Local TutorialBlender for the faint hearted - 04 : The Basics of Modeling, part 2

The Smooth function is pretty obvious A– it takes the vertices and faces selected and tries to physically 'smooth' the surface that they form. This is not a virtual smooth, like [Set Smooth] A– this really does move the faces and vertices. Again, demonstrating this with a cube is pretty pointless A– the [Smooth] function will work on a cube, but smooth all of the sides of a cube together and all you get is another, smaller cube. Keep smoothing the whole cube and all that you will see is a gradually shrinking cube (though it is useful to smooth parts of a cube individually...)

With the monkey from the last exercise, select the [Smooth] (or the [w][0] shortcut) function. You should see the Monkey change it's shape ever so slightly A– effectively, the angles between points are rounded off and vertices moved where appropriate to accomodate the smoothing process. Keep doing it, and while the Mokey will get smoother and smoother, it will also shrink in size.

There are two functions with similiar purposes, but also with subtle differences. While [Split] is in this button area and [Separate] is not, it seems logical to deal with them together so that we can look at the differences between the two.

If we take a typical cube, it is constructed of eight vertices and six faces. Each face shares a vertex with other faces. We can also look at this from the opposite point of view A– each vertex is shared by more than one face. As the vertex is physically the same vertex, Blender can look at all of the faces during, say a [Set Smooth] request, and continue its smooth curve across the vertex as if all of the faces were one and the same A– in fact, this is exactly how it is done.

But what if we did not want this to happen? Afterall, we are not living in 'bubble' land, where everything is nice and smooth. Curves come to an end. Sometimes we need a sharpe crease between curves.

To do this, we can split off faces. What implications does this have to the mesh? Well, the most obvious visual implication is that any curve created by a [Set Smooth] operation will stop at the edge that was split. It is also however, important to know how this affects the vertices and faces themselves.

Create a new Blend file and add a new primitive A– this time a Cone, with 24 segments. Enter Object mode using [Tab], select [Set Smooth] and press [z] to go into solid mode. You should now see a nice, smooth cone on the screen.


Hit [z] again to go back to wireframe, then [Tab] to go into edit mode. If anything is selected, press [a] to deselect everything, then press to enter box-selection mode. Your cursor should now change to be the centre of a cross-hair extending to the edges of the window in which you are operating. Go the top corner of one half of the cone, and press the left-mouse button, then drag the mouse to the bottom-right hand corner of the same half of the cone A– we want to include in our box all of the faces in this half of the cone. Release the mouse button, and all of the faces within the box should be selected.


Now select [Split] or the keyboard shortcut [y]. Press [Tab] to go back to object mode, and press [z] to go back to solid drawing. See the difference? You should now have a clean crease or seam down the two halves of the cone.

So what has actually happened? Before the Split, the cone was constructed of 26 vertices (24 around the base, 1 in the centre of the base, and one at the top) and 48 faces (24 in the base and 24 forming the cone). After the split, we still have 48 faces, but our vertex count has gone up by 4 points.

Why? This is because the vertices that formed the edges of the faces that we separated are no longer 'shared' between the two sets of faces. At the border between the two halves, the shared vertices have been duplicated, and one set of vertices are now owned by each of the two halves of the cone. We can't physically see the new points, as they both occupy the same points in space.

Go into edit mode using [Tab] then into [Vertex Selection Mode]. If anything is selected, press [a] to deselect everything. Now take a look at the top left of the main Blender title bar. It will say something like 'www.Blender.org 236' and then a series of numbers (the 236 by the way, is the revision number of the Blender application that you are using). What we want to look at here however, is 'Ve:0-30'. This is the number of vertices currently selected (0) out of the total number of vertices currently displayed in this mesh (30).


Randomly choose one of the vertices on one half of the cone (on the base A– deliberately choose a vertex that you know is in the middle of the faces and not one on the border between the two halves). Once it is selected, the title bar should change to 'Ve:1-30'. Press [a] to deselect everything again.

Now go into box-selection mode again by pressing and box-select the point of the cone. You should see the title bar change to 'Ve:2-30', and not 'Ve:1-30'. Why? This is because the cone has been split into two halves, and each half has its own copy of the pinnacle point A– they just happen to occupy the same point in space, so you can't visually see the division.

