FreeCAD deserves a lot more recognition. It is literally the only serious open source contender to commercial CAD softwares. There has been a lot of recent positive developments including Realthunder’s continuous effort to solve topological naming problem. I use FreeCAD as my daily CAD software to design some fairly complex assemblies and it has become a joy to use. Hope we can get more developers and companies behind it and witness the same revolution that happened with Blender.
That's fantastic to hear, I've been wanting to switch to it for a while, as I hate how closed the commercial CAD programs are, but I've never put in the effort due to its various issues.
Would you say it's as good as OnShape nowadays, in terms of ease of use/stability?
Definitely. I’d recommend using FreeCAD Linkstage3 for now as it has a fix for topo naming problem and a variety of other improvements. For tutorial, I highly recommend OficineRobotica youtube channel. His videos are the best educational stuff I know of. With some practice you can learn proper workflow which prevents a lot of problems down the line.
I noticed in the Assembly3 section that it uses SolveSpace as its solver. FreeCAD uses OpenCASCADE as its main CAD kernel, and this supports mating and constraints between parts (at least, CadQuery supports it, and it's essentially just wrapping OpenCASCADE) so I'd be interested to know why they chose to bundle and use a different kernel there.
Assembly3 support various backends including Solvespace, and another using Sympy and SciPy. Realthunder mentioned he was working on supporting FreeCAD interal solver so I guess it is still WIP.
>> I noticed in the Assembly3 section that it uses SolveSpace as its solver. FreeCAD uses OpenCASCADE as its main CAD kernel, and this supports mating and constraints between parts (at least, CadQuery supports it, and it's essentially just wrapping OpenCASCADE) so I'd be interested to know why they chose to bundle and use a different kernel there.
AFAICT OpenCASCADE does not include a geometric constraint solver like the one in SolveSpace. There are companies licensing such things for use in OpenCascade derived products. I think OCCT is used in FreeCAD for reading/writing files, handling all 3D geometry and NURBS operations and rendering. Only the constraint solver from Solvespace is being used in FreeCAD (and only in one workbench), as there isn't an equivalent in OCCT. The Solvespace geometric kernel is not as mature or complete as OCCT, but it is about 100 times smaller.
FreeCAD is a sprawling, incredibly feature filled piece of software that's slowly but surely becoming bug free enough to be usable. It's really impressive how many different aspects of it are under active development - and for simple things the part design workshop is really usable.
The biggest thing I'm looking forward to (assuming it ever happens) is the selection of a single "assembly" workbench. There's a whole load of assembly extensions, all of which take different approaches. A single "official" assembly workbench would give me the confidence to sink time into big, assembly heavy designs and know that it wouldn't all be wasted by the next release.
Same here — I taught myself CAD via FreeCAD tutorials and a month of making one-a-day household items, but ultimately I moved on to other tools (Fusion 360, Shapr3D) for better assembly support as I got into real-world uses.
I like FreeCAD a lot, but it's too hard to create assemblies and reference other bodies in an assembly. I gave up and moved to Fusion360 where all these things Just Work.
However, I miss the KiCAD plugin with really good integration for the updating board outline and component placements from CAD.
I find the FreeCAD shapebinder system for external references is a lot more stable and maintainable than, say, external references in SolidWorks. It takes more clicks and more work though.
I have canceled my auto renewal for Fusion 360 as I have moved completely to Linux and their product does not work without using an unofficial Wine setup (also my space mouse does not work in Wine, it works perfectly in FreeCAD and Blender). They do have a web version but it is not available to paying customers?! They also keep removing features from the paid version I have. I'm done with this garbage. I can't even easily export my work.
FreeCAD is "feature full", but ... bearer of bad news here (sorry):
1. The UI is simply *awful*. Reminds me of the bad old days of Blender. Feels like a Motif style UI from the 90's TBH. Freecad needs to do what Blender did, bite the bullet hard and redo the UI from scratch to bring it into the 21st century.
2. The underlying NURBS engine (OpenCascade) is what's best in the FOSS world, but unfortunately it can't hold a candle to the "kernels" (CAD lingo for engine) that power equivalent commercial offerings (Fusion, Rhino, SolidWorks, CATIA, etc...)
Finally, I really wish FreeCAD had an OpenSCAD-like domain specific language for modeling that can directly create NURBS (there is a kind of an OpenSCAD integration, but it can only do poly meshes).
I want to point out that FreeCAD receives $450/week in recurring donations (https://liberapay.com/FreeCAD/), and that that probably has some impact on how quickly things are moving now!
I personally do CAD less than once a month, but I still donate regularly. If open source CAD is a cause you care about, you can too!
I really wanted to like FreeCAD, but I’d say out of the five or so times I’ve attempted to use it, it crashed and lost my design four times. I picked up Fusion360 instead, then OnShape, and never looked back.
I just don't trust Fusion360 not letting me have my files on my own hard disk.
I'm too much of a beginner to drop thousands on the higher end professional software.
So FreeCAD sounded perfect.
But three main problems made me give up:
1) The "topological naming problem" / random crashing. Couldn't ask a team to put up with that.
2) I found there wasn't a clear path to navigate the choice between the competing versions / forks / assembly things, and learn CAD at the same time. (Don't know enough to know what I'm comparing)
3) The interface just isn't as easy to use or learn. There isn't a long and clear set of tutorials like there is for Fusion360 [1]
So I've decided to settle for learning CAD first via Fusion360 for now, in hope that by the time I've got a good understanding of it, the various FreeCAD tribes will have merged into one solution, or that the main fork has improved enough that someone makes great tutorials for it.
