Snowman – Modelling The Character (#2)

In the previous chapter, Part I, the importance of setting the scenes initial properties before was discussed. Whilst not generally an absolute requirement, it is an aspect of development that needs double-checking to make sure any subsequent work is correctly scaled with respect to the environment into which it is to be placed. Once this is done (where needed) it’s time to move on to the characters main construction.

In Part II then, continuing on from the Sphere objects initial placement the mesh can now be developed and shaped further using a number of Blenders core editing and manipulation tools – adding faces, filling gaps, mesh smoothing and so on.

Download: KatsBits – Snowman Character Source (c. 40 MB | *.blend, *.tga, *.psd, *.xcf). Zip file includes each stage of the process as indicated throughout.

Using Loop Cuts

With the two main spheres placed (“Shift+A“), positioned and joined together (“Ctrl+J“) as a single selectable Object, it’s time to start editing the mesh.

Using “Shift+MMB“, “Ctrl+MMB” and “MMB+Hold” to “Translate“, “Zoom” and “Rotate” the 3DView, manipulate the scene so its contents fit the height, and are approximately centre-screen of the window.

Design note: a majority of work done in the 3DView relies on its constant manipulation through the use of “Shift+MMB” (translate), “Ctrl+MMB” (zoom) and “MMB+drag” (rotate), making the need to have a firm grasp of the appropriate mouse and keyboard combinations essential.

Although the two spheres are now joined into a single mesh they need to be further edited so the two halves (upper and lower spheres) are ‘unified’, becoming a single whole object. This is done by removing some waste material and then filling the gap between the two halves. The first step to achieving this is to add a “Loop Cut” to the lower sphere as a means to balance out the distribution of faces between the two halves and make it easier to fill the gap.

Design note: adding a loop cut is generally a three step process; 1) the mouse is positioned over the face or group of faces to be cut; 2) the cut is made and then positioned within the loop; and finally 3) the action is confirmed and set.

To do this first RMB select the item and switch from “Object” (default) to “Edit” mode using the “Tab” key (or choose “Edit Mode” from the “Mode” selector in the 3DView Header). With the mouse cursor over the 3DView initiate “Loop Cut” using “Ctrl+R” (or click the “Loop Cut” button in the ToolShelf to the left). A pink guideline will appear, the loop selection marker, indicating where and in what direction the loop is to be cut. The position of the marker is determined by the mouse cursors location so move this to the lower of the two spheres and hover over the row of faces second from top – the guideline will encircle the mesh using the selected loop highlight this.

Design note: when moving the mouse to select a loop it generally needs to be positioned on or near an edge. Due to the fact that cuts are perpendicular to these a vertical edge needs to be used for a horizontal loop, and a horizontal edge for a vertical loop (shown below). In other words, a loop can only be cut across an edge, never down it.

With the mouse over the chosen loop LMB click to place the cut. The guideline will be replaced by an orange ‘cut’ line, the position of which needs to be finalised by again moving the mouse. Find the midpoint of the loop by holding “Ctrl” as the mouse is moved and/or watch the “Edge Slide:” indicator bottom far-left of the 3DView Header for it to display “0.00“, when this appears LMB click to confirm the action and set the cut in place.

Design note: as an alternative, because the cut is a ‘centre-line’ division, i.e. through the centre of an edge/face, once the loop has been selected (step 1) and the initial cut selection made (step 2), using RMB (at step 3) will confirm the action and place the cut dead-centre of the selected edge/faces automatically (removing the need to manually position the loop cut).

Select Mode & Border Select

In wanting to ‘unify’ both halves of the mesh into a single whole as discussed above, a specific area between the two first needs to be removed. However, doing this currently when ‘vertices’ are the selectable element, i.e. the type of element that highlights when using RMB, results in the removal of more material than necessary. To avoid doing this, using “Mesh Select Mode“, the type of element selection can be changed from the default “Vertex“, to “Face“, meaning only the necessary areas are subsequently affected.

Design note: the interdependent relationship between vertexes, edges and faces mean that whilst an individual face may be bound by four edges and/or four vertices, those same edges and vertices might also belong to neighbouring components, so changes to the one will likely affect the other.

Whilst remaining in Edit mode press “Ctrl+Tab” to access “Mesh Select Mode“. From the available options click “Face“. The mesh will update in the 3DView to display each face with a small dot in the centre indicating ‘faces’ are now the active element. Before making a selection however, switch to “Wireframe” using “Z” (or select the option from “Viewport Shading” in the 3DView Header), and make sure to be in “Front” Ortho view, “numPad5” -> “numPad1” – doing both these actions helps mitigate the problem of over-selecting (picking up elements that aren’t needed).

Now that the mesh is being viewed ‘flat’ and ‘front-on’, press “B” to “Border Select” or “C” to “Circle Select” and either LMB+drag a selection box or LMB+Hold to ‘paint’ a selection area which includes the lower two rows of the upper sphere and the upper two rows of the lower (shown below).