Just for completeness, lets cover some more things that lead on logically from what we've just done. We just used to enter box-selection mode. As we mentioned in a previous tutorial, this is one of those shortcuts that has a multi-use. If you instead press twice, Blender switches first into box-selection mode (displaying a cross-hair cursor), and then into brush-selection mode denoted by a circle around the mouse pointer. The brush-selection mode allows you to 'paint' the vertices or faces that you want to select. Just press [Escape] to turn it off.

Finally, up in the title bar you will also see a section displaying 'Fa:0-48'. Logically, this is the number of faces currently selected, and available for selection. Next to this is something like 'Mem:3.10M', denoting how much memory is being used by Blender for the current operation, data blocks etc. Typically this will grow as you increase the size of your file, but will also jump tremporarily when you are rendering. Finally, on the end, you will have the name of the object that is currently being addressed by all of this information.

One last feature to look at before taking a look at [Separate]. If you are not in edit mode, switch across using [Tab], go into [Face select mode] and deselect everything using [a]. Now select one face only (any face) in one side of the cone. It is possible to select 'more' or 'less' of the faces surrounding a currently selected face using the [Space] bar to bring up the main menu, then the menu items [Select] > [More] or [Select] > [Less]. It is of course much easier to simply use the keyboard shortcuts [ctrl+numpad +] and [ctrl+numpad -]. Hit [ctrl+numpad +] now and more of the faces around the one that you selected are added to the select.

Repeat the keystroke a few times. It selected quite a few faces, then stops. It does not select the whole cone. Why? Well, normally it would keep going to the entire mesh was selected, but the [More] and [Less] selection methods will only keep working within the current distinct group of faces. Because we split the cone in half, and we only selected a face in one half of the cone, Blender is forced to stop when it gets to the split edges.

OK, so what is [Separate], and how does it differ from [Split]? Its very subtle. Create a new blender file and add a new Cone primitive A– in exactly the same way as we did above. Follow the exercise exactly until you come to the part where you pressed [y] to split the faces. Instead, press [p] and you should get a context-menu asking whether you want to separate what you have selected or all loose parts. Choose [Selected].

Right A– the select turned black, but not much else happened, did it? Well, press [Tab] to go back into object mode and take a look at the you mesh that is selected. Now, the whole cone is not selected A– only half of the cone. Click on the other half and it gets selected, but the other half-deselects. What is going on?


Well, what has happened is that while [y] splits off the faces within a mesh, [p] splits off the faces and then separates them into a totally different mesh. What we have on screen now are two meshes, with two separate origins. Both origins may be currently located in the same place, but that's only because that's where it was when we separated the faces.

Select one of the meshes so that it goes pink, then press [g] to go into grab mode. The mesh will turn white. Move the mouse to another location and drop it by pressing the mouse button. You should now see two physically separated meshes, and two physically separated origins.

The [Flip Normals] button, also accessible using the shortcut [w][9] does exactly what it says A– flipping the normal of a selected face(s) into the opposite direction and effectively reversing the way the polygon is facing.

Why do this? Well, imagine creating a 'stage' for scene. One way of doing it would be to create a box, maybe removing one or two sides, and then placing all of the 'actors' inside the box. The problem is that the stage may well not be visible as the walls of the box are all facing outwards. To correct this, we could select the walls and flip their normals A– changing the walls to face inwards instead.

Two quick ways of ensuring the all of the normals of a particular object are facing in the same direct are using [Recalculate Outside] (shortcut [ctrl+n]) and [Recalculate Inside] (shortcut [ctrl+shift+n]). Blender will go through the entire mesh and order all of the normals to be pointing in the same direction. These last two should be used cautiously however, as what Blender regards as 'inside' and 'outside' can be debatable depending upon the logic being used and the shape of the mesh. You may have some tweaking to do with [Flip Normals] to get it perfect.

If you want to see exactly where the normal is facing, turn on [Draw Normals] in the [Mesh Tools 1] section.

[Remove Doubles] should be used cautiously as it can destroy hours of fine tweaking of a mesh A– trust me, I've been there. In itself, [Remove Doubles] is a very important function that will go though your mesh and find vertices and faces that share the same point in space (the Limit button sets the limiting distance that tells Blender what 'the same point in space' actually means). Once it finds vertices that 'share' the same point, it merges them into one vertex.

