Making Of 'Marcus'
Hello my name is Andrea Bruschi and I am an Italian self-taught artist. I am currently looking for a job so I appeal to the public to take advantage of this Making Of!
I don't pretend to teach anything with this article. It simply contains some of my thoughts about how to build a character and looks at ideas and comments that arose during the creation process. Because almost everyone knows about workflows, techniques etc., I will try to break the mould by specifically focusing on developing a character. I think the article will be of most use to those who are taking their first artistic steps and trying to make their first characters, and I hope this will help you on your artistic path.
The tris budget had been fixed to 9k and 1024px for texture resolution. Usually when I have to create a standard human character, I start to sculpt using a template to speed up the workflow. In this way the work becomes interesting and funny immediately, letting me leave the boring phase of translating the vertices for the future.
My preferred template is one made by Nick Zuccarello. It is downloadable from here: http://nickzucc.blogspot.com. The workflow that I prefer to follow is to sculpt every part of the model and then make a retopology with 3d Studio Max.
Here is the mesh template (Fig.03).
The first thing I do when I create a character is to define proportions. Looking at references is very important at this stage and I recommend working with a low resolution model. I use the Move tool and Move brush a lot, paying close attention to volume and shape. I think this is one of the most important steps if you're trying to create a believable and anatomically-correct character.
When I say anatomically, I'm not referring to the muscle position or their shape or their names; I'm refering to the shapes generated by them. Basically the silhouette assumed by body parts.
Often when artists approach character modeling, digital sculpting and anatomy studies, too much attention is given to the shape of a single muscle and this causes your model to look bumpy and rigid. The common error, in my opinion, is to study flat muscular system images from anatomy books that show the shape and position of muscles. The problem here is that looking at such images doesn't give us the exact perception of the shapes generated by muscles in 3D space. So being aware of this when using such references will help a lot. Unless you are making a bodybuilder or superhero here is my advice: try to reproduce the overall shape generated by the muscles rather than working muscle by muscle. Try to reproduce that fleshy feeling.
For 90% of the sculpting process I use the Clay and Claytubes brushes. I think that those brushes are very cool to sculpt organic shapes with. Combine such brushes with wise use of the Smooth brush and you can get some really nice effects.
Another very important thing is the character's pose. I tend to avoid the basic T pose and try to make a rigger-friendly pose instead, making sure it's naturalistic. I don't like rigid-looking characters. I prefer to bring down the arms to 45 degree angles as this helps to get a relaxed pose. Try to gently lift your arms up to 45 degrees using only your shoulder muscles. Leave your arms soft and your wrists without life... and that's it! In this way your character will have relaxed shoulders, dangling hands that give a light rotation to the forearm and the elbows will be bent, which helps to better define the arm and forearm shape.
Next I start to establish general proportions by doing a lot of tests. At the same time I change the pose to make it more convincing. If the character is going to be dressed then I won't pay much attention to the features of the body, but only create simple shapes and proportions. I like to do tests at this time because then no one will see my weird results... unless they ask me to write a Making Of!
Here are the sculpting proportions for this model (Fig.04).
During the modeling of the head I spend a lot of time on the shape of the eyes. I think it is very important to give the right thickness to the eyelids and try to avoid any flickering edges or outline. Pay attention to the lacrimal gland; this is a very important feature of the eyes and will help you to better define the shape of the eyes and position of the pupil. My advice is to put an eye on a background and follow its shape perfectly.
After I'd set up the body and modeled the head, I started sculpting the clothes (Fig.05).
Sometimes I start by building a base mesh in 3ds Max and other times I extract parts of my mesh inside ZBrush - it usually depends on the level of detail that I want to have at the beginning. For example, if I have to model a shirt with the collar bent in on itself than I usually build it in Max. This is simply to avoid adding subdivisions to my mesh and have a topology that makes me work easier.
In order to do the drapery I used the Clay brush for inner folds and a cutter alpha brush made by my friend Daniele Angelozzi (http://www.d3d.it) to create the outer folds. As always it is very important to look at some photo references and before you start sculpting folds it is important to bear in mind the shape that the fabric will assume. Here's a simple example: jeans that fall over the shoes. I start by sculpting a cylinder which fits my character's leg. I don't start sculpting the folds immediately, but I try to sculpt the shape first. I define the size that my fabric will assume in 3D space. Only when this is done do I start to sculpt the single folds.
