(It’s not really fair to call something with this many steps a “quick tip”.)

Blender is capable of some pretty impressive feats of “organic object” modeling, but to do organic modeling easily requires unlocking a popular tool among the 3D animator crowd known as “subdivision modeling”.  The idea behind subdivision modeling is that although the computer needs to draw many, many polygons to approximate a smooth surface, rarely are those smooth surfaces so mathematically complex that they can’t be more easily defined by a set of control points.

Many ways of doing this exist in the field of 3D modeling software, but one of the most intuitive, and one that Blender handles very well, is breaking the mesh down into smaller polygons and twisting them so that they reduce the sharpness of the corners of the original mesh.

If that seems like a bunch of jargon, that’s only because it is– let’s just do it and see what happens.

If you select the default cube, you should see this panel on the lower right:

There’s a long list of options available when you click the “Add Modifier” button, but we’re just going to use “subsurf”.

If you select the default cube and then select subsurf from the Add Modifier menu, you should see the following appear under the Modifiers tab, in what is called the “modifier stack”:

The cube, as you will see, has had each of its six faces subdivided into four smaller ones.  This represents one level of subdivision.  The “levels” item in the subsurf panel shows how many times this subdivision process will be performed by Blender while you’re editing it.  (For solid modeling purposes, we can ignore the “Render Levels” item, which is there so you can have Blender subdivide a model more thoroughly for rendering.)

For every level up we bump the Levels item, every face is subdivided into four new ones.  Below are cubes subdivided one, two, and three times:

The first has 24 faces, the second has 96 faces, and the last one has 384.  As you can see, going much higher than this will start to tax system resources at comparatively little benefit, especially if you select “set smooth” for these objects.

This is an interesting way to turn a cube into a sphere, but so what, right?  Well, the usefulness of this doesn’t become obvious unless we hit tab to switch to Edit mode:

As you can see, Blender’s edit mode is only using the original points.  For this reason, you can now distort the sphere into a bunch of other shapes just by moving the original vertices, which in a subdivided mesh are sometimes called control points.  What’s more, if you use some of the techniques from previous tutorials on this mesh, you can quickly produce some very organic-looking shapes with comparatively few operations:

This nifty teardrop shape was made with just two extrude operations and some scaling.  Pretty neat, huh?

Of course, if you use this very long, you’ll start to get frustrated by the way you have to pull verts through each other to get some effects, as well as how cluttered the view gets when you have a large, complex shape.  This is why the Blender developers have included the following option, although I may never know why they hid it so well.

See that little unassuming circle?  Click it.

NOW we’re talking!  With the “true” position of the original control points hidden, the view of the model is drastically simplified, but all the operations of extrusion, twisting, scaling and so on still can be applied to it.  The model we saw at the beginning of this tutorial was done in maybe three minutes of fiddling around!

The Wireframe “Draw type” can be terribly useful for building complex objects, since it automatically culls all co-planar face edges, which is to say that flat areas are drawn as though they were single complex faces instead of the cluster of simple faces they really are.

This has one obvious benefit and one subtle one.  The obvious one is that you can see what the heck you’re doing when looking at a complex mechanical design.  The less obvious one is that when something is wrong with your mesh, wireframe mode will often draw a big bright white line where otherwise Blender might not give you any clue that skeinforge wouldn’t like a mesh.

Take these two objects:

One of them skins, one of them doesn’t.  Blender’s solid mesh vier renders them exactly the same.  However, switching to wireframe mode makes it pretty obvious

The cylinder on the left will not slice in skeinforge and in fact will spray errors like the dickens.  The one on the right will slice with no problems.  Both look identical in solid view, but here in wireframe, Blender knows that the top faces are not truly coplanar (they have infinitely thin slices in them) and so does not render them as a single face.  I purposefully generated this particular example by extruding the cylinder on the left with the “individual faces” option instead of the “region” option, but this and a number of other operations in Blender (including booleans on some kinds of meshes) can create invisible but lethal (to skeinforge anyway) kinks in the mesh topology.  In predominantly flat geometry, as is common with mechanical designs, the wireframe draw mode makes these errors stand out like a sore thumb!

Precision mechanical CAD tools come with a wide variety of snap tools.  Blender is relatively light in this department (its primary users are animators, not engineers) but there is a snap menu and it can be pretty useful.

The snap menu is in the object menu in object mode, and in the mesh menu in edit mode.  It can be summoned in either mode by pressing shift-s, however.  Here are your available functions:

Selection to Grid – In edit mode, each selected vertex will move to the nearest grid node.  In object mode, object pivots are positioned to the nearest grid node.  The grid in these matters is defined as whichever grid is visible in the current view.  This means that zoomed well away, the grid might be one meter spacing, and zoomed in it might be one millimeter spacing.  This can be irritating, but it’s better than nothing.

