Making of 'Billy Bob Boone'

Introduction

Welcome to the making of "Billy Bob Boone".  This tutorial walks you through a brief explanation of the steps involved in creating this character, from start to finish.  It is recommended that you understand the fundamentals of computer graphics and animation, as this is meant for intermediate to advanced users.

"Billy Bob Boone" started off as an idea for the "Strange Behavior" competition, held in late 2007.  The idea was to create an image in which oddity is captured in any way, shape or form.  Since I specifically enjoy character development, my idea was to develop oddity through expression - to create a very odd-looking character with a very unusual expression.

Pre-production

As with any character, the very first incarnation is always created on paper.  Fig.01 shows a rough sketch of the initial idea, and it's interesting to see how it will evolve throughout the development process.

It is important to notice that the sketch is very loose, as are all my initial designs.  I find that gestural drawings tend to capture much more emotion and expression than a finely-tuned one.

Fig. 01

Production

As soon as I'm pleased with a concept, I immediately start working in ZBrush.  To me, ZBrush has proven to be the most artist-friendly modelling package I've ever used.  Since I am only an intermediate ZBrush user, I decided to improve my modelling skills by developing this character directly from ZSpheres - completely from within ZBrush, as opposed to importing a base mesh.

ZSpheres provide a very powerful and versatile method of modelling for artists using ZBrush.  Generally speaking, spheres are placed in any arrangement and connected together to form a rough skeleton.  An adaptive mesh can be displayed at any time over the spheres, thus allowing you to create base meshes very quickly and easily.  If you don't like what you see, you can simply re-arrange any of the spheres and see how this affects your adaptive mesh.  ZSpheres offer many more powerful features than just that, but since I'm using it strictly for modelling, I will limit the discussion to this.

Once I have arranged the ZSpheres to form a base mesh that I'm pleased with, I collapse the model in order to start working with it.  Fig.02 shows the very first mesh generated by the ZSpheres. Looking at this, you'd wonder how on earth any respectable model could be developed from it.  The poly-count is extremely low but this is important for us so that we have a lot of play with the forthcoming subdivisions.

Fig. 02

The unique feature of ZBrush is that it allows you to be able to use a low polygon mesh and continue subdividing it until you have enough polygons to fully-sculpt with it.  Once you're finished, you convert the high density mesh to a displacement map and achieve the same results but with a significantly lower-density mesh. What this means, for me, is that I will start with the lowest density mesh I can, sculpt it until I get a shape that I like, and finally subdivide it again.  This will continue until I reach a level in which I can sculpt all the details I would like to add to the character (or until your computer decides to explode!).

When the base mesh has been sculpted into a shape I am pleased with (Fig.02), I subdivide and continue my modelling process.


The next image shows the character with three levels of subdivisions (Fig.03).

Fig. 03

It's still a terrible mesh but that's not important right now.  What is important is to be able to continue shaping the model into a form that is similar to my initial sketch.  This is beginning to happen - the chin and mouth are formed and twisted, the nose is more distinct, the shape of the face has been refined and the cap has also been roughly created.  In this case, the cap is a "sub-tool", which means it's a separate mesh that can be worked on within the same scene using its own subdivision levels.  This is important because we will not require as many polygons for the cap as we will for the face, and sub-tools allow you that flexibility.  In addition, future objects will also be created as sub-tools, such as the eyes.

Once I have sculpted this level to a form I'm pleased with, I subdivide again and continue my modelling process. Fig.04 shows the character with five levels of subdivisions.

Fig. 04

As you can see, the character now has a significant amount of detail and is beginning to take a more distinct form.  Smaller detail areas, such as the nostrils, mouth, eyelids and ears, now have more definition because I have more polygons to sculpt with.

Generally, it's advisable to leave very fine details for last (such as skin pores, cuts, irregularities, etc.) since they require very high levels of subdivisions.  However, since our cap is smaller and therefore has a much higher polygon density than the face, we can start creating these details here.  The stitching on the sides of the cap are now roughed in, as are the rips and tears on the material.

The most obvious problem with this model right now is the severe stretching of the right eye polygons.  While this may look terrible right now, I know that I will be sculpting this area in more detail later on, and the eyeball will also be covering that up.  Therefore, I am not concerned with it at this time.

As I continue sculpting, my character continues to be subdivided until I reach the eighth subdivision level, which proves to be enough detail for me.  The image below shows the character with eight levels of subdivisions (Fig.05).

Fig. 05

 At this point, all my sculpting has been completed.  If you look closely, you can see skin creases and pores have been painted in and all the areas that were lacking detail now have it.  The previous problematic right eye looks fine at this level of detail with the eyeball sub-tools properly dropped in.  The cap has also had further sculpting details added, including cloth texture and more detail in the tear.

