The open-source parametric 3D modelling software, FreeCAD, is out in a release candidate for version 1.0. If you’ve tried FreeCAD before and found a few showstoppers, it might be a good time for you to test it out again because the two biggest of them have been solved in this latest version.
First, version 1.0 finally implements a solution to the “topological naming problem”. Imagine you want to put a hole into a surface. The program needs to know on which surface to put the hole, and so it refers to this surface by name / number. Now imagine you subdivide the surface, and both subsections get new names. Where does your hole go now? If you want to dig into the issue, the inimitable [MangoJelly] has a great video about the topo naming problem. Practically, there were workarounds, like only adding chamfers after the main design has stabilized, but frankly it was a hassle to remember all of the tricks. This is a huge fix.
The second big fix concerns assemblies. Older versions of FreeCAD were great for making single parts, but combining them all together inside the CAD program was always janky. Version 1.0 combines the previous two patchwork assembly workbenches into one, and it’s altogether more pleasant to use. The constraints of how two parts move when held together with an axle just works now, and this is a big deal for multi-part models.
If you’re coming from any other parametric CAD program, most of FreeCAD will seem familiar to you, but there will also be workflow differences that will take some getting used to. In trade, what do you get? Scriptability in Python, real open source software, and all of the bells and whistles for free. Now that its two biggest pain points have been addressed, FreeCAD has become a lot easier to love. We’re looking forward to some good V1.0 tutorials in the future, and we’ll keep you posted when we find them.
Great news for FreeCAD, especially for what concerns widespread adoption amongst hobbyists and professionals alike, especially considering the lack of viable alternatives in the Linux space for CAD tooling (not counting brittle Wine setups for running Fusion360, sorry not sorry).
Worth noting that although TNP has been vastly reduced, there’s still lots of work to do on that front – devs themselves never said anything about solving TNP, which btw is present in some form or another even in big name CAD packages.
However, coming from years of usage of those big names, I’m still kind of torn when using FreeCAD for even simple parts. I’m a big fan of symmetric designs and of reducing duplicating constraints manually when I can simply mirror features at sketch levels and let the solver do the rest, and after a little bit of toying with FC I’m still getting the usual PITA experience of constraints simply being ignored, or added and silently simmering up the usual explosion of invalid constraints whenever I close a chain.
I get that’s considered a “skill issue” and that’s simply part of “not knowing how FC wants you to behave”, but honestly for every other CAD program I’ve used, including SOLVESPACE, this is a non-issue.
Just my 2 cents.
I’m using big corp CAD all day long (SolidWorks, Solid Edge, Inventor) and you’re right regarding the TNP. It’s in all of them but the difference is all those packages let you repair your stuff by at least vaguely pointing at where the error is. Quick fixes most of the time.
That’s what I hope will be included in FreeCAD sooner or later.
I still use it for my private stuff and I like it! Assembly of parts was a main bummer though, that’s why I used it for parts only.
It’s not a 7k/yr CAD package but it does a very good job if you need to CAD up some stuff, even complicated parts.
You say:
That is the first and primary purpose of the new topological naming algorithm in FreeCAD 1.0. The second is sometimes suggesting a likely fix to accept manually; I see this often for dress-up ops. The third is, if confidently possible, to fix it automatically.
The gist i took after all those years of using them is: All CAD software sucks in one way or another, you just need to embrace the suck of your CAD package and work around it.
And that is true not only for serious CAD software, game editors all trough the bank have their “Why?!?” moments.
Very true, and in the case of FreeCAD that always used to be a very considerable change to the workflow for most who learned a different system. Seems like many of those are now resolved or at least mitigated which will help more folks learn and enjoy using it. I always found it good enough myself, as I put in the work to learn to ‘think FreeCAD’ and boy is its parametric spreadsheet type stuff worth the other pain points for what I’ve been doing, and for the price…
Seriously though you are very right everything sucks at least a little, and as its made by humans known for being really really flawless and reliably never make any mistakes….
The mirroring / circular / linear / etc. pattern in FC really only work on the extruded / substracted / etc. feature level (and nicely there in general), trying to do it on the sketch level is a world of pain indeed!
