I do not like comments on blogs. I think it cheapens the entire internet. I wrote about this here where I fully explain my thinking. I said that if you think you have something so clever and insightful to share that everyone should see it, send me an email and if I agree, I’ll write a new blog post so that everyone can. Here is a great example of that in action!

After writing about Blender For Engineers And Technical People a loyal and astute reader pointed out that LibreCAD and FreeCAD are options. Indeed they are! Good options. So much so that they deserve their own analysis.

I do like both of these programs a lot. Unfortunately I have to dismiss LibreCAD right away for my personal use however. In my CAD notes I say:

LibreCAD correctly emphasizes that it is "Open Source and GPLv2". This is very important! If you need some blueprints, take a look at this wholesome piece of software. My problem is that I believe that the entire concept of blueprints is moribund. Why are they even needed? Today, even the shop guys can simply reference a computer model of the part to manufacture. That may yet be a bit of science fiction, but it’s possible now and coming soon. Maybe 2d tool paths are a good application of this program. When it comes to 2d work, however, I’d probably just stick with what I know best which is…

Inkscape -- if you’re doing 2d work involving vectors, you owe it to yourself to properly understand exactly why Inkscape is deficient because in most cases, it is not. Inkscape is the straightforward GUI way to edit raw SVG (something that can be done with Vim and Unix tools too). Inkscape can make accurate geometry but it can be tedious. However, dashing off decent looking sketches and diagrams is very easy in Inkscape.

So if you’re just trying to replace an old-school draftsman, definitely check out LibreCAD.

FreeCAD is a serious tool for 3D modeling and I’m watching it closely. But mostly for my friends and colleagues. I’ve personally been focusing on Blender, but FreeCAD does have some advantages over Blender for many people and perhaps most engineers.

  1. It is designed for technical work so the developers will consider how to model some real world machine part accurately before considering how the final display of it will look. Blender can do proper precision modeling but it’s an afterthought for most Blender pros.

  2. It will be integrated with relevant analysis tools quicker than Blender will be. For example, FEA, SPICE, fluid flow, etc. One day I’m sure it will have modules for quirky things like groundwater dynamics, pipeline design, and naval architecture, etc. Oh, here we go — here is an impressive list of FreeCAD Workbenches that people are working on. Blender is similar but they tend to jump from one form of entertainment to the next and that is off topic for engineers. FreeCAD will be able to stick to engineering topics easier hopefully keeping the noise down. If you know you are really interested in, say, gear analysis, this active project proves that FreeCAD is a good platform for such things. If you’re willing to help contribute software, it is quite compelling.

  3. I don’t know this for fact, but I guess that FreeCAD is aping popular proprietary engineering software to a certain extent. Blender is doing that too, but for entertainment production software. People used to, say, AutoCAD or Creo will probably understand the FreeCAD workflow better.

  4. FreeCAD is likely to take more seriously the exchange nightmare between engineering software formats. You’ve been a bad DWG! Also from my CAD notes…

    As a quick reminder to myself why I am utterly correct in repudiating predatory proprietary CAD systems, here’s a classic example. I asked an intern to create a model on whatever system he liked. He used an expensive CAD package that only runs on terrible operating systems. Fine. Whatever. He then came to deliver the model on a different computer running a terrible operating system that also had this same company’s very expensive CAD system installed. I won’t link to that system but it rhymes with the Spanish word for "ugly". Shockingly to all of us, the intern’s model was not interoperable and could not be opened because the intern had used a "student" version of the program and the other computer had the "full" version installed which could not read those files! Mind blown!

  5. In theory their part libraries can be better integrated for technical and architectural purposes. This is just a matter of streamlining to reduce operator overhead. Blender can enjoy billions of available models of everything already, but FreeCAD could make, say, laying out HVAC ducting a little less of an adventure.

  6. There will be better support for things that are idiosyncratic to technical work. For example, the instruction manual to my miter saw has a diagram of all the parts in 3d, but exploded. FreeCAD should be good at stuff like that. Here is some work on that exact thing already.