Design note: both ‘border’ and ‘circle’ select are available from the “Select” Header menu when in Edit mode as “Select » Border Select” and “Select » Circle Select“.

Finally with the selection highlighted press “Del” to access “Delete“, then click “Faces” to remove the unwanted material (or from the “Mesh” menu in the 3DView Header, select “Mesh » Delete » Faces“). This ‘opens’ the mesh and allows it to be ‘unified’ without issue.

Design note: using “Del” won’t remove material directly because vertex, edge and face elements need to be treated in slightly different ways due to their respective relationship to mesh structures (see note above). Whilst a particular face might need to be removed, surrounding vertices and edges may need to remain in place for the integrity of other elements making use of them. Blender provides the option to make this choice rather than simply deleting everything outright.

Adding faces, using Fill

The gap between both spheres as a result of deleting needs to be filled. However, because of the way it loops around the mesh it’s not possible to do this in a single action. Instead it has to be done manually by selecting and closing a pair of edges.

To do this first switch “Viewport Shading” back to “Solid” using “Alt+Z” and select mode to “Edge” using “Ctrl+Tab” (select “Edge” from the list). Then using “Shift+RMB“, select a parallel and matching pair of edges across the gap and press “F” (or from the “Mesh” Header menu select “Mesh » Face » Make Edge/Face“), a new face will appear between the two. Deselect all, “A“, and “Shift+RMB” reselect another pair of edges and repeat, continuing around the mesh to close the gap entirely. The end result will be a new ‘loop’ of faces which will make the object look akin to a peanut shell.

Design note: be sure to properly clear selections and newly created faces, “A“, otherwise they may cause problems with subsequence selections/additions. Double-check mouse selections when selecting edge pairs to avoid picking up and edge at the back of the mesh, resulting in a warped face. “Ctrl+Z” to “Undo” creation in such instances and reselect.

Multiple Loop Cuts

The previously created loop of faces is to form a scarf wrapped around the Snowman’s neck. It lacks depth at present so to make it more substantial and scarf-like, two additional loop cuts need to be placed which will then be used to shape the particulars of the scarf. Remaining in “Edge” select mode, activate the loop cut tool, “Ctrl+R“, and when the initial pink guideline appears scroll the middle-mouse button “up” (“MMB+scroll up“) to add another loop marker. As the guidelines are automatically spaced it’s not necessary to adjust their position as per adding a single loop, so all that ‘s required is to click LMB confirming the action – both guidelines will be replaced by two highlighted orange edge-loops around the mesh. There are now three rows of faces and four loops (top, bottom and two new middle) defining this new area.

Design note: when placing multiple loop cuts it’s only necessary to use LMB once to confirm the action as opposed to twice as is needed for single loop placement, i.e. Ctrl+R » scroll » LMB to confirm versus Ctrl+R » LMB to place » LMB to confirm. To ‘add’ or ‘remove’ loop divisions, scroll the MMB up or down as needed once loop cut has been initiated. Note also that if the tool is repositioned over a different area of the mesh the number of the divisions and their distribution updates accordingly.

The new loops now need to be shaped to create the beginnings of the scarf. If they are not highlighted as a result of the previous step, remain in “Edge” select mode and use “Alt+Shift+RMB” to multi-select them again (i.e. central row of faces, top and bottom edge loops). Next press “S” and “Scale” the selection a short distance away from the main body (move the mouse a short distance, remembering not to engage the tool if the mouse cursor is too close to the mesh), then LMB click to confirm. Keeping the current selection active, press “S” again followed immediately be “Z” – this locks further manipulations to the ‘up/down’ axis only – and resize the loops, this time so the top and bottom row of faces form an ‘lip’ between the main body of the mesh and the central row of faces. Once done, LBM click to confirm the action.

Design note: alternatively switch selection mode to “Face“, select the central loop using “Alt+RMB” then scale as above. Switching between selection modes tends to make certain functions easier to carry out so it’s a useful ability to utilise regularly. Note also that scaling in this context can make use of the “Ctrl” key to ‘snap’ manipulations to fixed increments.

Using Extrude & Fill

Continuing on with the scarf, next the ‘drops’ need to be made (the bits that hang down when a scarf is wrapped around something, a ‘neck’ in this instance). A quick way this can be done is through the use of “Extrude“, pulling the necessary material from the main body of the mesh. To make this process easier, “Ctrl+Tab” and switch “Select Mode” to “Face” then “Shift+RMB” select a group of faces two columns wide by three rows down starting second row from the top of the lower sphere – ignore the narrow loop of faces running around the top of the lower sphere (directly under the scarf area just shaped).

Design note: the coloured grid lines, ‘red‘ and ‘green‘, are axis indicators – the “X” axis is ‘red‘ and typically represents ‘left/right‘, “Y” is ‘green‘ representing ‘front/back‘. To aid the selection and general editing process use MMB in combination with “numPad5” to toggle “Ortho/Persp” and press “numPad1” or “numPad3” respectively to reset the view to “Front” or “Right” orientation as needed.

