything useful. It will get rejected in favor of a paid option.
The whole thing is just an offensive anathema to open source.
I’ve been playing around with the weekly version and using the OpenDark theme. It looks pretty slick.
It seems like a really exciting time for open-source CAD (hopefully!).
Apparently, there are now > 500 bugs open.
When commercial software has bugs, you don't even know they exist often. Sure, but there are also ways to make bug free software, but hardly any open-source software exists that uses those development techniques.
It's still better than FreeCAD, though.
I've never once run into something that they didn't have a workaround for, and then eventually fix. Sometimes it takes a service pack or two.
They out out 4 or 5 service packs every year along with the major yearly releases, those service packs are basically entirely bug fixes.
You may want to look into switching VARs if your not happy with your support.
We've always lagged a year behind the current SW release to try to avoid getting hit by the latest and greatest bugs.
I think we just hit more edge cases than other companies. We have half a TB of stuff in our vault (about 250k files), going back to 2012 for the first import. Lots of legacy files, fairly complex assemblies, heavy use of some tricky stuff like the sheetmetal design tools (where we tend to find the most bugs). And then there's Solidworks Electrical, which I don't think anybody is set up to support well, as it's still a fairly recent acquisition from Trace Software.
If people have no alternative in their field or the alternative requires discarding years of experience, the incentive to fix bugs isn't as high as it could be. That's how you get problems like Excel invalidating data with automatic formatting for decades.
Particularly younger engineers have their mind blown over the concept of not immediately upgrading all your software just because there's a new version.
In the physical engineering spaces, it's more valuable to keep using the same software version you started a project with, rather than have hundreds of thousands of dollars in damages because of a dickhead bug.
You see the same problems with a lot of “flagship” oss ware. Gimp seems to have the same awkward feature set for 25 years. Apparently the only thing that has changed is they no longer compile against Gtk 1.0, and they decided to have all images in a single window.
Still can’t do smart background erase.
Inkscape kinda works, but it always seems like attachment points for easily alignment, and a non sucky text editor are forever out of reach.
I’m sorry. I know software is hard, but it’s been literally decades of stagnation.
Off the top of my head:
- Full color management
- Non-destructive filters
- 32-bit per channel precision
- Late binding CMYK support
- Unified transform tool
- Vastly better selection and cropping tools
- Vastly better text tool
Should I go on? :)
This is a simple tutorial task. Is it this easy in Gimp? No. It’s not. But hey, we have the same crappy ScriptFu plugins from a quarter century ago. Anyway, you have the source! Shut up and be grateful you insensitive clod!
Seems fair
No, not really. You did say 'feature set'.
> A bunch of data loading crap,
Sure, if you like editing with color fidelity loss.
> but nothing substantial on the usability.
Except non-destructive editing is a major UX improvement.
Except unified transform tool is a major UX improvement.
Except... Well, you get the idea. And I barely got started.
It's almost like you just hate the project and won't accept any changes they do.
I too have my fair share of frustrations with the project — enough to quit it as a team member a couple of years ago. But somehow I don't go around pretending big changes aren't big bc I'm feeling petty all of a sudden.
I gave a concrete example of a task that is trivial in 2024, and it just does not work nearly as well or as easily. It’s the same functionality of 30 years ago.
Instead, you’re sticking your fingers in your ears and screaming, “it is progress!” Yeah, well maybe technically correct, but the gap has widened. As a project it should be shutdown. It’s zombieware.
I think Kritta is the new flagship OSS image editing software. I don't use either, but that's my impression.
I basically want to extend the base of an Ikea lantern (Enrum); this require creating an (irregular) octagonal shape, and then creating a slot so that the lantern base can sit on it.
I had some distant memories of modeling stuff in Solidworks for a project back in university, so I'm not starting from zero.
I was able to finish my design in Fusion and redid it in one evening in Onshape. As for Freecad... I still don't know how to do it. Or rather: I could probably do it, but only with so much more googling on how to do basic stuff that it is discouraging to even thing about it.
I think the main issue is that Freecad generally expects you to know exactly what to do, and has very poor affordance.
Let's take the very few steps for instance. In order to create my lantern "shoe", I first need to recreate its base. It's some sort of octagon, so the plan is to draw an octagon, extrude it and then probably repeat the operation with several other octagon to carve a slot for the lamp.
So the very first thing is to create a sketch on the xy place. In both OnShape In both Fusion and OnShape, the very first menu item in the toolbar is the "create new sketch" one [1][2]. So you can immediately click on it, then chose you plane, then start to draw.