The use of this function is particularly important when meshes are imported from another application A– some conversions of models can leave vertices and faces in strange mathematical quandries, so much so that when you start to use them the application can crash. Using [Remove Doubles] first, can remove a huge number of these problems.

The reason that it can destroy hours of work however, is when you are not totally aware of what it is doing. Take our Cone exercise for example, where we split the mesh in two. Use [Remove Doubles] on that mesh, and the two distinct copies of vertices that form our crease down the side would be merged back into one set. As a result, the two halves would reform and the crease would disappear. If you have a model with thousands of such detailed creases in place, putting them back again can be a nightmare. As a result, seriously consider saving your work before performing a [Remove Doubles], just in case...

A closely connected function to [Remove Doubles] however, is [Merge Vertex], which merges selected vertices into a single point. To use [Merge], select two or more points to be merged then select [Edit] > [Vertices] > [Merge] or just [alt+m]. Blender displays a context menu asking whether the points should be merged at the centre (i.e. the average position) of the selected vertxes, or at the current location of the cursor.

Create a new Blender file, and add a cube primitive. Deselect everything using [a], then select two points in the same face. Now press [alt+m] and select [At Centre]. Blender will tell you how many points were removed (nearly always 1 less than the number selected) and the screen display will updated.

If you had wanted to merge the two points into one at the current location of one of those points, the easiest method of doing this is to select the appropriate point then use [shift+s] to display the [Snap] menu, and select [Cursor -> Selection]. When you perform the merge, as described above, you would select [At cursor] to merge into the selected location.

Notice in this example however, that we can now see one of the bad things about quads (four sided faces). Two of the faces are now 'bent'. They are made up of four vertices, but they are now no longer in the same plane. If we really wanted to keep our cube like this, we should really convert those 'bent' (or more accurately 'non-planar') faces into triangles.

The easiest way to do this, is to jump into [Face select mode] and select the two offending faces. Next, press [ctrl+t] or choose [Edit] > [Faces] > [Convert to triangles] from the main menu under [Space]. You will see how the faces get converted into triangle shapes. The reverse is to select two (or more, Blender can intelligently make the most appropriate decisions on this) and press [alt-j] to convert triangular faces into quads.

The Spinning and Screwing tools are all connected with a rotation of some sort, coupled with a duplication of selected meshes, faces or edges as the rotation progresses. You may hear these tools being called other things in other applications, such as 'lathe' in Lightwave, but essentially they all rely on three important values to do their work.

Degrees
The [Degrees] button sets the number of degrees around which the operation will turn. For some operations, the [Degrees] setting is used in opposition to the [Turns] setting and vice-versa.

Steps
While an operation is performed around a certain number of degrees or turns, Blender still needs to know how many individual steps to break that rotation into. An operation aroung 90 degress for example, specifying 9 steps, would result in clean breaks every 10 degrees. For a quick idea of what this means before we look at the operations themselves, think of a 36 sided cylinder. You could think of this as being a rotation of 360 degrees, with 36 steps A– each face occupying 10 degrees.

Turns
Some operations only operate on a full 360 degree rotation, so specifying a number of degrees is pointless. We can often however, specify how many rotations are to be made. This is what we do with the [Turns] button.

Keep Original and Clockwise
The [Keep Original] toggle button simply specifies whether the mesh, face or edge upon which we are operating should be considered part of the final result, or whether it should remain where it is and the final result is a new, separate mesh based on a copy of the original.

The [Clockwise] button simply specifies the direction of rotation when an operation is being performed. This is also a toggle button and is selected by default A– so any operation will rotate clockwise around the pivot point. Logically, if this button is turned off the rotation occurs counter- or anti-clockwise.

OK, onto the tools themselves. The [Screw] operation performs a repetative spin on a mesh or set or edges, usually moving downwards as the rotation progresses. Effectively, we can use this tool to create a 'screw' A– hence the tools name.

There are three things to remember. First, the [Screw] function only operates on complete cycles of 360 degrees A– we therefore use the [Steps] and [Turns] values. Next, the centre of the screw revolution is centred on the position of the cursor, so make sure that you place it in the correct location for the centre of the screw. Finally, the cross-section of the shape that you want to use in building the screw is important. The top-most vertex of the cross-section is always going to joined to the bottom-most vertex after one rotation A– as long as you keep this in mind, you can create some really great shapes and control the descent and shrinkage of the screw very precisely.