With the selection active, make sure the cursor is over the 3DView and press “E” to “Extrude“. Immediately upon doing this a new set of faces will appear simultaneously responding to mouse movement and traveling along a pale-blue axis marker. Move the collection a short distance from the mesh and LMB click to confirm.

Design note: the extrusion should be moved away from the mesh far enough that everything is easily selectable, including the sides. The pale-blue axis marker represents a directional orientation shared between a selected group of faces, edges or vertices. For a single face or other element, extrusion usually travels perpendicular to the selection (at right angles to it). For a group, it’s averaged and not always orientated to a regulated angular value (i.e. 30°, 45°, 90°). Extrude can also be accessed from the “Mesh” Header menu in the 3DView, “Mesh » Extrude Region” – note there are two types of extrude, the one discussed above and another which processes faces on an individual basis irrespective of their being part of a group selection or not (unused in this tutorial).

Before moving on to shape the scarf section further some material needs to be removed and then rebuilt so its more conducive to being manipulated without causing issues. The area in question belongs to the top of the section just created – due to the way the underside of the scarf section is inset, those faces can be awkward to try and extrude into shape directly. Instead it’s typical for faces in problematic areas like this to be manually created through deletion and recreation, or deletion and merging. In this instance the former has been chosen to highlight a more typical approach to fixing this type of issue.

Using “Shift+RMB” make a group selection consisting the upper two faces of the extrusion just made, two in the loop directly above that, and finally two more on the underside of the neck section. With the group active, press Del” or “X” to access the “Delete” menu and select “Faces” to remove the selection leaving a hole in the mesh which can now be filled.

Switch select mode, “Ctrl+Tab“, to “Edge” and employing the same technique used previously, “Shift+RMB” select the ‘top’ and ‘bottom’ edge of the gap, then press “F” to generate the new face. Deselect, “A“, then reselect another pair of edges, press “F” again and repeat until the gap has been ‘capped’ ready for the next step.

Design note: be sure to deselect all, “A“, when generating new faces, leaving ‘old’ selections in place typically results in nothing happening or the creation of a distortion as Blender tries to twist and shape the surface to fit. It’s also useful to switch “Mesh Select Mode” depending on what needs to be done – it’s easier to create new faces in “Edge” mode for example. Often, and depending on the shape of the area that needs to be filled, it’s not practical or possible to ‘group fill’ an area. Typically this means having to manually close gaps as described through the selection of edge pairs.

Group select tools & vertex manipulation

At this point it’s time to start ‘playing’ with the objects general appearance, that is, to make the shapes less regular and more organic. This is done laregly by manipulating, but not necessarily deleting or creating, various aspects of the mesh in Edit mode. Before doing so, and to ease the process, “Ctrl+Tab” switch to “Face” select mode; from “Persp” to “Ortho” view, “numPad5“; and change “Viewport Shading” to “Wireframe“, “Z” – this allows for much easier selections to be made by toggling between “Front“, “numPad1” or “Right“, “numPad3“, views absent of any perspective distortion.

Design note: an alternative to using the shortcut keys to switch views is to use the “View” header menu; “View » View Perps/Ortho“, “View » Front” etc.

First the overall height is changed. Press “B” to activate “Border Select” then LMB+drag a selection box around the upper sphere taking care to include the upper loop of the scarf. Once done press “G“, or LMB+drag the ‘blue’ handle of the “Transform” widget, and lower the selection down slightly to make the mesh a little more squat, be mindful not to collapse the mesh in on itself. LMB click to confirm alteration.

Design note: depending on the mechanism used to manipulate the mesh, once the action has been confirmed (LMB when using “G”, LMB release when dragging the widget), to “Undo” and retry, use “Ctrl+Z“.

Next, individual vertices need to be moved so the overall appearance of the mesh is more organic. The least problematic way to do this initially is to use the Orthogonal views. There are two main reasons for this; 1) selections ‘drill-down’ through the mesh relative to the mouse, i.e. ‘front’ and ‘back’ can be selected and manipulated at the same time; and 2) shaping and manipulation is easier when a limited number of axes are used, i.e. when the mesh is deformed left/right, top/bottom only, it’s easier to see what’s happening to the mesh as its shape is changed.

To do this first switch select mode, “Ctrl+tab“, to “Vertex” and make sure to be in “Wireframe” viewport shading, “Z“. Once done, press “C” to activate “Circle Select” and using either LMB or LMB+drag, ‘paint select’ individual or groups of vertices, press “G”, or use the manipulator widget handles, to move the vertices around so a slightly irregular flow of edge loops is formed (disrupting the original and very linear distribution) – switch between “Front“, “numbPad1” and “Right“, “numPad3” views making adjustments and changes in response to the developing shape.

Design note: both ‘Circle’ and ‘Border’ select are ‘multi-select operators’, i.e. “Shift” is not needed when making ‘group’ selections. This also affects de-selection. Whilst the tool is active MMB click a vertex or collection (area) to remove. In addition, using RMB will cancel the tool (whilst leaving any previously selected elements actively highlighted).