  7. This isn’t an advantage over Blender, but I must point out the insane situation where proprietary software packages are not cross platform. Not only will commercial products squeeze every dollar out of you they can, you’ll also be forced to use a shitty operating system. Yuck! Of course Blender is completely cross platform too. But let’s all just take this moment to appreciate how good the situation is today that free software exists to give one sensible options. I remember when AutoCAD went from DOS to DOS/Windows to Windows-only breaking all of my OS level keyboard re-bindings. I was not happy about that.

FreeCAD has some disadvantages too.

  1. The Raytracing Workbench suggests that models can be nicely rendered, but that is Blender’s insanely powerful forte. If you really want your FreeCAD models fantastically rendered, send them to Blender which is a completely sensible workflow.

  2. Same as #1 for architectural walk through animations and machine articulation demonstrations and the like. E.g. I’ve been trying to model a canal lock in Blender — I actually want the gates to open and the water (with boats on it) to rise! I’m confident FreeCAD will one day be able to do this (there is already an Animation Workbench) but I think it is a ways into the future yet.

  3. There is a Mesh Workbench but it is somewhat of an afterthought for FreeCAD. That’s fine and I deeply appreciate their more efficient exact modeling style. But for FEA and very complex (e.g. 3d printable) parts, that mesh power will come in handy. And Blender beats all comers with mesh awesomeness. I am regularly shocked that certain mesh operations in Blender are even possible with computer science.

  4. I feel like FreeCAD is an attempt at putting a sensible GUI over OpenCascade, a CAD library created by a proprietary company that, essentially, failed to become the 800lb gorilla. They then released their stuff to the public domain. That’s cool and that’s how that should go, but it turns out that Blender is also the result of a failed commercial attempt to compete with those dirty tricksters at AutoDesk. They failed, then opened their code up to GPL and it slowly crept forward. Now Blender is where Gimp was about 6 years ago when they had perfected everything except being usable. After fixing that problem too Gimp now goes head to head with Photoshop easily. (I’m shocked that people still use Photoshop. Or trust Adobe at all ever.) Blender is at this stage going through this now. FreeCAD is still chasing down core functionality and stability IMO. But they’ll get there. Frankly I’ll be surprised if big parts of OpenCascade aren’t jettisoned for some of Blender’s core internals which are fearsomely good.

  5. Blender is not very good at holes. But do a search here for the word "hole" and nothing comes up. So I worry that FreeCAD is not yet great at it either. Probably the hardest problem in BREP 3d modeling is having a nice thing and then riddling it with holes. And then tap those holes — ick! But with Blender’s massive mesh power, you can just brute force holes as if you had a sculptured artwork. And Blender has astonishingly good CGS tricks. (Glance at the pix here if that is unfamiliar.) I was modeling a building (in Blender) and came to a dead stop when trying to punch out a door. Once I figured that out, I was stopped even longer trying to figure out how to punch out the window. The normal Blender pro way is to just slice up the wall into a zillion quads and delete the ones over your window. But then there are 20 or so extra redundant quads which makes me queasy. But for Blender pros, as long as there are not 20,000 superfluous quads, it’s all good and no human will ever notice. That reflects my introduction to 3d modeling back before our current embarrassment of riches.

At the end of the day there is nothing profoundly different about any of these approaches to modeling. If you disagree, compare using any of these software tools to sitting at a large drafting board with a drafting arm and using pencils to put graphite lines on physical paper. That was a profoundly different way of modeling. Today we are going to get all our geometry into a computer model. Once there computers should be able to help us work with that data in whatever way is most convenient.

I think my main reason for going with Blender now (besides finding entertainment applications interesting) is that Blender is ready for serious professional use now. The details of whom it is targeted to are not so important to me. Remember that I have my own spartan geometric modeling system buried in my own programming language. I am flexible and can stay focused on the fundamentals of the model and the data. It is delightful that today we have some options in the free software world that can make that a lot easier than when I started my project 20 or so years ago.

I think FreeCAD is very interesting for engineers who are looking for a tractable alternative to whatever it is they use at work. I would say that one’s motivation to use free technical modeling software is inversely proportional to how long one should wait to start investing in FreeCAD. That is because FreeCAD has been steadily improving since I first started watching it many years ago. Properly motivated, it is usable today. One day in the next 3-10 years, people using it will be baffled at how proprietary alternatives can still exist.