During the modeling process, my advice is to take a break from time to time and go away and do something else. This is because of the effect it can have on the eyes. When you work for a long time on something, your eyes will become used to seeing what you are doing and you won't recognize any mistakes. What I do is try to be as diverse as possible. When I have completed something I move onto something else and then go back to the first thing a day later to make corrections.
Back to the sculpt and when I came to do the texturing I completely re-sculpted Marcus's head because previous imperfections were now visible. So I'd advise you to look at your model with a critical eye. Take inspiration from other models. Before starting a piece, I always look for something that can inspire me and I like to find models that are similar to the one I have in mind. This helps me improve mine and is very useful.
When the modeling process was complete, I created the low poly version with the help of Max and its Graphite tool. I used a mid-res model as a base to lean on and reconstruct the topology. During this process I didn't pay attention to the polygon budget. I tried not to exceed it too much, but what I wanted to do was make the normal map creation process easier and get better quality normal maps. More polys means greater accuracy and more similitude with the mid-res mesh. When the normal maps were made, I worked back over the low poly model and deleted some loops here and there to be back inside the polygon budget. In order to avoid texture problems you should not touch unwrap seams when you do this.
Usually I create normal maps with Xnormal. This software allows you to create normal maps by loading in the high poly and low poly models and projecting the details over a texture using the UV co-ordinates of the low poly. There aren't a lot of options to set up.
There is a quick, but less accurate, system that allows you to create normals simply by setting up the ray's distance. I don't want to go into detail about this, but imagine that you want to project a sphere in to a cube and have the length of the side equal to the sphere diameter. Well the distance that the projection ray will have to travel from the sphere to the cube will increase as we move to the cube edge, so if we want to project the whole sphere over the cube then we have to set up the distance data at a level that permits it to reach the cube's edge. This system is not very accurate, in particular when you have to create a normal map over geometry that folds back on itself as the rays could intersect each other and create artifacts.
The other, more accurate, system is to create a projection cage that establishes the maximum distance that projection rays will cover. Xnormal allows you to create and edit this cage and then I prefer to use 3ds Max and Projection Modifier to edit it. First of all you must convert your model to an editable mesh and then apply Projection Modifier. At this point the projection cage will appear; it looks like your model with a bigger size. By moving the vertices of this cage we can define what distance the projection rays will cover. At the end, just export an SBM file and load it as low poly into Xnormal. Check the Use Cage option and launch the calculation (Fig.06)!
I decided to create two single meshes, one for the head and the other for the body. This gave me a head texture with a bigger size and I didn't have to limit the number of textures. The head wireframe can be seen in Fig.07 and the full body wireframe in Fig.08.
I wasn't sure if this was the correct choice. While using one texture for the head gave me better quality, my advice after further testing is to divide the mesh in a different way. Splitting the head and arms into two different textures caused a different skin tone and this is because of different layer blending methods during the texturing.
Unwrap / Texturing
Unwrapping meshes for an artist is pretty boring. Luckily nowadays there are lots of different software packages to speed up the process, but this isn't enough to give you a good unwrap. If the character is intended for a video game then it is very important to use the whole UV space as much as possible. In this way the textures will have more resolution. Other than this, bear in mind that you need to place the unwrap seams in strategic places, like less visible points or in a way that permits you to paint the textures easily and without color edges.
For my texturing I used Photoshop with the help of Zapplink. I don't want to explain how Zapplink works but, quickly, what I do when I have the grabbed the ZBrush screen from inside Photoshop is to place my photo reference and adjust my reference over my model with the Liquify tool. This is a very powerful and quick way to texture a model and avoids cutting and pasting pieces of photos, smudges and spots over the texture. The clothes were just hand painted and there is a base color with some dirt blended over it (Fig.09).
The render is just a screenshot of the Max viewport. I used a DirectX shader made by Laurens Corijn - for more info just visit: http://www.laurenscorijn.com
And that's all! I want to thank everyone for the feedback received for this character and everyone before me that made tutorials and Making Ofs. I hope that this will help you. For all crits, suggestions, support (or job contract!) feel free to contact me at: firstname.lastname@example.org