In Freecad though, the first thing in the toolbar are.. the camera controls (which are somehow redundant with the interactive cube view of the main area). In fact, you cannot even create a sketch from the default toolbar; at best you can create a Part.
In order to create a new sketch, you need to go to the dropdown menu. This menu is ordered alphabetically. So the first item is "Arc", then "Draft", then "FEM", ... If you do not know the jargon and/or is just starting out, good luck finding where to next ("what's the difference between Draft and Sketch ?" they would say). And even select Sketcher from the combobox, the first controls you see are still the camera ones. Worse: the "create parts" button from earlier is still there ! [3] And so the "new sketch" button is hidden on the right side of while it's probably the very first thing you want to do.
And for the very short time I spend trying Freecad, issues like were constant.
There's seems to be some open issues related to this in the Github issues viewer, but no tangible commitment as far as I know.
Isn't that what OSS project always say?
About 10 years ago they released a new feature, slot sketch tool, so instead of drawing 2 circles, two lines, trimming, now it's one click.
If you use the new feature to save significant time, like me, it's wonderful.
The attachment editor is horrible, for instance. If you want to attach a sketch to a datum plane, it will be attached at a random orientation. There is no preview for this, since datum planes are just planes, except you know, the part where it actually is an LCS with a default attachment orientation internally. So now you have to go into the sketch and draw an arrow pointing up and then rotate the sketch into place so that it actually points up. You will have to do this for every single sketch. It is so painful and an every day occurrence that surely someone should have come up with the idea of fixing it years ago, right?
There is usually no preview for additive parts unless it successfully creates the final model. The idea of displaying a transparent preview of just the operation itself never occurred to them. Also, there are hardly any interactable UI features such as draggable arrows in the 3D editor, you mostly enter numbers in the left panel.
I haven't had much trouble with "bugs". The attachment editor thing is considered a feature by the way.
Let's just say I've seen their (internal) bug trackers.
TLDR; you couldn't be more wrong.
When OSS software has bugs, at least someone with an interest in the issue and the skills can do something about it.
For closed source software nobody can. Even the people working at the resp. company who are developers rarely or never decide about priorities.
And as an added hurdle, adding features always has higher priority as it drives sales.
Does parasolid not fit this requirement? The capabilities I've seen in Plasticity are very impressive. Or do you specifically mean FOSS?
The more I use CAD platforms, the more I develop the sense that general-purpose CAD is much less useful than single-purpose applications that provide tailored solutions to narrow problem domains. SVG/DXF/DWG output is a plus, but I think a drawing software that works for high-volume machine parts, one-of-a-kind architecture, highways, 50-mile pipelines, circuits, urban transit plans, and art is the wrong direction. I use industry standard roadway design software and a "road" does not exist in the object model. Horizontal, vertical, and sectional components are all independently defined despite these things being inseparable and having some very obvious and well-specified rules about their interaction. I really think designers should spend more time thinking about outcomes and less time telling the computer how to display them.
The user works with rich domain objects that limit the possibilities to what actually works or is practice; much like a game level builder. This allows non-experts to design solutions in this particular domain, which of course is great for business / sales.
So as a model builder you end up trying to build these higher-level abstractions into your model, where the parameters are your top-level interface.
I think CAD is so much like programming in those ways, although I could just be biased since I’m a software engineer.
Some of the wisdom from programming would do well to pass onto design software. Like how strong type systems can provide safety by pushing rules to compile time and make some errors unrepresentable. Meanwhile, I can fire up Microstation, draw roads crossing at the same elevation with no intersection (think stops and signals, not geometry) between them. Or a drain culvert could terminate in a big Hello Kitty picture. These things should be impossible. If the task is designing a road, I don't need the ability to draw anything I can possibly imagine. I really need the software to know what a road can and can't do, produce a model that obeys those constraints, and to give me files and documents I can give to an owner and a contractor which convey an accurate understanding of what that model is.
Then it is just up to you or your organization to come up with a design rule set that ensures your boards can be manufactured by the board house.
I'm just wondering why QCAD doesn't seem to get much if any mention on HN when CAD tools and open source comes up.
Contrast with Solvespace (or FreeCAD, or any of the popular commercial parametric 3D CAD programs), which lets you specify relationships (constraints) between elements in 2D and 3D. The shape of your part is the finished result, not the starting point.
