Getting Started, Documentation, Tips & Tricks

As someone who started learning 3D with AutoCAD, I understand your concerns about precision, Hamei. After years of typing in exact point coordinates and using highly accurate point-snap systems and dimensioning tools, the freeform geometry tools of 3D Studio felt very imprecise. Although it is possible to use exact coordinates in most 3D animation apps, the focus is definitely not on accuracy.

For me, what reconciled it up best was a passage in a 3D animation book (can't remember which one exactly), which explained that in 3D animation, numeric dimensions are irrelevant - what matters is what the viewer sees on screen. Surfaces don't have to align perfectly, so long as they appear to do so. This concept took some getting used to, but it does make sense: in CAD/CAM, the goal is not to make pretty images, but to produce accurate schematics; in animation, the goal is not to produce schematics, but to make pretty images. Kind of like the difference between traditional architectural drafting and oil painting.

Of course, when you start working with characters, any notions of precision go right out the window, as the process is basically akin to clay sculpting or figure drawing - it's entirely based on artistic judgments about form and proportion.

That said, it can sometimes be useful to use real dimensions and accurate modeling tools in 3D animation, especially when you're trying to model a man-made item based on its real-world measurements. In that case, all major 3D apps allow the user to set dimensions and type-in coordinates, and most have relatively robust snapping tools.

(BTW, you've made your feelings about AutoCAD clear - what would you say are the best advantages of Pro/E, Catia, or I-DEAS? I've never used any of them. And do you have any opinions on Microstation?).

I bet there's not a nickel's worth of difference between them, it's mostly personal taste. Kind of like Ford or Chevvy, they're all professional applications and they all do a very good job, whichever one you like is the one that best fits your brain wiring ....

I don't disagree with you about the differences between cad and animation, just think that maybe if people are looking at this as a possible career, don't overlook the boring mechanical engineering side. It's not so boring and there might be more jobs in that field. Making the coolest monster in the universe could get old when you're forty-five, where designing a boring refrigerator that you then see for sale at the Amana store, that can be pretty neat.


No, but I should try it some day. Those guys have been around a long time ....

I have never used Autocad for 3D-stuff or really done any 3D-CAD work, so can not comment on that. I would use 2D-CAD (or drafting app, as dc_v01 would say) for drawing something like that.

Here is an (bad)example to demonstrate that even if it is 2D-drawing, it does not have to look boring...

I was stubborn for a long time too. But now I'm easy-going and flexible

I've never liked 2D. Even DOS Cadkey was 3D, although wireframe. If you make the model in 3D then you have every view you need from one piece of work and there's a much reduced chance for errors.

The thing that is so appealing about a solid model is that you make one model, then everything is derived from that. (In China most factories make 2D drawings then create a 3D model from that. Talk about backwards ...)

Once your model is done, your drawings are (basically) done. Choose "New", select "drawing" then pull your model onto the sheet. Choose top, side, left, right, whatever views you want. The model is accurately dimensioned already, just choose which dimensions you want to show. Now here's the beauty of it : something is not so great about your model. You need to change the wall thickness or whatever. Open the model, change that feature. Click "regenerate" and the model will rebuild. Now open your associated drawings - yup. Drawing are all changed to match. Woo-hoo, no work required !

Need to know physical characteristics ? Weight, mass, center of gravity ? Click click. What if we want to make this lighter ? Change material to aluminum, click click.

Hmm, changed from steel to aluminum, should check for strength. Take this same model, drop it into an analysis program. Constrain it, apply a load. Ah ! Exceeds material strength in this area. Change model, everything regenerates once again.

Are you starting to see that the real beauty of these programs is not even the initial uses of the model ? It's how easy it is to make changes and have everything keep coordinated. Effortlessly ! Tnis is something a computer really is good at.

We think this thing is okay now, let's make a test part. Take the same model, use it for generating a machining program. Cut five, test to see if they behave as predicted ... if not, no biggy. Make your changes, everything you need to do changes along with.

You get everything from one model in a solid modelling program. For most things the 3d solid model is not that much harder to make than a simple 2d drawing. I admit that for some stuff I still use a wireframe program I've had for years. That's because it's really fast and I'm used to it. But for anything beyond totally trivial, that's stupid (Sometimes I qualify.) You get so much more from the same amount of effort put into a 3d solid model that anything else is retarded not a good idea

Ha! You're learning!