[1] Here's is a great, professional tutorial series for Fusion360 https://www.youtube.com/watch?v=qvrHuaHhqHI (Somehow it's almost too-professional unofficial-official. Does anyone know if this is actually Autodesk?)
I see the need for FreeCAD to form an organization or foundation as a non-profit. This is explicitly mentioned in the wiki. https://wiki.freecad.org/Donate
In FreeCAD, I see a product that is almost there, but suffers from a lack of creative leadership and business thought. Of course, it's being done by long suffering devs with little resources and time. But the truth is this could be a success on the level of Arduino or RPi if the right organizational structure was agreed upon and executed.
Devs need to be paid. Labors of love never turn out as well as well funded endeavors. There are many models to point to but the one I like is RedHat. RedHat serves corporate customers for money, and delivers free products to individuals. Open source software for free but with paid support and custom development. This is the way.
In my opinion this is the perfect model for a CAD system development organization. FreeCAD needs to be free for users to learn and modify. This is goes without saying and FreeCAD is popular with users IN SPITE of it's glaring flaws. If FreeCAD could be developed to a level that made it a viable choice for professionals like myself, a mechanical engineer, to trust enough to base a development toolchain on, then corporate money for enterprise level solutions would flow like water. The CAD market is ripe for disruption and a solid open source model could do that. SW costs ~$4000 plus ~$1000 per year in maintenance fees and let's be honest, it is broken but usable and has a network effect in play. But nothing they do is unique or even especially state of the art, it's stuck in the 80s.
How do we make this happen HN. I see a "FreeCAD is gr8" post like once every two months and looking at it we should be able to org our way to a better solution.
23 comments
[ 3.3 ms ] story [ 72.1 ms ] threadWould you say it's as good as OnShape nowadays, in terms of ease of use/stability?
AFAICT OpenCASCADE does not include a geometric constraint solver like the one in SolveSpace. There are companies licensing such things for use in OpenCascade derived products. I think OCCT is used in FreeCAD for reading/writing files, handling all 3D geometry and NURBS operations and rendering. Only the constraint solver from Solvespace is being used in FreeCAD (and only in one workbench), as there isn't an equivalent in OCCT. The Solvespace geometric kernel is not as mature or complete as OCCT, but it is about 100 times smaller.
The biggest thing I'm looking forward to (assuming it ever happens) is the selection of a single "assembly" workbench. There's a whole load of assembly extensions, all of which take different approaches. A single "official" assembly workbench would give me the confidence to sink time into big, assembly heavy designs and know that it wouldn't all be wasted by the next release.
However, I miss the KiCAD plugin with really good integration for the updating board outline and component placements from CAD.
I tried a few times to switch but now I have to.
What's really cool is how easily I can export Kicad projects and import them into blender. Check out my latest PCB render: https://www.youtube.com/watch?v=svU7NyfJJKY
I personally do CAD less than once a month, but I still donate regularly. If open source CAD is a cause you care about, you can too!
No wonder open source software takes so much time to develop.
Why are free software lovers such cheapskates?
I just don't trust Fusion360 not letting me have my files on my own hard disk.
I'm too much of a beginner to drop thousands on the higher end professional software.
So FreeCAD sounded perfect.
But three main problems made me give up: 1) The "topological naming problem" / random crashing. Couldn't ask a team to put up with that. 2) I found there wasn't a clear path to navigate the choice between the competing versions / forks / assembly things, and learn CAD at the same time. (Don't know enough to know what I'm comparing) 3) The interface just isn't as easy to use or learn. There isn't a long and clear set of tutorials like there is for Fusion360 [1]
So I've decided to settle for learning CAD first via Fusion360 for now, in hope that by the time I've got a good understanding of it, the various FreeCAD tribes will have merged into one solution, or that the main fork has improved enough that someone makes great tutorials for it.
[1] Here's is a great, professional tutorial series for Fusion360 https://www.youtube.com/watch?v=qvrHuaHhqHI (Somehow it's almost too-professional unofficial-official. Does anyone know if this is actually Autodesk?)
In FreeCAD, I see a product that is almost there, but suffers from a lack of creative leadership and business thought. Of course, it's being done by long suffering devs with little resources and time. But the truth is this could be a success on the level of Arduino or RPi if the right organizational structure was agreed upon and executed.
Devs need to be paid. Labors of love never turn out as well as well funded endeavors. There are many models to point to but the one I like is RedHat. RedHat serves corporate customers for money, and delivers free products to individuals. Open source software for free but with paid support and custom development. This is the way.
In my opinion this is the perfect model for a CAD system development organization. FreeCAD needs to be free for users to learn and modify. This is goes without saying and FreeCAD is popular with users IN SPITE of it's glaring flaws. If FreeCAD could be developed to a level that made it a viable choice for professionals like myself, a mechanical engineer, to trust enough to base a development toolchain on, then corporate money for enterprise level solutions would flow like water. The CAD market is ripe for disruption and a solid open source model could do that. SW costs ~$4000 plus ~$1000 per year in maintenance fees and let's be honest, it is broken but usable and has a network effect in play. But nothing they do is unique or even especially state of the art, it's stuck in the 80s.
How do we make this happen HN. I see a "FreeCAD is gr8" post like once every two months and looking at it we should be able to org our way to a better solution.
My DMs are open.