3D is where the excitement is, but also where the difficulty is from the kernel perspective.
I think it's a shame that LibreCAD has stagnated and diverted, particularly on usability. QCAD has one of the best 2D CAD UX I've used, akin to AutoCAD but more light and nimble. Exactly what is needed for most 2D CAD.
If you need a powerhouse, then you really need 3D anyway, and most likely you aren't a hobbiest if that's the case. For 2D, it's QCAD all the way. 3D is nice, but 2D gets the job done 9/10 times. 3D is overrated.
Indeed. This is a great application for Rust. It's something that's complicated, difficult, and has to work right.
For Rust itself, maybe, but not for its community.
SketchUp has a lot of community assets but these still leave you hanging almost immediately.
You need to be able to scrounge up drawings of certain parts by part number that can be inserted quickly.
This is surmountable but I think it would take at least a decade to reach parity...in a best case scenario where CADmium is readily adopted.
I would put in the time to learn CADmium. Having worked in FreeCAD some and SketchUp moreso.
I hate that I can't automate things easier or build my drawings from scriptable components--like a XAML file.
Interesting. Any chance you could explain why fillets are hard(er)?
1. Surface tangency matching - perfectly matching the tangent of the connecting surface on either side of the corner.
2. Edge tangency following - you select an edge, and the fillet should be able to follow along all connected tangent edges.
3. Edge reference tracking - when you modify the model further up the feature tree the kernel needs to keep track of that edge, even if the surfaces that make it drastically change or are split.
All three are hard problems on their own, once you connect them all it becomes a great indicator of the capabilities and stability of a parametric kernel.
The impression I’ve gotten while learning FreeCAD is that making a CAD ‘just work’ is incredibly difficult and the commercial packages probably have a lot of very messy workarounds and heuristics in place to work like they do.
FreeCAD’s (well, OpenCascade’s) issues with fillets and chamfers appear to be most acute with a subset of your third point: sometimes adding one involves the new feature fully consuming another edge. This is unsolved in OCC and it’s why sometimes even infinitesimally changing the depth of the fillet or chamfer is enough to stop an edge being destroyed.
My assumption (I've not looked) is that you could offset the surfaces from the two sides (say A and B), and then calculate the intersection of an offset of A with the original B. Then do the opposite for the other side. I think that will give you two edges that will produce a consistent radius.
Offsetting surfaced and calculating intersections aren't exactly easy problems to solve on their own. Fillets are hard!
(Solvespace is awesome BTW! If I had the time (or the expertise) it also makes a brilliant foundation for all this)
Yeah, offsetting surfaces is harder than curves ;-) This was my early stab at doing chamfers on extrusions: https://github.com/solvespace/solvespace/issues/453#issuecom...
I know how to do it better now, but ugh... not enough time.
Tried using some other program but I loved the way solidworks let's you create parts, and then some other features where its hard to exactly discribe the exact thing but for me solidworks seemed so much more intuitive.
I dropped my summer internship money on the version which was a few hundred bucks. ...pretty sure it was cuz I got the student or the hobbist deal.
https://www.solidworks.com/solution/3dexperience-solidworks-...
1) You don't want just 2D constraints, 3D is better. If you were writing in C++ I'd say just take our constraint solver (Like Dune3D did). Since you're loving Rust, I can point you to the work of Michael F Bryan who wrote one in Rust and blogged about it here: https://adventures.michaelfbryan.com/posts/constraints-part-...
I think his code is over at gitlab. I haven't looked at it in a couple years. He wrote that after I nerd-sniped him ;-)
2) For geometry kernels... I've got 3 classes of bugs I want to squash in the Solvespace kernel and then it should do booleans pretty reliably, but I haven't had the time. Ours is just under 6k LOC so you could learn a lot from it. My email is the same ID at gmail if you want to ping on this topic. Its been a while since I looked at Truck and I thought it had stagnated a bit. This is a really hard problem, which is why there are so few options out there even in the commercial world. Even triangulating a trimmed NURBS shell is tricky.
3) History/feature tree is closely related to the "topological naming problem" that FreeCAD has. Solvespace handles this by creating each entity from a set of known things. If you try to recreate (regenerate in our lingo) it will just return a handle to the existing entity rather than creating a new one. In other words, every entity "came from something" and that relationship is remembered. Where we handle topological naming it works perfectly. But not everything in solvespace is covered by this. You need to bake this in from the start, it's not something you can easily bolt on afterward.