Although most of the shaping process is generally done using the Orthogonal views, and done because it’s easier to see what’s going on as noted above, it’s also important to check the mesh for unwanted distortions in ‘Perspective’ view occasionally, “numPad5” – whilst Ortho is good for the initial phase of manipulation, it can be difficult to ‘see’ how such changes directly effect the overall shape as that relates to a 3D object which then necessitates switching back and forth between the two modes making adjustments as required.

Design note: the trick to making what might appear to be relatively abstract changes to a mesh, is to observe the way each manipulation affects the mesh and how those changes can then be used to inform subsequent actions. This may seem a somewhat nebulous concept but it is in fact the key to this type of organic modelling. It is very much one of constant feedback and revealing the character of the shape as opposed to necessarily constructing it (being a ‘sculptor’ rather than ‘engineer’).

Subdividing Edges

Whilst much of scarf’s general structure has been shaped it now needs a knot. Normally this would be done through extruding the front two faces that form the cross-over between the scarf drops and the neck wrap-around. Doing that here however results in a ‘lump’ of polygons that can be difficult to re-shape correctly relative to neighbours surfaces. Instead, using “Subdivide” the area can be split to help better position vertices to achieve the shape needed, one that gives the appearance of making sense visually relative to a simple scarf knot, effectively just a drape-over (the scarf tucked under then pulled up and out to drape over the top).

Using “Ctrl+Tab” switch to “Edge” mode and “Shift+RMB” select the top and bottom row of edges boardering the aforementioned faces which form the ‘cross-over’ between the neck wrap and drops (see below). In the “Add:” subsection of the ToolShelf on the left, click the “Subdivide” button (or use “Mesh » Edges » Subdivide” from the Header menu) to split the selection. The elements will divide in two along their respective centres forming an additional set of edges down each face.

Next use “Ctrl+Tab” again and switch to “Vertex” selection mode. Each new edge has a vertex at either end (accessible once in “Vertex” select mode, hence the requirement to switch), RMB click each in turn and position them to one side slightly to create a raised area with a distinct ‘left’ and ‘right’ side – the resulting shape should have a ‘box’ like appearance (shown below).

Once done the next step is to remove and rebuild a section of the mesh. Using “Ctrl+Tab” switch to “Face” select mode and then “Shift+RMB” select the top of the area just made plus the two adjoining faces directly above (where the characters ‘mouth’ would be if it had one). Press “Delete“, then click “Faces” to remove the highlighted elements only. This will create a hole which now needs to be filled.

Design note: when needing to remove specific surfaces whilst in either “Edge” or “Vertex” select mode, be sure to choose “Faces” from the “Delete” options list otherwise the wrong elements will be removed. If this happens, use “Ctrl+Z” to “Undo” accidental deletions.

To fill the gap, in “Edge” select mode, “Shift+RMB” select the top and bottom edge of the exposed hole and press “F” to “Make Edge/Face” (or use “Mesh » Faces » Make Edge/Face” from the header menu). The gap between the two edges will be closed. Deselect, “A“, and repeat this process for each upper and lower edge pairing to close the hole completely – this should create four new faces, two on top and two either side.

With the hole closed switch to “Vertex” select mode and as before, continue shaping the mesh by selecting individual or groups of vertices, positioning them to offset the still regulated structure of the mesh, i.e. keep the overall spherical shape of the object but make it appear slightly lumpy and misshapen (‘disturb’ the equidistant distribution of vertexes).

Design note: ‘shaping’ meshes like this is not specifically a ‘mathematical’ or ‘measured’ procedure because much of the process relies heavily on the artists ability to constantly assess the mesh as changes are made. This ‘visual feedback loop‘ (where the artists subsequent decisions are determined by the actions carried out) is often ignored or underrated as part of the process, it is in fact of primary importance, more so than necessarily knowing ‘how’ to do something.

Smoothing; Mark Sharp and Edge Split

Once the model gets to a point where the initial structure is built, it’s time to look at changing the way it’s displayed in the 3DView by activating “Smooth Shading“. Doing this serves two main purposes, 1) it helps define the overall shape, structure and general appearance of the mesh. And 2) it approximates the way light ‘shades’ a mesh, the ‘light and dark’ highlights that occur due to the influence of illumination.

Design note: “Smooth Shading” (“Mesh Smoothing” or “Smoothing” as its sometimes referred) is largely about the behaviour of light across surfaces. With the property applied objects typically appear uniformly lit because ‘shading’, the light and dark areas of a mesh, is evenly distributed (‘averaged’) across a collection of related (neighbouring) surfaces. Shading is related but separate to ‘shadows’ and other forms of depth, shape and form highlighting – it’s not as distinctly ‘directional’ as shadows, nor conversely, as indistinct as ambient illumination.

However, because ‘Smoothing’ is applied uniformly by default, it has a tendency to make meshes appear like amorphous masses. To solve this problem, edges can be marked with a particular property so they ‘break’ the surface through the formation of series of hard lines, a collection of which, placed strategically around the mesh, can be used to specifically delineate the object further – for example forming a distinct ‘head’ and ‘body’ area without necessarily physically splitting the mesh into separate sections (objects). The process of marking a mesh in this way is typically referred to as adding or marking “Smooth Groups“, and done through the use of “Mark Sharp” and the “Edge Split” modifier.