There are several ways of making a cross-section to use for this exercise A– but we'll just use a simple square plane and delete a few edges. First, we need to be in the front 3D view, so start a new Blender file and if you only have one 3D view in use, press [numpad 1] to get into this view.

Next, add a new mesh primitive A– in this case a simple plane. You'll just get a square shape in the centre of the 3D view. Zoom in so that you can see it better. Now, lets use a selection method that we've not used before A– [Edge Select mode]. Get into this mode by pressing the appropriate toolbar icon, then select the top and right most edges (remember that you'll need to use [shift] to select them both. Press [Delete] and select [Edges] from the context-menu that is displayed. We should now have an 'L' shaped edge on screen.


Switch across to [Vertex Select Mode] and select each vertex in turn, press [g] to go into grab mode, and drag them out to roughly the shape shown in the image.

Finally, jump back into [Edge select mode] and press [a] to select the entire shape A– this is important, as the [Screw] tool needs to operate on a curve (a series of interlinked edges in Blender). Make sure that you have 9 [Steps] and increase the [Turns] value to something like 4, then select [Screw].

If you've selected the button marked [Screw] instead of a keyboard shortcut, the cursor may change to a question mark and ask you to select the 3D view that we are operating in A– this will only happen if you more than one 3D view on screen. Select the correct one, and Voila A– the Blender screw tool.

The [Spin] tool is what other applications sometimes call 'lathe'. It effectively takes a cross section and 'spins' it around a central pivot-point, joining each segment as it rotates. It is this tool therefore, that can be used to create everything from tyres to wine-glasses, and rocket-bodies to gas-cylinders.

Create a new Blender file, and get into front view using [numpad 1]. Add a new primitive - we want lots of vertices to play with, so create a circle with 12 vertices then zoom in so that you can see them. As before, select each vertex, press [g] and drag them into roughly the shape that you can see below. If you can't remember how to get those points perfectly aligned in the centre of the view, remember the [n] key to bring up the 'Transform Properties' dialog and that x=0 is the perfect centre.


Once you have the shape, switch to top view using [numpad 7] A– remember that when we are spinning, we are spinning around the pivot point so we want to be able to place our cursor down the centre of the wine-class. You probably have not moved it however, so leave it in the centre of the 3D view. We still however, need to be in the top-view for this operation.

Change the degrees to 360 (the maximum that you can), and increase the steps to something like 10. Finally, make sure that you are in Edit mode and press [a] to select the entire mesh ([Spin] needs to know what vertices you want to spin A– it won't just work on the whole mesh), then select the [Spin] button. As with the [Screw] tool, if you have more than one 3D view on screen, the cursor will change to a question mark A– just click in the top-view to tell [Spin] where to operate. The cross-section should now be spun into a wine-glass, but were not finished yet A– there are still one or two things to do.

There are two things that we really need to do. Normally we'd do them automatically and immediately, but for this first time lets modify what we are looking at so we can see where the problems lie. Use your rotation keys to rotate the 3D view into a 3 / 4 view so that you can see the entire glass. Switch to object mode and make sure that the glass is selected, then select [Set Smooth]. Next, press [z] to display solid, then turn on [SubSurf] and take the [SubDiv] value up to 2 or 3. You may (not guaranteed) see two things A– first, there seems to be a sharp edge down one side and secondly, there seems to be some strange ridges around the glass.

First, while the [Spin] tool spins the cross section around, it does not join up the two ends. If we were to render this wine-glass therefore, we'd get a nice glass but a clearly obvious seam down the side. The start and end vertices should be so close however, that we should be able to correct this by just selecting [Remove Doubles]. Do that now, and that sharp edge should disappear.

Secondly A– those ridges. If you have them, its probably because some of the faces have their normals pointing in the wrong direction. This is typical in this type of operation and pretty simple to cure A– just jump into edit mode and choose [Recalculate Outside] from the [Edit] > [Normals] menu (or just press [ctrl+n]).

And that's that A– one wine glass.

[SpinDup] performs a similar action to [Spin], but instead of taking a cross section made of edges and lathing around a centre pivot point, [SpinDup] takes entire sections of a mesh and duplicates them at the centre points of the steps selected. One i