4) where is the link to try out CADmium?
I'm so grateful there's a tool like this out there - thanks for your efforts, they really are appreciated.
There is also a wrapper around SolveSpace's solver for Rust[1] but I dunno how hard that would be/is to compile for the webassembly target ...
The first line of the article says "We're building a new open-source CAD program"
Which is a shame because the AGPL has nearly the same effect & the bonus of unjustifiably terrifying big-corp legal teams.
Whereas Elastic license is about, "you must not deliver this software over the network as a cloud service".
The things the Elastic license is trying to prevent seem very different to me?
Cloud companies are unlikely to be comfortable doing that.
> Another downside is that solving this kind of matrix equation gets prohibitively slow when you have a lot of unknowns, which gives rise to the conventional wisdom that individual sketches should be small and simple.
I've gotten quite deep into this, and this is really not a problem in practice[1]
1. FreeCAD's main issues with constraint performance come from a redundant & unnecessary GUI layout algorithm, which falls over with just a few hundred constraints.
2. Eigen's sparse QR decomposition benchmarks at 18s for 2200 constraints, which is really not too bad
3. There are sparse QR decomposition libraries that can handle 500k-1M constraints in about 18s. I can't imagine a CAD sketch with more than a few thousand constraints.
[1]: https://github.com/FreeCAD/FreeCAD/issues/11498#issuecomment...
Awesome work! It seems like a hard problem
Ondsel is also increasingly good. They've released a new version recently: https://ondsel.com/blog/ondsel-es-2024-2/
Their website login-wall downloads for whatever reason, so go to github: https://github.com/Ondsel-Development/FreeCAD/releases
I've got no qualms with anything Ondsel is doing, except that their "releases" are based off of the dev versions of FreeCAD, which are still in active development where things are rapidly changing and are poorly tested as a whole package. This leads to the situation where, for example, very basic sketches will cause the most recent Ondsel version to puke in odd ways, but it's a problem that's already been fixed in the FreeCAD dev builds.
So certainly give Ondsel a try; there a lot to like, and they're adding some much needed features to FreeCAD, but until their releases are based off FreeCAD stable, treat it like the unstable dev build it is.
I pretty much only use the Part Design, Sketcher and Spreadsheet workbenches. Those mostly work OK together. I don’t dare branch out because of poor interoperability between benches; I sometimes use the Part bench as well but it’s so fiddly because you had to take it through two internal format changes to make it work in Part Design.
Hoping that one of the many open source CAD projects can supplant it eventually.
Another consideration for many is a good integrated CAM. This is where Autodesk Fusion, much as I dislike the company and the software, is unbeatable today.
I highly recommend the cadmium folks just soak themselves in onshape’s UX.
To illustrate why... There's a certain "artistic" FOSS project (before you ask — not GIMP) where the community, incl. former lead dev basically say this:
- UX is bullshit, it's just opinions (verbatim quote)
- No two people agree on what is the right way, so anyone willing to improve UX in this project will have to fight everybody else, don't hold your hopes high (almost a verbatim quote)
- We don't like drama in this community, so we tend to just accept patches that change UI (you'd think this contradicts the previous statement, but somehow the impossible is possible)
They'd also mock a person who wrote a very elaborate first impressions blog post about the software, claiming she is an idiot who didn't bother learning the software.
I don't think the FreeCAD community was ever this bad, but there certainly was a strong element of chaos before. Things are considerably better today. There's a design working group working its way through various incosistencies, for one. It's a long-term project, no doubt about it, they are slowly getting there.
Well, what else can you expect from the very same guy who came up with SolidWorks in the 1990s? :)
My ask for a free CAD software would really be for someone to go through and take the robustness and speed of Creo and combine it with the user friendliness of Solidworks, even if it were to have a fairly limited feature-set to start.
1) Watch and do one tutorial that gets you 80% of the way there.
2) Try to stumble together the rest of the way.
3) Find out the fundamental tutorial doesn’t set up my project in a way that allows me to finish the 20%.
4) Watch the whole project fall apart as I try to modify one of the previous steps.
5) Repeat.
Does the order typical matter that much in CAD? I went all in on this thinking it was a simple object and how hard can FreeCAD be. Turns out, very hard.
The Mango Jelly Solutions videos, the Joko engineering videos, the Adventures In Creation videos and thehardwareguy's videos are excellent.
There is no doubt that understanding how to mitigate TNP is important, though. Many of the videos do not go into this each time (which will be good once the issue is mitigated, very soon, but is obviously an issue at the moment).