Design note: in actuality “Smooth Groups” don’t exist in Blender (because it breaks surfaces based on hard/soft edges rather than defined groups of face elements), however, and for all intents and purposes, the net result of defining hard edges is the impression this is the case. Although all four terms tend to be used interchangeably, “Smoothing Shading“, “Mesh Smoothing“, “Smoothing” and “Smooth Groups“, the former and latter in particular should not be regarded as being the same thing.

To do this, remain in “Edit” mode (“Tab“) and using “Ctrl+Tab” switch to either “Edge” or “Vertex” selection mode. Using a combination of “Alt+RMB“, “Alt+RMB+Shift” and “Shift“, select the loops and edges which border the scarf area of the mesh. Once done, press “Ctrl+E” or use “Mesh » Edges » Mark Sharp” from the Header menu to access the “Edges” menu and select “Mark Sharp” from the list; this will mark the selection, indicated by a pale-blue overlay/outline.

Design note: when including or removing face, edge or vertex loops to or from previous selections, using “Shift” – so it becomes “Alt+RMB+Shift” – enables the appropriate level of multi-selecting. To include individual edges (or other sub-element) after making an initial loop selection, release the “Alt” key and just use “Shift+RMB“, this allows for specific elements to be included or removed without loosing the original selection. In this way it’s possible to use the different combinations – “Alt+RMB“, “Alt+RMB+Shift” and “Shift+RMB” – to add/remove various loops and/or individual elements, especially useful for awkward shapes. (note: the shortcut keys are not necessarily sequence specific, pressing “Alt+Shift+RMB” should work just as readily as “Shift+Alt+RMB” – what matters are that the correct keys are used).

To assign ‘Smoothing’ do one of the following, either 1) press “Tab” to exit Edit mode and RMB select the Object. Or, 2) whilst in Edit mode use “A” to select the entire mesh (every face/edge/vertex should be selected). Once done, in the ToolShelf on the left-hand side (it should be open by default, if not press “T“), find the “Shading:” sub-section (scroll down the tool list) and click the “Smooth” button, in the 3DView the mesh should change as the property is assigned (updating from its initial facetted appearance to looking like the lump of clay previously mentioned).

Design note: smoothing can be assigned in either “Edit” or “Object” mode. In Object mode it’s assigned uniformly across the mesh (by default). In Edit mode assignment is selection based so the entire mesh needs to be selected, hence “A” to “Select All” before the property can be assigned ‘globally’ (relatively) in that mode. Note also that if a limited number of faces are selected when Smoothing is active, it may give the impression that smooth ‘groups’ are being formed when this is not the case, any subsequent selections are added to the original assignment when-ever Smoothing is reapplied.

To display the previously marked hard edges (so as to give the appearance of smooth ‘groups’), in either ‘Object’ or ‘Edit’ mode, first click the “Object Modifiers” button (the ‘wrench‘ or ‘spanner‘ icon) in the “Properties” section on the right then click “Add Modifier” to access the list of available actions. From the second-column from the left, “Generate“, select “Edge Split” to assign the property.

Design note: the “Edge Split” modifier is assigned ‘globally’, it’s not applied based on selection. This means when in Edit mode the modifier will apply over the entire mesh regardless of any active selections.

On doing this the appearance of the mesh changes through the addition a series of hard edges which appear to break the surface in the series of delineated, and seemingly random, areas. These are the aforementioned mentioned ‘smooth groups‘ which are located and controlled based on an “Edge Angle” value or a series of marked “Sharp Edges“. For the former, “Edge Angle“, surface splits are formed based on an angle-of-incidence, in degrees ( “30°” being the default), between adjoining surfaces. For the latter, “Sharp Edges“, surface splits are based on edges being marked as “Sharp“.

Design note: using “Sharp Edges” tends to provide more control over smoothing where organic shapes are concerned because the nature of that type of material tends to mean a mesh splits in odd and apparently random locations.

Because the character was previously marked-up using the “Mark Sharp” property, only the latter modifier option, “Sharp Edges“, is needed, leaving “Edge Angle” to be deactivated and turned off by clicking the checkbox to the left of the listing. The mesh will update again in the 3DView showing fewer hard edges as a result (and confirming the change).

Design note: the process of making a model won’t always be quite so linear as described in this tutorial, meaning the addition of ‘Smoothing’ can be done at any time. The main point or principle to activating it now is to aid the production process so particular shapes, forms and behaviours can be seen as the mesh develops, rather than suggesting the property needs to be in place during the early stages of mesh development.

Additional mesh editing; adding a hat

At this point the character should be developed enough to build a ‘hat’. There are generally two ways to go about doing this; 1) add and shape a brand new mesh object; or 2) shape the necessary material from the original mesh structure. There are general advantages and shortcoming to using one or the other, however for characters and complex shapes it’s often better to use the latter, shape material from existing, because it ensures (amongst other things) the mesh is the type of ‘contiguous’ object real-time environments tend to prefer.