What's missing from this discussion of CAD vs. 3D modeling software is an appreciation of surfaces, and another class of CAD software. The 3D CAD software hamei is bringing up - Pro/E, CATIA, SolidWorks, etc. are parametric CAD programs. They are particularly excellent at modeling prismatic or rectilinear objects, with cylindrical surfaces thrown in for good measure. Just about everything is established through driving dimensions - and you can parametrically define different parts of the same family through tables of dimension values. The history of the modeling process is also preserved, and you can go back and make changes to earlier decisions.

However, this approach pretty much universally sucks at modeling freeform surfaces. Not that you can't, but it's usually painful. This is where Maya, Blender, etc. come into play. Come in, grab the vertices on the mesh directly, push them around until it looks right to the artistic eye. Not impossible to do on the other programs but you generally don't want to try. The 3D modeling software has staked out this ground, and they excel at freeform surfaces.

There is another class of CAD software, however - explicit modelers. These did not have parametric capability, do not preserve history. ME10 might be historically the most famous, and this was also the wasy AutoCAD/(maybe Inventor?) worked (OMG I wouldn't want to use those). They don't have all the power of the parametric modelers, but they don't have the restrictions, either, and they're generally much better at surface modeling. They're better than "3D modelers" at solid modeling, preserving more of the "CAD" approach. I believe ME10 (which was originally HP) then morphed into a product called CoCreate. PTC - PARAMETRIC Technology Corporation, the makers of Pro/E - recently bought CoCreate and appear to be pushing it quite heavily - even into areas that I'd think would want to be parametric. For those that are interested in this kind of software, they have free personal (evaluation?) licenses/downloads (for Windows) on their website. I haven't tried this, but I think they're pretty close to fully functional, little or no limitations.

This might be a good oportunity to compare your favorite 3D modeler with a CAD program, although I think you'd miss a lot of the power of parametric modeling. PTC seems like they were adding a lot of traditional parametric capabilities to CoCreate, but I don't really know the details.

It probably doesn't compare to Maya for ease of use but there are some "hybrid" modellers that supposedly work in solids or surfaces equally well. Varimetrix was one, (exclusively ?) on Irix until somewhere after version 4. I only got to play with it a little at shows but it was very nice. Now called VX and available only for Wincrap.


ME10 is/was the 2D drafting program. Really nice interface, almost as good as DOS Cadkey. It supposedly ran on Irix too, but I've never seen the Irix version anywhere in the real world. The 3D modeller was ME30. There was also a knockoff of ME30 done by some of the same people, and then that 3D Eye from Trispectives tried being a surfaces / solids hybrid. That one was sort of halfway between CAD and animation. It turned into IronCAD later on ?

http://www.thefreelibrary.com/3D%2FEYE+ ... a017435872

I wonder why it never took off ? It should have. Ran on Windows and everything


Yeah, what's up with that ? CoCreate was awful. PTC bought DesignWave which was really good and flushed it down the toilet. Then they buy this pile of poop and remake it then peddle the hell out of it. Weird.


Get BRL for free, it's the same crappy 'stick anything anywhere" system

I'm still surprised that there is no good CAD program for Linux. In fact, there's not even any potential projects working on it. There's a few goofy pretenders but nothing solid. Nothing except Blender in the animation type software, either, is there ? Any ideas why that would be ?

Linux users have a choice of Maya, Softimage, and Houdini, so there's no shortage of high-end 3D animation apps on the platform. Discreet has also transitioned most of their apps to Linux. It has basically replaced IRIX in the high-end VFX domain.

Yeah, uh, no, despite have never used CoCreate, I have touched BRL, so I'm fairly certain Co will spank it all over the place.

aj pointed out the "3d modeling" software. You're right about CAD area. Pro/E Wildfire used to be available for Linux, but PTC dropped support for it??? I would've thought some engineering types would love the power of Unix style command line scripting...

simply because writing a large software system requires more the efforts of a few monothematic open source hackers. it requires organization, and orchestration beyond the abilities of random aggregations of rogue programmers. its requires the repeated inspiration of a well organized, level headed, sustained effort of more than a 'couple of guys'. it's work, it's hard, and it takes years, more than most people take to grow out of the desire to waste your efforts on thankless toiling.