+1
I don’t even enjoy mirroring geometry in sketches in pro CAD software. Mirroring solids gives you other benefits, too. Like automatically merging overlapping geometry and so on…
While the tnp has pushed me away from FC in the past, sketch patterns in SW should be avoided unless you really know what you’re doing. They’re extremely fragile, it’s one region where I wouldn’t say the industry standard (SW) beats FC in usability. I also don’t love sketch symmetry, you need to fix it every time you change something small. Feature patterns/mirror are generally much more robust.
Linear arrays and polar patterns are vastly easier in FreeCAD in Sketcher now, and so is symmetry. Just try it in 1.0rc1.
The answer is: don’t mirror in the sketch. If there is symmetry, then sketch the lowest common denominator, then use pattern tools. This allows simpler sketches and simpler constraint schemes.
It also means sketches are easier to maintain and modify and are easier on the eyes.
Yeah I got that and that’s what I usually do, but it’s IMHO a non-solution. I’m not suggesting freecad should emulate other commercial CAD verbatim; but to offer a functionality and then when it never works to simply point users not to use it and act as if the feature wasn’t there to use in the first place is rather unpleasant.
I hope they will make it work eventually. When I started using FC 3-4Y ago, that was the case with many more buttons. Having pressed a few of them recently, to my surprise I found them working as expected! I have adapted my worklow now to no longer avoid them. I guess it’s what you might expect when the version is 0-point-something. One might argue that this is less expected in a 1-point-something version. OTOH, after they spent like 2 years on the topo fix (and many other improvements along the way), I can understand the desire to present FC as “no longer held together by prayer and string”
For every new FreeCAD user there is the first hurdle… Where can I extrude my scetch??
Look here for a step by step:
https://wiki.freecad.org/Basic_Part_Design_Tutorial
and here for just the button somewhat out of context:
https://wiki.freecad.org/PartDesign_Pad
I’m not sure if this is a serious ask for help or rhetorical comment.
For the record, when discussing FreeCAD and the generic term “extrude” it is best to be clear. In Part Design workbench the name of the operation is Pad. In Part workbench the name of the operation is Extrude. These operations are similar, but are different. Especially in the case of the Pad, which does not create an independent solid. Whereas in Part workbench Extrude does create an independent solid.
Don’t forget Draft Trimex op
It was a joke on a slightly confusing topic. Which you explained well.
I’ve had on/off relationship jumping between freecad / onshape for almost a decade on hobby projects.
Different learning curves. Love both.
The biggest hurdle (coming from Fusion) is that you can not use one sketch for multiple features (afaik [1]). That basically rules out working from a master sketch.
For the quick prototype work that I am doing the ability to quickly project anything into the current sketch makes working in Fusion a breeze.
I give FreeCAD another shot every once in a while, but so far I always fall back to Fusion :-/
(That said, FreeCAD 1.0rc1 is a lot more stable than previous versions)
[1] There is a workaround by shape binding the master sketch into the next part, then including features from it. But imo that’s too many hurdles.
Good news!
It has long been possible to use a single sketch for multiple features. What has not been possible using different elements in a single sketch for multiple features.
But now in 1.0, you can use separate elements of a single master sketch for different operations.
Downloaded and started using it a few days ago.
It feels slightly more polished than 0.21.1. I’m enjoying it so far.
Have they learned how to properly clean up a Windows uninstall yet? I still have v0.20.0 from 2022 showing up in my Windows app list and when I click on “uninstall” it errors out.
This is why i prefer portable versions. But in this specific case Freecad portable opens a full black screen for me, with only available options reload (the blackness) or close. Combined with my older experience of the same, they seem a very long way from my minimum.
It’s very unfortunate to suggest that the mitigations for the topological naming problem allow you safely to mix dress-up operations like chamfers and fillets into the middle of your design. That’s still fragile.
The topological naming algorithm does three things.
Most importantly, it usually can recognize if something is wrong topologically. (It’s not even theoretically possible to do this 100% reliably!) When it does this, it displays this at the operation that broke, rather than by the model turning to garbage potentially much later in the design tree.
Next most importantly, in some ambiguous cases it flags the error and presents a possible solution to the problem, which the user can either accept or change.
Sometimes (and, in practice, so often that many people expect this to be the only purpose!) it is so obvious even to the algorithm how to fix it up that it is able to fix it up automatically.
But failures to fix up automatically are not, strictly speaking, errors in the topological naming algorithm.
Over time, more automatic fixups and suggested fixups are likely to be implemented as more opportunities are discovered. What we have now is a framework and lots of use of that framework, but it’s not a one-and-done thing.
