Making Of 'Skullhead Mobster'
Hello My name is Harshdeep Borah. I'm originally from India, but did my CG studies from Vancouver, Canada. I am sharing my knowledge of how I made my latest character SKULLHEAD MOBSTER. I don't intend this to be an indepth tutorial, but more over a 'Making Of'. And I assume you have a good knowledge of your favorite 3d or 2d software.
I got inspired for this character when I saw a sketch somebody did in my life drawing teacher Stephen's sketchbook. It just clicked me and I thought of making it in CG. As always for my any project, I started off collecting some real life reference images and doing some research on how a human would look if his face skin is dried into the skull. Also I wanted to show the sheer power and confidence on his face. So what came in my mind was mobster. So I thought of mixing both the ideas and make it him as a Mobster. Here are some images I collected for inspiration.
Modeling the Skullhead
I started off with box modeling the head in Maya. I paid attention to the bone structure humans have, and also how the muscle flows. Then thought of how if skin dried, would look on the face. But I had to maintain the basic form of the human head. Then I modeled the basic polygon mesh for his clothes and hands. I wanted him to be in relaxed position and not in very T-Pose, because its always better to model the character in the closest pose your final render will be. So that you don't have to fight with rigging to reach you character from one extreme pose to another. I try to keep my polycage fairly simple and lowpoly, then export it into zbrush as .obj format and do a lot of tweaking and then import it back in Maya.
1. Its better if you install MJ Polytools before jumping into modeling, as its very handy for modeling and has been a great help for all Maya modelers.
2. I always model my characters using Smooth Proxy, so that I can work on my lowpoly half and see my smooth half as how it is going to be when smoothed.
At this stage I unwrap the complete model and arrange it properly, keeping my texturing needs in mind. So that I don't face any problems later while he is rigged or when I will be texturing him. Also note that it is important to unwrap your model before getting into Zbrush detailing. I mostly use Wings3D for unwrapping my most models and use planar mapping in Maya. Then export uvs at higher resolution of 4000 pix.
I made a basic bone setup in Maya to give him a pose. Rigging with lowpoly is better as while weighting you don't have to worry about heavy mesh and can work with it easily. Also to note, I always make animated frame so that I have basic T-pose in first frame and final pose in like 15th frame or so. This is good so as to interact with zbrush. As in, if done this way, I can export t-pose in zbrush, work on details and then re-export the final pose again to zbrush, and all the details done on t-pose will automatically transfer to posed model. This comes in very handy and you can go back and forth. I also smooth my model this time to see if all the deformations are good and if needs a change; do needed changes on lowpoly model.
1. For you final pose, always save your tpose keyframe and then another keyframe for desired pose.
Mesh detailing, Displacement and Bump maps in Zbrush:
Now comes the fun part. Beforehand I plan everything out. This is very important part of detailing. I plan how much detail is needed from displacements so that it would affect contours and how much from bump maps for fine detailing like skin wrinkles and pores. I export my posed lowpoly mesh to zbrush part by part and first start detailing displacements. Once done with displacements, I load the bump viewer material in zbrush and start painting finer details in projection mode. Note there are few settings you have to keep in mind for Maya and zbrush to impost the mesh before and while detailing.
Some things you need to keep in mind before exporting from Maya are:
1. There should be no overlapping in UV's of the mesh.
2. Try to make you model all quads. There should be NO n-gons in your model.
3. There should be no holes in your model. Though it works sometimes but try to ignore it.
4. All the uvs should be in 0-1 space of UV editor window.
Some things u need to keep in mind while importing your mesh in zbrush are:
1. Always check uvs overlapping in texture check.
2. Store your lowpoly mesh as morph target if you intend to export displacement maps.
Here are few renders I achieved using zbrush detailing and then rendering in Maya with a very basic light setup.
Once happy with detailing, now comes the texturing part. I love it and always been beaten up by mom for messing up my home walls. Also I can't stress on how much real life references are useful for texturing. Always refer to real life objects, their surfaces and how they react to light and environment.
So as Uvs are already laid out properly, I apply shader and start painting in Photoshop part by part. I hand painted all the textures from scratch in Photoshop. Started with the base color, made some custom brushes to add layers of dirt and weathering on objects. Then desaturated the color map and tweaked it to be used as bump and specular maps. Try not to paint too much of darker and highlights in color map, and let the light do that. Pay attention to finer details.
1. It's a good habit to bake your mesh's shadows from Maya and export it in Photoshop and used it as shadow base for painting darker parts.
2. Try to paint as big texture as your pc can handle, as you can always go smaller for texture size, but if needed to show more details can never stretch it out bigger. For my highpoly models I usually paint at least of 4k - 6k maps.
For shading I used a basic Blinn shader with color, bump and specular map attached to it. For leather jacket I used Maya's procedural leather texture in bump slot. Though the texture was handpainted and when mixed with procedural bump, it worked well. When ambient occlusion was not included in Maya, I used dirtmap plugin for every shader. It gives softer shadows and gives more depth to scene. Though it's a bit expensive in terms of rendering.
I always keep light setup very simple. Just one main source light, one fill light and one rim light sometimes. If there are other objects in the scene, I try to put one bounce light for each object. Cause all my soft shadows are generated by Ambient Occlusion/Dirtmap shaders itself, I don't need to put extra lights to cast softer shadows. I use mental ray for calculation as its really good renderer, though takes a bit more time.
Here are the final renders. Added smoke,beard and bg in Photoshop with some color corrections and post effects. Hope you enjoyed it.