The fundamental issue I see with bad Youtube tutorials is people skipping from Part Design to other workbenches without understanding that they are not wholly interchangeable. That is, all Part Design Body objects are Parts, but not all Parts are Part Design Body objects. Part Design is a subset.
There is at least one excellent video by Mango Jelly Solutions going into the detail of why you should choose one over the other.
But the guts of it is that Part Design does have an ordered flow -- it's a sequence of steps from a simple to a complete object, executed one after the other and implicitly booleaned together. So it has additive features and subtractive features.
Whereas in a non-Part-Design workflow, you are in charge of when something is cut from something else; there's no subtractive tooling, only boolean tools. More CSG-ish.
A good grasp of FreeCAD involves understanding Part Design's relatively special guided flow, and that outside of it is a more open-ended approach.
The RealThunder developments to Part Design after TNP was fixed do go a long way towards making this less relevant (Part Design can work with multiple bodies rather than being focussed on a single body, and you can "wrap" non-Part-Design objects in a Part Design Body object more freely without using a macro). And that way you get to use Part Design much more like Fusion 360.
But there are definitely FreeCAD proponents who would say that Part Design is training wheels for what is really going on, and that there's no substitute for many general concepts inside Part Design. The latter is true; also the implicit booleans in Part Design make some of the PD operations slower.
Yes, it does matter. I discovered this through my own self learning with OpenSCAD, FreeCAD and OnShape. I actually prefer the workflow of FreeCAD to OnShape for a lot of tasks.
FreeCAD isn't that hard, cad isn't that hard. Especially for any developer with some level of XP in breaking down a complex problem domain into smaller composable parts.
I feel it’s similar to having learnt Photoshop (7) and then struggle to use GIMP. I keep looking for tools that aren’t there, expect panels to show certain details and they come up short.
It reinforces for me how valuable usability and interface design is for power user tools. But also, how important training materials are, and how loyal a user will be once they are accustomed to completing a workflow with their tool of choice.
I also question what user problems this is trying to solve. CAD users don't necessarily need things to be open source. They may have limited budgets, but open source is not a user facing feature in this space the way it might be for some developers. Plasticity has shown that it is possible to license Parasolid and make money at a very low price point.
I've been using parametric CAD tools for 30+ years. I find that Onshape is a pretty amazing solution to many of the issues that exist with Creo or Solidworks. I never loose data. Having the equivalent of Google Docs for collaborative CAD with unlimited undo based on a ground up database instead of a file system is life changing. That said, its modeling capabilities are still playing catch up with other tools.
A big downside to Onshape (and Creo and Solidworks) is that it started as 2D sketches to extrude/revovle/loft etc into 3D objects. 3D tools are an add on and afterthought to the fundamentals. Plasticity (and Rhino and Alias) are all much more 3D first.
Onshape's FeatureScript is really cool and powerful, but it is not as good for CAD users to build their own tools as say Grasshopper in Rhino.
I would focus on what user problems are being solved. How will CADmium be a better CAD tool for a certain market than any of the others paid or free? Even though this is very early days in the development process, it's not too early to try to understand user needs. Open source is not in and of itself a goal.
I have no doubt that a small dedicated team could build a new parametric kernel and CAD package, particularly one that's open source.
It is an extremely tough market that many have been shaken out of.
Reliability is a must, particularly on the data representation and exchange front. And those assurances carry ENORMOUS costs. Big money at immediate risk in downstream, physical product if that goes south at any time - spread across multiple manufacturers and other product lifecycle parties.
Lots of workstation compatibility certification work done on CAD kernels.
PLM alone, embedded in many engineering organizations, is a very sticky element for retaining a certain CAD program and certain kernels internally - and it is my understanding that PLM is where the major money is made (not so much on the per seat CAD licensing costs).
New grads are coming out from college today after using CAD company-supplied software in much of their classwork for 4 years.
Many CNC controllers use Parasolid internally for certain visualization and programming operations - machines that will be working on the floor for decades from today.
The fact is that the per seat licensing cost and lock-in tradeoff is simply not a serious issue for many - and such costs have become arguably marginal for even small design houses and manufacturers.
I think CAD users would benefit greatly from good open source alternatives.