Great for free part design but free cad has limited utility in assembly design.
The third paragraph is all about how there will now be a unified, default assembly package.
Well I suppose I’ll need to take another look at it. As a long time CAD user (many different ones), I found FreeCAD very confusing and nothing at all like a typical CAD program. I get what they were trying to achieve with the parametrics but thought the implementation was horrible.
I have a personal vendetta against AutoDesk and their evil corporate overlords and licensing and refuse to use any of their products even though I admit their software works well.
Use OnShape?
Onshape is only free if you allow all of your files to be publicly accessible.
I have used it extensively for work and I consider it a half-baked piece of garbage. They refuse to fix simple issues (like adding text to a sketch dimension), hiding behind the premise that they focus on issues that get enough votes in the forums. This is a completely flawed approach to designing engineering software. It is one of the reasons people get so annoyed at solidworks. As if we spend all day combing the forums.
Yes, nothing is perfect, but onshape is not ready for primetime. It has some nice qualities but on the whole it is clunky. In addition to ignoring simple UI fixes (we’re talking actual UI, not 3D computational engineering feats), their drafting features are incredibly subpar. Don’t get me started on their refusal to allow sketching or planes in assembly. I feel like on the whole, onshape has pushed a lot of operational burden on the user to simplify computational complexity, i.e.- our internal engineering team’s desires are more important than the users.
Their yearly subscription price is the same as Inventor, which has 25 years of development behind it, and actively solicits individual user feedback. At this point, they are owned by a mega corp, and there is no excuse for not spending in engineering to make the product BETTER and EASIER to use, instead of focusing on the bare minimum feature add to sell more new licenses.
I’ve used AutoDesk Inventor for quite a while (while I got it for free as a student), then Fusion 360 and have switched to OnShape a couple of years ago. I found OnShape to be excellent for my needs (designing functional 3D printed parts) and much more robust and the AutoDesk software. It even works fine on Linux, as it’s just running in a browser (not that that’s perfect, but better than AutoDesk again). So is there anything better?
I’ve tried FreeCad quite a few years ago and it was just an utterly incomprehensible mess to me…
Assembly and referencing geometry in onshape is so convoluted and fragile it may as well not exist. While referencing part geometry is somewhat fragile and needs to be done with care no matter the CAD package, it is just unbelievably painful in onshape.
The TNP was the big thing keeping me away from regular use of freecad, and the lack of a standard assembly workbench sure wasn’t helping. I’ve been using onshape for things here and there, I suspect with freecad 1.0 I’ll finally find it personally acceptable and drop onshape.
I found Onshape very hard to use. The metaphors, the methods are all super dumbed down to the point of making no sense whatsoever. They won’t make it unless they redo that UI. It’s impossible to figure out.
I have been using FreeCAD as my go to CAD program since Autodesk bait-and-switched the Fusion360 licensing terms. It took some getting used to, but I’m now relatively proficient in it.
I’m pleased it’s seeing so much development and growth. I hope it remains open source and the good times keep coming.
Many years ago, Blender was laughed at as an option for 3D. It’s not the darling of the industry (yet), but it’s giving Maya a serious run for its money (literally). I see the same future for FreeCAD. And we all win when that happens.
I wonder if FreeCAD would have to move to GPL to pull off a Blender. Blender the project is hard to subvert as the software is licensed with GPL (v3 even!)
I’ll have to check this out. Fusion 360 has been the sole remaining reason for me to have my machine as a windows dual boot. Would be nice to break ties completely.
We use Alibre Atom 3D. It does 3d models and assemblies and drawings really well, (well, at least we think it does), and it’s not expensive, American based, and no resellers involved. Downside is boolean intersecting is somewhat limited until you pay for the Pro version and there’s an infuriating need to hook up the computer to the internet every 30 days for “licensing” purposes.
If a vendor (selling software OR hardware) requires that I connect to the internet, then I move on. It’s why I switched from windoze to linux, solidworks to freecad, and altium to kicad. Taken, the alternatives are not quite as polished, but for the price (dollar / ‘I own my data’ / privacy) I’m willing to sacrifice a bit of polish. True, internet connected is easier with the alternative products, but it is not forced upon you.
Hardware is sadly going the same way. Some 3D printers and other CNC machines require you to have the machine ‘internet connected’ otherwise they just don’t work.
Sure, if I want to connect – but don’t ram it down my throat buddy.
