Making Of 'Grandma'
*WARNING: CONTAINS NUDITY*
Hi my name is Arda Koyuncu, a 3D Character Artist/CG Artist currently living in San Francisco, CA.
I've always thought there are dark elements in fairy tales. With this idea in mind, I wanted to create all the characters of the famous Little Red Riding Hood story from a twisted perspective.
In this case, Grandma is the evil mastermind of the story. I wanted to use steampunk elements during creation of her character to enhance the creepy look with the steam powered machines, and dark and dull colors. In this Making Of I will try to explain the process I followed when creating Grandma.
After deciding on the concept, to be able to work on it more accurately in every aspect (modeling, texturing, shading, lighting etc) I spent some decent amount of time collecting a lot of reference images. I generally create different folders for reference images so I can browse my references faster. So in this instance I collected a lot of references for anatomy, fabric, wheelchairs, steampunk machinery etc.
When I had enough resources to start working on the model, I started creating a body base in Maya. I tried to keep the topology as good as possible, but since I was going to make major tweaks in ZBrush I retopologized it as I proceeded. I also built my model in a more natural pose instead of a T-shaped one (Fig.01).
The first thing I did when I imported the mesh in ZBrush was start working on the major shapes. I tried to create a solid structure and a silhouette first. I avoided focusing on one part and refining that area: I tried to build the whole model at the same time. Doing this helps to make the different forms and shapes work together. When I felt comfortable enough after sculpting the major forms overall, I exported the mesh for retopologizing.
After I'd retopologized the mesh I imported it back in ZBrush and started refining and sculpting details. This is generally when I turn Symmetry off to start working on the minor forms and create a more natural feel. One important thing to keep in mind is to have consistency overall. When sculpting major/minor forms age, gender, character and profession could be some of the factors to take into consideration. You can see my attempt to sculpt rheumatoid arthritis on Grandma's hands in Fig.02.
The face is probably one of the most important areas of focus for the majority of the characters. When sculpting age, sagginess and wrinkles on her face I made sure that I was not losing the female form and look because females lose their female form as they get older. I also sculpted a rough hair mesh to make the process easier (Fig.03).
When I started working on the wheelchair I started really simply, just working with primitives and making sure the shapes were working well together. As I felt comfortable I continued and kept refining the separate shapes. Most of the chair was modeled in Maya, but I sculpted the leather seats, wooden ornaments, wear and fine details in ZBrush and retopologized as needed. While I kept adding more details on both hard surface and organic meshes, I also started to work on the lighting setup and shaders. It helped me to look at something else for a while so that when I got back to detailing I was able to refine things further and better. It also helped me with texturing later on (Fig.04).
When I was done laying the UVs I tried to pack them by material. Doing this helps me adjust the colors and make major-minor changes in Photoshop later on when I start rendering. You can see some of the texture maps in Fig.05.
I wanted the fabric, the chair and basically the whole scene to look worn but not very old. I also wanted her skin to look aged. Since I was playing around with colors on the shaders in the meantime I had an idea about which colors to use as base colors for the desired look. There are a lot of different maps for the piece so I will not explain the whole process one by one, but I will try to explain the workflow instead.
For the wood, iron and leather materials I used high resolution images to use for projection in Mudbox (Fig.06 - 07). Before painting the base textures I had to bring the displacement maps into Mudbox, so I could paint on the high detailed mesh instead of the base mesh. To bring your displacement maps into Mudbox you can go to Maps > Sculpt Using Map > New Operation. In the new window that pops up, select the related mesh and path to the displacement file.
In both the hand-painting and projection approaches I always paint a base texture first. After painting the base I start adding some color variety and detail to the texture using the Mudbox layers. If you've sculpted detail on a certain mesh or area in ZBrush (Fig.08), you can export a cavity map and apply it on top of your texture layers in Mudbox to get some detail fast. I also like to separate certain things with layers to be able to tweak them further when I export all of the layers into Photoshop. The main reason I use Mudbox for texturing is because it is fast and it has layers that I can export for Photoshop.