While there are certainly challenges to open source development, there are no downsides to using open source software. It works just as well for people who don't know what it is or means and just as well for people who are indifferent to it. There are great reasons to avoid proprietary software, though, and they're even stronger when you're talking about SaaS (as we increasingly are nowadays, and specifically are in this case). CAD software is an area where many users have /already/ been burned by proprietary systems changing their offerings /after/ they've invested hundreds of hours learning and using their platform.
I'm aware I'm in the minority, but open source is absolutely a goal in and of itself for me. I've been using open source software exclusively for a very long time.
FWIW, Onshape was created for less than $10M to make a MVP in less than 3 years. I don’t think you need decades of PhDs to make a new CAD program nowadays.
Creo uses the Granite kernel, which was developed by PTC, starting in 1985. [1]
Onshape and Solidworks (and many others) also license the same 2D solver, D-Cubed, now from Siemens as well (originally from a company founded in 1989). [0]
None of this is to say that it is impossible to start fresh, just that there are insane numbers of person-years of development work embodied in these libraries. To get anywhere near the level of completeness and power will take a lot.
[0] https://www.engineering.com/story/parasolid-d-cubed-and-siem...
[1] https://www.shapr3d.com/history-of-cad/parametric-technology...
Favorite quote from this:
"The only popular open-source b-rep kernel is OpenCascade, which is the Pontiac Aztek of b-rep kernels: It is ugly, barebones, and it might break down on you, but it is drivable and you can get one for free."
SO true!
Solvespace has the benefit of being a single download/executable.
It also has a constraint solver which has been used in a couple of projects: CADsketcher as you noted, and Dune 3D: https://github.com/dune3d/dune3d where the author noted:
>I ended up directly using solvespace's solver instead of the suggested wrapper code since it didn't expose all of the features I needed. I also had to patch the solver to make it sufficiently fast for the kinds of equations I was generating by symbolically solving equations where applicable.
Any relation to: https://github.com/jay3sh/cadmium ?
Also, for CAD kernels, Manifold was not mentioned: https://github.com/elalish/manifold/wiki/Manifold-Library --- while I understand it to have many of the same disadvantages as OpenCASCADE, it does seem worth mentioning.
Interestingly the kernel was previously discussed here:
https://news.ycombinator.com/item?id=35071317
It seems really interesting/promising, esp. the compleat history and editability (I'd love to see that history listed in a pane which could be opened/closed --- add a series of disclosure triangles which would allow hiding finished elements so that one could focus on the current task and it would be a dream come true for me --- if I can puzzle out the 3D stuff, so far I've crashed and burned on all the apps I've tried (BRL-CAD, FreeCAD, Solvespace, Alibre Atom...) --- the only thing I've been successful w/ is OpenSCAD and similar coding tools).
Because the web browser did what systems developers have failed at for at least 60 years: creating a platform that can host an application that just works on many different devices.
Even XBox controller's drivers come baked into browser. Just check the number of APIs on MDN.
As for performance, that's already not anywhere into discussion. Check Google Docs. Everything single thing you are on screen is drawn using Canvas and it works just.... everywhere.
Because it means it'll work regardless of whether I have a Linux, Mac, Windows, ChromeOS, iPad, or just try to check some detail of my part on my phone.
When I design a part at home, need to make an adjustment at work, and then need to make it using whatever is available in the makerspace, not having to install something is a big win.
Manifold is a triangle mesh library. CAD should use NURBS surfaces and only triangulate them for rendering. I had looked at replacing the solvespace triangle mesh code with Manfold, but it's a lot of work and what we really need is to fix the NURBS bugs so we don't need to lean on the mesh code so much.
BTW the solvespace constraint solver is also used in the Assembly 3 workbench of FreeCAD ;-) It's really getting around these days.
I am also interested in having a simple constraint solver for some kind of constructive geometry useful for CAD modeling, but not necessarily a full-fledged NURBS kernel. I think it can be interesting for text-based modeling, probably similar to some of the CadQuery APIs.
It was just me. It might fix our remaining issues with meshes.
>> Feel free to open an issue or discussion if needed
I have an open issue for Solvespace here: https://github.com/solvespace/solvespace/issues/1371
It's a low priority though.
I agree.
> If you want to make a good one and you only have a small team, the framework had better do a lot of heavy lifting!
I am skeptical that general-purpose UI frameworks can be a good long-term solution for 3D CAD. At times, a nice UX will need to think about many of the following at once:
- the parametric 3D model
- its mesh approximation
- its hidden-surface projection into lines and patches
- the pixels in the frame buffer
- UI widgets.
