Not really a good candidate for a general-audience "easy to use" 3D design program (and thus doesn't preclude their plan to build something like that), but likely highly relevant for any HN readers who are looking for such a thing who are comfortable with code/markup and are wired to be able to pre-visualize things in their head prior to building them.
There are off-shoots like OpenJSCAD (see: http://openjscad.org/) as well, which actually give you more of a complete language. The OpenSCAD language is really simple and is more like a markup language with some limited flow-control than a real full-blown programming language -- but I kind of like it that way, and if you need to get really algorithmic with your designs you can always generate OpenSCAD "code" from other languages (of your choice) since it is all just text.
Designing and printing a 3D object is fundamentally much more complex than typing out an essay and hitting print. There's always going to be a big learning curve for someone trying to get into 3D printing/modelling for the first time. There's no way to get around that, it's especially hard for people who do not have a lot of spatial reasoning experience.
The most important point in the article, and I think the most valuable potential feature of new 3D printing oriented design software is this:
Communication of 3D printing limitations to the user at design time
For someone experienced with 3D printing (someone who's messed up a few prints and thought about why) it's easy to see why certain features are unlikely to print correctly because they have had the opportunity to build a mental model of the 3D printing process in their head. Helping people build up this mental model quicker, especially without having to actually print and iterate physically is invaluable to being able to efficiently design 3D parts.
For basic things like model checking, I'd recommend NetFabb model checker[0] and/or uploading the file to Shapeways[1].
The first will try to do basic model checking and close any holes it finds. Shapeways uploading process will go into much more detail, trying to find areas that might be too thin for the printer to print, or too close together...
Also, I long for the day where other 3D printing technologies reach the general public. FDM (the process all makerbots and clones use) is fine, but the quality is not as goood as it could be and things like support material are a necessity when doing complex pieces.
BTW, just in case anyone needs this, a great program to convert a 3D model to gcode is Cura[2].
I wonder if a simple way to teach people what are "printable" models is to build a library of good and bad examples , showing them some examples , and testing them on other examples whether they are good and why - while giving feedback after such example.
This should work because the brain is great at pattern recognition.
I'm surprised they don't mention OpenSCAD [1], which in conjunction with a decent vector graphics editor works wonders for a wide range of objects.
Yes, it has a programming interface, but I'd argue the visual/spatial context makes it much easier to pick up than most programming languages. I've seen it successfully taught to people who otherwise have little to no coding experience.
(If I had to gripe about something in OpenSCAD, it would probably be lack of spline support.)
OpenSCAD is great, I went from no 3D modeling experience to making the part I wanted in a about a weekend. I could see it being a bit cumbersome for large parts or dealing with assemblies of parts (can it even do that?), but it was really much easier than I thought it would be.
Not sure what the author means by "freeform". He seems to use that term to refer to mesh editors, which do surfaces for graphics, vs. solid geometry editors, which understand volumes. Mesh editors are more for 3D graphics.
Learning to use a constructive solid geometry system used to be hard, but Autodesk came out with low-end tools based on their their Inventor product line which make it easier. There's 123D, 123Make, 123Sculpt, Meshmixer, etc. The heavy CSG machinery of Autodesk Inventor is inside, but it's not dumped on the user all at once.
They also have Inventor Fusion, which is gratis (or at least used to be) on the Mac App Store - not sure what the Windows version costs. That and OnShape are probably the most capable $0 options for parametric solid CAD.
We used to teach little children to use OpenScad years ago along with programming(I volunteered for that). Kids learn fast.
Now we teach them FreeCAD, and python scripting for the advanced ones.
I have no idea what the author is talking about, designing a 3d editor is not trivial, most kids don't need very advanced 3D editing either. They will just ask the one who is good at it, or download it from the Internet from already made models.
There is a need for 3D curves in the Open source world. This will be worked on over time.
"The problem is that these requirements are not communicated to the user during the modelling process, so the model that the user designs is not well optimised to be printed, and must often be fixed with other software."
I couldn't agree more - finding out your design won't work after you've manufactured it is sort of like finding out your code won't compile after you've deployed. There needs to be better feedback during the design phase to make hardware as easy and fast as software.
This is exactly the problem we're trying to solve at Plethora, albeit with CNC milling instead of 3D printing. We have an add-in for 3D modelling programs that tells you manufacturability issues in real time as you design.
For parametric CAD, there's Solvespace. GPL'd as of version 2.0. I like it better than OpenSCAD, and the price is equally right. I don't know why, but it doesn't get much attention.
Just checked it out...on Linux. Cross-plateform packages have been posted to their forum. Solvespace is fairly intuitive and quite functional. Happy to have found it.
TinkerCAD is used a lot for 3D Printing, but at the low end.
At the higher end, there is http://onshape.com, an online professionally-oriented CAD tool that is aimed at mechanical engineers. Onshape is from the team that brought you SolidWorks.
While http://Clara.io does have some capabilities for 3D printing, it is best to think of Clara.io as like an online Blender with publishing. Like Blender, Clara.io is oriented towards media creation (stills, animations, videos, etc.) but not necessarily restricted to solid models.
We use SolidWorks at my school, and my students don't seem to have any issues. It's fairly easy to use and easy to generate files that our 3d printer can handle.
While the linked blog post focuses on usability rather than price, it is perhaps important to note that a single SolidWorks license costs about 8 times as much as you might pay for a pretty decent (if basic) FDM 3D printer.
I'm not arguing that the software isn't worth that price because its scope is vast, but that price makes it basically a non-starter as a suggestion for a software package for hobbyists.
SolidWorks is the industry standard, but the price is pretty far out of range for most hobbyists. If you aren't a student or pirating it, SolidWorks will exceed the cost of your printer. Most people are printing relatively simple things and don't need that much capability.
SolidWorks and Autodesk Inventor are the big two CSG CAD systems. If you know one, transitioning to the other isn't hard. Each reads the other's format, so they interoperate.
Yup SketchUp is pretty good, but it's really not ideal for 3D printable objects. And it's still not easy enough to help enable the widespread adoptions of 3D printing.
http://www.openscad.org/
Not really a good candidate for a general-audience "easy to use" 3D design program (and thus doesn't preclude their plan to build something like that), but likely highly relevant for any HN readers who are looking for such a thing who are comfortable with code/markup and are wired to be able to pre-visualize things in their head prior to building them.
There are off-shoots like OpenJSCAD (see: http://openjscad.org/) as well, which actually give you more of a complete language. The OpenSCAD language is really simple and is more like a markup language with some limited flow-control than a real full-blown programming language -- but I kind of like it that way, and if you need to get really algorithmic with your designs you can always generate OpenSCAD "code" from other languages (of your choice) since it is all just text.