When I start working on the textures in Photoshop I export a couple of maps and apply them to the related mesh to see how it looks in render. Sometimes, especially with subsurface scattering, the texture maps do not look like you painted them when you render, so I go back and forth between Maya and Photoshop to adjust the colors and layers I kept separate earlier until I feel happy with the results.
As I mentioned before, I start working on the shaders at the early stages of the whole process. I also keep tweaking them while I work on the textures. Here are some of the shaders used for this image.
Fabric / Velvet
When creating the shader for clothes I used the additive (shellac) mode of the VrayBlendMaterial (Fig.09). To blend I created two VrayMtls, one for the pattern and the specs and the other one for the velvet effect (Fig.10).
I used the fabric shader to procedurally tile the pattern. To be able to do that I created a seamless pattern in Photoshop first. Then I adjusted the color and plugged the files outColor value to the shader's diffuse channel. I also included the reflections in this shader as well. For the reflections I created a samplerInfo node and plugged its facingRatio to the uCoord and vCoord of a ramp. Using the ramp's values I adjusted the reflection on the surface.
For the velvet shader I used pretty much the same trick, but this time I adjusted the diffuse of the material instead of the reflection using a samplerInfo node and a ramp.
If you are using mental ray, you could try using Puppet Shaders to get this effect. p_MegaTK has a pretty decent fabric / velvet preset.
First of all, to be able to get decent results with Vray SSS your scene needs to be carefully set. Make sure that your units are set to centimeters. Then you can measure your model and see if it is scaled properly. If not, try scaling it before you start working on your SSS shader. Also after you decide what kind of lighting you will have in your scene make sure your camera's ISO, shutter speed and F-number are set to the proper values. Otherwise your light's intensity might not be enough (or too much), but your camera might show everything properly except SSS. I ran into this problem when I was trying to adjust the SSS; it took me a while to figure out the problem was the camera settings not the SSS.
For the subsurface scattering shader I again used a VrayBlendMtl in Additive mode (Fig.11). The reason for that is because I could not get decent specs with SSS specs. So I used a VrayMtl to get the specs instead. I created the VrayMtl for the specs and a Vray Fast SSS to get the color and the scattering and blended them in VrayBlend Mtl (Fig.12).
The good thing about Vray SSS is it has presets that make sense. I started with a preset and tweaked it for a while and also tweaked my texture maps in Photoshop and got decent results in the end. Again, if your scale, lights and camera are set properly, and if you do not have any problems in your mesh, you can start with a preset and get some good results in no time.
Lighting and Rendering
When I started lighting, the first thing I made sure was to set up V-Ray for linear workflow. Then after experimenting for a while I ended up using four V-Ray rectangular lights to light the scene (Fig.13). V-Ray lights are pretty straight forward and work pretty well by default, so I did not really tweak them that much. I mostly worked on the light color and intensity. I used temperature for the color mode in three of my lights, which are key, fill and bounce. Since the rim light has a high intensity, I wanted to give it a little more color so I did not use color temperature for this instance. The planks on the floor are lit by the rim light only. The spotlight effect was created by texturing.
When I felt comfortable with everything I started working on the fuzz for the clothing and the legs. For that I created a separate scene with the same V-Ray and lighting setup. Then I selected the surfaces I wanted to grow hair from. Using Shave & Haircut I quickly adjusted the look of the fuzz. One thing I also did was create a black hole shader to mask out the areas that were not visible to the camera and apply the shader to the surfaces that I grew the hair from.
When the fuzz was rendered I added it on top of the base render I got from V-Ray. I adjusted the backdrop, adding a smooth purple gradient to compliment the green chair and make the character pop out of the image a little more.
I added the smoke effect and hair in Photoshop too. I found some high resolution smoke and hair images and painted on top of those to get the desired effect (Fig.14).
These are more or less the steps I used to create this image (Fig.15). I've tried to answer the questions I was asked about Grandma and I would like to thank everyone that gave me feedback during and after the creation. I hope this tutorial is useful and I hope you can use the tips and tricks I covered in your workflow. Thanks for reading!