Maybe the frameworks will help for prototyping, but if the project grows, I expect at some point they will end up ditching frameworks for the core 3D viewport interactions, and have mostly their own code in between OpenGL (or similar) and mouse/keyboard events.
Frameworks for buttons, lists, etc. will probably be fine, but even those tend to be much more dynamic in 3D CAD than in average apps.
In general, egui IMHO has better ergonomics. Partly because of Rust and partly because it's newer.
I started with OpenSCAD but found FreeCAD to be way more productive for me when it came to actually producing 3d objects. Sometimes I want to start with a square, then delete a line, and add a semi circle. But I don't always know that until I start sketching. With OpenSCAD I feel you have to know the shape primitive you need, then the exact way to modify it, and the fact that it encourages 3D primitives over 2d w/ extrusion and rotating was kinda annoying imo.
Because it interprets the file as a sequence of Part object manipulations, it's a very useful way of getting a bRep representation of an OpenSCAD document, so you get the opportunity to generate a clean, minimal STEP file.
OpenSCAD cannot do this itself, and the other conversion approaches just get you a triangle mesh packaged in a STEP file.
It's not perfect -- you get a mesh if you use Minkowski or hull operations. But it's good.
But yes, the syntax is IMO much better because it isn’t trying to be so clever.
Many interesting features, and as it is bRep you have faces and edges, some access to constraints, the ability to import STEP etc. (oh and fillets/chamfers, with the usual OpenCascade caveats)
I have not got into it yet (mostly I am a happy FreeCADder) but your feelings about OpenSCAD resonate with me and this looks like it addresses many of them.
On the one hand OpenSCAD was the first thing that ever gave me any confidence that I would be able to make my own designs to print, and on the other the fact that it is wholly declarative is exasperating.
People make such amazing things in OpenSCAD (like Edgar Kech’s scalable large format field camera) but I do not have the tenacity to keep up with all the maths, or keep consistently avoiding coplanar issues in the preview.
As a trivial example, let's say I want a square and a diamond and I want the diamond's corner to touch the square on the side. It would look something like this:
square([10, 10], center=true); dx = 5 + 5 * sqrt(2); translate([dx, 0, 0]) rotate([0, 0, 45]) square([10, 10], center=true);
This is a simple case, but what if either shape is replaced with a hexagon, a gear, or a duck? What if I decide there should be some other shapes between them? What about...?
Ideally, it seems like I should be able to tell OpenSCAD that the square contains a reference point on one side, and one of the corners of the diamond is also a reference point, and OpenSCAD should translate the diamond however is needed to make those two reference points match.
Alternatively, I should use some other software that lets me do something like that.
cuboid([8,8,2]) align([RIGHT, FWD, TOP]) cuboid([5,5,2], spin=45);
You can't just write code to, for example, "put this line at a 56.7 degree angle and have it end 0.22mm from this surface". You have to break out the trigonometry to calculate how long that line needs to be.
We're also building an optional web-based file sharing and collaboration layer.
I'd be happy to support the project but current subscription levels are too high for a hobby shop. I wonder if there is a chance for a pathreon or similar type of support.
What you could do instead is donate to the FreeCAD Project Association. Part of the money goes into the development fund, that is, grants for developers.
As for the FreeCAD Project Association, thanks for the tip, that's already included. I prefer to spread the butter all over the bread I'm eating, if I can say it that way.
I am sure Ondsel will vary somewhat in UX over time (because they will be somewhat ahead of what they upstream) and because a few things may be unavailable for open-source licensing.
But most of what has changed in terms of UX is already in 0.22-dev. For example OpenTheme will work (don't know if it is included by default), the "glass" tree overlay from RealThunder is there, the overlay panels, the sketcher improvements, the new integrated Assembly workbench etc.
Ondsel is a commercial distribution of FreeCAD with a cloud-based engineering suite integrated -- file sharing, cloud compute (for parametric recalculations and STEP export) etc.
Here is how they describe themselves -- in some useful detail:
Ondsel are seemingly using that to integrate with recalculation of designs in the cloud compute facility (think Thingiverse customiser but a bit less basic).
I eagerly anticipate that in mainstream FreeCAD, because none of the alternatives are great -- e.g. Spreadsheet is simultaneously cool, powerful and utterly exasperating once it's well-integrated. I am sure we will see it.
Thank you for all of your work -- I am away from my CAD projects at the moment but I will be digging properly into 2024.2 as soon as I can.