You can see that the lower neck has the least amount of detail at this point.  This is because I intend to cover up that area with clothing so I will use my time effectively and only work on the areas that will actually be visible.

Once I'm completely satisfied with the model, it's time to texture it.  Although I used to formerly do all my texturing within Photoshop using unwrapped UV's, I've since changed my workflow to painting directly on the model, again within ZBrush.  The difference in quality by working in this manner is enormous.  You see exactly what your character will look like as you paint, instead of guessing where details should go and distorting them to make up for unwrapped polygons.

I start texturing "Billy Bob Boone" by applying large patches of colour and, just like subdividing the model for more detail, I add smaller and smaller areas of colour once the larger ones are laid in.  Again, I have to use my imagination to some degree and try to imagine what this will eventually look like once it has been properly lit.  It may not look as striking as I would like it to be in ZBrush, but eventually it will in the final render. Fig.06 shows the character with finished ZBrush textures.

Fig. 06

Not very exciting, is it.  That's okay.  We're dealing only with diffuse colour at this point and we're not seeing highlights, shadows, reflections, diffuse glows, ambient occlusion, GI, etc.  Even though I will be texturing some objects outside of ZBrush, such as the eyes, it still helps me to colour-code them white in ZBrush for a more accurate representation of my character.

Now the model gets exported from ZBrush and imported into 3ds Max where I will be using V-Ray for lighting and rendering.  In order to export the model, I derive a diffuse and displacement map of the highest detail level and save that out.  Then I decrease the subdivisions several steps and export a reasonably-dense model so that my 3D application can handle it.  The textures generally get exported at 4K since I will be rendering very high resolution images.  Once the model has been imported in 3ds Max, the diffuse and displacement map is re-applied, proper materials are assigned, and a good lighting setup is created.  Additional minor features are also modelled at this time, such as the pupils and teeth, for instance. Fig.07 shows the character rendered with V-Ray, using the same texture as above but with proper materials, lighting and shadows.

Fig. 07

Fig. 08

As you can see, the difference is enormous and this is still the exact same texture as before.  This is why it's important to know how to light properly.

Fig.07 is what I tend to call a 'raw' render.  That means that no colour correction or processing of any kind has been applied to the image - it's simply a direct output from V-Ray.  As a result, no highlights are present but that's okay because these will all be rendered out in separate layers to provide as much flexibility as possible during the composite.

Next, the various additional layers are rendered.  This means setting up numerous 3D files in order to output the required assets.  These assets will consist of highlights, shadows, reflections, diffuse glows, ambient occlusion, GI, and much more. Fig.08 is an example of the highlight layer.Most of these layers are rendered in greyscale since I will be colouring them in the compositing software and layering them on using various compositing modes.

Post-Production

As with all of my work, 50% of the image is generated in the 3D software and the other 50% in the composite.  This is the 'post-production' part of the work as you have already produced your main assets and it's the colour correction and image treatment that are important now.  In order to work on the composite I will use Photoshop as my primary tool.  If the render was a sequence instead of a still frame, I would most likely use After Effects.

About as much time is spent working in Photoshop as there was with ZBrush.  This is because of all the missing elements that still need to be added, such as the background, clothing, decals, touch-ups, etc.  Time-consuming details, such as hair, are also added at this time, and many of these elements are found in various public images scattered throughout the Internet.  As a result, a great deal of researching is also required in order to gather all the necessary images you will be cutting elements from.  Finding an image of a close-up of an old man with a good eyebrow to cut from may take over an hour - and this is just for one element which then needs to be properly colour-corrected and distorted to match my current palette!  Working with photographs in this manner is called working with a "photo-composite".

Image distortion is used extensively for my compositing work.  It is extremely rare, for instance, that one will find a photo of a chin with whiskers that will simply 'fit' perfectly on my character's heavily-exaggerated chin.  Therefore, these elements must be distorted to fit onto my character and this is generally done with Photoshop's "Liquify" tool.  The overalls, for instance, were pieced together from numerous photographic elements of clothing and then distorted to fit "Billy Bob's" thin neck.

Proper colour correction also takes a considerable amount of time; finding the correct balance of red, green and blue, deciding on how much saturation and contrast to add and figuring out what kind of colour grading I intend to have at the end.  The colour grading is the very last step involved in the production, but it is a crucial one.  This will tie all your elements together in a unified palette.  In this case, my colour grading is a subtle orange, in order to enforce the sunny afternoon setting.

At this point we have our finished image!

Final Image

The steps outlined above provided a general overview of the process involved.  Needless to say, there is much more work that goes into developing a character like this but cannot be fully covered in such a brief "making of".

I hope you've enjoyed this article.