Selection to Cursor – In edit mode, selected vertices move to the cursor position.  In object mode, selected objects move so their pivots are centered on the cursor.  (The cursor is that little life-raft looking thing.)

Selection to Center – Again, affects vertices in edit mode, objects in object mode.  Moves all selected items to their geometric center.

Cursor to Selection – Moves the cursor to the center of the selected objects.  (If one object is selected, cursor moves to it.)

Cursor to Grid – Moves the cursor to the nearest available grid node, again dependent on which grid is the smallest visible.

Cursor to Active – Moves the cursor to the active object.  This is the most recently selected object in the case of multiple selections.

These snap commands are a lot more limited than many CAD tools, but they’re enough that an experienced user can recreate many of the more advanced functions with some slight of hand.  For a great example of that, see rab3D’s tutorials on Blender for Precision Modeling.

The gear script is really terribly useful.  It gives you pretty much any kind of single gear you could want.  But what about rome gears?  The boolean union operation can do just that.

First a quick reminder on how to get a manifold gear out of the gear script.  First, generate your gear, switch to edit mode, then hold down alt and select an edge on those inner rims:

Hit alt-m to merge the selected vertices, collapsing the gear down to a single point:

Repeat this for the other side, recalculate the normals, and you’ve got a manifold gear.  Next, generate another, smaller gear and solidify it.  (Gear tip: two gears will mesh with one another if the adendum and dedendum are the same, and both have the same value for r/T where r is the radius and T is the number of teeth.)

Move the gears so they intersect where you want them to, and hit w -> union to create a merged copy of the two gears.  (I like to rotate one of the gears just a tad– this keeps the boolean operation from generating any zero-width faces, which can cause odd things to happen later.)

You’ve now got three objects in your scene, the two original gears and the unioned mesh underneath.  You can just hit the del key to get rid of the old parts or drag them off to the side for later use (I tend to do this to save time trying to remember what parameters I used to generate the gears.)  A recalculation of the normals is definitely in order after an operation like that, but otherwise this gear is ready to print, although a few extrude opearations might be in order to make the gear easier to interface…

There are lots of ways to go from simple to complex geometry in Blender.  You can use extrude operations to add faces, subdivide to increase surface complexity, and spinning operations to create radially symmetric objects.

But what if you’re trying to patch something up?  What if you don’t want to add any points at all, but rather want to join up two objects, or cover over the surface of an object so skeinforge will recognize it as a solid object?

Enter the “add face” command.

To add a face, shift-select a group of three or four vertices and hit the f key.  Blender will create a triangle or a quad that uses those three points as its vertices.  There’s only a few cases where you’ll want to create a single face; this technique’s real power is in adding many faces.

For example, take this RepRap logo I’ve prepared on a circle suitable for, say, a cool-looking custom washer:

(To do this yourself, just start with a circle, shift-d to duplicate, scale, then reposition the points.)  If I extrude the profile I’ve drawn straight up, this will create a non-manifold object that skeinforge won’t be able to use:

There are other ways to generate this shape, but working in 2D before extruding a profile is pretty comfortable, so I’d rather be able to extrude this and use it.  F key to the rescue!

We want to infill the region between the teardrop and the outer rim, so we select vertices between them for our faces:

Repeating this process all the way around the rim turned out to be a bit tedious, but I don’t think I needed that many vertices on my circle to begin with.  Note with profile editing: most of the time you don’t really need super-fine detail to get a working design.  I’m not sure how many points it takes to get skeinforge to draw a circular toolpath around a four-milimeter hole, but I’d bet it’s not more than twelve.  Once we’re done, we’ve got a solid profile instead of a hollow one:

So when we extrude it, we’ll get a solid object:

With a little patience and some practice, fairly complex 2D profiles can be created and given faces, allowing them to be extruded into complex and useful 3D objects:

Check out Syvwlch awesome youtube tutorial. He took Zach’s bracelet and redid it in sketchup in no time flat! I have to say, I’m impressed with the way that sketchup has improved since I used it 3 years ago!

To print out an object you need an STL file. You can create your own or download one from Thingiverse, or even scan one if you have a 3D scanner (WANT!)

Once you’ve got an STL file, you’ll need to get it to the right size. Watch the video to learn how. To do this you’ll need Blender, the open source 3D creative software. It’s great and it’s free!

I’ll be making more videos like this to walk you through the process of slicing the STL file with Skeinforge and sending the gcode from ReplicatorG to the MakerBot.