The shapebinder technique near the end of this video is amazing!
https://www.youtube.com/watch?v=Yp6cIMA7LsI
And there's another clever technique using variant links (which you can take a bit further with BaseFeatures):
https://www.youtube.com/watch?v=m9C_ahIVKOI
It's funny. People describe FreeCAD as maddening or sadistic, and for sure there are elements of the workflow that are frustrating (I'd be happy if I could do the five things I use the Draft workbench for without ever having to open it), but at the same time, it's so liberating and enthralling once you get your head into its way of thinking.
For me it's like QGIS or Inkscape: it's mindblowing that this tool is available to me for free. The trivial things I've been able to do have really changed my life (and I don't mean to overstate that -- these are things I never thought I'd learn and the impacts on my creativity have been striking).
The macro expect a part container named "export_family" with a single spreadsheet that is used as a template to set any properties and generate the part name.
It will also export any objects within a part container named "export". It also handles recursive part container.
All of that makes possible to export for 3D printing in a single click. Including handling multi material objects more conveniently.
I tried the variant link, but I could only have one variant per link somehow. I couldn't import the same part with multiple variant in the same assembly.
I am a mostly (and increasingly) happy FreeCAD user (though I agree with your framing) but OMG I am happy for you, and I will try to get involved when I can.
Best of luck.
I know that ISO 10303-243 uses JSON but in hindsight I think that was a mistake.
I saw “CADmium will use JSON for everything.” and cringed a bit.
This seems like a mistake.
He missed BRL-CAD, which also does brep. The interface is clunky but the kernel seems advanced.
> In practice many optimizations are made.
And I’m sure there’s a good reason to do physics. But, do iterative solvers like gmres or whatever have trouble with these kinds of problems?
See that discussion forum here:
https://github.com/tlalexander/open-cad-foundation/discussio...
Many times I have used traditional solvers to map out an old building where no angles are 90°. I measure a bunch of distances (with some inaccuracy of course) and spend hours fighting with the solver to get to something that satisfies most of the measurements.
This takes time because I am in essence doing exactly the converging spring forces simulation that you propose. Except much slower :-)
The good side is that all CADs have a very steep learning curve, with most tutorials made by people who are oblivious to the ignorant masses. The landscape is so complex today, that it's easier to use a vector drawing tool and transform it into a 3D model than to use any CAD to make a cylinder.
What would people recommend for a CAD program to run on Linux that I could use now (that is either free or cheap). I have simple requirements, I want to make a box for some electronics, but need the ability to import (and extrude etc) SVGs.
I couldn't see the SVG feature when I looked at dune, and I had some problems when I used freeCAD, but was planning on looking into it again.
What would people suggest?
FreeCAD has really good file format support in my XP.
On another note, please don't use Discord for development: https://drewdevault.com/2021/12/28/Dont-use-Discord-for-FOSS...
use IRC or Matrix but not Discord.
On a related note, we really, really need a worthy replacement for EAGLE. It's only a couple years away from EOL, and KiCAD is light years away from being feature-equivalent.
If you say KiCAD is not an professional EDA due to missing live synchronization, the same argument applies also to Altium, Cadence Allegro, and most other EDA systems which are considered as professional by most people.
Does CADmium suffer from a topological naming problem like FreeCAD?
Well this article talks about revisiting venture capital in the future. Which pretty much means the end of cadmium at that point haha
Extremely unlikely. Also, no such plans :)
Also, "no plans" doesn't really mean much. "I guarantee" is a lot more meaningful.
The best a startup company can do is work in quick iterations: throw things at the proverbial wall to see what sticks, then scale up development accordingly. "Consuming FreeCAD" is the opposite of that. It's how you paint yourself into a corner and die a horrible death.
A much healthier approach is to build a great working relationship with the community: lead the way in a few areas where you have great expertise and encourage development in other areas. It has worked nicely so far.
So no, sorry, I don't see Ondsel consuming FreeCAD over the next few years (or ever). But we can agree to get back to this conversation in, oh, 5 years time would be enough?
Why is it that open source programs always want me to use an awful (and proprietary) chat app?
JavaScript should be limited to 1997 Netscape 3.0 performance to stop this kind of garbage. And WASM should never have been attempted, and should be removed.
Write real native software if you’re trying to make something good for end users.
It might be sensible to run CAD in a web page hosted by a server on your local network? I have no idea if they added functionality to have the crunchy bits run on the server instead, that’d be a neat trick.
Personally, I think it's a catchy name.