I love Alibre. That is, I loved Alibre. I bought the personal edition many years ago through a reseller, because rest of world(outside of US) has to go through a reseller. Why? Just sell direct FFS. Anyway, shortly after purchasing, Alibre sold out to a big corporation(3DS?). The personal version got toy-ified to something else I can’t remember the name of, then again to ‘Cubify Design’. Then that mob didn’t want it anymore and Alibre became Alibre again but Cubify Design got left behind. So I’m now an orphan. Software still works, for now, is mostly Alibre like under the covers, but no upgrade path because it got orphaned and if I want to buy back into Alibre again it’s back thru resellers…. So…. Nope. Honestly, without reseller I’d probably have bought back in years ago.
“Version 1.0 combines the previous two patchwork assembly workbenches into one” — there are three actively developed 3rd party workbenches, and the integrated assembly workbench was developed from scratch anyway :)
I like CAM workbench. I can only imagine 3 improvements:
1.) Adaptive clearing only works in 2D, having some basic 2.5D functionality would be cool. (eg. combining 3D pocket and 2D adaptive into single 3D adaptive pocket)
2.) Tool management is bit clumsy to use
3.) I can’t figure out how to simulate externaly imported g-code using CNC simulator(s) provided by FreeCAD
Been using FreeCAD only by the past 2 years in all the projects in my company (that’s some 4 or 5 new designs monthly), and I’m in love with it.
That’s the one FOSS project I’m the most proud to be supportive.
For me it is KiCad. I make annual donations in place of the ridiculous licencing fees of altium.
FreeeCad is not quite at that level yet, but I’m hoping 1.0 will crack that egg.
I installed Ondsel (FreeCAD fork) literally two days ago.
I used a bit of the Realthunder fork before, but vanilla FreeCAD is way to buggy for me.
Sorry, I will wait for a real 1.0 release to try it out.
I need to try it again. I kept running into problems with the GUI. I was trying to follow a YouTube video course on how to use it, but tons of things were different between version 0.20 and 0.21, which made it very difficult to follow. I’m also not understanding the reasoning behind decisions made in the software. I spend several hours trying to figure out how to use it and I still have no idea what the point of restraints is, besides trying to annoy the user. I’m sure there is a point to it, but I have no idea what is is.
I’ve designed hundreds of things in TinkerCAD, which is my default “CAD” software. It’s probably not the thing I should be using, that’s why I wanted to use FreeCAD in the first place. And I couldn’t get Fusion360 to run on Linux.
Maybe CAD programs will work better for me once Elon gets the brain connector working properly.
I’m looking at using ollama and qwen2.5-coder constrained with a RAG on a vector database built from the FreeCAD python docs to help people make the jump to FreeCAD because the learning curve can be frustrating for people who just want to get a specific thing done. Having precomputed RDBs for each FOSS application release would be a huge help. Imagine that, hooking ollama or its kin into F1 level documentation, chat with the application to learn how to use it.
I don’t know why freecad can’t be as polished tool as blender.and on the top TNP just put off the whole thing.Devlopers are not focused on the real deal which in my opinion is to make a CAD software that works as good as others in fundamental design then add a hell lot of addon, plugins etc.I just want a simple CAD software that can work. Is this a lot I asked for? And the ui is just sucks lot of things in one window poorly organized.I mean there are lot of CAD software in the market to quantify what works what not.But I still have some hope let’s see what’s new in this version.
Posts like this make me wonder how much of a chance you actually gave FreeCAD.
Is FreeCAD the same as Fusion 360? SolidWorks? OnShaoe? Of course not. But they aren’t the same as one another either. And they all have strengths and weaknesses.
If you’re truly proficient in any other CAD package, getting up to speed in FreeCAD is really not hard. I suspect that a lot of the complaints come from users who just aren’t as experienced with mechanical design as they think they are.
And therein lies the ACTUAL weakness of FreeCAD: documentation.
FreeCAD has a wiki, but it’s (a) written rather tersely, mainly by programmers, and (b) unfinished and not all up to date. As a result, the best learning resources are actually YouTube videos and online forums.
Version 1.0RC1 is a fantastic package. Amazingly capable, and more than stable enough for almost anyone. It’s also pretty easy to use… BUT… without at an onboarding tutorial for newcomers, it’s not welcoming.
I hope that the development team addresses this weakness before the official release. If they don’t, they’ll be squandering a huge opportunity to win over the maker community.