Design note: having said the above however, the complexity of the required forms will determine to a large extent which approach is used. There isn’t really a ‘right’ or ‘wrong’ way to do this, only what best suits the general purposes of production.

Currently then, the ‘head’ section should consist of approximately four main rows of faces around the horizontal. As the hat will sit so it occupies approximately 1/3rd of this area, only the upper two loops of faces will be needed. The hat will be added above the middle row. This means placing a loop-cut marginally above the upper edge-loop of that central row of faces. First define the hats ‘brim’.

Press “Tab” and enter Edit mode (if not already in that mode), then “Ctrl+Tab” switch to “Edge” selection. Using “Ctrl+R“, in the row second-from-the-top (row directly above the middle), add a new loop and position it approximately 1/4 the way up from the bottom edge. Once placed, press “S” and “Scale” the still selected loop outwards a short distance. This forms the underside of the hats ‘brim’ and defines its ‘breadth’.

Design note: press “Ctrl+R” to active loop-cut, move the mouse to the second row from the top and LMB click to select. When the pink guideline changes to a yellow ‘cut’ line, move it towards to bottom edge and LMB click again to conform the new position. When scaling, make sure the mouse is a short distance from the object else movement may be overly sensitive.

To give the hat brim depth use “Ctrl+R” again, adding another loop-cut above the first. Scale this second loop outwards so it’s the same diameter and in combination with the previous cut, the two together give thickness to the brim – if this depth needs to be adjusted (because initial placement resulted in a brim that was too thick or thin), use “G” to move the selected loop (or “Alt+RMB” to select the loop again) and adjust its position ‘up’ or ‘down’, making the brim thicker or thinner as necessary.

Design note: a good sense of proportion is required most of the time when making 3D objects, in particular when resolving situations such as the above – typically there are no hard-and-fast rules about “placing ‘X’ 1.3 units above ‘Y'” when making shapes that are predominately organic, typically this means relying again on the key dynamic of personal observation. It cannot be stressed enough that this ‘dynamic’ flux of visual information is important for 3D design.

The next cut needed defines the breadth of the rim from the outer edge to the hats body. When adding this new loop, after the pink guideline changes to the yellow cut line, slide it back-and-forth along the edge-loop, LMB clicking when an appropriate size is reached to confirm placement – this new addition should take into account the curvature of the head so as not to appear to be sat atop a disproportionate head; greater latitude is afforded to making the hat larger, rather than smaller, than necessary.

Design note: if the cut is confirmed and set in place before being appropriately positioned, “Alt+RMB” select the loop then use “Ctrl+E” to access the “Edges” menu. Here select “Edge Slide” from the available options and ‘slide’ the loop back-and-forth (by moving the mouse in/out or side-to-side). LMB click to confirm and set a position.

Finally “Alt+RMB” select the edges around the top of the ‘hat’, press “S” and “Scale” outwards so the selection is the same circumference as the previous cut – the intention initially is to make the hats ‘body’ the same thickness top to bottom so it has a uniform appearance. LMB to confirm. Once done “Shift+RMB” select the vertex centre of the hat to include it in the selection just made and either LMB+drag the blue widget handle, or press “G” (then “Z” to constrain the axis) and move the mouse up, pulling the selection after it, creating a ‘top-hat’ style structure. LMB release or click to confirm the action depending on the way it was carried out (using the widget, release LMB; using ‘Grab’, LMB click).

Depending on preference, whether a Top or slightly floppy hat is preferred, for the latter the addition of one last loop-cut around the body of the hat finalises the basic structure – depending on how ‘soft’ or ’round’ the hat needs to be, the extra loop should be scaled outwards accordingly. So, using “Ctrl+R” add a new loop and position it closer to the pre-existing upper loop. LMB click to confirm. Once placed, whilst the loop is still selected, use “S” to scale outwards and again LMB clicking to confirm.

Design note: the loop can be repositioned after creation using “Edge Slide” again – with the loop selected, use “Ctrl+E” to access the “Edges” menu, select “Edge Slide” from the list as before.

Once the hat is done progress should then be checked against ‘Smoothing’ – the new area should be, and have, its own distinct ‘groupings’, separate from the rest of the mesh. Whilst still in Edit mode, “Alt+RMB” select the edge-loop where the body of the hat meets the brim, then press “Ctrl+E” to access the “Edges” menu. From the list select “Mark Sharp” (or from the “Mesh” Header menu select “Edges » Mark Sharp“), on doing this, because the “Edge Split” modifier was previously assigned to the Object, the newly marked edges will display immediately in the 3DView.

Design note: depending on the type of modifier applied to an object, it’s effect on the mesh in the 3DView can be instantaneous, as with marking ‘sharp’ edges. This way certain operations can be carried out in real-time.

