The Making Of Qunari
CG artist, Morten Frølich Jæger breaks down the key processes in the creation of his demon-inspired model, Qunari.
Earlier this year someone commented that I had never made a ZBrush demon before, so I thought it would be interesting to mix some fantasy elements with a face study and give it a try. I usually make sculpts of realistic people, and to be honest, a guy with horns is not as demon as it can get. Maybe next time I'll try something a little more extreme.
During this Making Of article, I will cover everything from research and the sculpting process to lighting and rendering. To start with, I will talk about gathering references.
As with any project, reference is key. I often meet people who will actively refuse to use reference, but in my opinion it's a necessity. I think some people feel like it's cheating, but it's just like looking at real life. The old masters never sculpted without having a live model. So first, I found as many useful references as I could from books and Google, especially in areas I'm unfamiliar with. Then I moved on to sculpting.
For this sculpt I started out with a basic sphere and used DynaMesh with a low resolution, in this case 64.
I then started blocking in the major shapes with the Clay Tubes brush, which was the brush I stuck with for most of the sculpting process. It's great for giving a more traditional feel to your sculpt, especially when used with its default alpha.
The first attempt
I got as far as the image shown here, but for this sculpt I actually started over – because my first sculpt was, let's just say, less than successful.
I think when you have moments like these it's important to stay on top of it, and not let yourself get too disheartened. Everyone has bad days, but you just have to keep going. It's also important not to get too attached to your work, and be able to let go when something just doesn't look right to you.
The second time I focused more on the basic shapes and proportions instead of moving into details too fast.
When I'd found the volumes I was satisfied with I started to refine the model more by putting more detail into the sculpt and increasing the resolution of the DynaMesh to 128.
I still mainly used the Clay Tubes brush and occasionally the DamStandard for making deeper groves.
I reached a point where I was happy with my volumes, so I attached the horns, which were made as a separate SubTool. I merged the 2 SubTools, and made them into 1 mesh with DynaMesh.
To make the horns blend in better, I used a little bit of the Inflate brush, making them look as if they had grown out of the head.
Fixing the topology
I then fixed the bad DynaMesh topology with the new ZRemesher. It's really amazing what results you can achieve even with just the default settings. All this was just one click! When you do this though, you have to re-project your details back onto the model. I essentially went from having 2 million polys to about 10,000. So, obviously, the new model can't support details like fine wrinkles.
Refining the details
When I sculpt, I hardly ever use the Smooth brush. I feel it takes character away from a raw sculpt like this. Instead I try to smooth the surface using a smaller brush size.
After this I started refining the shapes and adding details like wrinkles and hair, finally touching up the volumes. At this stage I usually also start to add a bit of asymmetry to make it look less CG.
A little trick with regards to the eyes – usually when I sculpt eyes I make a deep hole where the pupil would be. If the eye color is darker, the hole is quite a lot deeper than when the sculpt has lighter eyes.
Adding a slight tilt to the head also gives the sculpt a little more life, which is important when you want to present your model. Too often I see models that are very well done, but are just in a default straight-on pose, with no real expression. This usually does a poor job at conveying the character's personality, so try to avoid this.
The last thing is the pore detail. For this I used a custom brush just called Pores, which you can download for free at ZBrushCentral.
The final thing I did in ZBrush was to decimate my model for rendering, in order for Maya to be able to handle the polycount. With Decimation Master you can bring down the polycount of your model substantially without sacrificing detail. You can find this under your ZPlugin tab.
Now we jump into Maya for light setup and rendering. I'll be using V-Ray for rendering but you can easily do something similar with any other render engine.
I'm only using a few lights to create the mood I'm going for. A nice feature of VRayLights is their ability to create light color based on temperature (degrees Kelvin). This gives the lights a nice feel and you don't have to worry too much about the color being wrong.
Placing the lights
I'm basing my setup on a 3-point light setup, which usually consists of a Key, Fill and Backlight, but I've added a few extra lights for detail. The extra lights just serve as tiny bounce lights because I'm rendering without an environment. They also help to add a bit more color variation.
The color of the main lights shift between 3000 (yellow), to around 4200 (light-blue) degrees Kelvin.
On the extra lights, however, I go as low as 1500 (deep-red), and as high as 8000 (dark-blue) degrees Kelvin. I keep the Intensity multiplier very low on these lights to only add a subtle effect.
I always render with the VRayPhysicalCam just because it gives me real world options. So I try to stick to realistic apertures and shutter speeds. This is also important if you render things like motion blur and depth of field directly into the image. For this Image I just used depth of field.
I guess there are a lot of opinions about rendering effects like these directly into the picture, but the fact is, what you get is artifact-free. You might sacrifice a little rendering speed, but V-Ray is still pretty fast. This might not be ideal for everything, but in my experience nothing beats it for this kind of work.
For this particular setup I used a Focal length of about 70, which is great for portraits as it flattens the perspective a little. That way you avoid characters looking a bit like a fish.
Depth of field
Here's a little trick to help you figure out the distance to an object in Maya for the depth of field. In the Display tab in the top bar you can enable Object Details, which gives you the distance to your currently selected object based on your current camera.
First of all, I used the V-Ray Frame Buffer, as it has a nice feature that allows you to preview your render with Gamma Correction. After previewing, I always try to work linearly when actually rendering out. Linear workflow is a massive subject and would be too big to cover in this Making Of. Simply put, working linearly will get you better results in terms of noise, if you change the Gamma in Color mapping from 1.0 to 2.2. This tells V-Ray to render as if you are going to gamma correct the image later to 2.2. The reason for working linearly is because it gives your light more depth and you can achieve realistic results much easier.
Along with that, I rendered using a combination of Brute Force and Light Cache. This combination works very well for animation, but for just rendering stills you might be better off using a mix of Irradiance Map and Light Cache.
I find that Brute Force and Light Cache produce a satisfying result without too much noise. If you do have noise problems you can try and up your samples in your light, as well as in your Brute Force subdivision.
After the render was complete, I exported the image as an EXR and composited it in NUKE. This is just to add the final touches to the image. I usually adjust the contrast and color balance a little.
I think the best advice I can give is just to look at life and everything around you, and to use it as inspiration. I would also recommend practicing regularly, maybe even a fast sculpt every day, to improve your skills. I hope some of this was useful to you – it was definitely a fun little project and I hope to do more of these raw textural sculpts in the future.