Editing the mesh to ‘Ground’ the character

The final stage of modelling the main body is to do something called ‘grounding’, this simply refers to the way an object ‘sits’ atop another object or surface. Because real Snowman are generally made from rough piles of snow, some of which falls to the ground, gets trampled under-foot, or otherwise gathers, a concave gradient might form around the base which acts to anchors the character to the floor. Ideally this structural attribute should be present so the overall mesh appears correctly positioned relative to ‘grounding’ expectations in game. In other words objects typically should not be floating above the ground.

Design note: grounding simply means thinking about objects as if they had weight and/or mass; it’s the bulge at the bottom of car tyres, the depression under a heavy load, and the reason characters and game objects appear to ‘float’ when this isn’t taken into account. Depending on the complexity of the object (and whether it’s entirely practical) these types of nuances should be considered at the design stage rather than put in place as after-thoughts, as can sometimes happen.

To create the gradient mentioned above, in Edit mode “Alt+RMB” select the second from bottom edge-loop and using “S“, “Scale” outwards from the body a short distance. LMB click to confirm the action and de-select, “A“. Next, using “Alt+RMB” again, select the very bottom edge-loop and scale that outwards so, in combination with the previous manipulation, the two loops together form a concave incline (as viewed from the side). LMB click again to confirm, completing the action and formation of the gradient.

Design note: it helps to better gauge the shape of the slope by using “Front” or “Right” orthogonal views, so use “numPad5“, “numPad1” or “numPad3” to switch views, as well as using “Ctrl+Tab” to change “Mesh Select Mode” between “Face“, “Edge” and “Vertex“, depending on what action needs to be carried out.

Individual vertex position

After shaping the base a single vertex remains unaffected underside-centre of the mesh, a left-over from the original spheres. Because a certain degree of accuracy is needed for this type of manipulation, adjustments are best done using “Location:” coordinate data rather than widget manipulator or mouse.

First press “N” to make sure “View/Transform” Properties is available. Next, switch to “Vertex” select mode if not already in it, “Ctrl+Tab“, and from around the bottom edge of the mesh previously shaped, RMB select a single vertex. In the “Vertex:” sub-section of the “Transform” properties toolshelf, note the elements “Z:” coordinate value – in this instance “-0.924” – this is a ‘baseline’ for the bottom of the mesh. RMB select the centre vertex and from the “Vertex:” subsection of “Transform” properties type “-0.924” in the “Z:” input field, changing it from “-1.000“. The vertex should jump upwards to its new position in line with the other elements of the base area.

Design note: the values used in the above are applicable to the example file that accompanies this tutorial. Actual values will vary if the Snowman character is being built as the tutorial is being read based on personal interpretation.

Once the vertex is re-positioned, “Alt+RMB” select the outer edge-loop and use “Ctrl+E” to add one final “Mark Sharp” to the selected loop to correct smoothing around the base of the character. With that done, the main mesh is complete. Next additional detailing needs to be placed, the eyes and nose.

Design note: marking the edge ‘sharp’ forces a correction in the way shading is averaged as it ‘wraps’ over the edge of the base – the underside should be treated as a separate group in other words. Doing this also then means the interaction between the model and the ground upon which it’s standing is ‘normalised’ for that specific relationship, in other words, shading needs to be a “base<->ground” relationship, not “base/underside<->ground“.

Object Snap, adding new meshes

With the main torso built the characters eyes and nose can now be added. However, the low-poly nature of the mesh makes this slightly problematic because there aren’t enough polygons available from which the new elements can be extrude/made, or, if there are, they might need manipulating in a way that’s detrimental to the objects general appearance. So rather than shaping or forming the original mesh, instead, a new set of separate objects can be used, which will later be merged together before UVW unwrapping and texturing.

Design note: generally speaking reasons for deciding on whether to use extrusion versus placed object depends on ‘optimisation’ – extrusion typically requires more polygons but is easier to animate, whereas additional objects tend use fewer polygons but can be tricky to animate.

Because Blender loads objects into a scene using the 3D Cursor as a point of origin, it can be positioned such that any new objects appear approximately where they are needed. To do this first toggle “Orthogonal” view, press “numPad5” and then “numPad1” or “numPad3” depending on the ‘front‘ orientation of the mesh. Then enter Edit mode (“Tab“) and in “Face” select mode (“Ctrl+Tab“), RMB select the upper front face of the characters head and then press “Shift+S” to access the “Snap” menu, selecting “Cursor to Selected” from the list of options (or “Mesh » Snap » Cursor to Selected” from the 3DView Header menu). The 3D Cursor will instantly jump to centre on the face. Exit Edit mode (“Tab“).

Design note: when snapping the cursor in this way, it will typically centre itself at the selected elements ‘centre-of-mass‘; for an individual ‘face’ this is the ‘face normal‘; for an edge it’s the ‘mid-point‘ between either end, for a vertex it will be the vertex itself; and for two or more selected elements, an averaged mid-point between them. There is a distinct difference in behaviour between adding objects in Object mode versus Edit mode – in the former the item remains a separate and independently editable element of the scene. In the latter the item is ‘merged’ with the mesh it’s been added to making it editable as part of the overall mesh – for ease of editing adding in Object mode is preferable.

Once the cursor is positioned, using “Shift+A” to access the “Add” menu, select “Mesh » Sphere“. A new sphere object will appear in the scene centred of the cursor. In the “Add UV Sphere” section of the ToolShelf on the left (which automatically appears on adding a new sphere object), amend “Segments” and “Rings” to “7” and “4” respectively (“Segments: 7“, “Rings: 4“).

Design note: when reducing the density of the sphere to be used for the characters eyes, bear in mind that it’s not necessary for the mesh to remain particularly detailed or round in appearance; in particular the objects being simulated in this instance are usually rather rough misshapen objects (being lumps of coal) so the reduced poly count can be advantageous. In addition it’s worth noting that using ‘odd’ numbers (“5”, “7”, “9” etc.) can ‘hide’ the sometimes instantly recognisable ‘regularity’ of ‘even’ sub-divisions (“6”, “8”, “10” etc.).

After the initial reduction press “R” and “Rotate” the object onto its side (so the ‘peaks’ align left-to-right instead of top-bottom) and enter Edit mode (“Tab“). Press “Z” to toggle “Wireframe” and then using “B“, “Border Select“, select and then delete, “Del“, one half of the mesh.

Exit Edit mode then using “S“, scale the remaining half relative to the size of the head, or more appropriately, the face area. Rotate (“R“) and position (“G“) the mesh so it faces front (the ‘peak’ should point forwards from the mesh). Once done, use “Shift+D” and create a duplicate, move this (whilst active after duplication) to the other side of the face and LMB click to set it in place. The result should be two half-spheres, one on the left, one on the right. Then optionally, adjust the eyes (scale, rotate, manipulate – “S”, “R”, “G”) as needed so that perhaps one is smaller than the other.

Design note: as an alternative to the above, rather than deleting one half of the mesh and duplicating, press “P” to “Separate” the selection that would otherwise have been deleted, into two completely separate objects. Rotate and re-position each half of the original sphere as needed – two eyes from one object.

The character now has a pair of eyes, however, because they were placed largely in Orthographic view they don’t currently follow the curvature of the mesh. This can be clearly seen toggling back into Perspective mode, “numPad5“, where the problem is immediately apparent – the eyes are not properly positioned in relation to the mesh, they sit a good deal proud of the surface. To fix this problem, and taking advantage of the depth offered by being in perspective mode, each ‘eye’ simply needs to be repositioned.

Remaining in ‘perspective’ mode then, RMB select each ‘eye’ in-turn and using a combination of “R“, “S” and “G” to “Rotate“, “Scale” and “Move” the object, re-position them so they follow the general curvature of the head, and are buried a short distance into it. Once done, in the “Shading:” sub-section of the ToolShelf on the left, LMB click “Smooth” to remove the faceting.

Design note: each eye object has to be buried a short distance inwards to ensure there are no gaps between the flat underside of the mesh and the curvature of the head – this is one of the reasons why such objects are typically rather large, much of their structure may end up hidden from view whilst still leaving enough seen to serve the purpose for which they was placed, ‘eyes’ in this instance.

Finalising the mesh

The final object to add is the nose, a carrot in this instance. The easiest way to do this is to duplicate and edit one of the eye meshes. So RMB select either/or eye object and “Shift+D” duplicate it. Whilst still active (responding to the mouse after generation), position it above the scarf, between the eyes. LMB to confirm and then enter Edit mode, “Tab“. From here, and depending on which “Select Mode” is active, select the ‘pole’ vertex and edge-loop below it, or the faces around the inner ‘cap’ (the triangular faces pointing towards the centre of the mesh), and using “G“, pull the group a short distance away from the face to form the elongated shape of a ‘carrot’. LMB click to confirm. To change the width around the base (circumference), in Edit mode select, “Alt+RMB“, and then “Scale, “S“, to the desired size the bottom edge loop closest to the head. Or in Object mode, “Tab“, use “S” then “X“, “S” then “Y” to re-size the mesh along those specific axes. LMB click to confirm each change. The end result should be thinner and longer with a base circumference that’s smaller than the eye object the nose was duplicated from.

Design note: using “Shift+D” when duplicating these types of objects, instead of “Alt+D“, ensures the creation of meshes that can be edited independently of the original. Note also that when using sequential keyboard combinations to perform an operation, LMB clicking to confirm is usually required before being able to perform and action.

Once the nose is done that more-or-less completes the basic construction of the character. Before continuing however, make sure mesh smoothing is set up appropriately, so in Edit mode use “RMB“, “Shift+RMB” or “ALT+RMB” to select edges, press “Ctrl+E” to access the “Edges” menu and select either “Clear Sharp” or “Mark Sharp” to change assignments. Then give the mesh a final once over to make sure nothing has been missed and/or tweak the appearance slightly where needed (reshape/position vertices in Edit mode if necessary).

Design note: the meshes are not to be joined together yet (connected together to form a single, greater mesh object), this will be done during the next phase.